a] ay 4 4

Ay :
A WRT NK
‘yee st! ee
At
NS

oe

~S
—~ —

we ae

Pr sy eee
tote.

os

See

wy

at,
=

AAS,
Ss nate
te

nates

ang:
ehh

<0 tu >

ae
Rolie
SSE

Reekeatee
{oe

is Sa pT st
oft Fi A

oe i Ay

/ wee.

4 Pod ? Pe A hed

Wer Pe i ie ee ners te

rye herd ty , atitte ¢ +4 ty a8
i rat

. Panta ah a

PVE OD SD Vd

Hew teh it ae Ln ce ee Dat ie a
We te a * 4

-f
“dpi } il i ae
>

ee)
alse, SX,

ALF &
i |, x

»

x y, Si aie uf
~ ML cep My, %
Couey <H
S E> s, & =

SS
%,

WV 4

H,
y ‘i
n ea ni}

| | oe © |
| i i|
ee {
AMI coe sil
L x) = we 7>

ye
all a xy,
ih 4, x

4

%
Sep
Lotee i

ft Pat \
(Peay

Li
Hag ins |
R Be WA

Meg ed
Dati iy

om if 7 a
Ai, al ;
fine %

|

: : nu ; |
ae
ab ia.
i Le a nig
p :
Po
7 4
i

THE UNIVERSITY OF KANSAS

SCIENCE BULLETIN

A REVISION OF THE SPHENOPSID ORGAN
GENUS, LITOSTROBUS

By
Robert W. Baxter

VoL. XLVII Paces 1-23 Marcu 3, 1967 No. 1

ANNOUNCEMENT

The University of Kansas Science Bulletin (continuation of the Kansas Uni-
versity Quarterly) is issued in part at irregular intervals. Each volume contains
300 to 700 pages of reading matter, with necessary illustrations. Exchanges with
other institutions and learned societies everywhere are solicited. All exchanges

should be addressed to
Liprary OF THE UNIVERSITY OF Kansas,
LawreENce, Kansas 66044

PUBLICATION DATES

The actual date of publication (7.e., mailing date) of many of the volumes of
the University of Kansas Science Bulletin differs so markedly from the dates on
the covers of the publication or on the covers of the separata that it seems wise to
offer a corrected list showing the mailing date. The editor has been unable to
verify mailing dates earlier than 1932. Separata were issued at the same time as
the whole volume. Beginning with Volume XLVI, publication was by separate
numbers and the date on each number is the actual publication date.

Vol. XX—October 1, 1932. Vol. XXXIV,Pt. I—Oct. 1, 1951.
Vol. XXI— November 27, 1934. . Pt. Il—Feb. 15, 1952.
Vol. XXII— November 15, 1935. Vol. XXXV,Pt. I—July 1, 1952.
Vol. XxXIII—August 15, 1936. Pt. II—Sept. 10, 1953.
Vol. XXIV—February 16, 1938. Pt. I1I—Nov. 20, 1953.
Vol. XXV—July 10, 1939. Vol. XXXVI,Pt. I—June 1, 1954.
Vol. XXVI—November 27, 1940. Pr. Il—July 15, 1954.
Vol. XXVII,Pt. I—Dec. 30, 1941. Vol. XXXVII, Pt. I—Oct. 15, 1955.
Vol. XXVIII, Pt. I—May 15, 1942. Pt. II—June 29, 1956.

Pt. II—Nov. 12, 1942. Vol. XXXVIII, Pt. I—Dec. 20, 1956.
Vol. XXIX, Pt. I—July 15, 1943. Pt. II—March 2, 1958.

Pt. II—Oct. 15, 1943. Vol. XXXIX—Nov. 18, 1958.
Vol. XXX, Pt. I—June 12, 1944. Vol. XL—April 20, 1960.

Pt. Il—June 15, 1945. Vol. XLI—Dec. 23, 1960.
Vol. XXXI, Pt. I—May 1, 1946. Vol. XLII—Dec. 29, 1961.

Pt. II—Nov. 1, 1947. Vol. XLII—Supplement to, June 28, 1962.
Vol. XXXII—Nov. 25, 1948. Vol. XLIII—Aug. 20, 1962.
Vol. XXXIII,Pt. I—April 20, 1949. Vol. XLIV—Sept. 1, 1963.

Pt. II—March 20, 1950. Vol. XLV—June 7, 1965.

dtr Bein ee hae R. C. Jackson
Editorial Board ........ GeorceE Byers, Chairman

KENNETH ARMITAGE
CHARLES MICHENER
Paut Kiros
RICHARD JOHNSTON
DELBERT SHANKEL

4 : evi . ‘ a as
: “ rte.
re tte

THE UNIVERSITY OF KANSAS

SCIENCE BULLETIN

Vor. XLVIT Paces 1-23 Marcu 3, 1967 No. 1

A Revision of the Sphenopsid Organ genus, Litostrobus'
By

Ropert W. BAxTER

ABS RACd

A number of excellently preserved fragments of Litostrobus cones from coal
balls of the Desmoinesian Stage, Middle Pennsylvanian of Iowa are described.
Evidence is presented to show that a wide variation in number and arrangement
of sporangia and bracts occurs in different whorls of individual cones with an
apparent decrease in number towards the cone apex. Heteromorphism of 14 to 6
sporangia per whorl is accompanied by changes in the number and arrangement
of the subtending bracts so that apical whorls of 6 sporangia may have double the
number of alternate and opposite bracts while basal whorls of 14 sporangia may
have only an equal number of alternate bracts. The vascular system is shown to
be simple as originally described by Mamay (1954) rather than the triple
dichotomy described by Reed (1956). The cone of Mesidiophyton Leisman 1s
shown to be congeneric with Litostrobus but to constitute a distinct species, L.
paulus. New diagnoses are given for both L. zowensis and L. paulus. Possible
relationships of Sphenostrobus to Litostrobus are discussed.

INTRODUCTION

The organ genus Litostrobus was established by Mamay (1954) on a
single specimen of a sphenopsid cone found in a coal ball from the Urban-
dale Coal Mine, Urbandale, Iowa. The stratigraphic position of the type
species, L. zowensis, is Desmoinesian Stage, Middle Pennsylvanian. The
generic diagnosis was given as follows: “Cone verticillate, the appendages
produced in multiples of three. Whorls supeiposed, each consisting of ba-

‘This is part of a general investigation of the Pennsylvanian Coal ball flora supported by
National Science Foundation grant GB 4933.

SUITHSONIAN nb 06)

Tue UniversatTy ScIENCE BULLETIN

*

#
Re *

> ay
£

i

od

SpHENOPSID OrcAN GENUS 3

sally fused bracts and axillary sporangia. Bracts twice as numerous as spo-
rangia, arranged one opposite a sporangium, one alternate. Sporangia ortho-
tropus, each terminating a short axillary pedicel.”

Additional specimens from the Weir-Pittsburg Coal of Frontenac, Kan-
sas were described and placed in the same genus and species as Mamay’s
material by Reed (1956) who also emended the generic diagnosis to in-
clude an elaborate triple dichotomy of the vascular traces into the bracts
and axillary sporangiophores.

The present account is concerned with the description of a number of
exceptionally well preserved cones discovered in a coal ball from the Mich
Mine of Oskaloosa, Iowa, of the same stratigraphic position as the Urban-
dale Mine. These cones are clearly assignable to the genus Litostrobus, but
they show a number of variations from both Mamay’s and Reed’s specimens
which make it necessary to further revise the original concept of the cone
structure.

DESCRIPTION

Nine different fragments of cones were exposed in one coal ball and
one in another. Whether they all represented different cones or fragments
of a few cones is uncertain, but as the longest specimen exposed reached 25
mm with no sign of tapering at either end, it may be assumed that a number
of the transverse exposures probably represented the curving path of single
long cones. All of the specimens were close to 4.0 mm in diameter.

Three of the specimens were cut in the longitudinal plane which per-
mitted several complete series of tangential and radial sections, while the
other 7 specimens were worked out by extensive series of transverse sections.
All of the specimens appear to be identical to the type species described by
Mamay except for variations in the number and arrangement of sporangia
and bracts per whorl. Since these variations were observed within a single
cone, they are obviously of minor taxonomic significance. The present spe-
cimens do, however, show exceptionally fine preservation of almost all the
tissues and accordingly supply a number of new details regarding the gen-
eral cone anatomy which have been unavailable previously.

Tue Axis: A very small triarch, exarch, protostele only 320 » from one
protoxylem point to another occupies the exact center of the cone axis. The
protoxylem is situated at the three corners of the triangle and consists of
small tracheids with spiral and annular thickenings while the larger, thin
walled metaxylem tracheids show multiseriate bordered pits (Fig. 24) with
the narrow opposing appertures forming an X-shaped pattern in the center

Fic. 1. Litostrobus iowensis. Transverse section of cone with 11 sporangia per whorl. The
central triangular area in the axis is occupied by the triarch protostele(s) (here missing) and is
surrounded by the large cells of the melasmatic zone (M) which is enclosed by the smaller
celled cortex (C). The 11 sporangia are enclosed by the ascending bracts (b). X30.

A rm
. m= *

—

SPHENOPSID OrcAN GENUS 3

of the border. The pits are very small with the diameter of the entire border
averaging only 8 ». No secondary xylem is present in any specimen.

While the general character of the triarch protostele and the pitting of
the tracheids is typically sphenophyllalean, it is distinctive in its extemely small
size (compared to known species of other sphenophyllalean cones) and the
very fragile, thin-walled nature of the tracheids—particularly those of the
metaxylem. These seem to be so consistently delicate that, in spite of generally
excellent preservation of all other tissues, the area of the stele is usually only
evident as a triangular open area in the center of the axis. (Figs. 1, 2, 14, 15).

The triangular stelar area is immediately surrounded by a conspicuous
zone of large cells, tubular in character, reaching 1.5 mm in length and 90 p
in diameter. The zone is usually 2-4 cells thick with each cell filled with an
opaque “resin”. The cells and their contents are similar to the tissue described
by Leisman (1964) for the cone of Mesidiophyton paulus for which he used
the term “melasmatic,” a term which also seems appropriate in this case.
These “melasmatic” cells have been described previously in other sphenopsid
organ genera such as Asterophyllitis charaeformis (Thomas, 1911), in Sphe-
nostrobus thompsonu (Levittan and Barghoorn, 1948), and Calamostachys
americana (Arnold, 1958).

Directly outside of the melasmatic zone is the axis cortex consisting of
2-4 layers of cells with a vertical length up of 270 » and an average trans-
verse dimension of 40 x 65 » with the radial width the greater. These cells
have thin, smooth walls and are always devoid of contents (Figs. 1, 15).

Tue Bracts: The bracts of each whorl are fused internally to form a
broad, cup shaped receptacle enclosing the axillary sporangiophores and spor-
angia. They consist of the same cell type making up the cortex of the axis
which becomes gradually thicker in the upper part of the lower internode and
then flares out at the node at about a 45° angle to form the bract disc (Fig.
2). In the mature, central part of the cones the bracts average 3.7 mm in
length and extend about one half the way into the next internode above. The
individual bracts do not separate until a point about 0.8-1.0 mm above the
base of the disc so that the axillary sporangiophores and lower part of the
sporangia are completely enclosed (Fig. 14).

At the level of separation from the disc, the bracts are fusiform in trans-
verse outline and are 1.0 mm in tangential width at their broadest point (Fig.
23). The central area of each bract is traversed by a conspicuous strand of
melasmatic cells, continuous with the melasmatic zone of the cone axis (Fig.
3), which it is felt has been misidentified as the vascular trace in most pre-
vious specimens. The true vascular trace is adaxial to the melasmatic strand
and consists of never more than 2-3 very small tracheids about 6-8 » in

Fic. 2. Litostrobus iowensis. Median-longitudinal (radial) section of cone. S, sporangiophore;
b, bract; a, central area of axis stele. Note superposed position of sporangiophores and elongate
sporangia packed with spores. X25,

‘

(At KE

* ba B etm

So" ey

ve~ PXK

A ee i 5 a

“hie 4 x FR

= ASS
\% Ape Pod Pry

SPHENOPSID OrcAN GENUS 7

diameter with spiral and annular thickenings. These traces are so small that
only in a very few cases of exceptionally good preservation was it possible
to see them at all. However, all of the evidence points to the impression that
the trace enters the bract directly from the protoxylem area of the axis
stele without any intermediate branching.

The number of bracts, per whorl, varies from as few as 8 to more than
14 and their relation to the sporangia was equally variable being usually alter-
nate but occasionally also in an opposite position. The significance of this
variability is covered more fully in the discussion.

THE SPORANGIOPHORES AND SPORANGIA: The sporangiophores, or pedi-
cels, arise in the axil of the cone axis and the fused bracts. They average 0.5
to 0.7 mm in length with a radial width of 0.2 mm at the base to 0.4 mm at
their apex (Figs. 2, 12). Their tissues consist of elongated, thin-walled cells
similar to those of the outer cortex of the axis, traversed on the abaxial side
by a strand of 4-6 tubular cells of the melasmatic tissue and on the adaxial
side by a vascular trace (Figs. 14, 15). The vascular trace follows a direct,
ascending path from the exarch protoxylem strands of the axis stele up to the
base of the sporangium. There is no evidence, whatsoever, for the complex
triple dichotomy of a common bract and sporangiophore trace such as
described by Reed (1956). The sporangiophore trace is slender at its base
but widens out at its apex (where it comes directly in contact with the
sporangium wall) into an expanded wedge some 8-10 cells across in which
the tracheids become relatively much broader and shorter than in lower re-
pions (Figs, 12, 13).

Each sporangiophore bears a single, erect (orthotropous) sporangium.
The sporangia are elongate with average dimensions of 2.4 mm in length
to 1.2 x 0.8 mm in diameter, the radial width the greater (Figs. 1, 2, 10). The
total length of 3.0 mm for the erect sporangia on their sporangiophores is
such that the tips of the sporangia of one whorl extend well above the base
of the bract disc of the next whorl above (Figs. 2, 7, 22). Accordingly, the
general configuration of the cone (based on our 10 specimens) is like Ma-
may’s (1954) Fig. 2 rather than Reed’s (1956) Fig. 30 where the sporangia
are described and illustrated as “sub-globose.”

The outer sporangium wall consists of cells elongated in the plane of the
sporangium length with average measurements of 230 » long by 90 x 45
p. wide with the radial width being the larger. The cell walls are smooth
and straight in outline and lack the inwardly projecting buttresses found in
many calamitean cones.

An inner sporangial wall of several layers of “delicate, parenchymatous
cells” was described by Mamay but is only rarely present in our material. The

Fic. 3. Litostrobus iowensis. Tangential section of cone showing disc of fused bracts at top
and bottom with whorl of sprangia above lower disc. M, melasmatic strands in bract disc.

X50.

SPHENOPSID OrGAN GENUS 9

outer sporangium wall covers the entire surface of the sporangium and also
extends downwards along the abaxial side of the sporangiophore (Figs. 2,
8) with the result that some tangential sections of the cone seems to show
sporangia with considerable sterile tissue inside the wall at the lower end
(Figs 9, 11). The “sterile tissue” is, of course, only the obliquely sectioned
apex of the sporangiophore, which is here surrounded by that portion of
the sporangium wall which extends down its abaxial surface.

The number of sporangiophores and sporangia per whorl is apparently
subject to considerable variation, not only possibly in different cones but
also at various levels of the same cone. Among our 9 different specimens the
number of sporangia per whorl varied from 8 to 14 with some sections show-
ing the intermediate numbers of 9, 10 and 11 (Figs. 4, 5, 6).

In contrast to Mamay’s (1954) type specimen and Reed’s (1956) sub-
sequent description of additional specimens, none of the present material
shows the supposedly typical trimerous pattern of six sporangia and 12 bracts
per whorl. This may possibly be correlated with the fact that Mamay’s ori-
ginal cone (and possibly Reed’s also) was an apical fragment where the
number of parts per whorl seems to become smaller, possibly as a result of
diminishing, determinate growth of the apical meristem. For example,
one of our cones which was exposed for over 2.5 cm showed a whorl of 14
sporangiophores near the base, several whorls of 11 near middle and a
whorl of 8 at the top. As this fragmentary specimen of over 9 nodes and
internodes length showed no evidence of tapering at either end, it must
be assumed that its intact length must have been much more than 2.5 cm and
that the trend of variation in the entire cone might be expected to be also
greater than that shown in the fragment.

Tue spores: All of the sporangia of the numerous specimens studied
(Figs. 1, 2, 3, etc.) were packed with spherical spores averaging 90 » in
diameter. The spore wall is of two layers with the outer exospore consisting
of a tangled reticulum of threadlike thickenings covering a very thin,
smooth (almost transparent) endospore membrane.* The exospore threads
are sometimes fused as prominent ridges (Fig. 16) but their threadlike
nature may be seen in median views where the tangled loops of the thickened
strands are clearly evident (Figs. 17, 18). This exospore reticulum was ap-
parently laid down subsequent to the separation of the tetrads as the trilete

* While the terms exospore and endospore are used here (as in all prior descriptions), the
terms perispore and exospore, respectively, possibly would be more proper as the former appears
to have been formed after separation of the tetrads and the trilete scar is present only in the
latter.

Fics. 4-7. Litostrobus iowensis. Fig. 4. Transverse section showing 14 sporangia in a
whorl. X12. Fig. 5. Transverse section showing eight sporangia in a whorl. X12. Fig. 6.
Transverse section with 11 sporangia in a whorl. X12. Fig. 7. Radial section. b, bract; s,
sporangiophore; a, area of stele in center of axis. X12.

SPpHENOPSID Orcan GENUS 11

scar on the proximal surface of the spore appears to be only in the smooth
endospore wall although it may sometimes be visible through the exospore
reticulum as shown in Mamay’s Fig. 9.

The most unique character of the spores (remarked upon by Mamay
in his original species diagnosis) is the presence of a circular groove on the
proximal surface, clearly circumscribing the trilete scar, which seems to
function as a dehiscence mechanism as both the exospore and endospore
walls are ultimately shed in the area contained by it. The exospore wall is
the first to go, thus exposing a circular area of the endospore about 60
pv. in diameter. The exposed endospore membrane is marked by a centrally
placed trilete scar, with three equal arms, each arm measuring 21 p in
length (Figs. 19, 20, 21). The endospore membrane, in the circular area,
then breaks free leaving the round appertures shown in Figs. 20, 21. Most
of the hundreds of spores observed had lost both the exospore and endospore
walls within the area circumscribed by the dehiscence groove so that the
trilete scar was still present on only a relatively few spores.

A feature of the exospore reticulum, which also deserves mention, is
its apparent tendency to disintegration so that only rarely does it show the
detail illustrated in Figs. 16, 17 and 18. The more usual state of preservation
appears to be one in which the reticulate threads have become coalescent giv-
ing the impression of a membranaceous, ridged surface. In other cases the
exospore wall may be lost completely, leaving only the smooth walled endo-
spore, which in this condition resembles a spore of the Calamospora type
with an average diameter of only 70 vp.

DISCUSSION

There can be no doubt that the cone specimens described here are
assignable to the genus Litostrobus as established by Mamay (1954). There
is complete agreement in the orthotropous sporangia, each terminating a sin-
gle axillary sporangiophore, the character of the various cone tissues, and
the unique spores. The points of difference (primarily those of number and
arrangement of bracts and sporangiophores per whorl) are clearly variable in
a single cone and consequently cannot even be used for distinguishing species.

The complicated vascular anatomy described by Reed (1956) presents
more problems in that neither the present material nor the type specimen
show any such structure. It is also the more remarkable in that while the
specimens described here represent a stage of unusually fine preservation,
Reed constantly emphasized that her material was poorly preserved. State-

Fics. 8-9. Litostrobus iowensis. Fig. 8. Longitudinal section slightly tangential to cone axis
showing four erect sporangiophores with terminal sporangia. X25. Fig. 9. Tangential section
of cone showing apparent “sterile tissue(s)” in sporangium because of oblique sectioning of
apex of sporangiophore. X36,

i heal be a tt.

We £m
f j

: x
i Te &

SPHENOPSID OrcAN GENUS 13

ments such as, “Details of the stelar tissues are obscure because of poor pre-
servation...,” “The fragment is highly carbonized with the result that it
possesses almost no cellular detail...” and “...the axis of the strobilus is
poorly and incompletely preserved...” occur frequently. I have had the
opportunity also to examine a number of the specimen peels cited by Reed
and have been unable to see any such vascular branchings as were described.

It also seems significant that while Reed’s triple dichotomy with bracts
of the “first order” and “second order” is based completely on the concept of
a constant trimerous system, i.e., parts in multiples of three, our material
shows that no such constancy in numerical order exists. Certainly whorls of
8, 10, 11 and 14 could hardly be vascularized by a system based on a triple
dichotomy of traces.

Not only do the present specimens show clearly that the sporangiophore
trace arises directly from the stele axis and passes directly up to the base
of the sporangium, but the sporangiophore trace is shown to be a much larger
strand (8-10 tracheids vs. 2-3) than that in the bract, a condition which hardly
seems likely if the sporangiophore trace arose as the final branch of a triple
dichotomy. Accordingly, the simplicity of the structure of Litostrobus, first
emphasized by Mamay, seems to be verified.

While the evidence would seem to indicate that Reed’s concept of the
vascular anatomy of Litostrobus was in error, her reconstruction of the
cone showing short, sub-globose sporangia is certainly different from our
specimens and Mamay’s holotype. At the same time Reed’s specimens present
striking similarities to Leisman’s (1964) cone of Mesidiophyton which, pos-
sibly significantly, is from essentially the same Kansas locality and horizon.
The cone of Mesidiophyton, the stems of which Leisman (personal communi-
cation) now believes to be indistiguishable from Sphenophyllum, has the
same basic structure as Litostrobus with whorls of orthotropous sporangia
subtended by partially fused bracts. Leisman, however, felt that it was gen-
erically distinct in the following characters: (1) A trimerous symmetry for
Litostrobus vs. whorls of 8, 10 and 11 in Mesidiophyton. (2) Twice as many
bracts as sporangiophores in Litostrobus (both alternate and opposite) vs. the
same number of bracts as sporangiophores in an alternate arrangement in
Mesidiophyton. (3) Supperposed position of bracts and sporangiophores in
successive whorls in Litostrobus vs. alternate position in Mesidiophyton.

It is immediately apparent that the variability now known to occur in
Litostrobus is more than sufficient to account for all of the above supposed
generic distinctions. For example, we now know that Litostrobus shows a
variation in parts per whorl, of 6 to 14 with the entire range perhaps occurring
in a single cone. Possibly as a direct result of this variation in number of parts

Fies. 10-11. Litostrobus iowensis. Fig. 10. Transverse section of cone near base of bract
disc. X20. Fig. 11. Tangential section showing overlapping of bracts of successive nodes and
apparent “sterile tissues” in sporangia. X23.

SPHENOPSID OrcAN GENUS 15

from node to node there is also variation in the proportion and arrangement
of the bracts and sporangiophores in successive whorls, i.e., alternate vs. su-
perposed, etc. Figs. 1 and 6 show bracts equal in number to the sporangia
and alternate with them. Other whorls, however, show occasional bracts
opposite as well as alternate with the sporangia so that the number of bracts
was sometimes greater than the sporangia. It is also obvious that superposition
of parts in successive whorls is only possible where the number per whorl
remains the same, while a change in number of parts in successive whorls
would produce at least some alternate arrangement.

The Mesidiophyton and Litostrobus cones are, in addition to the axil-
lary, orthotropous sporangia, also essentially identical in all of the following
points: (1) Size of cone as a whole; (2) Size of the cone triarch stele; (3)
Adaxial position of sporangiophore trace; (4) Abaxial position of melas-
matic strand in sporangiophore; (5) Extension of sporangial wall down-
wards along abaxial surface of sporangiophore.

Accordingly there no longer seems to be any basis for not considering the
cone specimens described by Leisman as falling within the generic limits of
Litostrobus, although the following minor differences seem sufficient to
justify recognizing it as a distinct species: (1) Globose to sub-globose spor-
angia 0.4-0.7 mm in diameter vs. elongate sporangia 1.2 mm long by 0.8 mm
wide in L. towensis; (2) Cone axis with thick cortex with Sphenophyllum
like furrows and ridges vs. thin smooth cortex in L. towensis; (3) Spherical
spores with trilete scar but lacking circular, dehiscence groove found in
L. towensis.

If we now consider the above characters which seem to distinguish Leis-
man’s Kansas specimens of Mesidiophyton from the Iowa specimens repre-
sented by Mamay’s holotype and the present material, we find that the
cone fragments described by Reed (1956) from Frontenac, Kansas, agree most
closely with Leisman’s specimens and not Mamay’s L. zowensis. Accordingly,
based on the observations of Reed’s material and her own descriptions, it is
felt that her specimens and Leisman’s Mesidiophyton cone should be as-
signed to a distinct species of Litostrobus rather than the type species which
so far seems to occur only in Iowa coal balls. Here again, however, it should
be noted that Leisman’s several cones show a simple, direct passage of vascular
traces from the cone axis into the sporangiophores and bracts rather than
the triple dichotomy described by Reed. The following emmended diagnosis
and reassignments are as follows:

Litostrebus iowensis Mamay, emend. Baxter.

1956. Litostrobus 1owensis Mamay, emend. Reed. Phytomorphology, 6:261-272.

Fics. 12-13. Litostrobus iowensis. Fig. 12. Longitudinal section showing direct passage of
vascular trace into sporangiophore. b, bract; t, trace; c, cortex of axis; s, sporangium and spores.
X120. Fig. 13. Enlargement of portion of above. s, spore in sporangium; t, expanded apex
of sporangiophore trace with tracheids showing scalariform and spiral thickenings. X520.

a a
6: _ : Z 2 ah: Risks
j Sat . eee janet

AN

SPHENOPSID Orcan GENUS 17

GENERIC DIAGNosIs—Cone verticillate, each node bearing a whorl of bracts
and axillary, unisporangiate, pedicelate, orthotropous sporangia. Epidermal
cells of bracts showing sinuous interlocking outlines. Number and arrange-
ment of organs per whorl variable, decreasing in number towards the apex.
Triarch, exarch, protostele; protoxylem with annular and spiral thickenings,
metaxylem with multiseriate bordered pits.

SPECIFIC DIAGNosis—Cone small, up to 4-5 mm in diameter. Triarch pro-
tostele about 320 » measured from adjacent protoxylem points, metaxylem
with very thin walled tracheids. Melasmatic zone of tubular cells containing
opaque “resin” surrounding stele. Cortex consisting of empty rectangular
cells, 270 x 40 x 65 » with thin smooth walls. Bracts and sporangiophores in
whorls of 6-14 or more. Erect, axillary sporangiophores 0.4-0.7 mm long by
0.2 mm wide at base to 0.4 mm at flaring apex. Sporangiophore trace on
adaxial side with a melasmatic strand abaxial to it. Sporangia ovoid, up to 3
mm long by 1.2 x 0.8 mm wide. Spores spherical with average diameter of
90 ». Exospore of a tangled, threadlike reticulum, endospore memberane
smooth, translucent with a trilete scar surrounded by a circular “dehiscence
groove.”

Litostrobus paulus (Leisman) Baxter.

1956. Litostrobus 1owensis. Reed, Phytomorphology, 6:261-272.
1964. Mesidiophyton paulus (Cone). Leisman, Palaeontographica, 114B:135-146.

Cone similar to L. zowensis except in the following characters: Spor-
angia globose to sub-globose, 0.4 to 0.7 mm in diameter, cone axis with alter-
nate ridges and furrows in outer cortex, spores apparently lacking “dehiscence
groove” and having a sightly smaller size range of 55-80 » which possibly may
be due to loss of exospore wall.

Since it now seems certain that the stems attached to the Litostrobus
(Mesidiophyton) paulus cones are indistinguishable from Sphenophyllum,
there can no longer by any doubt of the sphenophyllalean affinity of the
cone genus. We are thus again confronted with the problem of the extreme
diversity of cone types apparently borne on Sphenophyllum stems, ranging
from the exceedingly complicated structure of Peltastrobus reedae (Baxter,
1950; Leisman and Graves, 1964) and Bowmanites fertilis (Leclercq, 1936)
to the relatively simple Bowmanites moore: (Mamay, 1959) and Litostrobus.
The problem is further compounded in that, not cnly do we have at least
four cone genera containing 35 species (Boureau, 1964) all attributed to one
or two species of Sphenophyllum stems, but the evidence of our present ma-
terial of Litostrobus seems to indicate a considerable variation in number

Fics. 14-15. Litostrobus iowensis. Fig. 14. Transverse section of cone near base of in-
ternode showing cone axis in center surrounded by whorl of 11 sporangiophores just below
their attachment to sporangia; the fused bract disc outside. X21. Fig. 15. Enlargement of top-
center area of above. s, area occupied by stele (here poorly preserved); m, melasmatic tissue of
cone axis; c, cortex of cone axis; x, adaxial xylem at apex of sporangiophore; ms, melasmatic
strand on abaxial side of sporangiophore trace. X65.

SPHENOPSID OrGAN GENUS 19

and arrangement of sterile and fertile parts (which variations have previously
been used as specific, or even, generic characters) within successive whorls of
individual cones.

While it has now been recognized for some time that such a basic anat-
omical structure as the stele in Lepidodendron might vary from a protostele
to a siphonstele in the same stem (Eggert, 1961) and the number of steles in
the pteridosperm, Medullosa, might vary from three to over 11 in the same
stem (Delevoryas, 1955), it was not uncommon for a number of early in-
vestigators to consider these variations as specific or generic differences.

In a similar way we now have a considerable assemblage of supposedly
distinct taxonomic entities based on sphenopsid fructifications in which it
may very well be that equivalent heteromorphic forms occur. For example,
the enigmatic cone, Sphenostrobus thompsonu (Levittan and Barghoorn,
1948), which has been recognized by both Mamay and Leisman as having
a number of features in common with Litostrobus, could conceivably be
merely a basal fragment of the latter genus. The main distinguishing char-
acters are the tetrarch stele in Sphenostrobus, its larger number of bracts
and sporangia per whorl, and the sessile nature of the sporangia.

It is not the intention here to do more than suggest that the range of
polymorphism which seems apparent in Litostrobus might include a larger,
more basal region with the characters of Sphenostrobus. It should be stated,
however, that there is also a (previously unreported) striking similarity in
the spores of the two genera as well as the distinctive melasmatic zone in the
cone axis.

While the triarch protostele has so far seemed basic to Sphenophyllum
axes, the number of protoxylem points of protosteles in living plants (roots)
is known to be subject to frequent variation. In the common Ranunuclus
acris root, for example, triarch, tetrarch and pentarch conditions may all

be found.

Fics. 16-19. Litostrobus iowensis. Spores. Fig. 16. Spore with reticulate exospore. X500.
Fig. 17. Optical-median view of spore showing reticulate strands of exospore in side view.
X500. Fig. 18. Enlargement of area of above. X850. Fig. 19. Spores with partial disintegration
and shedding of exospore. Bottom spore shows trilete scar on smooth endospore wall. X420.

SPHENOPSID OrGAN GENUS 21

Literature Cited

ArNoLp, C. A. 1958. Petrified Cones of the genus Calamostachys from the Carboniferous of
Illinois. Contr. Mus. Paleont. Univ. of Michigan 14:149-165.

Baxter, R. W. 1950. A New Sphenopsid Cone from the Pennsylvanian of Indiana. Bot. Gaz.
112:174-182.

Boureau, E. 1964. Traite de Paleobotanique. Vol. 3. Sphenophyta-Noeggerathiophyta, Masson
et@ies Paris:

Devevoryas, T. 1955. The Medullosae—Structure and Relationships. Palaeontographica 97B:
114-167.

Eccert, D. A. 1961. The Ontogeny of Carboniferous Arborescent Lycopsida. Palaeontographica
108B:43-92,

Leccerca, S. 1936. A propos du Sphenophyllum fertile Scott. Ann. Soc. Geol. Belgique 60:
170-172.

Leisman, G. A. ano C. Graves. 1964. The Structure of the Fossil Sphenopsid Cone, Pelta-
strobus reedae. Amer. Midl. Nat. 72:426-437.

LetsMAN, G. A. 1964. Mesidiophyton paulus gen. et sp. nov., A New Herbaceous Sphenophyll.
Palaeontographica 114B:135-146.

Levirran, E. D. ann E. S. BarGHoorn. 1948. Sphenostrobus thompsonii: A New Genus of
the Sphenophyllales. Amer. Jour. Bot. 35:350-358.

Mamay, S. H. 1954. A New Sphenopsid Cone from Iowa. Ann. Bot. (N.S.) 18:229-240.

Reep, Freppa D. 1956. The Vascular Anatomy of Litostrobus iowensis. Phytomorphology
6:261-272.

Tuomas, H. H. 1911. On the Leaves of Calamites (Calamocladus Section) Phil. Trans. Roy.
Soc. London (B) 202:51-92.

Fics. 20-21. Litostrobus iowensis. Spores showing “dehiscence area” formed by shedding
of circular section of exospore followed by underlying endospore. Spores with trilete marks still
have endospore(e). Those lacking scar have open “dehiscence” apperture. X500.

Kes RN

{

SPHENOPSID OrGAN GENUS 23

Fics. 22-24. Litostrobus 1owensis. Fig. 22. Transverse section of cone showing elongate
sporangia of lower whorl extending above the fused bracts of the next higher whorl. Compare
to Fig. 2. Sp, tips of sporangia of lower whorl; b, disc of fused bracts; s, sporangiophores of
next higher whorl; a, cone axis. X20. Fig. 23. Free lobe of a single bract. X25. Fig. 24.
Metaxylem tracheid with reticulate arrangement of round bordered pits. X375.

THE UNIVERSITY OF KANSAS

SCIENCE BULLETIN

FLORAL MORPHOLOGY AND SYSTEMATICS
OF PLATYSTEMON AND ITS ALLIES
HESPEROMECON AND MECONELLA

(PAPAVERACEAE: PLATYSTEMONOIDEAE)
By
Wallace R. Ernst

Voi. XLVII Paces 25-70 Marcu 3, 1967 No. 2

ANNOUNCEMENT

The University of Kansas Science Bulletin (continuation of the Kansas Uni-
versity Quarterly) is issucd in part at irregular intervals. Each volume contains
300 to 700 pages of reading matter, with necessary illustrations. Exchanges with
other institutions and learned societies everywhere are solicited. All exchanges

should be addressed to
LipraRY OF THE UNIVERSITY OF KANsas,
LawreENcE, Kansas 66044

PUBLICATION DATES

The actual date of publication (z.¢., mailing date) of many of the volumes of
the University of Kansas Science Bulletin differs so markedly from the dates on
the covers of the publication or on the covers of the separata that it seems wise to
offer a corrected list showing the mailing date. The editor has been unable to
verify mailing dates earlier than 1932. Separata were issued at the same time as
the whole volume. Beginning with Volume XLVI, publication was by separate
numbers and the date on each number is the actual publication date.

Vol. XX—October 1, 1932. Vol. XXXIV, Pt. I—Oct. 1, 1951.
Vol. XXI— November 27, 1934. Pt. II—Feb. 15, 1952.
Vol. XXJI— November 15, 1935. Vol. XXXV,Pt. I—July 1, 1952.
Vol. XxXIII—August 15, 1936. Pt. II1—Sept. 10, 1953.
Vol. XXIV—February 16, 1938. Pt. I1I—Nov. 20, 1953.
Vol. XXV—July 10, 1939. Vol. XXXVI,Pt. I—June 1, 1954.
Vol. XXVI—November 27, 1940. Pt. Il—July 15, 1954.
Vol. XXVII,Pt. I—Dec. 30, 1941. Vol. XXXVII, Pt. I—Oct. 15, 1955.
Vol. XXVIII,Pt. I—May 15, 1942. Pt. I1—June 29, 1956.

Pt. II—Nov. 12, 1942. Vol. XXXVIII, Pt. I—Dec. 20, 1956.
Vol. XXIX, Pt. I—July 15, 1943. Pt. II—March 2, 1958.

Pt. 1I—Oct. 15, 1943. Vol. XXXIX—Nov. 18, 1958.
Vol. XXX, Pt. I—June 12, 1944. Vol. XL—April 20, 1960.

Pt. Il—June 15, 1945. Vol. XLI—Dec. 23, 1960.
Vol. XXXI, Pt. I—May 1, 1946. Vol. XLII—Dec. 29, 1961.

Pt. II—Nov. 1, 1947. Vol. XLII—Supplement to, June 28, 1962.
Vol. XXXII—Nov. 25, 1948. Vol. XLIII—Aug. 20, 1962.
Vol. XXXIII,Pt. I—April 20, 1949. Vol. XLIV—Sept. 1, 1963.

Pt. II—March 20, 1950. Vol. XLV—June 7, 1965.

Eaton. < socks escaee R. C. Jackson
Editorial Board ........ GerorcE Byers, Chairman

KENNETH ARMITAGE
CHARLES MICHENER
Paut Kiros
RICHARD JOHNSTON
DELBERT SHANKEL

Vid Vries’
A t 4’ hie oe Hie | i! ar

THE UNIVERSITY OF KANSAS

SCIENCE BULLETIN

VoL. XLVII Paces 25-70 Marcu 3, 1967 No. 2

Floral Morphology and Systematics of Platystemon and its
Allies Hesperomecon and Meconella (Papaveraceae:
Platystemonoideae )

By

WaL.Lace R. Ernst*

ABSTRACT

The five species of subfamily Platystemonoideae, primarily in California, are
compared for morphological structure and variation. They are contrasted with
other Papaveraceae to show evolutionary relationships and as background for a
taxonomic treatment. Information is assembled from field, culture, herbarium,
and computer method studies.

INTRODUCTION

Comparative knowledge of form and structure, in addition to providing
the basis for logical systematics and taxonomy, also is provocative from an
evolutionary point of view. Many clusters of species of Papaveraceae can be
arranged in various sequences ranging from the complicated to the simple,
or the reverse, on the basis of morphological patterns in floral vasculature. As
for the actual direction of evolutionary changes among the ancestors of these
plants, the arguments often are about as valid for one sequence as for an-
other. Simplicity of structure in many cases may just as well be due to ad-
vancement or reduction as to antiquity; and complexity may as well be at-
tributed to multiplication of parts as to primitiveness. Surely, these sequences
imply evolutionary proximity among the taxa exhibiting them but many
questions concerning phylogeny cannot be answered without other kinds of
information.

*Department of Botany, Smithsonian Institution, Washington, D.C., and The University of
Kansas, Lawrence.

SMITHSONIAN aenna saree

26 Tue University SciENCE BULLETIN

The subfamily Platystemonoideae, comprised of Platystemon, Hespero-
mecon, and Meconella, is one of four major lines of development in Papa-
verceae. The other subfamilies are Chelidonioideae, Eschscholzioideae, and
Papaveroideae, which are distinguished on the basis of floral morphology,
pubescence, pollen, and seeds (Ernst 1962a, b). Such well known genera as
Corydalis, Dicentra, Fumaria, and their immediate allies are not included in
this conception of the family but form the related family Fumariaceae.

The objective of this paper is an improved taxonomy. The commentary
purports to explain evolutionary relationships among the taxa of subfamily
Platystemonoideae and the relationship of this complex of species to other
Papaveraceae. These taxa have been studied in conjunction with my morpho-
logical survey of the family. Since opinions have differed substantially on
systematic arrangement of these and other taxa of Papaveraceae, it was
thought that a detailed examination of floral morphology would bring to light
some less conspicuous structural features for comparison. Wild populations of
all species have been studied throughout much of their natural ranges and
representatives of all species have been examined cytologically and cultivated.
These observations have been supplemented by examination of herbarium
specimens from more than 20 herbaria. The following commentary con-
cludes with a taxonomic treatment of subfamily Platystemonoideae.

ACKNOWLEDGEMENTS

Among the many persons who have aided me in this study are M. A.
Canoso, R. H. Eyde, R.'S. Ferris, A. J. Hull, J. I. Howell, KR: @: Jacksansate
Koyama, H. Lewis, M. E. Hale, T. F. Niehaus, R. Ornduff, R. N. Philbrick,
P. H. Raven, R. L. Vaylor, J. Hy Thomas, Hi. J. Thompson, and eee
Twisselmann. I thank the curators of the herbaria where I have visited or
borrowed specimens. Part of the research presented in this paper was carried
out while I was a graduate student at Stanford University.

MATERIALS AND METHODS

Natural populations of the plants mentioned in this paper were observed
in Arizona and Oregon as well as in many parts of California. When feasible,
cytological fixations and herbarium voucher specimens were collected. A list
of collections studied in detail and cultivated for at least one season at Stan-
ford University appeared in Ernst (1958, p. 114). The first set of these col-
lections is deposited at Stanford University and duplicates also are available
at the University of California, Berkeley; California Academy of Sciences,
San Francisco; and Rancho Santa Ana Botanical Garden, Claremont. Be-
sides the herbarium materials in these institutions I also have examined

| |
| eunnieest | etaal aac)

PLATYSTEMON e@
MECONELLA OREGANA
CANBYA AUREA ° | |

| \

. CANDIDA * | | es ieee

C. CA | | esr ee

eres] ee tal | em oh

Fic. 1. Distribution of Platystemon californicus, Meconella oregana, Canbya aurea, and C.
candida. Northernmost locality is Jesse Island off Vancouver Island, British Columbia; northern
cluster of islands off Southern California, left to right, San Miguel, Santa Rosa, Santa Cruz,
and West Anacapa; next cluster south, left to right, San Nicolas, Santa Barbara, Santa Catalina,
and (no record) San Clemente; southernmost locality is Guadalupe Island, Mexico.

28 Tue University SciENCE BULLETIN

specimens from ARIZ, BH, BM, CU, E, GH, JEPS, K, LA, MO, NY, OBI,
ORE, OSC, ND, PH, POM, SBBG, SBM, SD, UCSB, US, WILLU, Wes}
WTU.

The chromosome numbers were reported earlier (Ernst, 1958) and no
new data are added here. It should be noted, however, that the chromosome
number for Hesperomecon linearis, n=7, is based upon earlier reports for
this species under the name of Meconella linearis. The illustrations are from
camera lucida tracings of materials fixed in 3:1 ethanol-acetic acid and
squashed in acetocarmine (Figs. 6, 7). Unless otherwise indicated, the
figures represent meiotic divisions of pollen mother cells.

The floral diagrams represent architectural patterns of primary xylem
and were prepared from young flowers cleared in lactic acid (Figs. 4, 5, 8, 9,
11). The materials used are from my own collections. The flowers were
cleared after some preliminary wetting or bleaching treatment such as carnoy
solution. If they were from pressed and dried herbarium specimens, they
were first moistened with alcohol or water. Larger or darkly colored flowers
sometimes required prolonged soaking in mild potasium- or sodium-hy-
droxide solution after which they were washed in water and placed in lactic
acid. It was advantageous to examine younger materials in polarized light as
newly formed secondary thickenings on cell walls glowed conspicuously and
developmental stages of tracheary tissues could be compared. The examina-
tion of whole cleared flowers often revealed details in structure and propor-
tion that are lost when one only studies sections of materials embedded in
parafhn.

The diagrams show the flowers as though cut longitudinally and flattened.
Unfortunately, the position and relative sizes of the organs are somewhat
distorted; however, an attempt was made to keep the distortions relatively
consistent. Certain problems must be acknowledged in this kind of two
dimensional representation of transparent or translucent three-dimensional
subjects. Among the foremost of these is that no two flowers, even from the
same plant, were precisely the same and some of the illustrations concern
very small structures which sometimes were difficult to observe. Since several
collections or several flowers have been examined for each species, it is
hoped that the diagrams are representative of the typical structure. There were
exceptions and not all of those observed have been reported. Floral mor-
phology should be thought of as dynamic in the sense that changes in struc-
ture and proportion are continuous from meristematic initiation until abscis-
sion. Even after cessation of active growth, changes that occur with aging
and drying-out must be expected. Eames (1961, pp. 87, 229) appropriately
cautions that floral organs, for comparative studies, only are differentially
mature at the time of anthesis but this should not discourage the study of
developmental phenomena. The distinctive developmental patterns in

Frorat MorpHoiocy aNp SysTEMAtics oF Platystemon 29

Platystemonoideae and especially in Meconella serve as examples since they
soon are obliterated and lead to rather similar knots of tracheary tissues in
the receptacles of older flowers.

A discussion of floral morphology is made more difficult by the problem
of applying consistent names to structures, particularly in the dynamic phases
of development. A clear account of trace in particular is not possible and one
must resort at times to vein as well as to bundle and trunk in an attempt to
distinguish more or less equivalent portions of the vascular supply. Con-
fusion arises in describing the vascular supply of the gynoecium when one
must compare a single (but theoretically duplex) placental bundle (or vein)
with a pair of adjacent ventral veins (or bundles) which might be somewhat
joined only basally. A placental bundle and a pair of ventral veins to ad-
jacent carpel margins are homologous except that the latter are not com-
pletely fused. A distinction between trace and vein (or bundle) sometimes
is dificult because the vein part often differentiates before the trace part can be
seen or before it is completely connected with the remainder of the vasculature
of the receptacle. It is hoped that the diagrams will clarify these points.

toe PEATYSTEMONOIDEAE

Over half a century ago a monograph of Platystemon, Hesperomecon, and
Meconella was published by E. L. Greene (1903), who included a total of
64 species, most of which he described as new. Soon afterward Fedde (1909)
revised these taxa and increased the number of species to 71 plus some infra-
species. When Jepson (1922) studied the same taxa he reduced the number
to three species and a few varieties and placed them in two genera. A more
satisfactory account was presented by Abrams (1944) who, for the first time,
adequately circumscribed the species. In my opinion there only are five
species and while I employ the same specific limitations used by Abrams, I
find myself more in agreement with Greene concerning the relegation of the
species to genera.

Platystemonoideae are indigenous to the western United States and, in
particular, to California where all occur and three are endemic or very
nearly so. They are colonial, low, tufted, obligate vernal annuals. When
elongated flowering shoots are produced, the leaves on these are opposite or
whorled and usually reduced in size. The lower leaves more or less are
alternate. The inflorescences are terminal, determinate, and more or less
scapiform. The buds nod and the erect flowers, which close at night, are
borne on relatively long peduncles and have trimerous perianths with three
sepals and a total of six petals inserted in two cycles; the outer petals are a
little larger than the inner ones. The staminal filaments frequently are ex-
panded or toothed; the pollen is tricolpate. The gynoecia are superior, com-
posed of three or more carpels and are syncarpous but with separate stigmas.

Tue University ScIENCE BULLETIN

ae =| =
J \
|
e OM | ‘
= 7 | :
% : | !
5 dic Att ae i,
SSS SS aa = SS i|
yo af = \\
Fadl t i 1
i peo mens i ‘
UN | | i |
hE Tet i
pp . | pa
fg (| vy i e 4 4 i
e atte .
e : NE “ ay 1
3 } \ ¢ \
+ ; yf IG | 1
t e = aaa Za =o ya8)
es oe ee aR | H
| ee. ’ | |
| a | ce ee |
| | e a
7 ele &
u 4 ——
| ——
| | SaSo | i] te 7 -
} | e e) 5 4 | \
| ~ ‘ 0 | |
| WED) °
| ae a
| | | eo e- E |
| tos e
| |
| | |
See °

% |
HESPEROMECON alt ae |

Fic. 2. Distribution of Hesperomecon linearis. Northernmost locality is in Butte County;
southernmost in Los Angeles County.

Many morphological features are shared by Platystemon, Hesperomecon,
and Meconella, the most important being the characteristics of the gynoecia,
which at once unite the subfamily and, at the same time, distinguish the
genera. Unlike all other Papaveraceae, the species in these genera do not
form intracarpellary valves in the fruit walls for seed dispersal, but rather
whole carpels, each with its discrete stigma attached, disassociate at maturity
by cleaving through the placental region of the fruits. The subfamilial rank
of Platystemonoideae is in recognition of the uniqueness of this situation in

Firorat MorpuHoiocy. AND SysTEMATIcs OF Platystemon Si

Papaveraceae. One cannot say from the evidence preserved in present day
plants whether the valves, so characteristic of the remainder of the family,
never developed in the phylogenetic sense, or whether they were developed
and have since been lost. Ontogenetically, the family is syncarpous in
respect to non-vascular tissues, and the apocarpouslike vascular structure
with the absence of valves in these species may be an example of a neotonic
innovation.

The only member of Platystemonoideae that is well known to taxonomists
is Platystemon. Its relative position within Papaveraceae has been given by
many authors, beginning with Bentham’s (1834) original description, as
borderline between Rhoeadales and Ranales for some combination of the
following characteristics: the perianth is trimerous; the anthers are numerous
and spirally arranged; the carpels are many and more or less separate or only
loosely coalescent, or essentially apocarpous. This allusion to its relationship
and alleged structure usually has not been accompanied by documentation,
and my observations do not support the above phylogenetic position as the
only possible interpretation. The insertion of the perianth with twice as
many petals as sepals and the petals in two cycles, as well as the specialized
and dissimilar margins of the sepals (Figs. 5A; 8E), clearly fixes Platystemon
in Papaveraceae. No convincing evidence is preserved in the floral mor-
phology of contemporary Papaveraceae that either two- or three-merous
forms, per se, are primitive. The abscission scars of the stamens externally on
the receptacle may seem to be spirally oriented but the traces to the scars
originate in no particular order (Fig. 4A). The gynoecium is syncarpous
ontogentically and its apocarpous-like appearance more or less is superficial,
becoming more conspicuous with later development and maturation (Fig.
5E, F, G). From the standpoint of the vascular supply to the carpels, the
apocarpouslike structure is much more highly developed in Meconella
oregana than in Platystemon. This situation is far more cryptic in M. oregana
and has gone unnoticed since there is no suggestion in the external form of
the gynoecium of the apocarpouslike internal pattern (Fig. 9G) and this
species, although distinctive, is not well known to botanists.

Nothing is peculiar to the structure of flowers to show taxonomic rank but
there is plenty of evidence for comparing degrees of similarity. On mor-
phological grounds there are five basic taxa (species) in the Platystemon-
Hesperomecon-Meconella complex. These occur only in western North
America and they have no close relatives anywhere except, possibly, for the
kind of limited parallelism in some genera of subfamily Papaveroideae. From
the taxonomic point of view, these taxa fall into three essentially coordinate
categories: there are three of Meconella, one of Platystemon, and one of
Hesperomecon. This conflicts with the system used in most contemporary
floras dealing with these species since Hesperomecon linearis is treated as a

32 Tue University SCIENCE BULLETIN

MECONELLA CALIFORNICA

M. DENTICULATA Om fecal

Fic. 3. Distribution of Meconella californica and M. denticulata. Northernmost locality is in
Butte County; the only insular locality is Santa Cruz Island; the southernmost locality is in San
Diego County.

species of Meconella, i.e., M. linearis. The critical element is Hesperomecon
linearis which has as much or more in common morphologically with Platy-
stemon as it does with Meconella.

A taxonomic system for these five species based on only two genera,
Platystemon and Meconella, disregards obvious morphological relationships.
Three coordinate genera or a single inclusive genus, Platystemon, is better.
With the remainder of the family in perspective, recognition of three genera
seems preferable to me but, in the final analysis, this is an arbitrary point and

W
Oo

FioraL Morpuorocy anp Systematics oF Platystemon

a single genus would be just as defensible on scientific grounds. My methods
have led me to believe that these are the practical units of Platystemonoideae
for taxonomic purposes. Additional taxa are not to be ruled out, but the
criteria for defining them will have to be more sophisticated than mine.

Pubescent, linear leaves and linear cotyledons distinguish both Platy-
stemon and Hesperomecon, but Platystemon has several carpels while Hes-
peromecon only has three. The species of Meconella are distinguished by their
glabrous and distinctly spatulate basal leaves and by their minute, spatulate
cotyledons. All species are diploid. The similarities between the species of
Platystemonoideae and the two species of Canbya, subfamily Papaveroideae,
will be discussed briefly but Candya is not included in the general remarks.
For brevity, since only one species is involved in either case, Platystemon
californicus and Hesperomecon linearis often are referred to only by generic
name. Synonomy follows Abrams (1944) except that Meconella linearis is
treated as Hesperomecon linearis. Contrary to some earlier treatments, I find
that the closest ally of Meconella oregana is M. californica and that these are
distinct species. The combination M. oregana var. denticulata is misleading
since M. denticulata stands somewhat apart from both M. oregana and M.
californica in structure of gynoecium and in morphology and/or number of
anthers.

Basic A fhnities

The Platystemonoideae are a distinctive group of species whose affinities
to other Papaveraceae are not altogether clear but there can be no question of
their being Papaveraceae. In superficial appearance, the cespitose, more or
less scapiform habit, the nodding buds with woolley sepals in Hesperomecon
and Platystemon seem somewhat similar to subfamily Papaveroideae, espe-
cially to Arctomecon, to some species of Papaver and Meconopsis, and possibly
also to Canbya. The closeness of the relationships, based upon these simi-
larities, are difficult to estimate. The multicellular-multiseriate hairs, the well
developed dissimilar margins of sepals, the tendency for suppression of the
dorsal bundle in the carpels, and the tricolpate pollen would seem to confirm
a relatively close afhnity with Papaveroideae. The basic nature of the carpels
with their free stigmas and total absence of commissural development, and
the dehiscence of the fruits without typical valves, however, clearly isolate
Platystemonoideae from Papaveroideae. Fruits predominately with only two
carpels, differences in dehiscence mechanisms of the fruits and structure of
pollen, hairs, and seeds also isolate Platystemonoideae from Eschscholzioideae
and from Chelidonioideae. The Platystemonoideae seem to represent a
separate line of development within Papaveraceae and, while the relationship
may not be very close, they show some phenotypic similarities with subfamily
Papaveroideae.

34 Tue University ScrENCE BULLETIN

Geographical Distribution

From the distribution maps (Figs. 1, 2, 3) it is obvious that Platystemon
is the most diverse ecologically since it covers the greatest geographical area
and the greatest variety of habitats and altitudes. The range extends farther
south and east than the other species, including at least seven of the Channel
Islands off California, and at one time also Guadalupe Island, Mexico. It is
not surprising that Platystemon is variable phenotypically and, apparently, it
is not uniform cytologically.

The distribution of Hesperomecon is much less extensive than, but falls
entirely within, that of Platystemon. Where Hesperomecon grows, one also
expects Platystemon to be nearby. The two often seem almost to imitate one
another in variability, and they are so similar phenotypically that specimens
of both have been mounted on the same herbarium sheet, in some cases
probably having been gathered in the same handful of plants.

The morphological gap separating Meconella from Platystemon and Hes-
peromecon is pronounced particularly in regard to leaf shape and, accord-
ingly, specific resemblences are more remote. The species of Meconella are
allopatric among themselves with one tentative exception. Some specimens
from a few localities east of Berkeley, California, in the vicinity of the
Alameda-Contra Costa county line, cannot be identified without some reserva-
tions. Perhaps the easiest solution to this taxonomic problem would be to con-
sider the small petals and the fewer stamens of these plants as representing
one of the extremes of M. californica. The variation within some of the col-
lections, however, seems to bridge the gap between M. californica and M. ore-
gana, whose next nearest locality to the north is in Oregon. Farther south in
California a few specimens have been collected on the eastern slopes of the
Mt. Hamilton Range in Santa Clara County, an area that is outside the range
of M. californica, and these plants are rather definitely equivalent to M.
oregana. Living plants in the Berkeley area have not been studied, but I sub-
mit that both M. californica and M. oregana might have occurred in this
region, perhaps within cross-pollination distance.

Populations of Meconella, with only two near exceptions, do not occur
in mixed colonies with either Platystemon or Hesperomecon, but they do
grow on neighboring hillsides. For the most part Platystemon and Hespero-
mecon inhabit more open and exposed regions such as slopes or fields, and
they span the communities from seaside dunes, woodlands, grasslands, to
meadowlike areas, and Platystemon also occurs in arid desert margins. In
those regions where the distribution of Meconella appears to overlap closely
that of Platystemon and/or Hesperomecon, the species of Meconella are more
likely to occur at somewhat higher altitudes on more moist to very wet banks,
in sunny openings to be sure, but usually near shrubs and often in association
with cryptogams such as terrestrial bryophytes and lichens as well as with

FioraLt Morruorocy aNnp Systematics oF Platystemon 35

species of Dodecatheon, Plectritis, Ranunculaceae, and Saxifragaceae. The
species often are found near colonies of Lasthenia. The dates recorded on
herbarium specimens for almost all species read from February to June. While
this establishes a general vernal growth and flowering period, the dates them-
selves probably are less important than the altitude and annual fluctuations
in seasonal rainfall and temperature. The ecological amplitude of Meconella
seems more restricted than that of Platystemon and Hesperomecon, and
Meconella occurs in the relatively earlier and wetter part of the spring-time,
regardless of the date. By appearing early, the species are able to survive
in plant communities such as chaparral or foothill woodland where digger
pine and/or douglas oak predominate and where the substrates soon become
hot and very dry.

Effective barriers between Platystemon and Hesperomecon are predicted
since the characteristics of the two species are maintained where the two occur
intimately in mixed colonies. They are similar but it is nonetheless difficult
to visualize what compromises in structure of the gynoecia might be brought
about by the impact of hybridity. Throughout most of its range Meconella
seems to have somewhat different ecological requirements than Platystemon
and Hesperomecon. It is notable, however, that Hesperomecon (normally
with many stamens) and Meconella californica (usually with about 12
stamens arranged in two cycles of six each) began to resemble each other
more closely in the foothills of the southern Sierra Nevada of California
where the number of stamens in Hesperomecon may be reduced to about 10
(almost in two series) or sometimes only six, and Meconella californica may
have as few as eight. Rather than being caused by the exchange of genetic
materials between these two species, I think the similarities in appearance of
the plants and reduced number of stamens is more likely due to parallel
responses in these species to the progressively more extreme environments in
which they are able to survive. One only can wonder if the conditions which
brought about the increased phenotypic similarity in otherwise distinctive
taxa might also enhance their potential for hybridization.

Under ordinary circumstances hybridization between Meconella and either
Platystemon or Hesperomecon would seem unlikely from the morphological
point of view; however, hybridization between any two species of Meconella
would not seem to require profound adjustments in basic morphology. Al-
though no interspecific crosses are available to demonstrate it, barriers may be
less well developed among the species of Meconella. This is suggested by the
allopatric distributions of the species, except for the single area where the
characteristics of M. oregana and M. californica appear to blend and where
plants of the two species might have occurred within pollinating distance. A
note of caution must be interjected into these speculations because the litera-
ture does contain references to hybridization in such unexpected places as

36 Tue University ScrENCE BULLETIN

between Papaver somniferum (a caulescent, obligate annual with chromo-
some numbers of n=11 or 22) and P. ortentale (an acaulescent herbaceous
perennial with n=?14 or 21) according to Kawatani & Asahina (1959). Other
hybrids equally surprising from both the taxonomic and morphological point
of view have been reported for Meconopsis.

Cultivation

Representatives of the five species of Platystemonoideae were cultivated
for at least three seasons in an open lathhouse at Stanford University. Like
most Papaveraceae, the plants usually do not survive transplantation and
new flowers seldom are initiated once the plants are disturbed. Transplants
are more likely to be successful if they are only a few mm tall. No particular
difficulty was encountered in growing either Platystemon or Hesperomecon
from seeds sown in pots or in nursery flats of soil, and the plants reappeared
year after year in the same containers without being resown. It was not pos-
sible to maintain cultures in the open ground.

The species of Meconella seemed more tempermental and frequently
damped-off from one cause or another. Cultures were established by potting
fruiting wild plants in a light, small gravel cover which seemed to provide
better aeration at the base of the plants and also served to prop the plants
upright. Seeds ripened from the transplanted specimens were allowed to fall
directly into the gravel where they remained through summer and winter in
the open lathhouse to germinate the following spring. The gravel seemed to
keep the very small seeds from being splashed out of the pots. Harvested seeds
sometimes required more than a year to germinate and more than once seed-
lings of Meconella appeared in flats or pots that had been abandoned the
previous year for lack of productivity.

The pollen often is loose before the flowers open for the first time. For
purposes of emasculating any of the species the small size of the buds makes
it difficult to be certain that pollen has not already reached the stigmatic re-
gions since some of the anthers are at about the same level as the base of the
stigmatic surfaces. The stigmas expand in a day or so after anthesis and then
the ovary elongates, lifting the stigmas above the anthers. No obvious pol-
linators have been observed in the wild state or in cultivation. A few inter-
specific crosses were attempted but none was successful. Some emasculated
flowers prepared for cross pollinations produced a few viable seeds that later
proved to be self-contaminants. In conjunction with the observation that
fertile seeds apparently were produced at times when only a very few indi-
viduals of a species were in cultivation, it is suggested that the plants probably
are at least partially self-compatible. The delay in the full expansion of the
stigmas probably enhances the opportunity for outcrossing, especially when
on bright days the petals and stamens are spread very broadly. Seed set

FiLorat MorpHoiocy aNbD SysTEMATIcs oF Platystemon 37

seemed improved in flowers which had been rubbed with those of another
plant. The flowers close at night through the duration of the petals, which
may persist for a few days. It seems likely that there is ample opportunity
for self-pollen to be deposited through purely mechanical means at least on
the bases of the stigmatic regions without the intervention of insects. Oc-
casionally, under the open conditions of the lathhouse, some fruits of all
species aborted or did not produce seeds, suggesting the importance of out-
crossing and/or pollinators. The most pertinent questions about the breeding
behavior in Platystemonoideae cannot be answered at this time.

Size

Among the prominent variables is the size of the plant which may vary
independently of the size of the flowers. Large and quite small plants fre-
quently stand side by side in nature. The circumstances of crowding, the
often substantial variation in seasonal rainfall or temperature, and the innate
plasticity of the plants to respond, no doubt strongly affect the size of the
flowers and of the plants beyond the generality that the last formed flowers
are smaller than the first. No unusually large plants were recovered from
wild seeds cultivated at Stanford although more or less giant plants occa-
sionally occur in nature.

In a general way, the larger the organ the greater the amount of vascula-
ture. Larger perianth segments usually have more veins than smaller ones.
Smaller flowers in Hesperomecon, also may have fewer stamens (from many
to as few as six); in Platystemon the larger flowers also may have more
stamens as well as more carpels. The number of carpels in Platystemon is
variable; in all others, with rare exception, the number of carpels is three.
Generally Platystemon has the largest plants with the largest flowers, but in
fringe areas such as some coastal strands and in drought the plants may be
quite small with only a few leaves and flowers. Both Platystemon and
Hesperomecon frequently occur in relatively rich grasslands where they must
compete with other plants, especially in those seasons with ample rainfall. If
climatic conditions are conducive to an unusually early and lush growth of
introduced weeds and grasses, the plants of Platystemon and particularly
Hesperomecon are likely to be overtopped and swamped. In drier years,
Platystemon and Hesperomecon may flower above and slightly ahead of the
other herbaceous vegetation. The frequency and conspicuousness of the
plants in the same localities may vary from year to year.

Herbarium specimens indicate that before the complete urbanization of
San Francisco moderate sized plants of both Hesperomecon and Meconella
californica from this area had unusually large flowers. Across the Central
Valley in the foothills of the Sierra of California, plants of both species seem

38 Tue Universiry ScrENCE BULLETIN

to become progressively smaller from north to south. Size alone must be
viewed as a treacherous taxonomic character.

Color

Color is a conspicuous variable in aerial portions of the plants. The petals
of Platystemon often are described as cream-colored. In more extensive
colonies of Platystemon some plants have bright yellow markings on the
otherwise pale petals, and no two plants will seem to have the same pattern.
The same is true for Hesperomecon and both sometimes have petals and/or
sepals marginally tinged with reddish-purple. Information on the geographi-
cal distribution of color and patterns in terms of frequency of individuals
exhibiting it should be freshly gathered since the color is lost on herbarium
specimens.

Similar yellow color also is present in the petals of some plants of
Meconella californica and a purplish tinge sometimes shows on the sepals.
In this species the yellow occurs on the upper (adaxial) surface of the outer
petals and the lower surface is quite pale. The flowers are striking with
their bright yellow, broad outer petals and contrasting pale, narrower inner
petals. This is the only example in the family of a truly two-colored corolla,
and it is remarkable that the yellow color in this case is present by day but
absent by night. This was confirmed over a period of three successive years
at Stanford University with plants cultivated from seeds gathered in the
western foothills of the Sierra Nevada of California. The color is present
when the buds open for the first time but with the waning of the daylight in
late afternoon and evening the color fades and in darkness the petals are
white. By morning of the next day the bright color returns and this diurnal
regeneration of the color continues in the same petals for a few days until
the petals wither and fall. This color phenomenon, which reminds one of
the presence and absence of scent, is analogus to other diurnal changes such
as the opening and closing of flowers. The yellow color in the petals of
Platystemon and Hesperomecon is similar but appears on both surfaces of
the petals and is unaffected by light.

Not all plants of Meconella californica have yellow outer petals. The
plants in the Coast Ranges only have white petals and plants with all-white
petals also occur in the foothills of the Sierra so that this is not merely a
difference between the plants of the eastern and western distributions. The
two phases of M. californica are very similar morphologically but are strik-
ingly distinct in regard to coloration; however, I do not feel that I can answer
the question at this time as to whether these two phases should receive sepa-
rate taxonomic recognition. The remaining two species of Meconella only
have white petals, sometimes with an obscure greenish or yellowish spot im-
mediately next to the attachment on the receptacle.

Fiorat MorpuHoiocy ANp Systematics oF Platystemon 39

The color of the vegetative portions of the plants depends upon the pres-
ence of glaucousness, which is quite common, and also on the amount of
pubescence, both of which are likely to be quite variable. There may be
variation in the balance between yellow and green within the cells of the
plant. Some of the plants of Platystemon with excessive, shaggy pubescence
appear to be yellowish. The plants have a somewhat bitter taste, probably
indicative of alkaloids. Under most circumstances the plants have a minute
amount of thin, watery juice, but in a few instances a thick, golden orange
sap, like that of many other Papaveraceae, was observed in some large and
succulent plants of Hesperomecon and a whitish sap in Platystemon. Little
importance can be attached to sap color as it has been found to be variable in
other genera among similar plants.

Pubescence

Pubescence in Platystemon and Hesperomecon usually is conspicuous,
particularly on the long peduncles, the sepals, and leaves. The fruits of
Hesperomecon are glabrous, but those of Platystemon sometimes have promi-
nent, stiff, upwardly curving hairs. The amount of pubescence is difficult to
quantify. It seems somewhat more constant but less abundant in Hesperome-
con, From time to time excessively shaggy plants of Platystemon are found
and, less frequently, very sparsely hairy plants. Both Hesperomecon and
Platystemon contrast markedly in this respect with Meconella which is
glabrous or with a few short warty hairs only on the apex of the sepals. In
all cases the hairs are multicellular and multiseriate. If the cells lie parallel
the hair appears smooth, but if the upper tips of the cells turn outward, the
hair appears rough. Rough and smooth hairs often occur on the same plants.
Unbranched multiseriate hairs also characterize subfamily Papaveroideae, but
the hairs of subfamily Chelidonioideae are uniseriate or branched. The hairs
of Eschscholzioideae are unicellular. Glabrous plants, of course, occur in all
groups. The amount of pubescence as a taxonomic character is unreliable
throughout the family; however, the cellular composition of the hairs is a
useful criterion for establishing higher taxa in Papaveraceae.

Cytology

The chromosomes of Platystemon, n=6, and Hesperomecon, n=7, are
relatively large and easily studied (Figs. 6, 7). Supernumerary chromosomes
have been observed in Platystemon californicus. Multivalents are unusual
features of Hesperomecon linearis. The chromosome number for H. linearts
was reported earlier (Ernst, 1958) under the name of Meconella linearts.
Persistent nucleoli seem to be frequent in both Meconella and Hesperomecon.
No obvious multivalents have been seen in Meconella although some ex-
amples of possible secondary attraction between bivalents has been noted,

40) Tue UNiversiry ScrENcE BULLETIN

The chromosome number for each of the three species of Meconella is
n=8 (Fig. 7). Supernumerary chromosomes and multivalents also have
been observed in Canbya. The chromosomes of Meconella seem smaller and
more difficult to study than those of either Platystemon or Hesperomecon,
and the chromosomes of M. oregana are the smallest of all (Fig. 7T). The
morphology of the chromosomes within taxa appears to be somewhat variable
and these peculiarities should serve as useful markers in studies of cytological
races.

Genomes of six or seven pairs of chromosomes (and multiples or other
numbers) are found in Papaveroideae and Eschscholzioideae, and six pairs
(and other numbers) in Chelidonioideae. Eight pairs of chromosomes, as in
Meconella, are unusual in Papaveraceae, being known only in Meconopsis
(Cathcartia) villosa and Canbya (here complicated by supernumerary
chromosomes) but are somewhat more common in Fumariaceae.

Floral Morphology

The peduncle and perianth. The bundles in the peduncles coalesce apically
to form a cylinder of tracheary tissues, here called the receptacle framework,
which frequently is interrupted by irregular gaps. The peduncles of Platyste-
mon and Hesperomecon contain six vascular bundles and each is directly
below a segment of the perianth (Fig. 4). Three lead directly to the sepals
and the alternate three directly to the outer series of petals. While each of
the sepals of Platystemon has three main traces, each trace divides below the
articulation of the sepal on the surface of the receptacle. The lateral traces to
adjacent sepals are branches from a common trunk. Each of the petals also
has three main traces which also divide before reaching the base of the petal.
The three main traces to a petal all originate from a single trunk, but the
three main traces to the sepals represent three different trunks. Each petal
trunk is directly associated with a sepal trunk, and the two emerge jointly
from the receptacle framework directly over a bundle in the peduncle. The
internal features of the perianth in Hesperomecon are very similar to those of
Platystemon with only a little less branching of the traces before they enter
the bases of the appendages. An alternative explanation of the vascular
pattern of the perianth is that each petal has a single basic trace (which
branches and looks like three traces) while a sepal has three basic traces
(which also branch). It is very difficult to standardize the numerical refer-
ences for petal traces since some petals, e.g., in some species of Eschscholzia
and Meconopsis, clearly have multiple traces. Three traces to the petals seems
to describe Platystemon and Hesperomecon, but in Meconella denticulata,
because of the relative depth of the receptacle, a petal would seem better
described as receiving a single trace.

Fiorat Morreuoiocy anp SysTEMArtics oF Platystemon 41

QO

EM ba

es Su 2) OS = =} N= a aie

Fic. 4. Vertical diagrams of vascular patterns in cleared flowers. —A, Platystemon callt-
fornicus, Ernst 108; diameter of receptacle ca 2 mm. —B, Hesperomecon linearis, Ernst 214;
diameter of receptacle ca 1.5 mm. —a, stamen traces; c, petal traces; d, dorsal bundles; k, sepal
traces; v, free ventral bundles of carpels; horizontal brackets enclose sepals.

There only are three bundles in the peduncles of Meconella (Fig. 9A, D,
G). The bundles that are lost are the ones that would alternate with the
sepals, i.e., those directly below the superposed placental regions, the outer
petals, and the lateral traces to the sepals. There are lateral traces to the
sepals in M. denticulata so that each sepal is well vascularized with the
branches from three main traces. The lateral traces to the sepals are deleted
in M. californica and M. oregana; thus, the sepals receive only a single trace
which usually remains unbranched.

42 Tue University SciENCE BULLETIN

The gathering together of the tracheary threads first into a fascicle of
petal traces and of sepal traces, and these joined ultimately into a common
trunk on the receptacle framework, may not be quite correct from the onto-
genetic point of view but this describes the end pattern of the perianth. For
petal traces to be intimately associated with sepal traces is common; how-
ever, the pattern exemplified in Platystemonoideae and Canbya (Fig. 11D)
seems, by comparison to other Papaveraceae, to be both simple and compact
and to represent a high degree of organization. Relative advancement espe-
cially is evident in Meconella with its reduction in number of bundles in
the peduncle, and reduction in amount of vasculature to the sepals and petals.

The androecium. The insertion of the stamens in Papaveraceae is diverse.
In most species with relatively large flowers and indefinite number of stamens,
the traces to the stamens are gathered into fascicles or some kind of com-
plicated internal branching system. In Dendromecon, a woody genus of
Eschscholziodideae, it is notable that the traces to individual stamens are
double and seem to be arranged in more or less vertical series. In both
Hesperomecon and Platystemon there normally are many stamens which
more or less are uniformly distributed around the periphery of the receptacle
(Fig. 4). The stamens in Platystemon sometimes are claimed to be spirally
arranged, but the traces to the stamens seem to be inserted at random and are
not in any particular order since they originate both singly and in clusters
on a receptacle framework which also is interrupted by gaps. In this instance
the pattern would seem better described as resulting from the crowding of
the maximum number of stamens into the available space on the surface of
the receptacle. The extent of truly spiral arrangement of floral appendages
and its significance requires further investigation. There are other examples
in Papaveraceae, i.e., Meconopsis, where the stamens superficially may appear
to be spirally arranged, but internally the staminal traces are associated with
very elaborate branching systems which are not spirally oriented.

There are only six or fewer stamens in Meconella denticulata and M.
oregana, inserted more or less alternately with the petals in a single series.
The stamens in M. californica are unequal in length and are inserted in two
cycles with a total of about 12 stamens (Fig. 9). The outer stamens are some-
what shorter and alternate with the petals; the inner ones are opposite the
petals. Reduction in the number of stamens is to be expected in the southern
regions of distribution in M. californica and also in Hesperomecon. Smaller
flowers in any case can be expected to have fewer stamens than larger ones.

The morphology of the filaments often is variable within a species and
within a flower. The filament frequently is expanded laterally and apically
dilated or sometimes toothed. The anthers contain four chambers of pollen
mother cells and become two-locular at anthesis. The anthers in Platystemon,
Hesperomecon, and Meconella denticulata (Figs. 5B; 8D; 9B) are distinctly

Fiorat Morrnoiocy anp Systematics oF Platystemon 43

Fic. 5. Diagrams of floral appendages of Platystemon californicus, Ernst 108. —A, sepal,
adaxial view, X 8; B, stamen, adaxial, X 9; C, cleared ovary shown as though partially transparent
over longitudinal section of receptacle, X 9; D, carpel showing attachment of ovules, X 7; E,
young gynoecium viewed from above showing lateral continuity of carpels, X 56; vertical
view of E with one stamen, shown as though transparent, the domed floor and the thick
placental regions darkened, X 56; G, transverse section through ovary showing position of
locules and ovules around a central cavity, X 20; H, traces to carpels showing attachment of
ovules and position of dorsal and ventral bundles. X 7. —a, stamen traces; a’, stamen; c, petal
traces; d, dorsal bundles; k, sepal traces; 0, ovules; v, ventral bundles.

44 Tue UNIverRsITY SCIENCE BULLETIN

oblong but those of M. californica and M. oregana (Fig. 9E, H) are orbiculate
to ovoid or deltoid. The anthers of M. denticulata frequently are as long or
longer than their filaments. The pollen is tricolpate in all species.

The gynoecium. The gynoecia of Papaveraceae are syncarpous and com-
posed of two to many carpels and, for the most part, the stigmas are united
or continuous from carpel to carpel by some kind of stylar union or by
duplex commissural development of the stigmas. Most of the family also have
intracarpellary valves by means of which the seeds are dispersed from the
fruits (Fig. 11). The Platystemonoideae have three carpels except in Platy-
stemon where there are several to many carpels; in all cases valves are lacking
and the stigmas are free. More than one gynoecium occasionally develops
in a flower of Platystemon resulting in two or more adjacent independent
whorls cf carpels (see Jepson, 1922, p. 555). The morphology of the gynoecia
in Platystemon, Hesperomecon (as Platystigma), and Meconella was de-
scribed clearly by Lignier (1911) and subsequently by others, most of whom
saw and illustrated essentially the same structure.

A morphological explanation of the carpel perhaps still is debatable, but
the conception of solid and open carpels discussed by Eames (1961, pp. 197, ff)
is no more enlightening than the earlier hypothesis of carpel polymorphism
of Saunders (1937) which it attempts to replace and both seem unnecessarily
elaborate for Papaveraceae. The more traditional Candollean interpretation
provides the best working hypothesis for all Papaveraceae, and it is essential
for the examples in Platystemonoideae which have free stigmas and as many
carpels as stigmas.

When the primordium of the gynoecium of Platystemon first is evident
on the floral meristem, it is a continuous, low, circular crest with undulating
margins both apically and laterally. Soon this structure somewhat resembles
the flutings on an old fashioned handmade pie shell (Fig. 5E, F). In the
horizontal plane the outer (abaxial) convexities become the dorsal regions
of the carpels; the inner (adaxial) convexities are the fused ventral margins
of the carpels (duplex placentae), and the indentations between them be-
come the locules. The developing carpels with their intruding placentae and
the indentations of their as yet undefined locules surround the open central
chamber of the gynoecium with its domed floor. In the vertical plane the
high parts of the undulations become the stigmatic lobes and the low ones the
placental regions.

From an ontogenetic viewpoint, the gynoecia of the five species of Platy-
stemonoideae, disregarding the greater number of carpels in Platystemon,
differentiate on the floral meristem into much the same original form. There
is complete continuity of the tissues from carpel to carpel around an open
central chamber into which each of the theoretically duplex future placental
regions intrudes. Alternating with the placentae are centrifugal outpocket-

Firorat MorpuHoiocy AND SysTEMATics oF Platystemon 45

»
wu” fo oe

is see aby dow jt yee acy)

wer gs og, ¥ 7 Vy 9 FG ees

A ae & % : A
g 1)

bit , Be Me gt, ter of
e- as a% -
+9103} u
ee rs
en a aia

my NO

Fic. 6. Pollen mother cell chromosomes of Platystemon californicus, first division A-V
except L, second division; W, premeiotic mitosis in anther. —A-C, Ernst 103: A, 6+6
chromosomes at anaphase I; B, bridge and fragment; C, delayed separation of | bivalent. —D-H,
Ernst 108: D, E, F, 6 bivalents; G, 6-++6 at Ar; H, delayed separation of 2 bivalents plus 1
supernumerary. —I-L, Ernst 164: I, 6 bivalents plus 2 supernumeraries; J, K, 6+6 and
supernumeraries at Ar showing division of 2 supernumeraries; L, showing supernumeraries at
second telophase. M to S, examples of irregular divisions with supernumeraries. M-P,
Ernst 111; —Q—S, Ernst 176: —T-V, each showing 6 bivalents: T, Ernst 234; U, Ernst 491;
V, Ernst 515. —W, Raven 15213: showing 12 somatic chromosomes. —Arrows indicate some
of the supernumerary chromosomes.

ings of the central chamber in the position of each of the dorsal trace regions
of the carpels. When viewed in the transparent condition, the ovary wall has
alternating thick and thin regions in transverse section (Fig. 8H-J). In
Meconella and Hesperomecon, development leads to a gynoecium of three
parietal placentae around a single central locule and the thickness in the
placental regions diminishes.

Unlike the gynoecium of Hesperomecon, which is deformed into a three-
cusped shape in transverse section with diffuse placental regions, the gynoecia
of the three species of Meconella remain essentially round in transverse out-
line and the placental regions are vertically restricted (Fig. 8F, G). The
ovary wall in Platystemon, however, appears peculiarly abaxially deformed in
such a way that the several intrusive placentae continue to jut into the original
central chamber while at the same time the outpocketings of the central cham-

46 Tue University ScrENCE BULLETIN

s don <4, {facets

(4 0 oe, SE

D

| . ‘ EN 5 oy 1 pido
8 —— \/
{35 | DIPW Se iS % ss SsB1dhoe

Fic. 7. Pollen mother cell chromosomes of Hesperomecon linearis, A-L; M, premeiotic mitosis
in anther. Meconella, N-T. —A, B, E, Ernst 235a: A, 7 bivalents; B, delayed separation of
bivalent with ?fragment; E, 7 bivalents. —C, D, F, Ernst 2356: C, D, 5 bivalents plus 1
tetravalent and nucleoli; F, 7 bivalents. —G-I, Ernst 235c: G, 5 bivalents plus 1 tetravalent; H,
3 bivaients plus 2 tetravalents; I, 7-++-7 at anaphase I. —J, K, L, Lewts s.n.: J, K, 7 bivalents; L,
bridges at anaphase I. —M, Ernst s.n.: 14+14 somatic chromosomes at anaphase showing
satellites. —-N-P, M. denticulata: N, Ernst 160, 8 bivalents; O, Ernst 158, 6 bivalents plus
?tetravalent; P, Ernst 158, 8 bivalents and nucleolus. —Q, R, S, M. californica: Q, Ernst 230, 8
bivalents and nucleolus; R, Ernst 210, 8 bivalents; S, Ernst 214, 8 bivalents. —T, M. oregana,
Ernst 243, 8 bivalents.

ber enlarge and become enclosed by what amounts to a kind of conduplica-
tion of the carpels (Fig. 5G). This leaves all of the duplex placentae in
close proximity adaxially and the outpockets are pinched off from the central
chamber as lateral locules. The lateral margins of adjacent carpels, of course,
are fused from the beginning in this syncarpous gynoecium. Later develop-
ment of the gynoecium, which intensifies the peculiar folded aspect of the
ovary wall, causes self-margins of the respective carpels to be closely appressed
internally but they do not unite. The conspicuous characteristics of the
gynoecium of Platystemon have led botanists to consider it to be essentially
apocarpous. These characteristics are superficial features of later rather than
of earlier ontogeny and seem like exageration of the same basic development
which in Hesperomecon and Meconella only can be interpreted as genuinely
syNcarpous,

Frorat Morpuoiocy anp Systematics oF Platystemon 47

Fic. 8. Diagrams of floral appendages of Hesperomecon linearis, Ernst 214, A-E; and of
Meconella, F-J. —A, vasculature of stigma showing apical bifurcation of dorsal bundle, X 15;
B, transverse section of ovary showing ovules and major vascular bundles, X 20; C, ovary
shown as though transparent over longitudinal section of receptacle, and lateral view of
stamen, X 13; D, adaxial view of stamen, X 13; E, adaxial view of sepal, X 12. —F, trans-
verse section through ovary, M. denticulata, Ernst 158, X 27. —G, transverse section through
ovary M. californica, Ernst 215, X 27. —H, I, J, M. oregana, Ernst 252: H, cleared view of
I and J from above, X 100; I, meristematic ovary shown as though transparent, view through
placental region, thicker portions darkened, X 100; J, vertical view of 1 turned through 90°,
observed through dorsal region. —a, stamen trace; a’, stamen; c, petal traces; c’, petal; d, dorsal
bundles; k, sepal traces; 0, ovules; v, ventral bundles.

48 Tue University ScrENcCE BULLETIN

By the time that the pollen mother cells have divided, the lateral margins
of the several carpels in Platystemon are marked by deep external depres-
sions. The ovules are attached alternately to the opposing internal margins
of their respective carpels in a single vertical series. Transverse sections of
the ovary embedded in paraffin taken as late as the pollen tetrad stage in the
inner anthers, which still is before the formation of the embryo sacs, clearly
demonstrate the continuity of the epidermal and parenchymatous tissues
from carpel to carpel in Platystemon.

The development of the dorsal bundle in the carpel is of interest. In
Platystemon and Hesperomecon, the dorsals are well developed, differentiat-
ing upward from the bases of the carpels to the tips of the stigmas. The
dorsals have been observed to reach the stigmatic lobes in Platvstemon long
before the ventrals were visable. In Meconella, however, the dorsals are de-
layed and it is the ventrals that reach the stigmas first. The dorsal usually
extends to the stigmas in M. denticulata, but in M. californica and in M.
oregana, it frequently is incomplete, stopping before reaching the stigma
(Fig. 9A, D, G). The dorsal sometimes was absent in materials of both
species when studied as late as the time of anthesis. Examination of the
fruits on herbarium specimens, however, shows that the dorsal sometimes is
completed and that the delayed development must be observed in younger
ovaries.

The dorsals in Chelidonioideae and Eschscholzioideae normally are
strong and complete, but in Papaveroideae they frequently are vestigial (as
in Canbya, Fig. 11), absent (as in Argemone), and sometimes (as in Papaver
and Roemeria) are replaced by strong pseudodorsal bundles which originate
from the placental bundles. Reduction of the dorsal bundle, its complete
absence, and particularly, its replacement by a pseudodorsal, certainly provides
insight into the evolutionary advancement of the carpel in these taxa, but,
obviously, it is not possible to erect a phylogeny for these plants merely on
this basis and to the exclusion of other morphological phenomena.

The placental regions of the ovary in Meconella denticulata contain a
single strong placental bundle which, near the apex, divides, sending a
branch to the stigmatic lobe on either side (Fig. 9A). In Hesperomecon,
there is a fascicle of traces toward the bottom in the ventral position, a more
or less strong, single placental bundle through the lower portion of the
ovary, but toward the top the placental bundle divides into a pair of veins
which pass to the stigma on either side (Figs. 4B; 8B, C). Each placental
region in Platystemon has a pair of ventrals, but these are joined basally at
their point of insertion on the receptacle framework (Figs. 4A; 5G, H). In
transverse section each carpel appears to be folded inwardly (ventrally con-
duplicate) so that the margins oppose one another. The duplex nature of
the placental regions (which are the fused margins of adjacent carpels) is

Fiorat Morpuoiocy aNnp SysTEMArTics oF Platystemon 49

emphasized by the paired (as well as seemingly inverted) ventral bundles in
transverse section (see Arber 1938, pp. 662, 663; Lignier 1911).

The placental regions in Meconella californica and M. oregana contain a
pair of veins from bottom to top (Figs. 9D, G; 8G). It is notable that the
ventral veins of a given carpel in these two species arise very near to their
dorsal on the receptacle framework which strongly enhances the apocarpous-
like appearance of the vasculature because the vasculature of each carpel is
totally independent of the others. The adjacent ventrals of adjacent carpels
in Platystemon actually are inserted closer to one another than they are to
their respective dorsal and they are joined basally (Figs. 4A; 5D, G, H). In
Hesperomecon and M. denticulata the ventrals of adjacent carpels clearly
are united (Figs. 4B; 8C; 9A). Subsequent connecting veins between dorsal
and respective ventrals are established in all species.

The fruit cleaves at maturity through the placental region so that whole
carpels separate partially or completely. In Platystemon, Meconella californica,
and M. oregana, the line of cleavage passes through the placental region be-
tween the paired ventrals of adjacent carpels. In M. denticulata and Hes-
peromecon the cleavage is the same in the apical portion, but below this there
is a single placental bundle rather than a pair of ventrals. Separation of the
carpels is arrested by the bundle or the bundle must be torn free from one or
the other of the adjacent carpels. The ovary wall closely invests the seeds
in Platystemon and the carpels finally fragment into one-seeded segments.
The seeds in the fruits of other species of Platystemonoideae are loose.

The dehiscence of fruits in other Papaveraceae is exemplified by Canbya
(Fig. 11). The stigmas in Canbya are permanently united into a stigmatic
mechanism with elaborately developed commissural appendages, and each
of the placental regions contains a single well developed placental bundle.
Valves are formed in the ovary wall between the placental bundles. The
valves open basipetally in subfamily Papaveroideae, creating a pore toward
the top of each carpel through which the seeds are liberated. The placental
bundles and the stigmatic mechanism are persistent and remain united. The
seeds are dispersed, in this case, through intracarpellary valves in the fruit
wall whereas in subfamily Platystemonoideae seeds are scattered by the dis-
association of whole carpels.

A few points can be assembled to summarize this discussion of the
gynoecium. The terms apocarpous and syncarpous normally describe totally
different morphological conceptions. The Papaveraceae are syncarpous but
the contrast between apocarpous and syncarpous in terms of the vascular
supply to the carpels becomes somewhat ambiguous in the Platystemonoideae.
The gynoecia of all Platystemonoideae, in all cases, clearly are syncarpous
ontogenetically with complete continuity of the non-vascular tissues from
carpel to carpel. The vascular patterns in both Meconella denticulata and

50 Tue UNiversiry ScIENCE BULLETIN

Hesperomecon are so conventional as not to require comment, but the vascu-
lature in the other three taxa, i.e.. M. californica, M. oregana, and Platy-
stemon, somewhat resembles apocarpous patterns. The apocarpouslike con-

Fic. 9. Diagrams of vascular patterns and appendages in flowers of Meconella. —A-C, M.
denticulata, Ernst 158: A, vertical diagram of receptacle, diameter of receptacle ca 0.5 mm;
B, cleared ovary on longitudinal section of receptacle showing 1 stamen, X 13; C, adaxial view
of sepal showing venation, X 8. —D-F, M. californica, Ernst 215: D, vertical diagram of
receptacle, diameter of receptacle ca 1 mm; E, cleared ovary on optical section of receptacle
showing two unequal stamens and receptacle rim, X 9; F, adaxial view of sepal showing single
vein, X 8. —G-I, M. oregana, Ernst 252: G, diameter of receptacle ca 0.5 mm; H, cleared
ovary on optical section of receptacle showing stamens and receptacle rim, X 13; I, adaxial view
of sepal showing single vein, X 10. —a, stamen traces; c, petal traces; d, dorsal bundles; k,
sepal traces; r, receptacle rim; v, placental or ventral bundles.

Fiorat Morpuoiocy aNnp Systematics oF Platystemon 51

dition of the vasculature in M. californica and M. oregana is inconspicuous
externally although internally it is more strikingly developed than in
Platystemon which previously has been considered essentially apocarpous on
the grounds of its external appearance.

DISCUSSION

A glance at the floral diagrams (Figs. 4, 9) suggests a simple evolutionary
reduction series such as from Platystemon—Hesperomecon—Meconella
denticulata—>M. californica>to M. oregana. An increase in chromosome
number from Platystemon to Hesperomecon is accompanied by a reduction
in the number of carpels and stamens but this relationship will be recon-
sidered shortly. From Hesperomecon to Meconella there is a reduction from
six to three bundles in the peduncles, a reduction in number of stamens as
well as a change in arrangement from indefinite to definite and cyclic, and
an increase in chromosome number. More important, perhaps, are the
modifications in M. californica and M. oregana, including a reduction in
vascularization of the perianth, particularly in the sepals; a tendency for late
appearance and incomplete development of the dorsal bundle of the carpels;
and spatial separation of the vascular supply of the carpels which increases
the apocarpouslike structure of the gynoecium. The double cycle of stamens
in M. californica might be a simple multiplication of parts in an otherwise
diminishing sequence. In M. oregana there is an increase in the frequency
of teretological fusions, additions and deletions.

The reduction series from Hesperomecon through Meconella oregana
seems clear and well preserved in contemporary plants. The highly reduced
M. oregana, probably one of the most highly advanced species of Papavera-
ceae, culminates a tendency for loss of parts in peduncle, perianth, an-
droecium, and gynoecium. The most interesting aspect of this morphological
sequence is the development of the strongly, although concealed, apocarpous-
like vascular supply of the gynoecium, apparently through reduction within
a typically syncarpous family accompanied by the simultaneous deterioration
of the dorsal trace.

The five species of Platystemonoideae, after consideration of many aspects
of the plants, seem to conform to a linear sequence, i.e., Platystemon, Hespero-
mecon, Meconella denticulata, M. californica, and M. oregana (Fig. 10).
The distinctive spatulate leaves clearly separate the species of Meconella.
The remainder of the arrangement is influenced by a few structural features
of the carpels and the reduction-multiplication series in vascular supply in
the flowers. The relationship of Platystemon and Hesperomecon to the three
species of Meconella seems well established. The individual position of the
three species of Meconella is less exactly fixed since there are not a large
number of characters for consideration; however, M. californica and M.

Wi
i)

Tue UNiversiry SCIENCE BULLETIN

Fic. 10. Intuitive linear arrangement of the five species of Platystemonoideae based upon
estimated phenotypic and morphological similarities.

organa must be adjacent, and it is clear that M. denticulata stands somewhat
apart. On external features M. denticulata and M. oregana are phenotypically
close, but internally they are more remote on the basis of presence of re-
ceptacle rim, relative length to width of receptacle, length of anther to fila-
ment, and the distributional disjunction. It could be argued that pheno-
typically M. californica with its more numerous stamens and sometimes
colored petals should receive a position closer to Hesperomecon. Alternative
arrangement of the species is possible, but I believe that a linear one as
presented above is the most representative of the relationship. In this system,
the direction of diminishing complexity of floral structure clearly is from
left to right. The size of the flowers usually is largest in Platystemon and
smallest in M. oregana, and generally the amount of vascularization varies
more or less directly with the size of the flower or organ in Papaveraceae.
This sequence, when read from left to right, also reflects the increased local-
ization of the plants to more mesic sites and the increase in number of
chromosomes.

It is tempting also to read this morphological series from left to right as
the phylogenetic history of these species. I hesitate to accept this as a phlogeny
in the historical sense because, morphologically, the sequence reads just as
well in the opposite direction and, phylogenetically, we are left without a
reliable corner stone. Reading from left to right is more appealing because
it places M. oregana with its reduced dorsal bundle in the carpel, along with
general diminution of floral structure and size, in the most derived position.
If it merely is the apocarpouslike structure per se of Platystemon which
causes it to be thought of as a prototype, then M. oregana with its even
greater apocarpouslike structure, is just as likely a protoype. If either extreme,
1.e., Platystemon or M. oregana, is the prototype and if this sequence is
phylogenetic, then apocarpouslike plants at one end have generated typically
syncarpous plants in the middle only to revert once again to producing apo-
carpouslike plants at the far end.

The morphology of either Platystemon or M. oregana seems too queer
for a direct comparison with other Papaveraceae under the circumstances of
this discussion. The morphology of Hesperomecon is more general and,

Fiorat Morpuoiocy aNnp Systematics oF Platystemon 53

perhaps, in the abstract sense, it might require the least amount of modifica-
tion to be even more nearly like other Papaveraceae. While it is speculative,
I could readily believe that Hesperomecon is a kind of model on which evolu-
tion may have pivoted, giving rise through a reduction series leading to the
apocarpouslike structure of M. oregana, on the one hand, and to the rather
different apocarpouslike structure of Platystemon, on the other. As an
abstraction, this scheme seems the most reasonable means to explain the
evolution of form but it relies too heavily upon the symbolism that some
living species are directly derived from other contemporary species.

The basic taxonomic units of Platystemonoideae already are known, and
their general similarities should be evident from the foregoing discussion.
The first of two objectives now is to compare these similarities by some
impartial method. The second objective is to attempt to interpret the simi-
larities phylogenetically. Computer methods provide the appropriate tech-
niques for the first purpose.

Table 1 introduces 18 morphological characters having two different
states which are coded by the numerals 1 and 2. A cytological character is
added which has three states. All characters have been discussed in the
preceding text.

Table 2 presents a matrix showing the coded state of each character for
each species. This table also indicates the mean and the standard deviation
for each character. From these data the character states were standardized
and the Mean Character Difference, the Taxonomic Distance, and the Cor-
relation Coefficients were calculated by H. J. Thompson and A. J. Hill, fol-
lowing the methods of Sokal & Sneath (1963), and using the Engineering
Computing Facility at the University of California at Los Angeles. The de-
tails of these calculations could be supplied to interested readers.

Table 3 shows the Mean Character Differences (MCD) matrix in which
the lowest value (0) indicates the lowest difference, i.e., the closest similarity.
These data were the basis for the comparisons shown in the five-sided figure
of Table 4C and for the dendrogram shown in Table 4B.

The rounded-off Mean Character Difference values (Table 4C) indicate
that Meconella californica and M. oregana must be adjacent and that M.
denticulata clusters with them as shown in the dendrogram (Table 4B). It
is also evident that Hesperomecon and Platystemon must be adjacent and
that the next closest affinity is between Hesperomecon and M. denticulata.
This lends credibility to the order of the names on the dendrogram (Table
4B) for Platystemon, Hesperomecon, and Meconella denticulata and for Fig.
10. Table 4C shows that M. oregana is equidistant from both Platystemon
and Hesperomecon; also that M. californica is equidistant from both Platy-
stemon and Hesperomecon although slightly closer to them. The values in
the five-sided figure (Table 4C) demonstrate that the spatial relationships

514]

Tue University SciENCE BULLETIN

Tasre 1. Characters that show different states in the five species of Platystemo-
noideae and their codes. All are two-state characters except the last which is

three-state.

Ovary wall circular

Chromosome number six
Chromosome number seven

Characters State Code
il. Plants pubescent 1
Plants glabrous 2
We Shoots elongating 1
Shoots not elongating a
Be Leaves linear 1
Leaves spatulate 2
tic Peduncle bundles six 1
Peduncle bundles three 2
De Receptacle rim minute 1
Receptacle rim absent 2
6. Sepal traces one 1
Sepal traces three 2
Te Petals only white ]
Petals sometimes yellow 2
8. Stamens many, indefinite 1
Stamens few, definite 2
9. Stamens single series, equal 1
Stamens more than one series, and unequal 2
10. Anthers long 1
Anthers short 2
ill. Carpels three ]
Carpels more 2
12% Dorsal carpel trace well developed 1
Dorsal carpel trace vestigial or incomplete 2
13. Ventrals of carpels separate 1
Ventrals of carpels united 2
14. Ovules in single vertical series 1
Ovules in more than a single series 2
1% Fruit elongating, twisted l
Fruit not elongating, not twisted 2
16. Stigma short 1
Stigma long 2
IVE Ovary not fragmenting transversely, seeds free 1
Ovary fragmenting transversely, seeds not free 2
18. Ovary wall deformed 1
5
]
2
3

Chromosome number eight

among the five species, if precisely plotted, would describe a peculiar three-
dimensional polygon rather than the straight line shown in Fig. 10. Not
supported is my contention that M. denticulata stands at least as far removed
from M. oregana as from M. californica. This contrasts the freedom of
thought allowed in an intuitive method with the rigorous limitations imposed
by the precision of computer techniques.

The dendrogram (Table 4B) is redrawn from a printout obtained from
the Mean Character Difference matrix of Table 3 by the method of clustering

Frorat MorpuHotocy anp Systematics oF Platystemon 55

Taste 2. Data matrix with mean and standard deviation for the five species of
Platystemonoideae coded from Table 1.

OTU
Char. Platys. Hesper. M. dent. M. calif. M. oreg. MEAN ST. DEV.
Il 1.000 1.000 2.000 2.000 2.000 1.600 0.548
2 1.000 2.000 1.000 1.000 1.000 1.200 0.447
Se 1.000 1.000 2.000 2.000 2.900 1.600 0.548
4. 1.000 1.000 2.000 2.000 2.000 1.600 0.548
3) 1.000 2.000 2.000 1.000 1.000 1.400 0.548
6. 2.000 2.000 2.000 1.000 1.000 1.600 0.548
7. 2.000 2.000 1.000 2.000 1.000 1.600 0.548
8. 1.000 1.000 2.000 2.000 2.000 1.600 0.548
9, 2.000 2.000 1.000 2.000 1.000 1.600 0.548
10. 1.000 1.000 1.000 2.000 2.000 1.400 0.548
11. 2.000 1.000 1.000 1.000 1.000 1.200 0.447
IP. 1.000 1.000 1.000 2.000 2.000 1.400 0.548
il 3. 1.000 2.000 2.000 1.000 1.000 1.400 0.548
14. 1.000 2.000 2.000 1.000 1.000 1.400 0.548
Sy. 2.000 2.000 1.000 1.000 1.000 1.400 0.548
16. 2.000 1.000 1.000 1.000 1.000 1.200 0.447
Fhe 2.000 1.000 1.000 1.000 1.000 1.200 0.447
18. 1.000 1.000 2.000 2.000 2.000 1.600 0.548
19. 1.000 2.000 3.000 3.000 3.000 2.400 0.894

by weighted pair-groups (Sokal & Sneath, 1963). The MCD values are shown
at the left of the dendrogram and by the solid horizontal bars. In actual prac-
tice the data of Table 3 were inverted for convenience of the operation. The
arrangement of names of the taxa at the top of the dendrogram is according
to my preference, a value judgment, but is supported in part by the relation-
ship exhibited by the five-sided figure of Table 4C. A similar dendrogram
was obtained from the Taxonomic Distance matrix which was calculated
from the standardized character values (not shown). The two dendrograms
were nearly identical in appearance but their vertical scales were different
and do not have a clear relationship. The clustering in the two dendrograms
was the same and in the same order. The dendrogram obtained from the
Correlation Coefficients matrix also agreed in general with the Mean
Character Difference dendrogram.

The dendrogram obtained by computer methods (Table 4B) is similar to
that obtained from a simple matching analysis of the characters listed in

Table 1. The coding of the characters in Table 2 is the data for Table 4A.

Tasie 3. Similarity matrix of Mean Character Difference for the five species of
Platystemonoideae, calculated from Table 2.

OTU Platys. Hesper. M. dent. M. calif. M. oreg.
Platys. 0. 0.818 1.528 1.336 1.528
Hesper. 0.818 0. 0.945 1.330 1.522
M. dent. 1.528 0.945 0. 0.769 0.577
M. calif. 1.336 1.330 0.769 0 0.192

Raoree: 1.528 1.522 0.577 0.192 0.

56 Tue University SciENCE BULLETIN

All matching characters between any two possible combinations of species
was tallied; thus, each time a character state was the same for any two species
it was counted. The values in the lower left of Table 4A represent the sums
of matching character states for each pair of species. The values in the upper
right are equivalents expressed in percent. For convenience 20 rather than
19 was selected for the denominator in computing percent.

The values in Table 4A could be inserted into the appropriate places in
the five-sided figure of Table 4C and would express in percent the amount of
similarity between any two species on the basis of the 19 characters of Table
2. The higher values, in this case, mean the higher similarity between species.
The percent values for similarity (Table 4A) appear on the right side of the
dendrogram (Table 4B) and are indicated on the dendrogram by the broken
horizontal bars. The clustering of the first two species, Meconella oregana
and M. californica, is at 85 percent similarity. The next clustering on the
dendrogram is at 60 percent and is obtained by averaging the similarity values
between Meconella denticulata and M. californica (55 percent) with that be-
tween Meconella denticulata and M. oregana (65 percent). The clustering
of Platystemon and Hesperomecon is read directly from the table at 55 per-
cent. The value between each combination of Platystemon with the three
species of Meconella (averaging 23.3 percent) and between each combination
of Hesperomecon with the three species of Meconella (averaging 28.3 per-
cent) is reaveraged for the final clustering at 26 (25.8) percent similarity on
the dendrogram.

The superposed dendrograms (Table 4B) are similar but, since the
methods of analysis were different, the scales of the vertical lines also are
different and cannot be compared directly. The information content in
either case is relative and the order of the clustering of the species, which
is the same in both cases, seems more important than the lengths of the
vertical lines. The dendrogram provides an impartial and graphic estimate
of similarities among the five species of Platystemonoideae based upon com-
parative information. It indicates the closeness of the phenetic relationship
among the species of Meconella and the disparity between these species and
both Platystemon and Hesperomecon. It also shows that Hesperomecon
(sometimes treated as a species of Meconella) has slightly more in common
with Platystemon than with Meconella.

It remains now to offer an interpretation of the dendrogram as a possible
phylogeny. My assumptions are that all Platystemonoideae are highly ad-
vanced organisms, they ultimately had a common ancestor, the species with
most similar morphology likely have diverged from one another most re-
cently, and the least similar have been separated for the longest period of
time from a common ancestor. Rates of evolutionary changes probably were
not uniform, we have no idea of extinct forms, and we are at a loss to detect

Fiorat MorpuHoiocy AND SysTEMATIcs OF Platystemon 57

mega-mutations of the past which might have given rise to intersecting lines
or development leading to cryptic convergences and/or parallelisms. We only
can deal with what we have, not with what we wish we had.

Taste 4. Similarities among the five species of Platystemonoideae—A, hand
tabulation of matching similarities from Table 2: smaller sized numerals in upper
right are percent equivalents for larger numerals in lower left. —B, superposed
dendrograms: values shown at left and with solid horizontal bars redrawn from
computer printout calculated from Table 3 by weighted pair-group method; values
shown at right and with broken horizontal bars are hand calculated averages from
percentages shown in A. —C, comparison of rounded-off Mean Character Differ-
ences among species from Table 3; the lowest value indicates greatest similarity;
highest indicates least similarity.

M. M. M. M Meconella

Platys. Hesper. dent. | calif. | oreg. CD Platys. Hesper. dent. calif. oreg. °fo
0.07 7100

0.27
7 85

|

|

M. |

!
M. al +60
5 0.8 155

|

!

8 |

|

|

A tii |
al +26

B

M. denticulata

58 Tue UNiversiry SciENCE BULLETIN

Table 4B is a dendrogram of similarities and not a phylogeny but let it
serve as one for the purposes of this discussion since it is as good as any that
might be proposed. As an hypothetical phylogeny the dendrogram ad-
vantageously avoids the symbolism that one modern species is directly
descended from another. It suggests that the progenitors of Platystemonoideae
probably were somewhat different from present-day taxa and supports my
hypothesis that Platystemon and Meconella oregana probably are phylogenetic
dead ends, that neither are prototypes for, nor ancestral to, other living
Papaveraceae, and that neither are connecting links to other families. The
implications of this conclusion are important in view of the significance
previously attributed to the unusual structure of Platystemon since I believe
that phylogenetic attention should be shifted from Platystemon to something
perhaps closer to Hesperomecon. The reason for this is that the apocarpous-
like structure, which has gone unnoticed in Meconella californica and M.
oregana, is just as well developed in these species as it is in Platystemon.
Within Platystemonoideae, however, the apocarpouslike structure seems
more like an innovation than a phylogentic relict. Hesperomecon is a con-
venient model from which both apocarpouslike extremes might have been
derived and, at the same time, also serves as an abstract link relating Platy-
stemonoideae to the remainder of the family.

I believe that phylogenetic interpretations based primarily on Besseyanlike
dicta (see Thorne 1958, 1963, and many other authors beginning with Ben-
tham 1834) lean too heavily on Platystemon as primitive for Papaveraceae and
as a connecting link to Ranales without careful examination of Platystemon
and without attention to the relationship of Platystemon to its closest allies.
Traditional phylogenetic interpretations of this kind, which suggest that
somebody really knows what primitive structure is like, might be correct—
but, at the same time, where does one turn for the evidence to support them?
In the case at hand, it is not the queer morphology of Platystemon but the
more conventional morphology of Hesperomecon that seems important to
me. Apparent sequences in morphological complexity among similar species
surely imply relative evolutionary proximity and provide sturdy criteria for
taxonomic purposes, but speculations on evolutionary direction or phylo-
genetic history from morphological data alone are projections into a dimen-
sion where too little is known. Furthermore, it is difficult to discuss hypo-
thetical phylogeny and evolutionary similarity without introducing a quag-
mire of semantic difficulties. The low chromosome number and the nature
of the gynoecium (or of other morphological phenomena) do not seem
sufficient reason to call Platystemon primitive. No fossils are recognized
and, regrettably, no inter-taxa hybrids are available for cytological clues to
the more recent past of Platystemonoideae.

Firorat MorpuHoiocy aNp Systematics oF Platystemon 59

CONCLUSION

Platystemonoideae is one of four coordinate clusters of taxa within
Papaveraceae and whether it has any claim to phylogenetic antiquity rela-
tive to other Papaveraceae certainly is not clear. The most generalized ele-
ment of Platystemonoideae is Hesperomecon which seems satisfactory as an
evolutionary model for this subfamily, but it does not seem to be a very
promising connecting link to other families. It is significant that the phylo-
genetic position of Platystemon, depending on the facts one tends to recog-
nize, can be rationalized either as primitive or advanced, and in the end we
know very little more about phylogeny than in the beginning.

THE GENUS CANBYA (SUBFAMILY PAPAVEROIDEAE)

For the sake of completeness in this review of Platystemon and its allies
it seems appropriate to discuss the genus Canbya Parry ex A. Gray, at least
briefly, since Greene (1903, p. 156) and some other authors have thought it
“intimately related” to Platystemonoideae. The genus Canbya is typical of
subfamily Papaveroideae (Ernst 1962a, b) and the two-fold purpose of in-
cluding it now is to illustrate the unique features of Platystemonoideae and
to eliminate Canbya from Platystemonoideae. The species names appear in
the following key but are not treated taxonomically.

The genus Canbya is composed of two species and is distinguished vegeta-
tively by the diminutive size of the plants, the linear terete and compactly
arranged leaves, and the absence of elongated flowering shoots. The plants
are glabrous. The flowers are borne on relatively long peduncles and are
similar to those of Platystemonoideae with trimerous perianth, twelve or six
stamens in a single series, and a gynoecium of three carpels. The genus funda-
mentally is different from Platystemonoideae, however, in that the stigmatic
regions of the three carpels are fused into a single structure with greatly de-
veloped, downwardly directed appendages in the commissural position (Fig.
11). The glossy black seeds are dispersed by means of intracarpellary valves
in the wall of the fruit which open basipetally along regular lines of de-
hiscence. The pollen is tricolpate and the chromosome number is n=8, but
multivalents and extra chromosome material sometimes are present. The
cytology of both species needs to be restudied.

The distinguishing characteristics of Canbya aurea S. Wats. are the bright
golden yellow petals which normally are quickly deciduous, the approxi-
mately 12 stamens, and the distribution in the sagebrush plains of southern
Oregon and extreme northwestern Nevada (Fig. 1). In contrast, the petals
of C. candida Parry ex A. Gray are white and somewhat more persistent,
there only are about six stamens, and the species is limited to the Joshua Tree
woodland of the western Mojave Desert in Southern California (Fig. 1).

60 THe UNiversiry SciENcCE BULLETIN

Fic. 11. Diagrams of vascular patterns in flowers and the capsular fruit of Canbya (Subfam.
Papaveroideae). —-A-C, C. aurea, Ernst 191: A, ovary drawn as though partially transparent
showing duplex commissural appendages of stigmas and longitudinal section of receptacle, X 27;
B, same as A turned through 120°; open fruit showing united stigmas, intracarpellary valves
reflexed, and persistent placental bundles, X 23. —D, C. candida, Ernst 138:, vertical diagram
of vascular pattern, diameter of receptacle ca 0.5 mm. —a, stamen traces; c, petal traces; d,
dorsal bundles; k, sepal traces; m, duplex commissural appendages of stigmas; 0, ovules; v,
placental bundles; x, margins of intracarpellary valves; horizontal brackets enclose sepals.

Floral morphology. The peduncles contain six vascular bundles in Canbya
aurea but only three in C. candida (Fig. 11D); in both species the receptacle
framework is dissected by irregular gaps. The vascular pattern of the perianth
is similar to that of Platystemon and Hesperomecon, but the sepals of
Canbya, for their relatively small size, are uncommonly thickly vascularized.
The stamen traces in Candya are arranged in a single cycle for both species.
The vestigial dorsal bundle of the carpel is represented only by a few short,
often branched threads of tracheary tissue. The placental bundle is well de-

Fiorat MorpuHoiocy anp Systematics oF Platystemon 61

veloped, consisting of three main branches near the base. The central branch,
near the top of the ovary, divides into two portions which continue into the
adjacent stigmatic regions to either side. The lower main branches divide
repeatedly, sending veins inwardly to the ovules and peripherally to the ovary
wall. The basic branching of the placental bundle recalls somewhat that of
Hesperomecon. Toward the upper part of the ovary in Canbdya, the main
branch of the peripheral system curves toward the median portion of the
carpel. As the fruit matures, a line of cleavage separates a valve in the carpel
wall from the heavy middle branch of the placental bundle. The funda-
mental difference between Canbya and Platystemonoideae thus is demon-
strated by the persistent, fused stigmatic mechanism which is permanently
united to the persistent, main portion of the placental bundle, both remaining
behind when the valves in the ovary wall open basipetally (Fig. 11C).

When the diagrams of Meconella, Hesperomecon, and Canbya are com-
pared, there is correspondence between the position of the bundles in the
peduncles and the position of the sepals and dorsal traces to the carpels (Figs.
4,9, 11). The relative arrangement is preserved even though alternate bundles
in the peduncles are lost. The bundles that are lost in Meconella and in
Canbya candida are the ones that are alternate with the dorsal traces to the
carpels. The apparent reduction series observed in the two species of Canbya
recalls the reduction series preserved in Platystmonoideae since there is a
loss of vascular bundles in the peduncle, reduction in number of stamens,
and a reduction to obsolescence of the dorsal trace to the carpel. It is notable
that differences in internal structure between the two genera Hesperomecon
and Meconella also exist between the two species of Canbya.

Synopsis of Canbya. A discussion of the position of Canbya within
Papaveraceae only can be meaningful in a full account of subfamily Papa-
veroideae. It is coincidence, I believe, that Canbya and Meconella appear
to have so much in common. The valvular dehiscence within the carpel wall
of the fruit which, at maturity, leaves behind the persistent placental bundles
and the fused stigmatic mechanism, clearly isolates Canbya from Platy-
stemonoideae. In taking into account the whole of the family, it would
seem that C. candida may culminate a reduction series within Papaveroideae
while Meconella oregana culminates a reduction series within Platystemono-
ideae. This view supports my hypothesis that the similarities between Canyba
and Platystemonoideae are manifestations of highly advanced form in
Papaveraceae and that Canbya probably can have played no direct role in
the evolution of Platystemonoideae. Whether Platystemonoideae might have
played a part in the ancient evolution of Canbya is less clear.

KEY £O) THE SPECIES

A. Gynoecium of 3 carpels, the stigmas united and with a prominent re-
flexed lobe over the placentae; fruit wall opening by valves between

62 Tue UNniversiry SciENCE BULLETIN

the persistent placentae; leaves linear, terete; plants glabrous | Subfam.
Papaveroideae—Canbya |.
B. Petals yellow; stamens about 12, the anthers shorter than fila-

ments; mostly southern Oregon. = 16. C. aurea. |
BB. Petals white; stamens about 6, the anthers as Icng as filaments;
Southern Calitormray i 2 «2 ee ee [7. C. candida. |

AA. Gynoecium of 3 or more carpels, the stigmas free; fruit wall cleaving
through placentae without formation of intracarpellary values (Subfam.
Platystemonoideae ).

C. Basal leaves broadly linear without petiole; plants pubescent;
stamens usually many; fruits not narrowly linear.
D. Gynoecium of more than 3 carpels, each forming a separate
locule around a central chamber, the fruit shattering into
l-seeded segments, the fruit wall adhering to the seeds.
ee ee eee ee ee eee 1. Platystemon californicus.
DD. Gyncecium of 3 carpels, the seeds glossy black, free of
foot walltse2. Ce te sere eae 2. Hesperomecon linearis.
CC. Basal leaves distinctly spatulate, narrowed at base; plants
glabrous or with only a few hairs on sepals; fruits narrowly
linear (Meconella).
E. Receptacle about as broad as long, without rim; stamens 6,
the anthers frequently as long or longer than filaments.
vo SS ee Coed Ts iE ee ee liad, ee Dae aes 3. M. denticulata.
EE. Receptacle broader than long, with small rim beneath in-
sertion of sepals; anthers very much shorter than filaments.
F. Stamens about 12 (sometimes fewer), biseriate or un-
SGU peat re cat cee eo eR 4. M. californica.
FF. Stamens 4-6, in one series, about equal in length.
a' ple BIG), Jha. sacdsets he ecue Oe Od Pee ea ee oe 5. M. oregana.

SUBFAMILY PLATYSTEMONOIDEAE ERNST,
Jour. Arnold Arb. 43: 317. 1962.

Colonial vernal herbs of the western United States, at first caespitose or
rosette forming, often with elongated flowering shoots from the base, some-
times decumbent. Leaves + alternate below but opposite or whorled above.
Flowers terminal on long peduncles. Sepals 3, petals 6, stamens hypogynous,
frequently with expanded or toothed filaments. Gynoecium syncarpous but
with discrete stigmas; median carpellary traces present reduced or sometimes
absent. Carpels 3 or more, disassociating in fruit without the formation of
intracarpellary valves, dehiscent through the placentae. Hairs multicellular-
multiseriate; pollen 3-colpate; seeds many, small, without arils. Type genus:
Platystemon Benth.

Platystemon Benth., Trans. Hort. Soc. London II. 1: 405. 1834.

Plants villous, leaves broadly linear, not narrowed at base. Stamens many,
carpels more than 3, each forming a separate locule around a central chamber.
Ovary + oblong, the wall constricted between and + adnate to seeds; carpels

Frorat Morpuoiocy anp Systematics oF Platystemon 63

disassociating in fruit and shattering into 1-seeded nutlets. Cotyledons linear.
Peduncles with 6 vascular bundles. A monotypic genus. Type species: P.
californicus Benth—Cream cups. (Fedde recognized 57 species, mostly de-
scribed by Greene, but the characters seem too indistinct for specific segrega-
tion.)

1. Platystemon californicus Benth., Trans. Hort. Soc. London II. 1: 405. 1834.

Flowering plants 3-30 (or 60) cm tall, erect to decumbent; leaves 1-9 cm
long; peduncles to 25 cm long; flowers to 3.8 cm diam, petals white to cream
colored, sometimes with bright yellow marking and/or reddish-purple mar-
gins. Fruit to 1 cm broad, to 2.5 cm long; stigmas linear, to 1 cm long; carpels
rarely fewer than 5, frequently about 20 (or more), sometimes inserted in
more than | whorl per flower. Chromosome number n=6 (plus occasional
supernumeraries).

Type: Dried plants and plants grown from seeds gathered in California
by David Douglas are mentioned in the original commentary. The specimen
at Kew collected by Douglas and bearing the annotation of Bentham and the
stamp of the Bentham Herbarium should be accepted as the lectotype for this
species. The locality of the collection in California is unknown. In the
original publication, the epithet californicum was used.

This species was presented in a paper under the title “Report on some of
the more remarkable hardy ornamental plants raised in the Horticultural
Society’s garden from seeds received from Mr. David Douglas, in the years
1831, 1832, 1833,” and was “Read before the Horticultural Society, January
21, 1834.” The title page for this volume of the Transactions is dated 1835 but
circumstantial evidence in addition to the early presentation date suggests
that the paper should be dated 1834. A separately repaged reprint of this
paper is dated 1834 (original at Dudley Herbarium, Stanford University) and
it clearly is stated by the printer as “From the Horticultural Transactions.”
The renumbering of the pages suggests that the printer may not have known
what the pagination would be for the formal binding of the Transactions.
The paper also is reproduced in the French language in Ann. Sci. Nat. II.
2: 80-89, with 1834 as the title page date for the volume and a printer’s date-
line on page 81, of “Aout,” August. All of the included review papers are
dated for 1834 or earlier. Whether it was the original printing of Bentham’s
paper for the Transactions or the repaged reprint that was circulated first is
anybody’s guess. For convenience, it is easier to cite the reference in the
Transactions than the reprint, and the date for both probably is 1834 rather
than 1835. Bentham’s paper includes the original descriptions for nine new
species of Papaveraceae from California, some of them representing new
genera.

64 Tue University ScrENCE BULLETIN

Distribution: OREGON (Coos Bay, Coos Co.), southward through
CALIFORNIA including the islands of S. Miguel, S. Nicolas, Sta. Barbara,
Sta. Catalina, Sta. Cruz, Sta. Rosa, and West Anacapa, into northern BAJA
CALIFORNIA; and discontinuously in ARIZONA and local in south-
western UTAH (Fig. 1). Primarily below 3000 ft. alt. in California. Slopes,
fields, seashore, sand dunes, grasslands, open oak and/or pine woodlands, to
desert. Flowering from February until June.

The altitudinal range in California is from near sea level to 100-200 feet
both coastally and inland, north as well as south. The occurrence at 7400 ft alt.
in the Panamint Mts., Inyo Co. (Hall & Chandler 6958, UC), is remarkable.
Other highest records in mountains of Southern California are Kern Co.,
4200 ft.; Mt. Pinos, 5500 ft.; S. Antonio, 5700 ft.; Cajon Pass, 3800 ft.; near
Victorville, 3200 ft.; Deep Springs, 4550 ft.; and S. Jacinto, 4600 ft. These
compare favorably with altitudes beyond the California boundaries such as
Baja California, to 3450 ft.; Arizona, 1350-4350 ft.; Utah, to 4500 ft. The
southernmost locality in Baja California probably is at 30°2’ N. Lat. (Raven,
Mathias, & Turner 12664, UC). The species also is recorded for Guadalupe
Island, Mexico (Brandegee, 20 March 1897, UC), which straddles 29° N. Lat.
about 160 miles west of Baja California, but it has not been recollected from
this island and probably is extinct there now. The populations in grasslands
of California may extend over some acres with much variation in form, size,
pubescence, and color. Besides the occasional yellow markings on the petals,
the most obvious variants are plants with succulent, broad leaves (occasion-
ally maritime); with particularly shaggy pubescence (Ventura and Kern
Cos.) ; with nodding fruits (Baja California, San Diego Co., and insular) ; or
plants nearly glabrate.

Hesperomecon Greene, Pittonia 5: 146. 1903.
Platystigma Benth., Trans. Hort. Soc. London II. 1:406. 1834, not R. Brown, 1832.

Plants villous, leaves broadly linear, not narrowed at base. Stamens many
(to few), carpels 3. Ovary with single locule, urceolate to ellipsoidal, the
carpels partially disassociating from the top in fruit, the seeds lustrous black
and free. Cotyledons linear. Peduncles with 6 vascular bundles. A mono-
typic genus. Type species: Platystigma lineare Benth.—Hesperomecon
linearis (Benth.) Greene. (Fedde recognized 9 species, mostly described by
Greene, but the characters seem too variable for specific segregation.)

The name Platystigma of Bentham is rejected for Papaveraceae because
Robert Brown used the name earlier for an Old World genus of Euphor-
biaceae which now is submerged in the genus Platea Bl. of Icacinaceae. In
creating Hesperomecon, Greene (1903, p. 139) observed that his new genus
and Platystemon “have often been seen to be so exactly alike in habit, foliage,
pubescence, color of flowers and form of stamens, that the best botanists, in

Fiorat MorpuHoiocy anp Systematics oF Platystemon 65

order to be able to say which was... which .. . would be obliged to examine
the pistils; even these, at the first flowering stage
different at first glance, as they are destined to appear when mature.”

are| not always so very

2. Hesperomecon linearis ( Benth.) Greene, Pittonia 5:146. 1903.

Platystigma lineare Benth., Trans. Hort. Soc. London II. 1:407. 1834.
Platystemon linearis (Benth.) M. K. Curran, Proc. Calif. Acad. Sci. Il. 1:242. 1888.
Meconella linearis (Benth.) A. Nels. & Macbr., Bot. Gaz. 61:31. 1916.

M. linearis var. pulchella (Greene) Jeps., Fl. Calif. 1:558. 1922.

Flowering plants 5-30 cm tall; leaves 1-8 cm long; peduncles to 16 cm
long; flowers to 3.5 cm diam; petals white to cream colored, sometimes with
bright yellow marking and/or suffused with reddish-purple margins. Fruits
to 0.6 cm diam and to 1.5 cm long; stigmas to 4 mm long, + deltoid, revolute
margined. Chromosome number, n=7.

Type: Described from dried material collected in California by David
Douglas, the locality unknown. The specimen at Kew collected by Douglas,
bearing the annotation of Bentham, and the stamp of the Bentham Her-
barium, should be accepted as the lectotype for this species. (See bibliographi-
cal note under Platystemon californicus concerning date of publication for
original description.)

Distribution: CALIFORNIA. San Francisco, Contra Costa, and Tuo-
lumne counties southward to Los Angeles Co., and discontinuously in Butte
Co. (Fig. 2). Primarily below 3000 ft. alt. but as low as 75 ft. (Contra Costa
Co.) and to 3300 ft. (Kern Co.); generally lower in northern and coastal
regions, higher (above 1000 ft.) in the south. Sand dunes, grasslands, open
oak and/or pine woodlands. Flowering from February to June.

The distribution is entirely within that of Platystemon, with which Hes-
peromecon often grows, but the range is less extensive and the plants are
less common. The localities in the San Joaquin Valley such as near the towns
of Minturn, Madera, Tulare, as well as near McKittrick are remarkable.
North of San Francisco the species is quite rare with the northernmost record
in Butte Co. (Heller 14537, UC, US). The following dubious localities are
noted: Siskiyou Co. (Rattan, DS); Fort Ross (Wrangell, US); Tomales
(Andrews, JEPS). If this species occurs in Marin, Sonoma, or Siskiyou
counties it must be exceedingly rare and newer collections with definite
locality would be most welcome. I doubt very much that the species occurs
in Oregon as given by Peck (1941, ed. 1, p. 320) who may have included it
on the basis of Fedde (1909, p. 102) who referred to an ancient specimen at
the Gray Herbarium bearing no other information than “Oregon ex D. C.
Eaton.” In Peck’s second edition (1961, p. 350), the notation was changed
to Meconella californica, which also is an error for Oregon.

The closeness between Bentham’s Platystemon californicus and_ his
“Platystigma lineare” was recognized by Curran (i.e., K. Brandegee) as early

66 Tue UNIversiry ScIENCE BULLETIN

as 1888 when she made the combination Platystemon linearts. It is ironic
that Hesperomecon linearis, the type species for the genus, has been sub-
merged in Meconella since 1916 even though the more obvious morphological
relationship is with Platystemon. No one could mistake the plants of
Hesperomecon for a species of Meconella; however, their similarity to plants
of Platystemon is pronounced and confusing at first glance. Plants of Hes-
peromecon tend to have more compact rosettes, are less likely to have elon-
gated flowering shoots, and the leaves are narrower and frequently seem
somewhat more acute but usually are minutely truncated. A positive identi-
fication, however, requires an examination of the gynoecium which, in this
case, is urn shaped to obpyriform or ellipsoidal, three lobed, and composed
of only three carpels, each with distinct, deltoid (seldom linear) stigma with
revolute margin.

Sometimes Platystemon and Hesperomecon are found in close proximity
as indicated by the frequent mixture of the two on herbarium sheets, perhaps
having been gathered in the same handful of plants. In the La Panza Camp-
ground, San Luis Obispo County, Hesperomecon and Platystemon grow
side by side and are very similar except for the ovary. Variation in Hespero-
mecon parallels, more or less, the variation in Platystemon although there
does not seem to be a form with nodding fruits or an unusually pubescent
phase. Smaller plants of Hesperomecon occur in Kern County where nearby
populations of Meconella californica seem unusually similar. In this region
Hesperomecon, normally with many stamens, may have as few as 10 or 11
arranged, more or less, in two series or sometimes as few as six stamens;
however, this similarity is unusual and the shape of the ovary and the basal
leaves serve to distinguish these plants from Meconella, which in this region
also may have a reduced number of stamens relative to the plants of more
northern localities. Reduction in number of stamens to about seven also has
been noted in San Luis Obispo County. Plants with petals marked with
yellow are occasional.

Meconella Nutt. in Torr. & Gray, Fl. N. Am. 1: 64. 1838.

Plants glabrous or with a very few short hairs on sepals, basal leaves spatu-
late, distinctly narrowed at base, the limb more or less deltoid to orbiculate,
the upper leaves more or less linear. Stamens 4-6 in one series or about 12
and biseriate; carpels 3. Ovary linear-oblong with a single locule. Fruit nar-
rowly linear, elongating to 10 or 15 times the length at anthesis, frequently
spirally twisted, the carpels disassociating from the top, the seeds lustrous
black and free. Cotyledons spatulate. Peduncles with 3 vascular bundles. A
genus of 3 species. Type species: M. oregana Nutt. in Torr. & Gray. (Fedde
recognized 6 species.)

Frorat Morpuoiocy anp Systematics oF Platystemon 67

3. Meconella denticulata Greene, Bull. Calif. Acad. Sci. 2:59. 1886.

Platystemon denticulatus (Greene) Greene, Fl. Franciscana 283. 1892.
M. kakoethes Fedde, Rep. nov. spec. 3:275. 1907.
M. oregana var. denticulata (Greene) Jeps., Fl. Calif. 1:599. 1922.

Flowering plants 3-21 cm tall, sometimes decumbent; leaves occasionally
denticulate, the basal distinctly spatulate, to 3.5 cm long, the blade to 5 mm
wide, the petiole to 3 cm long; upper leaves linear to 3.8 cm long; peduncles
to 4 cm long; receptacles relatively deep and broad but without rim below
sepals; flowers to 1.4 cm diam; petals white, occasionally with small obscure
greenish or yellowish spot near base; stamens 6, anthers linear-oblong, fre-
quently half as long or longer than filaments. Fruits to 2.5 mm diam, to 3 cm
long; stigmas to 1 mm long. Chromosome number, n=8.

Type: Greene, 27 March 1885, Temecula Canyon, north of San Luis Rey,
San. Diego Co.; only known specimen at GH.

Distribution: CALIFORNIA. Monterey, Santa Barbara, Ventura, Los
Angeles, Orange, Riverside, and San Diego counties, also Santa Cruz Island
(Fig. 3). Altitude from about 1000 to 3000 ft. (Los Angeles Co.); between
450-1200 ft. Sta. Cruz Island. Moist slopes in partly shaded opening of
chaparral or oak-pine woodlands, frequently with cryptogams. Flowering
from March to May.

Not known to me from north of Tassajara Hot Springs, Monterey Co.,
but dubiously occurring as far north as Castroville (Brandegee, April 1889,
US). The absence of a receptacle rim, the well vascularized sepals, the often
elongated anthers on short filaments, the occasionally denticulate margins of
the leaves, and the distribution in coastal Southern California distinguish this
species. Toward the southern portion of the range the plants tend to have
longer anthers and shorter filaments.

4. Meconella californica Torr. & Frém., Report. Expl. Expd. Rocky Mts.
312. 1845:

Platystemon oreganus (Nutt.) M. K. Curran, Proc. Calif. Acad. Sci. II. 1:242. 1888, in part.
Platystemon torreyi Greene, Fl. Franciscana 283. 1892.
Meconella oregana var. californica (Torr. & Frém.) Jeps., Fl. Calif. 1:558. 1922.
M. collina Greene, Pittonia 5:143. 1903.
M. octandra Greene, Pittonia 5:142. 1903.

Flowering plants 3.5-18 cm tall; basal leaves 0.5-2.5 cm long, the blade to
5 mm broad, the petiole to 17 mm long; peduncles to 12 cm long; receptacles
with small fleshy rim beneath insertion of sepals; flowers to 2.2 cm diam;
petals white or cream colored or inner petals white and outer petals yellow;
stamens about 12, biseriate, unequal in length, anthers ovoid to orbiculate,
much shorter than the filaments. Fruits to 5 cm long and to 1.5 mm in diam;
stigmas to 5 mm long, linear. Chromosome number, n=8.

Type: Frémont, probably in 1844, American fork of the Sacramento
River; two sheets, the only known specimens, deposited at NY.

68 Tue University ScrENcE BULLETIN

The name of the species usually is given as Torr. in Frém. but the original
publication attributes the name to Torr. and Frém., although there is every
reason to believe that the technical botanical work was exclusively that of
Torrey.

Distribution: CALIFORNIA. Sonoma to Santa Cruz counties coastally,
and discontinuously from Butte to Kern counties in the western foothills of
the Sierra Nevada (Fig. 3). In costal regions occurring from an altitude of
about 250-1000 ft. and from about 500 ft. (Butte Co.) to 2800-3000 ft. (Kern
Co.) in interior. On sunny moist slopes in oak- douglas fir association
coastally or oak and/or pine woodland interiorly and sometimes on or near
serpentine soil. Flowering from February to June.

The inclusion of this species in the flora of Oregon by Peck (1961, ed. 2,
p. 350) probably is an error as there are no specimens cited for documentation.
The species is distinguished by the presence of the small receptacle rim, the
single unbranched vein in the sepals, the biseriate and unequal stamens with
small anthers, the slow development of the dorsal trace to the carpels, and the
paired ventrals in the placentae. The plants in the Coast Ranges seem only to
have white petals. Some of the plants of the interior have white petals and
others have inner petals white or cream colored and the outer petals yellow,
the only example in the family of a truly two-colored corolla. The yellow
color, as discussed previously under the heading of Color, is sensitive to day-
light, fading at night to white but regenerating in daylight to yellow. The
colored phase possibly deserves taxonomic recognition but the plants do not
seem very distinctive morphologically, and it does not seem possible to tell
whether the type collections was colored or not.

5. Meconella oregana Nutt. in Torr. & Gray, Fl. N. Am. 1: 64. 1838.
Platystemon oreganus (Nutt.) M. K. Curran, Proc. Calif. Acad. Sci. I. 1:242. 1888, in part.

Flowering plants 2-10 cm tall; basal leaves to 1.8 cm long, petiole to 1 cm
long, blade to 3 mm wide; peduncles to 5 cm long; receptacles with small
fleshy rim beneath insertion of sepals; flowers to 11 mm diam; petals white;
stamens 4, 5, or 6, the anthers minute, ovoid and much shorter than the
filaments. Fruits to 2.5 cm long, to 1.5 mm diam, stigmas to 1 mm long.
Chromosome number, n=8.

Type: Nuttall, in 1835 on “open plains of the Oregon [now Columbia
River| near its confluence with the Wahlamet [now Willamette River ].”
This would seem to place the type locality somewhere near Portland, but
the only specimens known to me with definite locality for this region are
from considerably further east along the Columbia River. Two sheets col-
lected by Nuttall in Oregon are preserved at Kew and three sheets, probably
of type material, are deposited at the Gray Herbarium. This is the type
species for the genus and the only one known to occur in Oregon.

FioraL MorpHoLocy AND SYSTEMATICS OF Platystemon 69

Distribution: BRITISH COLUMBIA (Vancouver Island), and south-
ward through WASHINGTON, including Orcas, San Juan, and Whidbey
islands, and discontinuously southward to Jackson and Josephine counties
in OREGON (Fig. 1). Local and rare in CALIFORNIA as noted below.
Most plants are presumed to occur below 1000 ft. alt. Sandy bluffs, meadows,
and partly sunny moist banks. Flowering from March to June.

The northernmost limit probably is Jesse Island, Departure Bay, Van-
couver Island (W. Spreadborough, 17 April 1910, GH, CAN) at about
49°12’ N. Lat. The occurrence in California is unexpected but documented.
The species has been noted by Rossbach & Rossbach (1940) on the eastern
slopes of the Mt. Hamilton Range, Santa Clara Co., and I have collected it in
this region in the Arroyo Bayo at an altitude of about 1500 ft. (Ernst 413,
509, DS; J. T. Howell 4662, CAS; Rossbach & Beaver 665, DS). My living
plants as well as herbarium specimens were indistinguishable from a culture
of plants originating from the high banks of the Columbia River, Oregon,
between Mosier and The Dalles at an altitude of about 700 ft. (Ernst 252,
DS) when compared at Stanford. The airline distance between these two
localities is almost 600 miles and it nearly is 300 miles from Mt. Hamilton to
the nearest locality in Oregon. Some other collections from the vicinity of
the Alameda-Contra Costa county line near Berkelely, California, may be
referable to M. oregana and are noted (Chandler 854, Helsley 163, Mason
3666, Tracy 1796, all UC; J. W. Blankenship, 3 Mar. 1892, GH, with 6 very
unequal stamens). Some of these plants, however, are confusingly similar to
M. californica (e.g., Tracy 1796 has up to 9 stamens and they are unequal)
and it is possible that both species are represented in these collections. I have
not been able to study living plants in this region.

The species is distinguished by the small receptacle rim, the single un-
branched vein in the sepals, relatively narrow petals, and about 4 to 6 stamens
whose anthers frequently appear to be inserted somewhat obliquely on the
filaments. Frequent irregularities such as fused parts (stamens to petals) ;
fewer parts than expected (missing petals or stamens); or irregularities in
the symmetry have been observed. The receptacle rim is sufficient to separate
this species from M. denticulata, but in the case of incomplete or depauperate
specimens, it only is the number and unequalness of the stamens which
ultimately distinguishes M. californica from M. oregana. This accounts for
the confusion in the identity of the specimens from the Alameda-Contra
Costa region since otherwise the distributional pattern of M. oregana and
M. californica are completely allopatric.

EVPERATURE Clk D

Asrams, L. 1944. Illustrated flora of the Pacific States. Vol. 2 Stanford University Press.

ArpBer, A. 1938. Studies in flower structure. IV. On the gynaeceum of Papaver and related
genera. Ann. Bot. II. 2:649-664.

70 Tue University SciENCE BULLETIN

Eames, A. J. 1961. Morphology of the angiosperms. McGraw-Hill, New York.

Ernst, W. R. 1958. Chromosome numbers of some western Papaveraceae. Contr. Dudley Herb.

5:109-115.

. 1962a. A comparative morpholgy of the Papaveraceae. 213 pp. + 202 figs. Doctoral

dissertation, Stanford University. Reproduced by Microfilm-Xerography, University

Microfilms, Inc., Ann Arbor, 1963.

——. 1962b. The genera of Papaveraceae and Fumariaceae in the southeastern United States.
Jour. Arnold Arb. 53:315-343 [see extensive bibliography ].

Feppe, F. 1909. Papaveraceae-Hypecoideae et Papaveraceae-Papaveroideae. Jn: Engler, A. Das
Pflanzenreich 40 (IV. 104) :1-430.

GreENE, E. L. 1903. Platystemon and its allies. Pittonia 5:139-194.

Jepson, W. L. 1922. A flora of California. Vol. 1. Assoc. Students Store, Berkeley.

Kawarant, T. anp H. Asanina. 1959. External characters and alkaloids of the artificial inter-
specific F, hybrid between Papaver orientale L. (9) and P. somniferum L. (6).
Jap. Jour. Genet. 34:353-362.

Licnier, O. 1911. Notes anatomique sur l’ovaire de quelques Papavéracées; Platystémonées.
Bull. Soc. Bot. France 58:279-283.

Peck, M. E. 1941. A manual of the higher plants of Oregon. Binfords & Mort, Portland. Ed.
B WSo\e

Rosspacu, G. B. anv R. P. Rosspacu. 1940. Southern occurrences of Allium crenulatum and
Meconella oregana. Madrono 5:240.

Saunpers, E. R. 1937. Floral morphology. Vol. 1. W. Heffer & Sons, Ltd., Cambridge.

SoxaL, R. R. anp P. H. A. SNeatH. 1963. Principles of numerical taxonomy. W. H. Freeman
& Co., San Francisco.

Tuorne, R. F. 1958. Some guiding principles of angiosperm phylogeny. Brittonia 10:72-77.

. 1963. Some problems and guiding principles of angiosperm phylogeny. Amer. Naturalist
97 :287-305.

K2K32

THE UNIVERSITY OF KANSAS

SCIENCE BULLETIN

|

A REVIEW OF THE SUBFAMILY

| CYLINDROTOMINAE IN NORTH AMERICA
(DIPTERA: TIPULIDAE)

:

By

Fenja Brodo

Vor. XLVII Paces 71-115 Aprit 14, 1967 No. 3
| 3

ANNOUNCEMENT

The University of Kansas Science Bulletin (continuation of the Kansas Uni-
versity Quarterly) is issued in part at irregular intervals. Each volume contains

300 to 700 pages of reading matter, with necessary illustrations. Exchanges with
other institutions and learned societies everywhere are solicited. All exchanges

should be addressed to

LiBRARY OF THE UNIVERSITY OF KAnsas,
LawrENcE, Kansas 66044

PUBLICATION DATES

The actual date of publication (1.e., mailing date) of many of the volumes of
the University of Kansas Science Bulletin differs so markedly from the dates on
the covers of the publication or on the covers of the separata that it seems wise to
offer a corrected list showing the mailing date. The editor has been unable to’
verify mailing dates earlier than 1932. Separata were issued at the same time as_
the whole volume. Beginning with Volume XLVI, publication was by separate

numbers and the date on each number is the actual publication date.

Vol. XX—October 1, 1932. Vol. XXXIV,Pt. I—Oct. 1, 1951.
Vol. XXI—November 27, 1934. Pt. II—Feb. 15, 1952.
Vol. XXII—November 15, 1935. Vol. XXXV,Pt. I—July 1, 1952.
Vol. XxXIII—August 15, 1936. Pt. II—Sept. 10, 1953.
Vol. XXIV—February 16, 1938. Prt. II—Nov. 20, 1953.
Vol. XXV—July 10, 1939. Vol. XXXVI,Pt. I—June 1, 1954.
Vol. ©XXVI—November 27, 1940. Pt. II—July 15, 1954.
Vol. XXVII,Pt. I—Dec. 30, 1941. Vol. XXXVII, Pt. I—Oct. 15, 1955.
Vol. XXVIII,Pt. I—May 15, 1942. Pt. II—June 29, 1956.

Pt. II—Nov. 12, 1942. Vol. XXXVIII, Pt. I—Dec. 20, 1956.
Vol. XXIX, Pt. I—July 15, 1943. Pt. II—March 2, 1958.

Pt. 1I—Oct. 15, 1943. Vol. XXXIX—Nov. 18, 1958.
Vol. XXX, Pt. I—June 12, 1944. Vol. XL—April 20, 1960.

Pt. II—June 15, 1945. Vol. XLI—Dec. 23, 1960. ;
Vol. XXXI, Pt. I—May 1, 1946. Vol. XLII—Dec. 29, 1961. e

Pt. II—Nov. 1, 1947. Vol. XLII—Supplement to, June 28, 1962. —
Vol. XXXII—Nov. 25, 1948. Vol. XLIII—Aug. 20, 1962.
Vol. XXXIII, Pt. I—April 20, 1949. Vol. XLIV—Sept. 1, 1963.

Pt. II—March 20, 1950. Vol. XLV—June 7, 1965.

Bator la erate R. C. JacKson
Editorial Board ........ GerorcE Byers, Chairman

KENNETH ARMITAGE
CHARLES MICHENER
Pau Kiros
RICHARD JOHNSTON
DELBERT SHANKEL

i ae Snag at

ee eee ee ee ee

ee

—— —

THE UNIVERSITY OF KANSAS

SCIENCE BULLETIN

VoL. XLVII Paces 71-115 ApriL 14, 1967 No. 3

_ A Review of the Subfamily Cylindrotominae in North America
(Diptera: Tipulidae)’
| By
Fenjya Bropo

INTRODUCTION

The Cylindrotominae are the smallest subfamily of the Tipulidae. Mem-
bers of the group are not commonly collected and are therefore unlikely to
appear in general insect collections. Literary references are scattered, with
no all-inclusive keys existing to the North American species.

__ The purpose of this paper is to redescribe the North American species of
Cylindrotominae, to construct comprehensive illustrated keys for their
identification, to describe and illustrate representative immature stages, and
_ to bring together the pertinent literature about this group.

There is still a difference of opinion concerning the taxonomic rank of this
group. Schiner (1864) first recognized these flies as forming a natural taxon,
-Limnobinae Cylindrotomaeformes, within the tribe Limnobina. Later
workers (Brunetti, Needham) retained Cylindrotomini as a tribe of the
Limoniinae (Limnobiinae). Others (Kertész, Peus) have considered them
}as a distinct family. Alexander, Rogers and Takahashi relegate them to
_a subfamily of the Tipulidae. This latter classification is adopted here.

One North American species of Cylindrotominae was described (from
Europe) by Linnaeus (1758), four were described by Osten Sacken in 1865,
one by Johnson (1912), and the last to be described were Phalacrocera occi-
dentalis and P. vancouverensis by Alexander (1927a, b). Immature stages of
Phalacrocera replicata (from Europe) were first described by De Geer
| (1773); Cameron (1918) described the larva and pupa of Cylindrotoma
| distinctissima americana (as C. splendens), and Alexander (1914b) des-
_cribed the immature stages of Liogma nodicornis. The descriptions of the

* Contribution No. 1341 from the Department of Entomology, The University of Kansas,
Lawrence, Kansas.

ye Tue UNIversiry ScIENCE BULLETIN

larva and pupa of Triogma exsculpta and of the pupa of Phalacrocera tipulina
are included in this paper. The immatures of Cylindrotoma tarsalis and
Phalacrocera occidentalis are still unknown.

After comparing the North American cylindrotomines with those of
Europe and Asia (by the examination of some European specimens as well
as the literature), I suspect that the genera are ill-defined, as suggested by
Alexander (1949), and further study of the subfamily on a world-wide basis
may result in the lowering of some of the genera to subgeneric level.

ACKNOWLEDGEMENTS

I wish to express my sincere thanks to Dr. George W. Byers who sug-
gested this study, made much of the material available and advised and en-
couraged me. To Dr. Charles D. Michener I extend my thanks for his
friendly advice and interest. Thanks are due Dr. C. P. Alexander, Amherst,
Massachusetts, for his graciousness in loaning me valuable specimens, and
to Dr. J. R. Vockeroth, Entomology Research Institute, Canada Department
of Agriculture, for descriptions and drawings of the holotype of P. vancouver-
ensis and for the loan of many European and American specimens from the
Canadian National Collection. I wish also to thank Dr. T. E. Moore, Uni-
versity of Michigan, Dr. P. J. Darlington, Jr., Museum of Comparative Zoo-
logy, Harvard University, and Dr. Selwyn Roback, Academy of Natural
Sciences of Philadelphia, for the loan of additional specimens.

METHODS AND MATERIALS

Adult specimens either pinned or preserved in alcohol were used for
most of this study. Wherever possible, the genitalia of at least one male and
one female of each species were boiled for a few minutes in a weak solution
of KOH, washed, and placed in glycerine, permitting detailed study with
magnifications as high as 54X. The genitalia were eventually placed in micro-
vials pinned under the respective specimens.

All measurements were taken with an ocular grid and are therefore
rough, serving only to indicate the general size range of the structures in
question. Body length refers to the distance between the vertex of the head
(disregarding antennae) to the tip of the abdomen, regardless of the cur-
vature of the body. Wing length is the straight-line measurement from point
of attachment to tip.

KEY TO THE SUBFAMILIES OF TIPULIDAE
IN NORTH AMERICA (ADULTS)
(modified from Alexander, 1942)
1. Terminal segment of maxillary palpus elongate, whip-like; nasus usually

distinct; antennae usually with 13 segments; vein Cu, deflected at m-cu,
the latter at or close to fork of M344; body size usually large. ............ Tipulinae

CYLINDROTOMINAE OF NortH AMERICA 73

Terminal segment of maxillary palpus short; no distinct nasus; antennae
usually with either 14 or 16 segments; vein Cu, straight, not deflected at
m-cu, the latter placed far before fork of Mg 44, usually at or close to
FOmaOrMe body size usually small or tedium. <2 eens ec coco ene ecensce cn ecen enense cee 2

2. Male: aedaegus tripartite (Figs. 1-8), often extruded in dried specimens
(Figs. 9-20); female: cerci short, broad (Figs. 21-27); mesonotal suture
distinct only in median third of thorax, fading out laterally (Fig. 28);
wings mot patterned except for pale stigma. 22. nccccdeecececccens Cylindrotominae

Male: aedaegus having one or two openings; female: cerci elongate,
pointed; mesonotal suture distinctly “V-shaped,” characteristic of the
STIL NE Vas NOXHCON Hes lg oP (rg 026 arm tee es ne en eae eee ire eo Limoniinae

SUBFAMILY CYLINDROTOMINAE ALEXANDER

Limnobinae Cylindrotomaeformes Schiner, 1864: 560-563.

Cylindrotomaeformia Osten Sacken, 1865: 234-342.

Cylindrotomina Osten Sacken, 1869: 289-308; 1897: 362-366.

Cylindrotomini Scudder, 1894: 189.

Erucaeformia Bengtsson, 1897: 1-102.

Cylindrotomidae Kertész, 1902: Peus, 1952: 1-77; Takahashi, 1960: 81-91.

Cyttaromini Meunier, 1915: 229-230.

Cylindrotominae Alexander, 1914a: 603-605; 1919c: 926-928; 1920: 959-974; 1927a: 1-16;

1942: 292-296.

This is a curious little subfamily of the Tipulidae comprising 46 species
in nine genera. In contrast to the widespread and diverse distribution of the
majority of crane flies, the Cylindrotominae are, in general, sparsely scat-
tered over the Holarctic Region, although they also include 14 species in four
genera extending southward into the Oriental and Neotropical regions. They
usually occur in small, scattered populations in wooded situations at high
altitudes, where conditions are typically cool.

The genera within this subfamily cannot yet be satisfactorily placed in
any kind of evolutionary sequence. However, they do fall neatly into two
tribes, Cylindrotomini and Stibadocerini, on the basis of the much longer
antennae in the latter, as well as several other morphological characters.

The Stibadocerini appear to be the more specialized group morphologi-
cally, but their ecology and biology are very inadequately known. The scanty
data accompanying specimens indicate that these flies are usually found near
small waterfalls and mountain streams at altitudes of 3,000 to 11,000 ft. The
combined range of the 14 species of Stibadocerini extends along the southern
border of the distribution range of the Cylindrotomini, with nine of these
species concentrated in the Oriental Region. All these species (as far as the
records indicate) occur at high altitudes where conditions are very similar to
those of North America and Europe. Three species are found in the Austra-
lian Region (one each in New Guinea, New South Wales and Tasmania),
and one isolated species is found in the mountains of Patagonia in southern

Chile.

74 Tue Universiry SciENCE BULLETIN

Fics. 1-8. Aedeagus and accompanying structures, dorsal aspect. ej—ejaculatory duct;
pa—paraphyses (gonapophyses, of most authors). Fig. 1. Cylindrotoma distinctissima americana.
Fig. 2. Cylindrotoma tarsalis. Fig. 3. Phalacrocera occidentalis. Fig. 4. Phalacrocera replicata
(=neoxena). Fig. 5. Phalacrocera tipulina. Fig. 6. Phalacrocera vancouverensis. Fig. 7.
Liogma nodicornis. Fig. 8. Triogma exsculpta.

CYLINDROTOMINAE OF NortH AMERICA 75

The Cylindrotomini consist of 32 species arranged in five genera (see
Peus, 1952; Alexander, 1956b). Only eight of these, representing four genera,
occur in North America: Cylindrotoma tarsalis, C. distinctissima americana,
Phalacrocera replicata (=neoxena of other authors), P. occidentalis, P. tipu-
lina, P. vancouverensis, Liogma nodicornis, and Triogma exsculpta. The

Fics. 9-20. Hypopygia of males. ae—aedeagus; 9—ninth sternum. Fig. 9. Cylindrotoma
distinctissima americana, ventral aspect. Fig. 10. Cylindrotoma distinctissima americana, \eft
lateral aspect. Fig. 11. Cylindrotoma distinctissima borealis, left |ateral aspect. Fig. 12.
| Cylindrotoma distinctissima distinctissima, left lateral aspect. Fig. 13. Cylindrotoma distinctissima
americana, ninth tergum. Fig. 14. Cylindrotoma tarsalis, left lateral aspect. Fig. 15. Phalacroera
tipulina, \eft lateral aspect, showing position and outline of aedeagus. Fig. 16. Phalacrocera
replicata (=neoxena), left lateral aspect. Fig. 17. Phalacrocera occidentalis, left dististyle, outer
aspect. Fig. 18. Phalacrocera vancouverensis, \eft dististyle, outer aspect. Fig. 19. Liogma
nodicornis, left lateral aspect. Fig. 20. Triogma exsculpta, left lateral aspect.

76 Tue University SciENCE BULLETIN

richest cyclindrotomine fauna is in the Orient (China, Japan and India),
where there are nine species of Cylindrotoma, three Phalacrocera, seven
Liogma, two Triogma and two Diogma. The European species are Cylindro-
toma d. distinctissima, C. d. alpestris, C. d. borealis, Phalacrocera replicata,
Diogma glabrata and Triogma trisulcata.

GEOLOGICAL RECORD

The earliest information concerning the geological history of the Tipu-
lidae comes from the Mesozoic; however, this is rather inconclusive, and it
is not until the lower Tertiary that undisputed tipuline forms occur. In the
North American Eocene (such as the White River and Green River beds)
and in the Oligocene (the Florissant shales, etc.), there is evidence of an
extraordinary development of the Cylindrotominae. In the White River basin
they almost dominate the known crane fly fauna, in sharp contrast to their
paucity and irregular distribution within the fauna of today.

Many fossil finds have been erroneously placed in this subfamily, but the
only fossil genus now generally accepted as belonging to the Cylindrotominae
is Cyttaromyia, with Cyttaromyia fenestrata Scudder (1877) the type species.
This genus is extraordinarily similar to the modern genus Cylindrotoma,
which is also confined to the Holarctic and is distinguished from the
former only by having an additional crossvein in cell Rs forming another
closed cell in the wing. Cylindotoma specimens having this additional cross-
vein are occasionally found (Fig. 46). Cockerell (1920) described a new
species from the Eocene (White River, Colorado) which appeared to lack
this vein, and he therefore placed it with question in the genus Cylindrotoma
as C. veterana.

Of the five genera in this holarctic group, it is generally accepted that
Cylindrotoma is probably the most advanced genus, although a form very
similar to this genus is postulated as having given rise to the other four gen-
era (Alexander, 1927a).

Fossil record of North American Cylindrotominae to date:

From the Eocene:

Cyttaromyia fenestrata Scudder, 1877, White River, Utah.

C. fuscula (Cockerell 1920, as Asilopsis), White River, Colorado.

C. reclusa Cockerell, 1924, Green River, Colorado.

? (Cylindrotoma) veterana Cockerell, 1920, Roan Mountain, Colorado.

From the Oligocene:

Cyttaromyia cancellata Scudder, 1894, Florissant, Colorado.

C. clathrata Scudder, 1894, Florissant, Colorado.

C. oligocena Scudder, 1894, Florissant, Colorado.

C. princetonmiana Scudder, 1894, Florissant, Colorado.

The above species seem to be most closely related to North American
species of Cylindrotoma.

CyYLINDROTOMINAE OF Nortru AMERICA Te

27

Fics. 21-27. Female terminalia. 9—ninth tergum; 10—tenth tergum; en—entire edge; se—
serrated edge. Fig. 21. Cylindrotoma distinctissima americana, dorsal aspect. Fig. 22. Cylindro-
toma distinctissima americana, right lateral aspect. Fig. 23. Cylindrotoma tarsalis, right lateral
aspect. Fig. 24. Phalacrocera tipulina, dorsal aspect. Fig. 25. Phalacrocera replicata (—=neoxema),
dorsal aspect. Fig. 26. Liogma nodicornis, dorsal aspect. Fig. 27. Triogma exsculpta, dorsal aspect

78 Tue UNiversiry SciENCE BULLETIN

” aw
ee =
fi<

r)
psc
pasc = [
J
——— 45||
28 N p
Z \
32 33 1
\ Ss Vr
< 36
34 35
37
Va
4 HE
ng y 38
V7 /
“4 ag L
Se t'{ <C

rg wy 42

< " 4
y

Fics, 28-44. Structural details of Cylindrotominae. p—pronotum; pasc—parascutellum;
psc—prescutum; sc—scutum; scl—scutellum. Fig. 28. Cylindrotoma tarsalis, thorax, dorsal
aspect. Fig. 29. Cylindrotoma distinctissima americana, male, scape, pedicel and three flagellar
segments. Fig. 30. Cylindrotoma distinctissima americana, female, three typical flagellar seg-
ments. Fig. 31. Cylindrotoma tarsalis, male antenna. Fig. 32. Phalacrocera occidentalis, male,
2 typical flagellar segments. Fig. 33. Phalacrocera occidentalis, female, 2 typical flagellar seg-
ments. Fig. 34, Phalacrocera tipulina, male, scape, pedical and first flagellar segment. Fig. 35.

CYLINDROTOMINAE OF NortH AMERICA 79

It is curious that no Cylindrotominae have been found in the lower Oligo-
cene Baltic Amber of Europe. According to paleoecologists, the Baltic re-
gion in the lower Oligocene was covered with Pinus succinifera, an amber-
producing pine growing on very mountainous terrain cut by rushing moun-
tain streams. This was a humid and densely shaded environment, judging by
the total lack of any specimens requiring arid or sunny conditions and the
presence of Trichoptera and other insects which thrive in cool, damp situa-
tions. The dipterous fauna in the Baltic amber most closely resembles that
which now inhabits eastern North America between 32° and 40° latitude

(Alexander, 1931).

TRIBE CYLINDROTOMINI

Diagnosis: The Cylindrotomini (which include all North American spe-
cies of the subfamily Cylindrotominae) are characterized by medium body
size (8-16 mm) and legs which are considerably shorter than those of Li-
moniinae of equivalent body size. The wings are slightly suffused with
brown and sometimes have in addition a pale oval stigma but never possess
any other kind of pattern. The antennae are 16-segmented. The thorax has
at least a faint indication of 3 broad longitudinal stripes, giving a characteris-
tic pattern to this group of flies. The mesonotal suture is curved instead of
“V-shaped” and does not extend beyond the median prescutal stripe. The
male hypopygium is broad and bears only 1 pair of dististyles. The aedeagus
divides into 3 tubes distally and is often found protruding in dried specimens.
The female terminalia are short and broad in contrast to the much longer and
more slender external genital apparatus found in most species of the other 2
subfamilies. The blade-like cerci are dorsal to and partly overlap the hypo-
valves.

Description: Antennae 16-segmented, longer in male than in female;
scape with small, toothlike lateral projection (Fig. 34); palpi 5-segmented,
last segment sometimes lengthened, often almost as long as first 4 com-
bined but never exceeding these segments in length; clypeus short, with apical
tuft of hairs but no “nasus”; first cervical sclerite very tiny or apparently ab-
sent; second cervical sclerite large, quadrate.

Pronotum typically divided into 3 areas by two longitudinal furrows, an-
teriorly a ridge, medially a more or less raised crest, posteriorly a ridge split
and flattened medially. Pretergite glabrous, generally paler than rest of
thorax; mesonotum with 3 more or less distinct broad longitudinal stripes,

Phalacrocera tipulina, male, 4 typical flagellar segments. Fig. 36. Phalacrocera tipulina, female, 2
typical flagellar segments. Fig. 37. Phalacrocera replicata (=neoxena), male, 2 typical flagellar
segments. Fig. 38. Phalacrocera vancouverensis, male, 6 typical flagellar segments. Fig. 39.
Liogma nodicornis, male antenna. Fig. 40. Liogma nodicornis, female antenna. Fig. 41. Triogma
exsculpta, male, 3 typical flagellar segments. Fig. 42. Triogma exsculpta, female, 3 typical flagellar
segments. Fig. 43. Phalacrocera tipulina, male, tarsal segment and tarsal claw. Fig. 44.
Phalacrocera vancouverensis, male, tarsal claw.

80 Tue University ScrENcE BULLETIN

median stripe extending from pronotum to mesonotal suture, lateral stripes
commencing slightly below median stripe, swelling outwards on scutum,
narrowing, almost merging behind mesonotal suture; mesonotal suture
curved behind median stripe, fading out laterally (Fig. 28) (in Cylindro-
toma apparently continuing forward along sides of median stripe); anterior
edges of scutum raised, more or less giving impression of continuation of
mesonotal suture across thorax.

Wings pale brown due to microscopic brown hairs; veins darker brown,
costa and distal branches of radius and media bearing brown macrotrichia,
other veins bare; pale, oval stigma usually present. Subcosta ending un-
branched before stigma or joining costa or radius, or dividing, Sc: ending
in costal cell, Scz turning posteriorly (Fig. 49, Sc2a) to join Ri for a short
length, then anteriorly (Scr) to join costa; branches of Sc often fragmented.
Radius with 2 or 3 branches reaching costa; tip of Ri+2 usually obliterated,
Re considered to have fused with Ri just beyond area of stigma, often present
only as a very short, almost obliterated crossvein (Fig. 53); Rs and Ra+s
reaching costa independently. Three or 4 branches of media reaching costa:
Mi; and Mbp not separated in Liogma and Triogma; Mz and Mz not separated
in Phalacrocera replicata, P. tipulina and P. vancouverensis. Cubitus straight,
not bent at junction of m-cu, but sharply curved distally, joining wing mar-
gin perpendicularly. Two anal veins, as characteristic of the family.

Abdomen long, broad posteriorly, more slender and darker anteriorly.
Ninth tergum variously lobed or notched. Basistyles large, broad, indistinctly
separated medially by membranous region; single pair of dististyles, simple
or armed with small projections. Aedeagus? tripartite distally (bipartite in the
European Diogma) enclosing 3 ejaculatory ducts (lateral ducts in Cylindro-
toma apparently vestigial). Variously shaped scleritized plates closely as-
sociated with basal end of aedeagus (lacking in Phalacrocera occidentalis and
Cylindrotoma tarsalis, but probably represented by basal developments of
aedeagus in the latter). Paraphyses (Figs. 1, 2, 3,5, 7,8) 2 variable sclerites on
either side of or above aedeagus.

FEMALE terminalia: Ovipositor short, broad; cerci broad, flat, entire or
with serrated dorsal edge; hypovalves lightly sclerotized, fused to eighth ster-
num or separated by thin membranous region. Tenth tergum fragmented:
large flat sclerite with median and often with two lateral projections caudally
(in Cylindrotoma median projection much prolonged and forked distally),

“The morphology of the aedeagus and its accompanying structures is confusing. Snodgrass
(1904), studying the hypopygium of Phalacrocera tipulina, concluded that the penis (aedeagus)
might actually represent a fusion of the penis and its guard (adminiculum). The aedeagus of
P. tipulina has a bifurcate, dorsal, scleritized prolongation, reminiscent of the adminicular rods
of Dolichopeza (see Byers, 1961).

CYLINDROTOMINAE OF NortH AMERICA 81

2 small sclerites on either side (attached in Phalacrocera replicata and Triog-
ma trisulcata of Europe), and narrow sclerite between bases of cerci. Three
sclerotized spermathecae visible in cleared specimens.

One suspects that the cylindrotomine genera are ill-defined when noting
the striking similarity between Triogma and Liogma, both of which genera
include only one North American and one European species. The insecure
taxonomic position of Phalacrocera occidentalis, which has been shifted be-
tween Phalacrocera and Cylindrotoma, testifies to the inadequate definition of

¥ =e
48

66
63 64 65

Fics. 45-66. Wings of Cylindrotominae. M—media; R—radius; Sc—subcosta; arrows—call
attention to variations in venation. Figs. 45-48. Cylindrotoma distinctissima americana. Figs. 49-
50. Cylindrotoma tarsals. Figs. 51-52. Phalacrocera occidentalis. Figs. 53-54. Phalacrocera
replicata (=neoxena). Figs. 55-56. Phalacrocera tipulina. Figs. 57-58. Phalacrocera vancouver-
ensis (58=holotype). Figs. 59-62. Liogma nodicornis. Figs. 63-66. Triogma exsculpta.

82 Tue University SciENcCE BULLETIN

DORON
\

Lge

Fics. 67-76. Figures showing immature stages.

d—dorsal aspect; 1, 2, 3 etc.—refer to
abdominal segments.

Fig. 67. Phalacrocera tipulina, pupa, right lateral aspect. Fig. 68.
Phalacrocera tipulina, pupa, ventral aspect of typical abdominal segment. Fig. 69. Phalacrocera

CYLINDROTOMINAE OF NortH AMERICA 83

these two genera. A world-wide study of this group with particular emphasis
on the more diverse Asian species should lead to a better delineation of the
genera.

KEY TO THE GENERA OF CYLINDROTOMINAE
OF NORTH AMERICA (adults)

1. Stout, dark flies, distinctly rugose or punctate on head and
eT Ipc ele pees Ad NEw ea Pe RE Seb 1 ee coe eh a adeno Triogma

Slender bodied flies without such punctatioms: <.22.2/0 2.2.25 oe cceeeee Z

2. Thoracic dorsum with whitish or yellowish-grey pubescence; pleurae
fuloesce nts dull black OL DLO Wo sana tS nee eine Phalacrocera

At least part of thorax pale yellow, 3 more or less distinct black stripes on
prescutum, thoraxsdullior shiny but mot pubescent... 3

3. Shiny mahogany-brown or black stripes on prescutum; 3 branches of
media reaching wing margin (Fig. 59); cerci smooth, rounded blades
(Fig. 26); antennae nodulose, almost cordate in male (Fig. 39)... Liogma

Dull brown or black stripes on thorax; 4 branches of media reaching
wing margin (Fig. 45); cerci with serrated edges dorsally (Figs. 21, 22,
75) antennal scements cylindrical (Figs. 29, 30). 22 oe Cylindrotoma

The following keys to the larvae and to the pupae are provisional since
the immatures of several species remain unknown. The diagnosis of Phala-
crocera is based on the characteristics of P. replicata, P. tipulina and P. van-
couverensis. The immature stages of P. occidentalis are not known. The
diagnosis of Cylindrotoma is based solely on C. distinctissima subspecies since
the immatures of C. tarsalis are not known.

KEY TO THE LARVAE OF CYLINDROTOMINAE
OF NORTH AMERICA
(modified from Alexander, 1927a)

1. Body appendages very long, filiform and dichotomously branched
Bareticrl ype nee ct eA S1e—s Peae hed. 4 a 2 ob Rese eae Biv aes Phalacrocera

ErcravecloCMaa ees SIOnty etree seer ee pee eee Cane eee ee cree ena 2

2. Dorsal appendages simple, in a single median row on abdominal segments
Bo/ sterrestiial OM spermatophytic plants. 2.200... tcc es cee ee Cylindrotoma

Dorsal appendages in pairs with teeth or with small protuberances
RIENCE) oa ea a Tc RE ei Pee CoD 3

tipulina, pupa, ventral aspect of posterior segments. Fig. 70. Triogma exsculpta, larva, ventral
aspect. Thoracic segments indicated as pro, meso, meta. Fig. 71. Triogma exsculpta, larva, dorsal
aspect. Fig. 72. Triogma exsculpta, larva, right lateral aspect of posterior segments. Fig. 73.
Triogma exsculpta, spiracular disk of larva. Fig. 74. Liogma nodicornis, spiracular disk of larva
(redrawn from Alexander, 1920). Fig. 75. Triogma exsculpta, female pupa, ventral aspect.
Fig. 76. Triogma exsculpta, female pupa, dorsal aspect.

84 Tue University SctENcE BULLETIN

3. Two pairs of processes on spiracular disks (Fig. 74); dark, oblique,
lateral «hacks on. abdomiunalitetoas J) .0. 5. ee Liogma

Three pairs of processes on spiracular disks (Fig. 73); abdominal seg-
ments browa dorsally, whitish at‘edoes....2 ee Triogma

KEY TO THE PUPAE OF CYLINDROTOMINAE
OF NORTH AMERICA

(modified from Alexander, 1927a)

l” Abdominal terga 3:to 7 without processes or lobes, 2.2...) 2
Abdominal terga 3 to 7 with lobes on each segment. .....0....-22-2-22:--0--ece-eeeeeeeeeees 3

2, Abdominal sterna bearing small spines; mesonotal* respiratory horns
elongate ditected backwards (Figs: 67,05 es ee Phalacrocera
Abdominal segments unarmed; mesonotal respiratory horns small,
ditectedishizhtly forward. 78-2). ee ee Cylindrotoma

3. Two of 3 pairs of lobes on each abdominal tergum (Fig. 76). —.......... Triogma
A single pair of long slender lobes on each abdominal tergum. ............ Liogma

Cylindrotoma Macquart

Cylindrotoma Macquart, 1834: 107; Walker, 1856: 268-337; Schiner, 1863: 217-226; Osten

Sacken, 1865: 234-237; Alexander, 1920: 927; 1942: 294; Peus, 1952: 67.

Type: Limnobia distinctissima Meigen.

Diagnosis: Cylindrotomini with 3 dull, dark brown or black stripes on
pale yellowish prescutum. Four branches of media reaching wing margin.
Female cerci distinctly serrate dorsally, partially covered by distal branches of
median prolongation of tenth tergum.

Description: Head smooth, bare, very light to dark brown; scape short,
round; flagellar segments cylindrical; no hairs anteriorly on pronotum; pre-
tergites pale yellow; mesonotum hairless, pale yellow, often dusky anteriorly,
with 3 dark longitudinal stripes on prescutum; mesonotal suture apparently
“U-shaped”, flanking median stripe, slightly raised edges of scutum giving
only faint indication of continuation of mesonotal suture across thorax; scu-
tum partially divided by longitudinal suture; scutellar sutures clearly defined;
pleural sclerites pale yellow, black areas on anepisternum, postepisternum,
and pleurotergite; mesepisternum with transverse suture; longitudinal suture
on pleurotergite; wings faintly suffused with brown, oval stigma slightly
darker; Sci fading just before stigma, Sc2 weakly indicated or absent; radius
with two branches (Rs and Ra45) reaching costa; r-m usually present (oc-
casionally lacking in one or both wings); 4 branches of media present, M1
branching at distal edge of discal cell, Ms together with short medial cross-

* Byers (1961: 769) noted that the breathing horns of the Tipulidae are mesothoracic not
prothoracic as propounded in the literature. I have verified this for the Cylindrotominae.

CyYLINDROTOMINAE OF Nortu AMERICA 85

vein forming distal edge of cell; tibial spurs small, indistinct, close together,
covered with appressed microscopic hairs; tarsal claws small, slender. Abdo-
men varying from dark to light brown, lighter median longitudinal band on
dorsum and venter; distal edge of each segment pale.

MALE terminalia: Eighth tergum narrow, entire or partially divided dor-
sally by narrow membranous area; ninth tergum large, squarish, rectang-
ularly notched medially, produced into 2 lateral lobes distally; aedeagus with
three tubes, only middle tube showing complete ejaculatory duct leading to
semen pump.

FEMALE terminalia: Eighth tergum not continuous, reduced to 2 narrow
lateral sclerites; ninth tergum produced into 2 large lateral areas narrowly
joined anteriorly and dorsally; tenth tergum elongate, divaricate distally,
with 2 small lateral projections near base, 2 triangular fragments in mem-
brane between tenth tergum and cerci; cerci broad, dorsal margins serrate.

KEY TO ADULTS OF NORTH AMERICAN SPECIES
OF CYLINDROTOMA

1. Male with flattened, gently curved aedeagus, aedeagal tubes short, of
equal length (Figs. 1, 9); 2 dorsal edges on each female cercus, outer
edge entire, inner edge serrate (Figs. 21, 22); tarsal segments brown,
Pipountlya Gan ken, CiStally. 22. Ue Co a distinctissima americana

Males with broadly curved aedeagus, aedeagal tubes elongated, middle
tube consideraly longer than lateral tubes (Figs. 2, 14); 2 dorsal edges
on female cercus as in C. distinctissima americana, but the outer non-
serrated edge vestigial, represented only by anterior and posterior vestiges
joined basally (Fig. 23); second and third tarsal segments abruptly paler
CUESEL, GAS Eo AS oft wT) Cpa a ea Baal oll ne ad a ee ee tarsalis

Cylindrotoma distinctissima americana Osten Sacken
(Pigss Uy 9310513 625322529, 30, 20-48 977

Limnobia distinctissima Meigan, 1818: 131.*

Cylindrotoma distinctissima; Macquart, 1834: 107.

Cylindrotoma brevicornis Zetterstedt, 1838: 846.

Cylindrotoma americana Osten Sacken, 1865: 236-237; Alexander, 1942: 294.

Cylindrotoma splendens Doane, 1900: 197, pl. 8, fig. 21 (wing); Cameron, 1918: 67-89;
Alexander, 1920: 708-710, 967-969, pl. 84 (larva and pupa); 1927, pl. 1, fig. C (wing),
fig. 8 (female genitalia), fig. 9 (male genitalia). (New synonymy.)

Cylindrotoma juncta Coquillett, 1900: 401.

Cylindrotoma pallescens Alexander, 1930: 280.

Cylindrotoma splendens pallescens Alexander, 1954: 42. (New synonymy.)

Cylindrotoma distinctissima americana; Alexander, 1956a: 177.

There is much confusion concerning the taxonomic extent of this species.
Osten Sacken (1865) characterized americana from eastern North America
by its three black thoracic stripes, the central stripe being longitudinally di-
vided by a thin yellow line (Fig. 28). He concluded with the observation:
“The European C. distinctissima seems to be very like americana in its color-

“Only the basic references to the European C. d. distinctissima have been cited.

86 THe Universiry ScieENcE BULLETIN

ing.” In 1900, Doane described a western species, C. splendens, having black
thoracic stripes; no mention was made of a median yellow dividing line on
the prescutum, but he noticed three distinct dark spots on the pleurae and
minor variations in the radial and medial fields of the wing. My study of
splendens and americana revealed that the thin yellow stripe along the mid-
line of the thorax occurs occasionally in both groups of flies and that both
exhibit the same pleural markings.

In 1930 Alexander described pallescens, which he differentiated from
splendens by its much paler and rufous appearance. Later, in 1954, he came
to the conclusion that pallescens was merely a subspecies of splendens. 1 was
able to compare several hundred specimens ranging in the west from Un-
alaska, Alaska, southward to Colorado and in the east from Nova Scotia
southward to New York and westward to Michigan (see Fig. 77). It soon be-
came apparent that coloration is of little use taxonomically, as the same sort of
variation exists throughout, and the same held true for slight differences ob-
served occasionally in wing venation. The details of the genitalia are virtually
identical throughout. Despite the geographic division of these specimens into
eastern and western populations, I consider them to represent one species.
There is insufficient evidence or justification for the retention of subspecies
within this group.

In a relatively recent paper, Alexander (1956) gave his opinion that
americana is merely a subspecies of the European distinctissima. | am in-
debted to the Canadian National Collection for the loan of 11 specimens of
distinctissima from Sweden and the opportunity to compare these in the
minutest detail with representative specimens of americana from all parts
of its range. In almost every structural detail, including those of the genitalia
in both sexes, no differences could be found. The Swedish specimens, how-
ever, lacked the very pale and rufous form characteristic of some of the Ameri-
can populations, but this may simply be a function of the smaller sample size
in the former. Peus (1952) distinguished three subspecies of the European
distinctissima, and with the help of his drawings and descriptions I was
able to differentiate the americana group by the shape of the posterior edge
of the ninth tergum in the male (Figs. 10-12). Unfortunately this character
is of limited use as the exact shape of the tergum is often difficult to deter-
mine, due to distortion in drying of the insects.

Diagnosis: Males of this species have a flattened, slightly curved aedea-
gus, divided distally into 3 short tubes of equal length. Females have a
pair of cutting edges dorsally on each cercus, the outer edge smooth, the in-
ner edge serrate. All tarsal segments are brown, but the fourth and fifth are
slightly darker.

Description: MALE: Body length 10-12 mm; wing length 8-10 mm. Head
pale yellow, dark spot behind antennae, almost square in dorsal profile,

CYLINDROTOMINAE OF NortH AMERICA 87

slight indentation behind eyes; no pubescence on head, several light hairs
posteriorly; palpi, labium, and labrum medium to dark brown, membranous
areas lighter; clypeus medium brown, irregularly fringed with hairs, nar-
row laterally (slightly broader than in tarsalis), defined by deep, horizontal
fronto-clypeal suture; antennal bases raised anteriorly, close together, narrow
furrow between them widening and fading out behind scapes; scape short,
brown, lighter apically, slightly longer medially but lacking distinct tooth-
like projection; pedicel slightly broader than scape; flagellum dark brown
with cylindrical segments densely covered with erect hairs as long as width
of segments proper; 2 to 4 longer darker hairs on each segment, those more
dorsal slightly longer (Fig. 29); eyes not protruding but accented by slight
indentation behind; ommatidia small (smaller than in tarsalis), black; first
cervical sclerite small, narrow; second cervical large, squarish, dark brown.

Pronotum pale yellow to brown, anterior edge slightly curved, broader
and higher medially creating depression behind; 2 more or less fused basal
spots on postscutellum; halteres dark on knobs, stems lighter, fringed with
hairs; coxae pale, inner distal edge black; femora, tibiae brown; tarsi brown,
apical segments slightly darker. Mi+2 branching near distal edge of discal
cell.

Terminalia of male: Ninth tergum broad, slightly produced laterally into
broad conical lobes (Fig. 13); dististyles heavily sclerotized, haired, broad,
with denser tufts of hairs basally, narrower ridged shaft widening anteriorly
into round concave head (Fig. 9); aedaegus a flattened, slightly curved blade,
divided distally into 3 short tubes of equal length, basally supported by a
ventral sclerite and a sclerite on each side (Fig. 1); paraphyses small, slender
sclerites lying transversely in membrane between aedeagus and basistyles.

FEMALE: As for male, except for following characteristics: thoracic stripes
never attaining dark brown-black characteristic of many males; flagellar seg-
ments (Fig. 30) shorter than in male, covered with shorter hairs, several
longer hairs dorsally. Terminalia (Figs. 21, 22): Cerci broad flat blades bear-
ing two dorsal cutting edges, separated by narrow sclerotized trough, outer
edge smooth, inner edge serrate; ventral edges of cerci sinuous; tenth ster-
num 2 lightly sclerotized plates with hairs, lying in membrane ventral to
cerci; hypovalves membranous and narrowly cleft at tips; 3 identical, spheri-
cal, heavily sclerotized spermathecae.

Variation: Variation in amount and distribution of pigments has been
the sole basis for differentiating the western group pallescens from splendens
and from americana as noted above. Among the specimens which I have
studied, three from Nova Scotia (Baddeck, 6 July 1928, D. M. Bates) in
addition to 19 from Colorado (Kenosha Pass, 20 June-10 July, 1919) were
mottled light orange on the thorax, with the longitudinal stripes barely dis-
tinguishable from the background, thus fitting the description of pallescens.

88 Tue University SciENcE BULLETIN

The eastern specimens, with the exception of the paler ones from Nova
Scotia, had distinct dark stripes (but females never attained the full black
of some of the males), and the western specimens from Washington and
Oregon varied from almost white to very black. Occasionally fully pigmented
specimens of both the eastern and western populations show the yellow divid-
ing line down the middle of the thorax. A series from Klamath Falls, Oregon,
23 June 1959 (G. W. Byers), was intermediate in coloration, containing most-
ly pale brown specimens and several darker males. Strikingly similar variation
in pigmentation has been studied in other insects (Ross, 1956), and experi-
ments have shown that varying humidity and temperature caused the pro-
duction of different amounts of carotinoids and melanins. Apparently this is
a basic physiological reaction, the effects of which are masked in many
insects by other pigments or by structural colors.

The wings also exhibit much variation (Figs. 45-48). Traces of Sc2 are
variable, Mi+2 may divide before or beyond the distal edge of the discal
cell, and parts of the media or radius may be weak or almost obliterated.
Short spurs are often present, particularly on branches of the media, and one
specimen was found with an extra crossvein in cell Rs, creating another
closed cell, as in the fossil genus Cyttaromyiza.

Larva: Living specimens are light chlorophyll green with 2 narrow,
pale brown longitudinal lines on dorsum, extending from the posterior end,
above spiracles anteriorly, becoming more expanded and diffuse on the
fore part of the body. Preserved specimens are brown to light yellow, the
longitudinal lines remaining visible. Dorsal processes are simple, in a
single median row on abdominal segments three to seven.

Pupa: The live pupa is green. Preserved specimens are pale yellowish. The
small mesonotal breathing horns are directed slightly forwards.

Detailed descriptions of larvae and pupae were given by Cameron (1918),
Alexander (1920) and Peus (1952).

Ecology: The life history of this species is well-known, having been de-
scribed in detail by Cameron (1918) and summarized by Alexander (1920).
The females have a unique ovipositor equipped with two sawlike blades,
enabling them to make incisions in the upper epidermis of leaves of small
land plants upon which their larvae will later feed. In the western range of
this species, the larvae have been found in rich woodlands, feeding on
Allium, Anemone, Trautvetteria, Stellaria, and Viola. The larvae cease feed-
ing after one molt, hibernate among fallen dead leaves, and finish feeding and
molting in the following spring. Females emerge in large numbers before
males are seen. Dr. George W. Byers observed near Anchorage, Alaska, on
28 June and 3 July, that this species was locally very abundant in cow par-
snip (Heracleum) thickets in deep shade and that the adults rest on the

CYLINDROTOMINAE OF NortH AMERICA 89

upper surface of the leaves or hang from the edge with wings outspread.
They were also taken in Klamath County, Oregon, in a swale where Smila-
cina sessilifolia was a common plant.

Cylindrotoma distinctissima americana occurs in northern regions at
relatively high elevations. In Colorado, the southernmost tip of its known
range, it has been collected at 10,000 ft.; in Oregon and Washington at
3,000-4,000 ft.; up to 7,000 ft. in Montana and British Columbia; 5,000 ft. in
New eemresiiic? and almost at sea level in Unalaska, Alaska, and Nova
Scotia.

Flight records: 6 May-31 Aug.

Distribution (Fig. 77): ALASKA: Admiralty Island, 23 June-26 Aug.; Cascade (15 mi.
SE of Anchorage), 28 June -3 July; Cold Bay (163° W 55° N, on tundra), 26-27 July; Igiak
Bay, 17 July; Kenai Peninsula (22 mi. N of Seward), 1 July; Unalaska, 24 Aug.; Virgin Bay,
(Prince William Sound), 26 June. ALBERTA: 39 mi. SSE of Valley View, 10 July; Waterton
Lakes National Park, 23-24 June. BRITISH COLUMBIA: Vancouver District, Squamish Dia-
mond Head Trail (3200-4000 ft.), 5-19 Aug.; Caribou District, Lillooet, Mt. McLean (7200

ft.), 12 June; Kootenay District, Mt. Revelstoke, 26-30 Aug.; Nanaimo District, Vancouver Is.:
Forbidden Plateau, Victoria, Westholme, 10 May-18 Aug. COLORADO: Grand Co., Ute

Fic. 77. Range of Cylindrotoma distinctissima americana. Each spot represents one or more
collections within a county or at a locality.

90 Tue University ScIENCE BULLETIN

Pass (Williams Fork Range, 10,500 ft.), 16 July; Gunnison Co., Gothic (9500-10,000 ft.), 16
July; Park Co., Kenosha Pass (10,000 ft.), 20 June-10 July. LABRADOR: Nutak, 26 July.
MAINE: Piscataquis Co., Capens, 11 July. NEW BRUNSWICK: (no further data). NEW-
FOUNDLAND: Bonavista North and South Districts, Terra Nova National Park, 8 July. NEW
HAMPSHIRE: Coos Co., Bretton Woods, Glen House, Mount Washington, Lake of the Clouds
(5000 ft.), Tuckerman’s Ravine (5400 ft.), White Mts., 28 June-31 Aug. NEW YORK: Essex
Co., Mt. Marcy, 21-28 July; Fulton Co., Gloversville, 17 June. NOVA SCOTIA: Victoria Co.,
Baddeck, 6 July. ONTARIO: Parry Sound Co., Burks Falls, Sand Lake, 5-14 Aug. OREGON:
Clackamas Co., Stull Creek, 16 July; Clatsop Co., Saddle Mt. (Boyer Station), 6 May-18 June;
Hood River Co., Cascades (Mt. Hood, Hood River Meadows, 4800 ft.), 17 July-8 Aug.; Klamath
Co., SE edge of Crater Lake National Park (4350 ft.), Crater Lake (6200 ft.), Odell Lake (4760
ft.), 23 Junme-2 Aug.; Linn Co., Big Meadows, North Santiam, 3 Aug.; Morrow Co., Castle
Rock, Sourgrass Creek, 13-17 May; Umatilla Co., Blue Mt., Spring Creek (3900 ft.), Little
Antone Creek (4100 ft.), Little Phillips Creek (3800 ft.), 5 June-3 July; Wallowa Co., Wallowa
Creek (4635 ft.), above Lazy T Ranch (5500 ft.), Lostine Valley Guard Station (4900 ft.), 29
June-17 July; Yamhill Co., High Heaven, 4 May. PENNSYLVANIA: (no further data).
QUEBEC: Bonaventure Co., Escuminac, 2-31 July; Indian House Lake (56° 08’ N Lat., 64°
44’ W Long.), 18 July. WASHINGTON: Pierce Co., Cayuse Pass (Mt. Rainier, 4650 ft.),
Nisqually Glacier Trail (4000 ft.), Wonderland Trail (above White River, 4900-5000 ft.), 23
July-1 Aug.; Pacific Co., Ilwaco, 5 May. WYOMING: Lincoln Co., Jenny Lake (6780 ft.),
Leigh Lake (6870 ft.), 3-12 July; Yellowstone National Park (Northeast Entrance 7200 ft.),
Obsidian Creek (7300 ft.), 24-27 June.

Types: C. americana Osten Sacken, two male syntypes (one with broken
abdomen), White Mountains, New Hampshire, end of June, 1836 (no.
10237, Museum of Comparative Zoology, Harvard University).

C. splendens Doane, three male syntypes, Unalaska, Alaska, 24 Aug.
1897 (Kincaid) (no. 7,051, U.S. National Museum).

C. juncta Coquillet, female holotype, Virgin Bay, Prince William Sound,
Alaska, 26 June (no. 5,204, U.S. National Museum).

C. pallescens Alexander, female holotype, Ute Pass, Williams Fork Range,
Grand Co., Colorado, 10,500 ft., 16 July 1929 (C. F. Clagg), (C. P. Alexander
Collection, Amherst, Massachusetts).

Cylindrotoma tarsalis Johnson
(Figs. 25 145235285 S495 50), 7/8)
Cylindrotoma tarsalis Johnson, 1912: 2, fig. 4 (wing); Alexander, 1919c: 963, plate 30, fig.

7 (wing); 1927a: plate I, fig. 1 (wing); 1942: 294, fig. 8 (antenna), fig. G (wing).
Cylindrotoma ? anomala Johnson, 1912: 2-3, fig. 3 (wing).

Diagnosis: In the male the aedeagus is strongly curved, and its median
tube is greatly prolonged. The female has a serrated dorsal edge on each
cercus, lateral to which lie anterior and posterior vestiges of a smooth edge.
The second and third tarsal segments are abruptly lighter than the other
leg segments.

Description: Mate: Body length 8-9 mm; wing length 7-9 mm. Head
dark brown dorsally, lighter ventrally and laterally, grayish pollinose with
fine light hairs posteriorly, broadly oval in dorsal profile, no indentation be-
hind eyes; palpi, labium, labrum medium brown, membranous areas lighter;
clypeus light brown, irregularly fringed with. hairs, delineated by arched

CyYLINDROTOMINAE OF NortH AMERICA 9]

frontoclypeal suture; clypeus narrow laterally (narrower than in C. d.
americana); antennal bases raised slightly and somewhat further apart than
in americana, separated by narrow dark brown sclerotized strip; scape and
pedicel of equal length, pale yellow or brown; broad tooth on scape; flagellum
(Fig. 31) brown, cylindrical segments densely covered with erect hairs the
length of diameter of segments, 2 to 4 longer hairs dorsally on flagellar seg-
ments, none ventrally; head oval in dorsal aspect, with eyes forming round-
ed ends ommatidia larger than in americana, appearing grayish white within
black network; first cervical sclerite small, narrow; second large, approxi-
mately square.

Pronotum hairless, pale to dark brown, anterior edge slightly curved,
broader and higher medially, creating depression behind; pale brown
spot on postscutellum; halteres dark on knobs, stems lighter and fringed with
hairs; coxae pale, inner distal edge black; femora, tibiae brown; first, fourth,
fifth tarsal segments brown, second and third abruptly lighter. Mi +2 branch-
ing close to distal edge of discal cell.

Terminalia of male: Ninth tergum with 2 narrow lobes (Fig. 14); dis-
tistyles sclerotized, simple structures narrowing and curving distally, fewer
hairs than in americana; median tube of aedeagus much prolonged, curved
dorsally; baso-lateral portions of aedeagus much produced on either side;
paraphyses large sclerotized “C-shaped” structures on either side of aede-
agus (Fig. 2).

FEMALE: Only 2 females seen. Similar to male except for usual sexual
dimorphism. Head slightly indented behind eyes; flagellar segments shorter
than in male, first 2 segments broadly joined, others showing small basal
“neck” region; segments clothed with short hairs, less dense than in male,
appressed rather than erect; several longer, darker hairs on dorsal surface of
segments. Terminalia (Fig. 23): Cerci broad, flat; serrate dorsal edge on
each cercus lateral to which lie anterior and posterior vestiges of smooth blade
joined basally, entire central portion of blade apparently never developed;
ventral edges of cerci straight; hypovalves membranous at tips, cleft some-
what more deeply than in americana; (tenth sternum and spermathecae not
observed due to insufficient material).

Variation: The same differences in the amount and quality of pigmen-
tation exist here (at least in the males) as in C. d. americana. Occasionally
the crossvein r-m is obliterated.

Immature stages: Not seen.

Ecology: These flies apparently frequent the same sort of habitats as
C. d. americana. They are found at high altitudes, in Hudsonian and high
Canadian zones, in small, widely scattered populations, flying from May to
September. The adults can be swept from rank vegetation shaded by trees
in close proximity to water. The females have serrated edges on their cerci;

92 Tue University ScrENcE BULLETIN

Fic. 78. Range of Cylindrotoma tarsalis. Each spot represents one or more collections within
a county or ata locality.

thus in this species the eggs are probably laid in incisions of the epidermis of
food plant leaves in much the same manner as by C. d. americana.

Flight records: 30 May-12 Sept.

Distribution (Fig. 78): CONNECTICUT: Litchfield Co., Norfolk, 12 June-12 Sept. NEW
BRUNSWICK: (no further data). NEW YORK: Fulton Co., Bleecker (Woodsworth’s Lake,
1650 ft.), Gloversville, 9 June-19 Aug.; Herkimer Co., Indian Castle, 13 June. PENNSYL-
VANIA: Lycoming Co., 30 May. VERMONT: Lamoille Co., Stowe, 15-24 June.

Types: Male holotype, 6 male paratypes, Woodsworth’s Lake, Bleecker,
New York, 19 Aug. 1909, (no, 7480, Museum of Comparative Zoology, Har-
vard University); 7 male paratypes, same data (C. P. Alexander collection,
Amherst, Massachusetts).

C. ? anomala Johnson. Female holotype, Woodsworth’s Lake, Bleecker,
New York, 1 Aug. 1909, (no. 7481, Museum of Comparative Zoology, Har-

vard University).

Phalacrocera Schiner

Phalacrocera Schiner, 1863: 224; Alexander, 1927a: 9-10; 1942: 292-293; Peus, 1952: 64.
Type: Tipula replicata Linnaeus, 1758 (=Limnobia nudicornis Schummel, 1829).

CYLINDROTOMINAE OF NortH AMERICA 93

Schiner established this genus on the basis of the presence of Riy2 and
the sparsely haired flagellar segments. However, neither of these characters is
useful, even when only the Palearctic species are considered. The four North
American species, Phalacrocera occidentalis, tipulina, replicata (=neoxena)
and vancouverensis, form a reasonably homogeneous group, but characters
employed in this paper are probably of limited use on a world-wide basis.

Diagnosis: This genus may be recognized by the light pubescence on a
black or dark brown head and thorax, almost obscuring the 3 mesonotal
stripes, which are only faintly outlined by a thicker pubescence. Either 3 or 4
branches of the media reach the wing margin. The tenth tergum in the fe-
male is not greatly elongated and extends only to the base of the cerci. The
cerci do not have a serrated dorsal edge.

Description: Head smooth, black or dark brown, covered with whitish or
yellowish pubescence; scape longest antennal segment; flagellar segments
cylindrical or subcylindrical, sparsely haired or densely covered with ex-
tremely short erect hairs; prontum haired; pretergites pale yellow to brown;
mesonotum black, pubescent, pubescence thickest along longitudinal lines
delimiting thoracic stripes; pleural sclerites black or dark brown; scutellar
suture usually indistinct; transverse suture on mesepisteri'um; longitudinal
suture on pleurotergite; wings faintly suffused with brown, oval stigma
barely darker; Sci and Sc2 usually present; radius with 2 or 3 branches
reaching wing margin; r-m present; 3 or 4 branches of media reaching
wing margin; if only 3, Ms44 not branching but continuing directly to
wing margin; if 4, pattern as for Cylindrotoma. Tibial spurs large, strongly
curved and divergent, haired, tips bare; tarsal claws larger than those of
Cylindrotoma.

MALE terminalia: Eighth tergum entire, narrow; ninth tergum large,
squarish, notching at distal edge variable; lateral lobes variable; aedeagus
with 3 slender tubes each sheathing an ejaculatory duct.

FEMALE terminalia: Eighth tergum narrow, may be reduced to 2 lateral
sclerites; ninth tergum narrow, not produced laterally, tenth tergum vari-
able but never prolonged and forked over cerci; cerci broad, flat, with no
distinct serrated edge but occasionally several microscopic, irregularly formed
teeth distally.

KEY TO ADULTS OF NORTH AMERICAN SPECIES
OF PHALACROCERA
1. Pronotum with distinct transverse crest in center; Ry 42 present. --....-....... ps

Pronotum almost flat except for shallow horizontal furrow; tip of Ri+2
(ESAT I yao 0] el 6 es te a a alee Rs Spelt ke Aner i Soe ne aS are 3

2. Flagellar segments covered with dense coat of short erect hairs, with a
few longer irregularly placed hairs (Fig. 32); mesonotal suture broad,
SHinyero wine seeeee 2 eee Uwe ls ica tld aa ie eh eee kee replicata

94 Tue University SciENcE BULLETIN

Flagellar segments with verticils, not covered with short dense hairs
(Fig. 38); mesonotal suture narrow, dull and darker than rest of
DIRGEARS a2 ornare Eee es, ee oe ye vancouverensis

3. Three branches of media present (Fig. 55); median aedeagal tube
strongly arched dorsally, lateral tubes parallel and close together (Fig.
13); pronotuim: slabrous. \...2 22-2 ee ee tipulina

Four branches of media present (Fig. 51); 3 aedeagal tubes short,
straight and parallel (Fig. 3); pronotum pollinose. —............. a6 aha occidentalis

Phalacrocera occidentalis Alexander
(iigsys, lies25 55529)
Phalacrocera occidentalis Alexander, 1927a: 10.
Cylindrotoma occidentalis Alexander, 1954: 41-42.

This species is of particular interest because it resembles both Phalacrocera
and Cylindrotoma. Alexander (1954) placed occidentalis in Cylindrotoma
because it has four branches of the media, although he later (Alexander, 1965)
listed this species in Phalacrocera. 1 am recognizing Phalacrocera occidentalis
because this species seems to resemble the genus Phalacrocera more closely
in the following respects: The thorax is blackish with a grayish-yellow pubes-
cence. (Cylindrotoma species have three distinct black prescutal stripes on a
pale thorax.) The mesonotal suture does not flank the median stripe. The
longest antennal segment is the scape, and the whorl of hairs on the basal
enlargement of the flagellar segments most closely resembles that of P.
tipulina. The pronotum is flat like that of P. tpulina, lacking the median
crest of Cylindrotoma. The cerci of the female are simple rounded blades,
lacking the serrated edge found in Cylindrotoma, and the ninth tergum is
not prolonged nor divaricate distally. Lastly, the tibial spurs are long and
slender, characteristic of Phalacrocera. Except for characters of the male
genitalia, this species most closely resembles P. tipulina.

Diagnosis: This species is black with grayish-yellow pubescence on the
head and thorax. The pronotum is broad, flat and pollinose. Four branches of
the media reach the wing margin. The semen pump and its apodemes and
paraphyses are extremely large in comparison to the three very small tubes
of the aedeagus.

Description: Mate: Body length 12 mm; wing length 8-10 mm. Head
dorsally black with whitish pubescence, ventrally dark brown glabrous; dor-
sal profile subtriangular, no indentation behind eyes, eyes confluent with
head contour; palpi, labial segments dusky brown, haired, membranous re-
gion lighter brown; labrum shiny brown; clypeus dusky brown, pubescent,
broad and flat, bearded distally, lateral portions long, at least two-thirds
length of anterior region; frontoclypeal suture deep, horizontal; antennae
about 5 mm, dusky brown, bases not raised anteriorly; scape longest, broad-
est segment, lateral teeth narrow; pedicel globular; flagellar segments sub-

CYLINDROTOMINAE OF NortH AMERICA 95

cylindrical, covered with extremely short, erect hairs, whorl of longer hairs
on enlargement near base of each segment (Fig. 32); first cervical sclerite
not seen (insufficient material), second large, black, convex.

Pronotum almost flat, dark brown with yellowish pubescence, short erect
hairs anteriorly, shallow furrow bisecting it horizontally; pretergite pale
yellow to brown. Mesonotum black with yellowish pubescence; 3 stripes
indicated vaguely by denser pubescence in longitudinal furrows flanking
region of stripes; narrow line of lighter pubescence longitudinally bisecting
mesonotum; mesonotal suture small, almost straight, glabrous brown,
scutum black, pubescent, pubescence whiter at scutal suture; scutellum very
pale, pubescent, with 2 shiny brown pits anterolaterally; postscutellum with
2 black basal areas, whitish pubescence medially; pleurites light brown to
black, pubescent, membranes dusky; pleurotergal suture and suture between
anepisternum and preepisternum lacking.

Coxae dark brown, pubescent; trochanters lighter, edged with black dis-
tally; femora, tibiae, tarsal segments brown, darker distally; tibial spurs
long, diverging, clothed with short brown appressed hairs, darker and
bare distally; last tarsal segment cylindrical, claws long, slender and broadly
curved, only slightly widened at base. Knobs of halteres dusky black, stems
lighter, edged with hairs.

Wings (Figs. 51, 52) pale brown, stigma slightly darker, oval; Sci fading
out just before stigma, Scza and Sczr absent or merely suggested by presence
of pale lines in wing; 2 branches of radius (R3 and Ri+5) reaching wing
margin, (a very small portion of Ri+2 found on 1 specimen); r-m in line
with radial crossvein, joining media near proximal corner of discal cell;
4 branches of media reaching wing margin; Mi+2 branching at or just be-
yond distal edge of discal cell, Ms +4 branching at posterior distal edge of dis-
cal cell; short medial crossvein plus section of M3 form distal edge of
discal cell.

Abdomen dark brown to black, posterior segments bordered anteriorly by
a colorless band. Terminalia of male (Figs. 3, 17): Ninth tergum with
rounded notch medially on distal edge; basistyles without lateral projections;
dististyles simple narrow structures, slightly widened at base, yellowish, black-
ened distally, covered with short fine hairs; aedeagus with 3 short, parallel
tubes, large, broad semen pump, but no basal sclerites; paraphyses large,
“mitten-shaped” (Fig. 3).

FEMALE: Differs from male in the following characters: antennae shorter
(255 mm), flagellar segments shorter (Fig. 33), cylindrical, without basal
enlargements, clothed with erect hairs slightly longer than in male, whorl
of longer hairs basally. Terminalia: (Character of ninth tergum not ob-
served on the one dried specimen); tenth tergum a large sclerite with 3 distal
projections, two small triangular sclerites in membrane between projections;

96 Tue University ScIENcCE BULLETIN

cerci short, broad, haired structures, rounded distally; hypovalves weakly
sclerotized, incompletely fused to ninth sternum, narrowly notched and mem-
branous distally.

This description is based on only two specimens, a male and a female,
borrowed from Dr. C. P. Alexander, Amherst, Massachusetts.

Immature stages: Not seen.

Ecology: This species has been found in highest, humid Transition or
low Canadian life zones in Washington State at 2,800 ft., and at 4,475 ft. in
Oregon. Individuals were found in high alpine meadows overgrown with
grasses and sedges in the vicinity of a stream. The characteristic plants in this
region were clumps of scrubby willow, Veratrum viride, Habenaria, Caltha
leptosepala, Pedicularis groenlandica surrecta, Dodecatheon and alpine fir,
mountain hemlock and pines along streams (see Alexander, 1949).

Flight records: 10 June-8 Aug.

Distribution (Fig. 79): OREGON: Hood River Co., Cascades, (Hood River Meadows, Mt.
Hood, 4475 ft.), 17 July-8 Aug. WASHINGTON: Pierce Co., Longmire Springs (Mt. Rainier,
2800 ft.), 10 June, 17 July.

Type: Male holotype, Longmire Springs, Mt. Rainer, Washington, 10
June 1917 (H. G. Dyar), (C. P. Alexander collection, Amherst Massachu-
setts).

Phalacrocera replicata (Linnaeus)
(Figs. 4, 16, 25, 37, 53, 79)

Tipula replicata Linnaeus, 1758: 587.°

Limnobia nudicornis Schummel, 1829: 122.

Tipula brevirostris Zetterstedt, 1838: 844.

Phalacrocera replicata; Schiner, 1863: 224; Grinberg, 1910: 33-35, fig. 33 (antenna), fig. 34
(male terminalia), fig. 35 (female terminalia), fig. 36 (wing); Peus, 1952: 1-77, fig. 2
(head), fig. 4 (thorax), fig. 10 (wings), fig. 17 (tarsal claws), fig. 18 (female terminalia),
fig. 23, 29 (male terminalia), fig. 33, 36, 37a, 39, 40, 41, 42, 43, 47, 50 (larvae), fig. 52,
56a, 57a (pupae), fig. 61 (eggs).

Phalacrocera neoxena Alexander 1914a: 603-604, pl. 25, fig. 10 (wing); 1919: 30, fig. 9 (wing);
1927a: pl. 2, fig. 6 (wing); 1942: 293, 295, fig. 33, E (wing).

A comparison of American specimens of neoxena with five Swedish
specimens of replicata indicates that only one species is involved. Alexander
(1914a) noted a similarity between these two groups but also pointed out
several differences such as color, variations in the radial-medial crossvein and
differences in the male genitalia. However, my material afforded detailed
comparisons, including comparable preparations of the genitalia of both
sexes, and I found approximately the same range of variation in both popu-
lations, albeit my European sample was considerably smaller and limited in
geographic range.

Diagnosis: This is one of the more robust species in the subfamily. It
is black with grayish pubescence on the head and whitish yellow pubescence

° Only the basic European references to this species are given.

CYLINDROTOMINAE OF NortH AMERICA 97

Fic. 79. Ranges of Phalacrocera occidental:s (triangles) and Phalacrocera replicata (circles).
Each spot represents one or more collections within a county or at a locality.

on the thorax. The pronotum is pollinose with dark crest in the center, ac-
cented by a furrow behind and flanked by 2 dark spots on either side. Three
branches of the media and 3 of the radius reach the wing margin. The 3 ae-
deagal tubes are long and slender in comparison to the semen pump and
its apodemes. The flagellar segments have several long, scattered hairs in
addition to a dense covering of very short hairs. The mesonotal suture is
broad and shiny brown.

Description: MALE: Body length 12-15 mm; wing length 10-11 mm. Head
black, grayish pubescence particularly evident laterally and bordering eyes,
several light hairs posteriorly; dorsal profile broadly rectangular with short,
narrow “neck”, slight indentation behind eyes; labium, labrum, and palpi
medium brown, clypeus brown, broadest and most densely haired distally,
slight horizontal distal groove giving impression of broad rim; lateral ex-
pansion of clypeus somewhat less than two-thirds its length; frontoclypeal
suture strongly arched, obscured by coloration; antennae (Fig. 37) about
4.0 mm, dark brown, bases not raised but separated by dark narrow groove;
scape subcylindrical, slightly widened apically, about length of first few
flagellar segments, lateral teeth broadly triangular; pedicel globular; flagellar
segments subcylindrical, each gently tapered proximally, truncate apically,
covered with very short, dense, erect hairs with several sparse, irregularly

98 Tue University ScrENcCE BULLETIN

placed longer hairs; first cervical sclerite brown, small, squarish; second
cervical sclerite larger, rectangular, anterior corner narrowly prolonged.

Pronotum broad, almost flat, dark brown with grayish pubescence; short
row of erect hairs anteriorly, raised crest in center accentuated by furrow
behind, flanked by dark spot on either side; pretergite pale yellow. Meso-
notum black, smooth, with whitish yellow pubescence; pubescence thicker
along 2 longitudinal furrows, lighter along narrow mid-line; mesonotal su-
ture curved, polished brown, extending to longitudinal furrows; scutum
black, pubescent; scutal suture obliterated basally; scutellum brown, pubes-
cent, finely pitted, with 2 larger pits anteriorly; postscutellum light yellow,
smooth, with narrow ridge along midline; pleurites with minute white
hairs, no discernible suture between the anepisternum and postepisternum;
pleurotergal suture also lacking.

Coxae, trochanters pale; femora, tibiae, tarsi uniformly darker brown,
heavily haired; tibial spurs stout, sharply pointed, covered with appressed
hairs, tips bare and black; tarsal claws sharply curved, brown with black tips.
Halteres with light brown knobs, stems lighter, edged with hairs.

Wings (Figs. 53, 54) strongly suffused with brown, stigma prominent,
oval; Sci fading out before reaching stigma; Sci;Scza usually reaching ra-
dius, presence of Scr variable; 3 branches of radius reaching margin: Ri+2,
Rs and R445; r-m present, or occasionally almost obliterated; 3 branches of
media reaching margin: Mi+2 branching just beyond discal cell, therefore
cell Mi: petiolate, single crossvein closing cell distally; M3+4 not branching,
curving slightly at margin.

Abdomen brown, with darker median longitudinal streak on dorsum and
venter. Terminalia of male (Figs. 4, 16): Eighth tergum entire; ninth tergum
with round medial notch; basistyles with narrow fingerlike projections la-
terally; dististyles mostly hairy, bare at tips, longitudinal central ridge
broader distally, small sclerotized lobe and notch on outer edge, broad con-
cavity basally, aedeagus with three long slender tubes, each enclosing ejacula-
tory duct, tubes narrowed at tips, close together, outer pair more curved and
slightly shorter than median tube; 1 small basal sclerite enveloping semen
pump region of aedeagus; paraphyses (not shown in figure) long, curved,
arms attached to inner sclerotized edges of basistyles; fleshy anal segment
often visible in dried specimens.

FEMALE: As in male except for following characters: antennae shorter
(25mm), dark brown, scape cylindrical; flagellar segments small, sub-
cylindrical; covered with minute hairs, several longer hairs on each
segment, shorter, stiffer ones ventrally. Terminalia (Fig. 25): Ninth tergum
broad; tenth tergum large with 3 distal extensions plus a small sclerite be-
tween cercal bases; cerci broad, entire; hypovalves deeply notched distally,
fused with broader eighth sternum, thin projection distally on dorsal edge.

CyYLINDROTOMINAE OF NortH AMERICA 99

Larva: Distinctly greenish when alive. “Body covered with numerous,
elongate, trachea-bearing filaments, the posterior pair on dorsal segments
deeply forked, the others simple. Spiracular disk with dorsal pair of lobes
formed by rudimentary posterior branch of branched filaments of eighth
abdominal segments” (Alexander, 1927a).

Pupa: “| Mesonotal| breathing horns long, almost straight. Dorsal abdo-
minal segments with tubercles, those of sixth and eighth segments enlarged
into spinous hooks; 2 pointed tubercles on seventh sternite” (Alexander,
1927a).

More complete descriptions of larvae and pupae are given by Alexander
(1920).

Ecology: Rogers (1942) reports this species is “locally common to
abundant about the moss and Myriophyllum choked pools of unshaded
seepage marshes”. They are often taken around the “margins of grass-sedge-
fern marshes” but never found in association with Sphagnum.

Flight records: 12 May-30 June.

Distribution (Fig. 79): MASSACHUSETTS: Berkshire Co., Mt. Greylock, 15 June. MICHI-
GAN: Livingston Co., E. S. George Reserve, Honey Creek, Putnam Twp., 12-20 May; Oscoda
Co., 18 June. NEW YORK: Cayuga Co., North Fairhaven, 17 May. ONTARIO: Thunder Bay
Co., Nipigon, 17 June; Cochrane Co., Fort Albany (James Bay), 10 June. PENNSYLVANIA:
Luzerne Co., Hazleton, 30 June. QUEBEC: Huntingdon Co., Hemmingford, 23-27 June.

Types: P. neoxena Alexander. Male holotype, Nipigon, Thunder Bay
County, Ontario, 17 June 1913 (Dr. E. M. Walker). Female allotype and
one male paratype no. 1 same data as holotype; no. 2, male, type locality, 18
June 1913; no. 3, female, North Fairhaven, Cayuga Co., N. Y., 17 May 1913.
(Type and paratype no. 1 in University of Toronto collection; paratype no.
2 in C. P. Alexander collection, Amherst, Massachusetts.)

Phalacrocera tipulina Osten Sacken
(Figs. 5, 15, 24, 34-36, 43, 55, 56, 67-69, 80)

Phalacrocera tipulina Osten Sacken, 1865: 241-242; Needham, 1908: pl. 11, fig. 2 (wing);
Alexander, 1914a: pl. 25, fig. 11 (wing); 1919: pl. 30, fig. 8 (wing), text fig. 125 J
(male antenna); 1920: 961; 1942: 293; Dickinson, 1932: 210, hg. 112 (wing).

Diagnosis: Vhis species is dark brown to black with a tawny pubescence.
The pronotum is broad, flat and glabrous. Two branches of the radius and 3
branches of the media reach the wing margin. The 3 long, slender aedeagal
tubes are readily visible, the median tube arched over the parallel lateral tubes,
with the tips of all 3 converging.

Description: Mae: Body length 10-14 mm; wing length 8-11 mm. Head
dorsally dark brown to black, with tawny pubescence, several light hairs
posteriorly; dorsal profile triangular, narrow “neck” with 2 or 3 encircling
ridges, no indentation behind eyes; palpi, labial segments dark brown, hairy;
labrum and surrounding membranes orange-brown, clypeus darker brown,
relatively large, rectangular, distally bearded, giving appearance of short

100 Tue University SciENcE BULLETIN

nasus, lateral expanse about half of frontal length; fronto-clypeal suture hori-
zontal, groove-like; antennae 3.5-4.5 mm, dusky brown, bases not raised
(or only slightly so); scape longest segment, cylindrical, irregularly haired,
often secondarily grooved or ringed, lateral teeth narrow (Fig. 34); pedicel
globular, pubescent, with several long scattered hairs; flagellum (Fig. 35)
slightly narrower than scape, segments sub-cylindrical, covered with short
dense hairs, whorl of longer hairs on slight proximal enlargement, last seg-
ment attenuated medially, 4 hairs at tip; first cervical sclerite brown, small,
narrow; second cervical sclerite large, rectangular, anterior corner medially
prolonged.

Pronotum broad, flat, narrowed anteriorly, shiny brown except for nar-
row band of pubescence anteriorly before shallow groove; more distinct
groove before paler pretergites. Mesonotum black with yellowish pubescence,
pubescence thickest on longitudinal furrows, narrow bare line often dividing
mesonotum longitudinally; mesonotal suture a polished brown shallow
curve widening to shiny brown pits at longitudinal furrows; scutum black,
pubescent; scutal suture indistinct or absent; scutellum brown, pubescent,
with 2 pits anteriorly; postscutellum smooth, brown, whitish pubescence
medially; pleurites brown, thickly pubescent, suture present between anepis-
ternum and postepisternum; pleurotergal suture lacking.

Coxae, trochanters light brown; femora, tibiae, tarsi darker, thickest
and darkest apically, covered with hairs; tibial spurs long, diverging, with
short appressed hairs, bare at tips; last tarsal segment concave ventrally (Fig.
43); tarsal claws almost length of preceding segment, slender, strongly
curved, several short teeth basally. Halteres brown at knobs, stems lighter,
edged with hairs.

Wings (Figs. 55, 56) strongly suffused with brown, stigma darker, oval;
Sci fading out before stigma, Scza present, Scez variable; 2 branches of
radius reaching costa, Rs and R445; r-m always present, usually touching
radial crossvein and joining media close to proximal end of discal cell; 3
branches of media reaching wing margin: Mi+2 forming rectangular edge
of discall cell, cell Mi usually petiolate; Ms+4 not dividing, curving slightly
at margin.

Abdomen dark brown, segments edged distally by darker band next to
outer colorless band. Terminalia of male (Figs. 5, 15): Eighth tergum entire;
ninth tergum rectangular, with narrow medial notch (smaller than in repli-
cata); basistyles without lateral projections; dististyles haired at least basally,
bare at tips, slender, curved and tapering distally; aedeagus with 3 long
narrow tapered tubes, each enclosing ejaculatory duct, median tube arched
over parallel lateral tubes, sclerotized bifurcate projection at juncture of
three tubes; large basal sclerite enveloping basal half of aedeagus; paraphyses
(Fig. 5), triangular plates dorsal to aedeagus.

CYLINDROTOMINAE OF NortH AMERICA 101

FEMALE: As in male except for following characters: antennae (Fig. 36)
considerably shorter than in male (2-2.5 mm), flagellar segments shorter,
basal enlargement more pronounced, short hairs less dense; tarsal claws
shorter, broader basally not strongly curved, lacking teeth. Abdomen broader
than in male. Terminalia (Fig. 24): Ninth tergum a narrow sclerotized ring;
tenth tergum of 4 sections: large sclerite with 3 short distal extensions, 2 small
triangular sclerites in membrane between extensions, small narrow sclerite
between and just anterior to cercal bases; cerci short, broad, with rounded
caudal edges, smooth lateral edges; hypovalves weakly sclerotized, deeply
notched at tip, thin projection distally on dorsal edge.

Variation: The tip of Ri+2 is occasionally present on one or both wings,
confusing this species with P. replicata. A pair of slender horny outgrowths
of the pronotum were found on two specimens of approximately 300 studied.
Another specimen had only one horn with no indication of its mate on the
other side of the pronotum. This is probably nothing more than an anomaly.

Larva: Not seen.

Pupa (Figs. 67-69): Only pupal skins were available for study. Specimens
were light brown, wing pads and leg sheaths darker, darker brown stripe
on dorsal mid-line, pair of small spots laterally on venter of each segment.

Length 13.5 mm; width 3 mm. Cephalic part of head broad, without spines
or lobes. Prothorax lacking processes. Mesothorax broad, irregular, with
a slender pair of curved breathing horns anteriorly, devoid of other processes.
Broad wing sheaths extending down two-thirds of second abdominal segment.
Metanotum devoid of spines; haltere sheaths lying alongside first abdominal
segment, partially covered by wing sheaths. Leg sheaths extending almost
to end of third abdominal sternum. First abdominal segment short, without
processes. Abdominal segments 2 to 8 as follows: 4 pairs of spots dorsally,
posterior 2 bearing short hair on small prominence; ventrally (Fig. 68) with 2
oval spots antero-laterally, 3 pairs of tiny spots medially, a thin transverse
division behind which lies a roundish spot on each side of mid-line, and more
posteriorly, a pair of small spines on each side; laterally produced into a
broadly bi-lobed flange, a pigmented spot at each tip. Eighth segment with
a broad lateral flange, a small pigmented spot at broadest portion on each
side; 4 small spines ventrally, middle 2 larger; posteriorly, 2 rounded pro-
cesses, each tipped with a small hair. Ninth segment terminating in several
lobes: dorsal pair of lobes longest, 2 rows of spines dorsally; 6 ventral lobes,
middle lobe smallest; deeply bilobed lateral process, outermost lobes broader,
tipped with 3 small hairs.

This pupa is quite similar to that of P. replicata, differing only in the
absence of a pair of terminal processes on the dorsum of the sixth abdominal
segment and on the seventh abdominal sternum.

Ecology: This species is not as difficult to find as are most Cylindroto-
minae. It occurs in or near sphagnum bogs in northern or mountainous locali-

102 Tue University SCIENCE BULLETIN

Fic. 80. Ranges of Phalacrocera vancouverensis (inverted triangles) and Phalacrocera tipulina
(circles). Each spot represents one or more collections with:n a county or at a locality.

ties. Good series have been taken in Michigan (J. S. Rogers) and Pennsyl-
vania (G. W. Byers). Byers recorded in his field notes (Centre County, Penn-
sylvania) taking these flies “in low-hanging branches of alder and hem-
lock along the edges of a marsh and in clumps of Carex. Several mating pairs
were collected. The female supports or suspends the male or both may cling
to adjacent leaves of Carex for support. The flies hang up with wings out-
spread and may let hind legs hang free.” A live male pupa and several male
skins were found in low hummocks of Sphagnum. Wing remains of this

species have been found in the pitcher plant, Sarracenia purpurea (see Alex-
ander, 1920).

Flight records: 9 May-6 Sept.

Distribution (Fig. 80): CONNECTICUT: Windham Co., Putnam, 15 June; (?) Co.,
Manitic Lake, 8-9 June. MAINE: Aroostook Co., Ninemile (near Lac Frontiére, Quebec), 2 June;
Hancock Co., Bar Harbor, Mt. Desert, 4-23 July; Penobscot Co., Bangor, May-July; Washington
Co., Machias, May-July. MARYLAND: Garrett Co., Cranesville Pine Swamp, 18 June. MASSA-
CHUSETTS: Essex Co., Beverly, 3 June. MICHIGAN: Gogebic Co., 15 June-16 Aug.;
Houghton Co., 20 June; Livingston Co., 12 May; Oscoda Co., 19-25 June; Schoolcraft Co.,
Floodwood, 20 July-14 Aug. NEWFOUNDLAND: Avalon Peninsula, Cochrane Pond (450
ft.), Mackinson’s, St. John’s (Karwood Cabins, along shore of Neville’s Pond), Cataracts of
North Harbor River near Colinet, 22-30 June; White Bay District, St. Anthony, 2 July.
NEW HAMPSHIRE: Cheshire Co., Mt. Monadnock area, May-July; Coos Co., White Moun-
tains, May-July; Rockingham Co., Hampton area, May-July. NEW YORK: Essex Co., Ausable
Lakes, Hampton, Lake Tear of the Clouds, Mt. Marcy, 10-30 July; Fulton Co., near
Sacandaga Park, Canada Lake (1600 ft.), 18 June-10 July; Hamilton Co., Lake Pleasant, 20-21
June; Herkimer Co., Old Forge, July-3 Aug.; Suffolk Co., Riverhead, 2 June; Tompkins Co.,

CYLINDROTOMINAE OF NortH AMERICA 103

Ringwood Hollow, 3 July. ONTARIO: Carlton Co., Mer Bleu, Hawthorne, Ottawa, 20 May-6
Sept.; Georgian Bay, 13 July. PENNSYLVANIA: Centre Co., Bear Meadows (1820 ft.), 25-
26 June; Luzerne Co., Hazelton, 19 May-4 June; Sullivan Co., 27 June. QUEBEC: Huntingdon
Co., Covey Hill, Hemmingford, 27 June-13 July; Kamouraska Co., Andreville, 25-31 June;
Laurentides Park, 24 June-16 Aug.; large bog 2 mi. N Lake Mistassini, Rupert River, 10 June-
13 July. VERMONT: Lamouille Co., Stowe, 24 June. VIRGINIA: Giles Co., Hawthorne, Little
Stoney Creek, Mountain Lake, 29 May-21 June. WISCONSIN: Vilas Co., July.

Types: Two syntypes (male and female), White Mountains, New Hamp-

shire. (no. 10238, Museum of Comparative Zoology, Harvard University.)

Phalacrocera vancouverensis Alexander
(Figs. 6, 18, 38, 44, 57, 58, 80)

Phalacrocera vancouverensis Alexander, 1927b: 189-190; Spencer: 1930: 15-16.

Diagnosis: This species is dark brown and robust, with whitish pubescence
on the head and golden pubescence on the thorax. The pronotum has a
prominent transverse crest. Three branches of the radius and 3 of the media
reach the wing margin. The flagellar segments have verticils. The mesonotal
suture is narrow, dull and darker than the rest of the thorax.

Description: Mate: Body length 12-13 mm; wing length 10-12.6 mm.
Head dark brown to black, with whitish pollinosity of short hairs, most
dense ventrally and around eyes, several brown hairs on vertex; dorsal pro-
file of head broadly rectangular, with short narrow “neck” region; labium,
labrum, and palpi brown, membranes yellowish; clypeus large, trapezoidal,
brown-black, with whitish pubescence, several brown hairs distally, lateral
extensions narrow; fronto-clypeal suture broad, horizontal (not as well de-
fined as in tipulina); antennae dark brown, scape longest segment, cylindri-
cal, bearing scattered hairs, lateral tooth obscured on available specimens;
pedicel globular, with scattered hairs; flagellar segments (Fig. 38) subglo-
bular, slightly longer than pedicel, slightly produced ventrally, bare except for
whorl of short dark hairs on each segment; first cervical sclerite brown, small,
narrow; second large, rectangular, its curved edge without narrow anterior
prolongation.

Pronotum dusky black, broad, with prominent transverse crest, bordered
anteriorly with erect hairs; pretergites brown. Mesonotum dull black, with
golden pubescence; longitudinal furrows barely suggested by slightly heavier
pubescence, no median longitudinal line on thorax; mesonotal suture dull
brown, narrow, slightly curved, visible only behind median stripe of thorax;
scutum dark brown; scutal suture distinct; postscutellum and pleurites dull
brown with light grayish pollinosity, membranes dusky beige; suture pre-
sent between anepisternum and postepisternum; pleural suture lacking.

Coxae, trochanters obscure yellow; femora brownish yellow, lighter ba-
sally, tips widened and dark brown; tibiae, basitarsi light brown, tips nar-
rowly darkened; other segments of tarsi dark brown; tibial spurs lighter
brown, black tipped, conspicuous; tarsal claws (Fig. 44) sharply curved, with

104 Tue University SciENcE BULLETIN

small bump near broad base. Knobs of halteres dark brown, stems dusky
yellow, haired.

Wings (Figs. 57, 58) strongly suffused with brown, stigma barely dis-
cernible; Sci fading out before stigma, no evidence of Sc2; 3 branches of
radius reeaching costa: Ri+2, Ra, Ra+5; Ri+2 very short, sharply diverging
from Rs; r-m in line with base of discal cell; 3 branches of media reaching
margin: Mi+2 branching just beyond edge of discal cell, cell Mi petiolate.
M3+4 not branching, slightly curved at margin.

Abdomen dark brown, segments edged distally by narrow colorless bands.
Terminalia (Figs. 6, 18): Character of eighth and ninth terga obscured in
specimen examined (on slide). Dististyles irregularly shaped, with flat round-
ed head, a notch and a small tooth along inner edge, several small teeth along
outer of broad basal region; 3 tubes of aedeagus on same plane, outer two
curved, inner straight, each narrowed distally; basal sclerites “H-shaped,”
lying dorsal to aedeagus; paraphyses long, narrow sclerites attached to in-
ner sclerotized edges of basistyles (not shown in figure). Holotype differs
in the following characters: antennae slightly shorter than in male, scape and
flagellar segments oval to subcylindrical, not produced at all ventrally nor
truncate apically; last segment elongate, nearly twice length of penultimate;
each segment with a whorl of hairs; terminal segment elongate, relatively
long, conspicuous. Wing of holotype possesses r-m crossvein. (Probably this
character is quite variable in vancouverensis.) Abdomen light yellowish
brown, with conspicuous black median stripe on dorsum and venter; first
segment darkened, pruinose; caudal margins of segments narrowly pale.
Terminalia (observations incomplete due to insufficient material): Base of
cerci obscured by rather large, distally pointed hairy tenth tergum. Tips of
cerci subacute, caudal margins roughened, almost serratulate. Hypovalves
deeply notched.

Immature stages: Not seen. The larva is bright green and has bifurcated
filamentous dorsal and lateral processes (Spencer, 1930).

Ecology: Spencer (1930) observed that the larva was green and resem-
bled the stems of the pond weed Nitella, to which it was clinging. Its move-
ments were sluggish, barely perceptible. The long bifurcated dorsal and later-
al “gills” were very slowly waved back and forth. After pupation, the last
laval skin remained loosely attached to the pond weed and the pupa floated
at an angle of 30° to the water surface.

Flight records: 10 April-9 Aug.

Distribution (Fig. 80): BRITISH COLUMBIA: Vancouver Island, Forbidden Plateau, 9
Aug.; Vancouver District., Vancouver, 10 April.

Type: Female holotype, Vancouver, 10 April 1922 (W. B. Anderson),
in (Canadian Nat. Collection). Male allotype (genitalia on slide), For-
bidden Plateau, Vancouver, British Columbia, 9 Aug. 1950; (in Museum of
Zoology, University of Michigan, Ann Arbor).

CYLINDROTOMINAE OF NortH AMERICA 105

Liogma Osten Sacken

Liogma Osten Sacken, 1869: 298; Peus, 1952: 66.

Type: Triogma nodicornis Osten Sacken.

Liogma is a poorly defined genus represented by only one species in North
America. It shows many striking similarities to Trzogma in wing venation
and genitalia, yet superficially these genera are quite dissimilar.

Diagnosis: This genus is distinguished by having strongly nodulose, al-
most cordate antennal segments which are much more pronounced in the
male than in the female. The prescutal stripes are shiny black.

Description: Head shiny black, with fine light hairs laterally and posteri-
orly; scape long, cylindrical; flagellar segments nodulose or cordate; pronotum
bare; pretergites pale yellow; prescutum pale yellow with 3 shiny black
stripes; mesonotal suture a shiny brown curve ending in small pits on either
side of median stripe; scutum divided longitudinally by distinct suture; scu-
tellar sutures obscured in dark brown, spindle-shaped pits; pleural sclerites
pale yellow, brown-black areas on anepisternum, postepisternum; pleuroter-
gite brown-black, wrinkled; wings suffused with brown, stigma oval, very
pale; Sci fading before stigma; Scza present, Scer obliterated; 2 branches of
radius reaching margin (Rs and R445); r-m obliterated by fusion of Ra+5
with Mi+2 at corner of or before discal cell; 3 branches of media reaching
wing margin (Mi+2, Ms, Ms); medial crossvein plus part of Ms closing
discal cell distally; tibial spurs very short, stout, haired; tarsal claws long,
slender, broadly curved. Abdomen dark brown, distal edge of each segment
colorless.

MALE terminalia: Eighth tergum entire; ninth tergum squarish, produced
into 2 conical projections distally; aedeagus with 3 tubes each enclosing eyjac-
ulatory duct leading from semen pump.

FEMALE terminalia: Ninth tergum very narrow; tenth tergum divided
into 4 sclerites: large median sclerite, pair of triangular lateral sclerites, one
narrow anterior sclerite; cerci short, broad, not distinctly serrate.

Liogma nodicornis (Osten Sacken)
(Figs. 7, 19, 26, 39, 40, 59-62, 74, 81)

Triogma nodicornis Osten Sacken, 1865: 239-240.
Cylindrotoma nodicornis Osten Sacken, 1869: 301.
Liogma nodicornis Osten Sacken, 1869: 298; Alexander, 1914b: 105-118; 1927a: 10-11.
Liogma nodicornis flaveola Alexander, 1919b: 195.

Diagnosis: This species is intermediate in length between Cylindro-
toma and Phalacrocera. The thorax is pale yellow with shiny black or dark
brown stripes. The antennal segments are strongly nodulose or cordate.

Description: Mave: Body length 9-12 mm; wing length 8.5-9.5 mm. Head
shiny black, broadly oval, tapering behind eyes; labium, labrum medium
brown, haired distally; palpi darker brown; membranous areas tawny; cly-

106 Tue UNIversity SCIENCE BULLETIN

peus dark brown, short, narrow, convex, with long hairs distally extending
over labial segments; fronto-clypeal suture arched; antennal bases close to-
gether, separated by narrow incision-like groove; scape long, cylindrical,
slightly narrower than flagellum and roughly twice the length of 1 flagellar
segment, bearing several hairs; lateral teeth not pronounced; pedicel sub-
ovate; most flagellar segments expanded ventrally (Fig. 39), distal segments
sub-ovate, last segment attenuated, slender; dense, erect, short hairs covering
each flagellar segment, verticillate, dorsal bristles longer.

Pronotum pale yellow to dark brown in center, anterior region reduced,
prominent narrow transverse crest near mid-length; pretergites pale yellow
with short erect hairs; prescutal stripes shiny black, median stripe divided
posteriorly by short narrow longitudinal furrow; scutellum light brown;
postscutellum shiny brown posteriorly, paler and wrinkled anteriorly. Hal-
teres dusky, lighter at base of stem; thick row of hairs on anterior edge.

Coxae brown, trochanters light yellow, edged with black; femora, tibiae
very pale, brown at tips, covered with short, stout hairs, each in small brown
spot; tarsi dark brown, hairs long.

Terminalia (Figs. 7, 19): Dististyles light brown, slender, widened at
base, covered with short hairs except for bare tips, denser hairs along inner
edges; aedeagus branching into three tubes beyond junction of ejaculatory
ducts, outer tubes slightly more curved than median tube; ventral side of
each tube prolonged distally, tips flattened; basal sclerite apparently fused
to aedeagus and joined to paraphyses by a long narrow extension on each
side; paraphyses irregular sclerites (Fig. 7) lying dorsal to aedeagus; anal
segment lightly sclerotized.

FEMALE: Differs from male in the following: antennae (Fig. 40) some-
what shorter, pedicel and flagellar segments nodulose, not cordate, verticils
shorter than in male. Terminalia (Fig. 26): Large median sclerite of tenth
tergum with three sharp points on caudal margin. Cerci with very small irreg-
ular terminal teeth; hypovalves short, broadly notched distally, fused to
slightly broader eighth sternum.

Variation: Considerable variation has been noted in the extent of color
on the mesothorax. The black mesonotal stripes may merge anteriorly or be
distinct along their entire length. The background color varies from very
pale to quite dusky yellow. Pleural sclerites and postnotum are very pale in
some specimens. Wing variations include the occasional presence of r-m,
occasional absence of Sc2a and m-cu joining M before, beyond or at the
base of the discal cell. One specimen was found with an additional crossvein
in cell Rs (Fig. 62).

Larva: Living specimens are light green. Numerous slender lobes cover-
ing body are darker. Preserved specimens are pale whitish-yellow with dark
oblique marks laterally on abdominal segments. Tuberculate paired processes

CYLINDROTOMINAE OF NortH AMERICA 107

on abdominal terga each bearing a small sharp tooth anteriorly near base.

Pupa: Living pupae are light yellow to brown (probably depending upon
age). Preserved specimens are brownish with an interrupted dark longi-
tudinal line along either side of the middle of the dorsum. Each abdominal
segment bears a single dorsal pair of slender lobes.

Detailed descriptions of the immature stages are presented by Alexander
(1914).

Ecology: The adults of this species are relatively abundant although quite
local in distribution. They are very sluggish flyers and may be swept from
rank vegetation in cool, shaded, or partially shaded woodlands, in tama-
rack and sumac swamps, and along wooded margins of marshes. The Alberta
specimens were collected at the edge of spruce and tamarack “muskeg” (Alex-
ander, 1927c). Immature stages of these flies have been found feeding on
Mnium punctatum, Hypnum cupressiforme and related species of mosses,
and in shallow pools.

Flight records: 7 May-28 July.

Distribution (Fig. 81): ALBERTA: Bilby (30 mi. W. of Edmonton), 19 June. CONNECTI-
CUT: Fairfield Co., Redding, 2 June; Hartford Co., Grandby, W. Grandby, Hartland, 8-9
June; Litchfield Co., Cornwall Bridge, Kent Falls, Norfolk, Riverton, 31 May-13 June; New
Haven Co., Hamden, 2 June. DISTRICT OF COLUMBIA: Washington (no date). GEORGIA:
Union Co., Neel Gap, 22 May. ILLINOIS: Carroll Co., Savanna, 15 June. INDIANA: Jefferson
Co., 25-27 May; Owen Co., McCormick’s Creek State Park, 6 June; Parke Co., Turkey Run
State Park, 28 May-11 June. MAINE: Franklin Co., Rangeley Lakes, June-July; Hancock Co.,
Mt. Desert area, 7 June-4 July; Penobscot Co., Bangor, Ellsworth, Orono, 8 June-July; Piscataquis
Co., Capens (Moosehead Lake area), June-July; Washington Co., Machias, 25 July. MANITOBA:
Portage la Prairie Distr., Aweme, 25 May-16 June; Souris Distr., Ninette, 12 June. MARY-
LAND: Montgomery Co., Glen Echo; Prince Georges Co., Beltsville, Bowie, 29-31 May.
MASSACHUSETTS: Berkshire Co., Mt. Greylock, N. Adams, 14-18 June; Hampshire Co.,
Amherst, 3-16 June; Worcester Co., Petersham, June. MICHIGAN: Clinton Co., Beulah, Cusino,
Rose Lake, 24 May-27 June; Crawford Co., 4 June; Delta Co., 11 June; Ionia Co., Ionia, 31
May; Iron Co., 13 June; Keweenaw Co., Copper Harbor, 19 June; Houghton Co., 20 June;
Lake Co., 8 June; Livingston Co., E. S. George Reserve, 26 May-27 June; Mackinac Co., 7
June-5 July; Marquette Co., Huron Mts., 13-15 June; Newaygo Co., 21 July; Oakland Co., 20
May; Ogemaw Co., Rifle River area, 3-10 June; Ontonogon Co., 18 June; Oscoda Co., 14-15
June; Roscommon Co., 4 June; Schoolcraft Co., 8 June-5 July; Washtenaw Co., Ann Arbor,
Cascade Glen, 28 May-10 June. MINNESOTA: St. Louis Co., Eagle Nest, 4 June. NEW
BRUNSWICK: Charlotte Co., Waweig, 23 May. NEWFOUNDLAND: Bonavista North and
Bonavista South Distrs., Middle Brook, Southwest River near Lethbridge, Terra Nova Nat'l Pk.,
3-6 July. NEW HAMPSHIRE: Cheshire Co., Jaffrey, Mt. Monadnock, 19 June; Coos Co., Bretton
Woods, Randolph, White Mountains, May-July; Grafton Co., Franconia, Hanover, June-July.
NEW JERSEY: Essex Co., Orange Mts., West Orange, June; (?) Co., Hemlock Falls, May.
NEW YORK: Erie Co., Hamburg, 28 May-20 June; Essex Co., Mt. Marcy, Heart Lake (2150
ft.), 29 June; Fulton Co., Canada Lake, Gloversville, Johnstown, Sacandaga Park, Vanderburgs,
15-27 June; Hamilton Co., Lake Pleasant, 20 June; Herkimer Co., Indian Castle, 9 June;
Onondaga Co., Green Lake; Orange Co., Bear Mt., 31 May; Brookview, 7 June; Tompkins Co.,
North Lansing, Ithaca, 1-14 June; Washington Co., Hampton (no date); Westchester Co., Dobbs
Ferry, Tarrytown, 9 June; (?) Co., Misake, 20 June. NORTH CAROLINA: Buncombe Co.,
Swannanoa, 23 May; Burke Co., Linville Falls (3250 ft.), 3-10 June; Haywood Co., Black Mt.,
Mt. Pisgah (3000 ft.), 28 May; Jackson Co., Cashiers (3800 ft.), 12 June; Macon Co., Highlands
(4000 ft.), Van Hook Glade (3800 ft.), Nantahala Gorge, May-June; Swain Co., Great Smoky
Mountains National Park, Forney Ridge trail (6000-6300 ft.), 18 June; Transylvania Co., Camp
Toxoway (3200 ft.), 9-10 June; Wake Co., Raleigh, May; Yancey Co., Mt. Mitchell Game
Refuge (3100 ft.), 22 May-9 June; (?) Co., Conestee Falls, 12 June. NOVA SCOTIA: Kings
Co., Kentville, 4 July; Victoria Co., Baddeck, 6-27 July. OHIO: Hocking Co., Goodhope Twp.,
20 May-6 June, Wayne Co., Wooster, 13 May. ONTARIO: Carleton Co., Britannia Bay, Ottawa,
1-12 June; Cochrane Co., Fort Albany (James Bay), 15-19 June; Kent Co., Bothwell, 13 June;

108 Tue University SciENcE BULLETIN

Fic. 81. Range of Liogma nodicornis. Each spot represents one or more collections within
a county or at a locality.

Nipissing Co., Algonquin Pk., 3 June-7 July; Norfolk Co., Normandale, Simcoe, 9-20 June;
Parry Sound Co., Burks Falls, Kearney, Sand Lake, 29 June-16 July; Simcoe Co., Orillia, 28
June-2 July; (?) Co., Fishers Glen, 2 July. PENNSYLVANIA: Carbon Co., Palmerton, 10
June; Luzerne Co., Hazleton, 22 May-26 June; Philadelphia Co., Roxborough, 7 June. QUEBEC:
Bonaventure Co., Escuminac, 2-31 July; Brome Co., Knowlton, 14-29 June; Gaspé East Co., Gaspé,
26 June-28 July; Gaspé West Co., Anse Pleureuse, Mt. Lyall (1500 ft.). 27 June-29 July;
Gatineau Co., Hull, Meach Lake, 28 May-21 June; Huntingdon Co., Covey Hill, 17 June; Lake
St. John West Co., Mistassini, (?) June-20 July; Matane Co., Metis-sur-mer, 26 June-29 July;
Megantic Co., 6-16 July; Richmond Co., Stoke Centre, 26 June; Shefford Co., Abbotsford, 17
June; Stanstead Co., Georgeville, 14-19 June; Terrebonne Co., Montreal, (?) June. SOUTH
CAROLINA: Spartanburg Co., Spartanburg, 13 May. SOUTH DAKOTA: Pennington Co.,
Harney Peak (6700 ft.), 15 July. TENNESSEE: Great Smoky Mountains National Park, Siler’s
trail (6000-6500 ft.), Indian Gap (5300-5500 ft.), Anakeesta Ridge (4500 ft.), Leconte Lodge
(6400 ft.), Greenbrier, Brushy Mt. trail (3500 ft.), 7 May-15 June; Green Co., Camp Creek, 7
May; Scott Co., 29 May. VERMONT: Bennington Co., Manchester, 6 June; Franklin Co., St.
Albans, 19 June; Washington Co., Montpelier, 25 June. VIRGINIA: Alexandria Co., Glencarlyn,
7 June; Arlington Co., Chain Bridge, 28 May; Fairfax Co., Dead Run, Falls Church, Great
Falls, 19 May-18 June; Giles Co., Kire, Little Stoney Creek, Mountain Lake, 27 May-17 June.
WISCONSIN: Dane Co., May-June; Sauk Co., 28 May.

Types: 8 syntypes (5 males, 1 female, 2 broken), Washington, D. C.,
Dobbs Ferry, New York; White Mountains, New Hampshire; Illinois; New
Jersey (no. 10236, Museum of Comparative Zoology, Harvard University).
One male paratype, New Jersey, May (in Academy of Natural Sciences of
Philadelphia).

Liogma nodicornis flaveola Alexander, male holotype, Great Falls, Vir-
ginia, 19 May 1915; allotype, Great Falls, Virginia, 7 June 1915 (in C. P.
Alexander collection, Amherst, Massachusetts).

CYLINDROTOMINAE OF NortH AMERICA 109

Triogma Schiner

Triogma Schiner, 1863: 223; Osten Sacken, 1865: 237-241.

Type: Limnobia trisulcata Schummel.

This is a very small genus consisting of four known species with only one
from North America. Peus (1952) places the two Oriental species of Triogma
(nimbipennis Alexander, kuwanai Alexander) “incertae sedis” pending more
detailed studies, particularly of the genitalia. Thus this genus is actually based
on only two species (trisulcata of Europe and exsculpta of North America).

Diagnosis: A heavily rugose and pitted head and thorax and dull dark
coloration distinguishes this genus in North America.

Description: Head bare, dull dark brown, pitted; scape long, cylindrical;
flagellar segments subcylindrical; pronotum bare; pretergites glabrous, dark
brown; prescutum concolorous dark brown to black, two longitudinal series
of pits trisecting mesothorax; mesonotal suture pitted, ending at and joining
with longitudinal rows of pits; scutal suture pitted; scutellar sutures obscured
by two deep pits; all pleural sclerites dark brown to black, pitted; wings
strongly suffused with brown, stigma absent; Sci extending to juncture of
Scea with radius; Scen usually present; 2 branches of radius reaching margin
(Rs and Rais); r-m usually obliterated by short fusion of Ri+5 with
Mi +2 at corner of discal cell; three branches of media reaching wing margin:
Mi +2 not dividing, forming basal and anterior edges of discal cell; Ms +4 at
posterior edge of cell, branching, Mz and Mz attaining margin independently;
medial crossvein plus part of Ms closing cell distally; tibial spurs short, prom-
inent, haired; tarsal claws short, broad, abruptly curved distally. Abdomen
dull dark brown, posterior edge of each segment pale.

MALE terminalia: Eighth tergum entire; ninth tergum squarish, 2 small
broad caudal projections laterally; aedeagus with 3 tubes, each enclosing an
ejaculatory duct leading from semen pump.

FEMALE terminalia: Ninth tergum very narrow; tenth tergum divided
into four sclerites: large median pentagonal sclerite, 2 small triangular scle-
rites, 1 on each side, one narrow anterior sclerite; cerci short, broad, not
serrate.

Triogma exsculpta Osten Sacken
(Eigsnd, 20) 275 ily 12109-00.9/0-755 />s0 0,102)

Triogma exsculpta Osten Sacken, 1865: 239; 1869: 304-305.
“ied es isre]|s Weenies, NOI WMojs IPOs Byaie Teyzie ile MEAS AMG lkeyereies,
Diagnosis: Triogma exsculpta is the stockiest and darkest species in this
subfamily. It is dull dark brown to black with characteristic pits or puncta-
tions on both the head and the thorax.
Description: Mate: Body length 7-8 mm; wing length 7-8 mm. Head
dull dark brown, subspherical, rugose; labium, palpi dark brown, haired;

110 Tue UNiversiry ScrENcE BULLETIN

labrum dark brown, glabrous; membranes around mouth parts dusky; cly-
peus very short, dark, convex, not extended laterally, with small hairs distally;
fronto-clypeal suture arched; antennal bases very close together; scape cylin-
drical, longest and broadest segment, sparsely haired, bearing small tooth on
each side, inner tooth broader than outer tooth; pedicel subglobular, sparsely
haired; flagellar segments subcylindrical, covered with long erect hairs, verti-
cils in an uneven whorl.

Pronotum dark brown to black, a shiny brown furrow separating anterior
pitted ridge from shiny, lighter posterior portion; prescutum black, with
tawny pubescence, trisected by 2 longitudinal rows of deep pits, a second-
ary row of pits along mid-line; scutum, scutellum black with tawny pubes-
cence; postscutellum smooth, with 2 black spots posteriorly, anteriorly lighter,
wrinkled, longitudinally divided by shallow furrow; pleural membranes
dusky orange-brown. Stems of halteres pubescent, dusky yellow, knobs
darker.

Coxae blackish, pubescent; trochanters lighter, pubescent, sparsely haired;
femora, tibiae, tarsi light brown, haired, segments progressively more haired
distally.

Terminalia (Figs. 8, 20): Dististyles short, simple, curved, haired, but
tips bare; aedeagus branching into 3 tubes beyond junction of ejaculatory
ducts, middle tube more curved than lateral tubes; ventral side of each tube
slightly prolonged distally (less pronounced than in Liogma), tips flattened;
basal sclerite apparently fused to aedeagus; paraphyses irregular sclerites lying
dorsal to aedeagus (Fig. 8); anal segment lightly sclerotized.

FEMALE: Differs from male in the following: antennae (Fig. 42) shorter,
flagellar segments subspherical, almost moniliform, hairs extremely short,
verticils just below mid-length of each segment. Terminalia (Fig. 27): Cerci
flat, broad, smoothly rounded, notched laterally; hypovalves short, broadly
notched posteriorly, narrower than and fused to eighth sternum.

Variation: This species is quite uniform; however, variations in wing
venation are found, e.g., the absence or only traces of Scz and of the tip of
Sci, occasional presence of r-m; also, the position of m-cu is quite variable,
and the medial crossvein is often lacking.

Larva (Figs. 70-73): Only preserved specimens have been studied. This
larva is distinguished by having short, paired tergal and sternal lobes, no
distinct color pattern and 3 pairs of lobes (2 large, 1 small) surrounding
spiracular disk.

Length 16 mm; width 3 mm. Head completely retractile into prothorax,
the latter having neither hairs nor bristles. Numerous conical lobes cover
larva. Prothorax with 2 pairs of short, simple, dorsal lobes, anterior pair
smaller and closer together than posterior pair; 1 pair of lateral lobes; no
ventral lobes. Mesothorax with 2 pairs of dorsal lobes, posterior pair slightly

CYLINDROTOMINAE OF NortH AMERICA 111

larger, lobes with a small anterior protuberance at base; 2 pairs of simple
lobes on each side. Metathorax dorsally as for mesothorax; 3 simple lobes on
each side and 3 pairs of ventral lobes, second pair largest, placed further apart.
Dorsal abdominal lobes as follows: first abdominal segment with 2 pairs
of lobes, each with an anterior protuberance; 4 pairs of lobes on segments 2 to
7, first pair very small (largest on second segment, becoming progressively
smaller towards posterior end of larva), second pair of lobes slightly larger,
third pair with two dorsal tooth-like protuberances, fourth pair longest, with
one dorsal “tooth”. Lateral abdominal lobes simple; first abdominal segment
with two on each side; second to seventh abdominal segments with 3 on each
side. Ventral lobes simple, conical projections, arranged in pairs, posterior
pair largest: 2 pairs on first abdominal, 5 pairs on second to seventh abdom-
inal segments. Probably 3 or 4 segments involved in caudal region of larva.
Spiracular disk (Fig. 66) small, the 2 rounded spiracles situated side by side,
inclined toward each other; 3 pairs of lobes surrounding spiracle, a small
median dorsal pair, a longer dorso-lateral pair, and a ventral pair with a
conspicuous black line on inner surface of each lobe, tip ending in a sharp
recurved hook. Ventral surface of terminal segment with protuberances
(Rig. 67).

Triogma exsculpta \arvae closely resemble those of Liogma nodicornis
but differ from the latter by having no color pattern, a hairless pronotum,
and an additional small pair of dorsal lobes on the spiracular disk.

Pupa (Figs. 75, 76): Only preserved female specimens have been studied.
This pupa is distinguished by having no color pattern and by 2 short, dorsal
pairs of lobes on each segment, each with a very small, basal protuberance.

Length 12 mm; width 3 mm. Cephalic part of head flat, broad, without
lobes. Prothorax without processes. Mesonotal breathing horns small, directed
dorsad and laterad, terminal half bent cephalad. Mesonotum with two small
slender lobes caudally; broad wing sheaths laterally reaching posterior margin
of second abdominal segment. Metanotum with 2 pairs of simple lobes, an-
terior pair very small, posterior pair longer; lateral haltere sheaths lying along-
side first abdominal segment, obscured by wing sheaths. Leg sheaths ending
just before posterior margin of third abdominal sternum. First abdominal seg-
ment short, 2 pairs of dorsal processes posteriorly, each with small sharp tooth
basally; no ventral nor lateral processes. Abdominal segments 2 to 7 with 3
pairs of dorsal processes, anterior pair small, simple (not always figured,
obscured by posterior pair of processes of preceding segment), middle pair
longer, bearing 2 sharp spines anteriorly, posterior pair largest, 2 teeth an-
teriorly; laterally 3 simple lobes on each side; ventrally 3 pairs of short simple
spines, anteriormost pair extremely tiny. Eighth abdominal segment smaller
than seventh, terminating in lateral lobes, with a small medial pair of spines
on dorsum and on venter. Ninth abdominal segment narrower than eighth,

112 Tue UNiversiry SciENcE BULLETIN

Fic. 82. Range of Triogma exsculpta. Each spot represents one or more collections within
a county or at a locality.

terminating dorso-laterally in a pair of lobes, between which the developing
female cerci can be seen in mature specimens.

This pupa closely resembles that of Liogma nodicornis but may be dif-
ferentiated from the latter by having no dorsal pattern and by having 3
pairs of dorsal lobes on all but the first abdominal tergum.

Ecology: Rogers (1942) reported Triogma exsculpta as very local yet
numerous in quite limited areas. The flies may be found in seepage areas and
along short marshy spring rills and are often found in wet mossy meadows
where there is no woody vegetation. The larvae and pupae have been col-
lected on the aquatic moss Fontinalis antipyretica, and they have been
found on floating mosses of small marsh pools.

Flight records: 20 April-19 June. i

Distribution (Fig. 82): CONNECTICUT: Fairfield Co., Stamford, 15 May, Litchfield Co.,
Norfolk, 9 June. MASSACHUSETTS: Hampshire Co., Amherst, 20 April-25 May. MICHIGAN:
Arenac Co., 25 May; Crawford Co., Branch of Big Creek, 23 May-19 June; Livingston Co., E. S.
George Reserve, Honey Creek, Putnam Twp., 15-20 May; Otsego Co., 24 May; Washtenaw Co.,
15 May. NEW HAMPSHIRE: (no further data). NEW JERSEY: Essex Co., Waverly, May.

NEW YORK: Washington Co., Hampton, 20 May. PENNSYLVANIA: (no further data).
QUEBEC: Gatineau Co., Hull, 17 May. WISCONSIN: Dane Co., 10 May.

Types: Female holotype, Pennsylvania, (ANSP Type no. 6058, Academy
of Natural Sciences, Philadelphia).

CYLINDROTOMINAE OF NortH AMERICA 113

LITERATURE’ CITED
BIBLIOGRAPHY

ALExANDER, C. P. 1913. Report on a collection of Japanese crane-flies (Tipulidae). Canadian
Ent. 45(10) :313-322.

—. 1914a. New or little-known crane-flies from the United States and Canada, Tipulidae.
Proc. Acad. Nat. Sci. Philadelphia 66:579-606.

—. 1914b. Biology of the North American crane-flies (Tipulidae, Diptera). II. Liogma
nodicornis Osten Sacken. Pomona Jour. Ent. and Zool. 6:105-120.

—. 1919a. Undescribed species of Japanese crane-flies (Tipulidae, Diptera). Ann Ent.
Soc. Amer. 12:327-348.

—. 1919b. New Nearctic crane-flies (Tipulidae, Diptera). Part VIII. Canadian Ent. 51:
191-199,

——. 1919c. The crane-flies of New York. Part I. Distribution and taxonomy of the adult
flies. Cornell Univ. Agr. Exp. Sta. Memoir 25:766-993, plates XXX-LV.

—. 1920. The crane-flies of New York. Part II. Biology and phylogeny. Cornell Univ.
Agr. Exp. Sta. Memoir 38:695-1133, plates XII-XCVII.

—. 1922. Undescribed crane-flies (Tanyderidae, Tipulidae) in the South Australian Museum.
Rec. South Australian Museum 2:250-251.

——. 1924. New or little-known crane-flies from northern Japan (Tipulidae, Diptera). Philip-
pine Jour. Sci., 24(5):531-611. rm

—. 1927a. Cylindrotominae. Genera Insectorum. 187:1-16, plates I, II.

—. 1927b. New Nearctic crane-flies (Tipulidae, Diptera). Part XII. Canadian Ent. 59(8):
184-193.

———.1927c. Records and descriptions of crane-flies from Alberta (Tipulidae, Diptera).
Canadian Ent. 59(8) :214-225.

—.. 1930. New or insufficiently-known crane-flies from the Nearctic Region (Tipulidae,
Diptera). Part I. Bull. Brooklyn Ent. Soc. 25(5) :276-282.

—. 1931. Crane-flies of the Baltic Amber (Diptera). Bernstein-Forschungen (Amber
Studies), Vol. 2, ed. by K. Andree. Berlin and Leipzig: Gruyter and Co. 135 pp.,
illus.

—. 1932. New or little known Tipulidae from the Philippines (Diptera). Part XIII.
Philippine Jour. Sci. 47(1):163-195.

—. 1933. Records and descriptions of Japanese Tipulidae (Diptera). Philippine Jour. Sci.
51(2):216-219.

—. 1940. Records and descriptions of North American crane-flies (Diptera). Part I.
Tipuloidea of the Great Smoky Mountains National Park, Tennessee. Amer. Midland
Nat. 24(3) :602-644.

—. 1941. Records and descriptions of North American crane-flies (Diptera). Part II.

Tipuloidea of mountainous western North Carolina. Amer. Midland Nat. 26(2):

281-319.

. 1942. Family Tipulidae. Im, The Diptera or true flies of Connecticut. Conn. State

Geol. and Nat. Hist. Survey, Bull. 64:196-509.

.1943a. Records and descriptions of North American crane-flies (Diptera). Part III.

Tipuloidea of the Upper Gunnison Valley, Colorado. Amer. Midland Nat. 29(1):

147-179.

. 1943b. Records and descriptions of North American crane-flies (Diptera). Part IV.

Tipuloidea of the Yellowstone National Park. Amer. Midland Nat. 30(3):718-764.

. 1945. Records and descriptions of North American crane-flies (Diptera). Part V.

Tipuloidea of the Grand Teton National Park and Teton National Forest, Wyoming.

Amer. Midland Nat. 33(2) :391-439.

. 1949. Records and descriptions of North American crane-flies. Part VIII. Tipuloidea

of Washington. Amer. Midland Nat. 42(2):257-333.

. 1954. Records and descriptions of North American crane-flies (Diptera). Part IX.

Tipuloidea of Oregon, 1. Amer. Midland Nat. 51(1):1-86.

. 1956a. Undescribed species of crane-flies from the eastern United States and Canada

(Diptera, Tipulidae). Part XII. Ent. News 67:177-185.

.1956b. Undescribed species of crane-flies from the Himalaya Mountains (Tipulidae,

Diptera), I. Jour. New York Ent. Soc. 64:137-147.

. 1965. Family Tipulidae, p. 16-90. In, A catalog of the Diptera of America north of

Mexico. Agriculture handbook No. 276. Agriculture Research Service, U.S. Dept.

Agr., Washington.

114 Tue University ScrENcE BULLETIN

, and W. L. McArer. 1920. Diptera of the Superfamily Tipuloidea found in the

District of Columbia. Proc. U.S. Nat. Mus. 58(2344) :385-435.

Benotsson, S. 1897. Till kannedomen om larven af Phalacrocera replicata (Lin.) Acta Reg.
Soc. Physiogr. Lund. 8:1-117, 4 plates.

BrimLey, C. S. 1938. Insects of North Carolina. North Carolina Dept. Agr., Div. of Ent.,
Raleigh. 560 pp.

Brunetti, E. 1912. Diptera Nematocera (excluding Chironomidae and Culicidae). In, The
fauna of British India, as one volume in a series. London, Taylor and Francis. 581
pp., 12 plates.

Byers, G. W. 1961. The crane fly genus Dolichopeza in North America. Univ. Kansas Sci.
Bull. 42 (6) :665-924.

Cameron, A. E. 1918. Life-history of the leaf-eating crane fly Cylindrotoma splendens Doane.
Ann. Ent. Soc. Amer. 11:67-89.

CockERELL, T. D. A. 1920. Eocene insects from the Rocky Mountains. Proc. U.S. Nat. Mus.
Washington 57(2313):233-260.

Coguittett, D. W. 1900. Papers from the Harriman Alaska expedition. Entomological results
(3). Proc. Washington Acad. Sci. 2:389-464.

Dickinson, W. E. 1932. The crane-flies of Wisconsin. Bull. of the Public Museum of the
City of Milwaukee 8(2):139-260.

Doane, R. W. 1900. New North American Tipulidae. Jour. New York Ent. Soc. 8:182-198.

ENpDERLEIN, G. 1912. Studien tiber die Tipuliden, Limoniiden, Cylindrotomiden und Pty-
chopteriden. Zool. Jahrb. Syst. 32:83-85.

Grinserc, K. 1910. Diptera, Zweifligler, Part I. Die Stisswasserfauna Deutschlands 2A:1-312.

Hennic, W. 1950. Die Larvenformen der Dipteren. Part II. Berlin: Akademie-Verlag, 458 pp.,
10 plates.

Jounson, C. W. 1912. New North American Diptera. Psyche 19(1):1-5.

Kerrtsz, C. 1902. Catalogus dipterorum hucusque descriptorum. Vol. II. Cecidomyiidae,
Tipulidae, etc. Leipzig. 359 pp.

Leonarp, M. D. 1928. A list of insects of New York. Cornell Univ. Agr. Exp. Sta. Memoir
101:1-1121 (Tipuloidea, pp. 688-701).

LinnarEus, C. 1758. Systema Naturae. Editio decima, reformata, Tomus I. Laurentii Salvi,
Holmiae. 824 pp.

MacGinitize, H. D. 1953. Fossil plants of the Florissant beds, Colorado. Carnegie Institute of
Washington. Washington ii + 198 pp.

Macquart, P. J. M. 1834. Histoire naturelle des insectes. Diptéres. I Tom. Paris, Librairie
Encyclopedique de Roret, 578 pp., 12 plates.

MEIcEN, J. W. 1818. Systematische Beschreibung der bekannten europaischen zweifligeligen
Insekten. 1:1-269.

Meunier, F. 1906. Monographie des Tipulidae et des Dixidae de l’ambre de la Baltique. Ann.

Sci. Nat., Zool. (9)4:349-403.

.1915. 9. Uber einige fossile Insekten aus den Braunkohlenschichten (Aquitanien) von

Rott im Siebengebirge. Zeitschr. Deutschen Geo. Ges. 67:219-230.

NeepHaM, J. G. 1908. Report of the entomologic field station conducted at Old Forge, New
York, in the summer of 1905. 23rd Rept. of the State Entomologist, 1907:156-248,
15 figs.

OsTEN SackeNn, C. R. 1865. Description of some new genera and species of North American
Limnobina. Part 1. Proc. Ent. Soc. Philadelphia 4:224-242.

—. 1869. Monographs of the Diptera of North America. Part IV. Smithsonian Misc. Coll.
8(219):1-345, 4 plates, 7 figs.

——. 1878. Catalogue of described Diptera of North America. Second edition. Smithsonian
Misc. Coll. 16(270) :1-276.

—. 1887. Studies on Tipulidae. Part 2. Review of the published genera of the Tzpzlidae
brevipalpt. Berliner Ent. Zeitschr. 31:163-242.

—. 1897. Remarks on the literature of the earlier stages of the Cylindrotomina, a section
of Tipulidae. Trans. Ent. Soc. London 1897:362-366.

Peus, F. 1952. Cylindrotomidae. Im, Lindner, Erwin. Die Fliegen der palearktischen Region.
Lfg. 169(17):1-80, 2 plates, 83 figs.

Rocers, J. S. 1918. Collection of Tipulidae from Schoolcraft Co., Michigan. Occ. Papers Mus.

Zool. Univ. Michigan 55:1-4.

. 1930. The summer crane-fly fauna of the Cumberland Plateau in Tennessee. Occ.

Papers Mus. Zool. Univ. Michigan 215:1-50, 5 plates.
. 1942. The crane-flies (Tipulidae) of the George Reserve, Michigan. Univ. Mich. Mus.
Zool. Misc. Publ. 53:1-128, 8 plates.

CYLINDROTOMINAE OF NortH AMERICA 115

Ross, H. H. 1956. A textbook of entomology. 2nd edition. New York, John Wiley & Sons,
Inc., xi + 519 pp., 402 figs.
Scuiner, I. R. 1863. Vorlaufiger Commentar zum dipterologischen Theile der “Fauna austriaca.”
V. Wiener Ent. Monatschr. 7:217-226.
. 1864. Fauna austriaca. Die Fliegen. 2:1-658.
ScHuMMEL, T. E. 1829. Beschreibung der in Schesien einheimischen arten einiger dipteren-
gattungen. I. Limnobia. Beitr. z. Ent. (Breslau): 97-201.
ScuppErR, S. H. 1877. The first discovered traces of fossil insects in the American tertiaries.
Bull. U.S. Geol. Surv. Terr. 3 (4) :741-762.
. 1894. Tertiary Tipulidae, with special reference to those of Florissant Colorado. Proc.
Amer. Philos. Soc. 32:163-245.
Snopcrass, R. E. 1904. The hypopygium of the Tipulidae. Trans. Amer. Ent. Soc. 30:179-235.
. 1957. A revised interpretation of the external reproductive organs of male insects.
Smithsonian Misc. Collections 135(6):1-60.
Spencer, G. J. 1930. Notes on Phalacrocera species, an aquatic crane fly (Diptera, Tipulidae).
Proc. Ent. Soc. British Columbia 27:15-16.
Takanasut, M. 1960. A review of Japanese Cylindrotominae (Diptera, Tipulidae). Trans.
Shikoku Ent. Soc. 6(6-7) :81-91.
Wacker, F. 1856. Insecta Britannica. Diptera III:268-337.
ZETTERSTEDT, J. W. 1838. Insecta Lapponica descripta. Lipsiae, Voss. Dipt.:477-868.

ny

4

K 33

THE UNIVERSITY OF KANSAS

SCIENCE BULLETIN

SPERM TRANSPORT FERTILIZATION AND
PREIMPLANTATION LOSS IN
PHH AND PHL MICE

By
John E. Lothers, Jr.

Vou. XLVII Paces 117-144 Aprit 14, 1967 No. 4

ANNOUNCEMENT

ee

The University of Kansas Science Bulletin (continuation of the Kansas Uni-—

versity Quarterly) is issucd in part at irregular intervals. Each volume contains
300 to 700 pages of reading matter, with necessary illustrations. Exchanges with
other institutions and learned societies everywhere are solicited. All exchanges
should be addressed to

LIBRARY OF THE UNIVERSITY OF KANsaAs,
LAWRENCE, Kansas 66044

PUBLICATION DATES

The actual date of publication (z.e., mailing date) of many of the volumes of
the University of Kansas Science Bulletin differs so markedly from the dates on
the covers of the publication or on the covers of the separata that it seems wise to
offer a corrected list showing the mailing date. The editor has been unable to
verify mailing dates earlier than 1932. Separata were issued at the same time as

the whole volume. Beginning with Volume XLVI, publication was by separate ©

numbers and the date on each number is the actual publication date.

Vol. XX—October 1, 1932. Vol. XXXIV, Pt. I—Oct. 1, 1951.
Vol. XXI— November 27, 1934. Pt. II—Feb. 15, 1952.
Vol. XXJI-—-November 15, 1935. Vol. XXXV,Pt. I—July 1, 1952.
Vol. XXIII—August 15, 1936. Pt. II—Sept. 10, 1953.
Vol. XXIV—February 16, 1938. Pt. II—Nov. 20, 1953.
Vol. XXV—July 10, 1939. Vol. XXXVI,Pt. I—June 1, 1954.
Vol. XXVI—November 27, 1940. Pt. II—July 15, 1954.
Vol. XXVII,Pt. I—Dec. 30, 1941. Vol. XXXVII, Pt. I—Oct. 15, 1955.
Vol. XXVIII, Pt. I—May 15, 1942. Pt. II—June 29, 1956.

Pt. II—Nov. 12, 1942. Vol. XXXVIII, Pt. I—Dec. 20, 1956.
Vol. XXIX,Pt. I—July 15, 1943. Pt. II—March 2, 1958.

Pt. II—Oct. 15, 1943. Vol. XXXIX—Nov. 18, 1958.
Vol. XXX, Pt. I—June 12, 1944. Vol. XL—April 20, 1960.

Pt. II—June 15, 1945. Vol. XLI—Dec. 23, 1960.
Vol. XXXI, Pt. I—May 1, 1946. Vol. XLII—Dec. 29, 1961.

Pt. II—Nov. 1, 1947. Vol. XLII—Supplement to, June 28, 1962.
Vol. XXXII—Nov. 25, 1948. Vol. XLIlI—Aug. 20, 1962.
Vol. XXXIII, Pt. -I—April 20, 1949. Vol. | XLIV-—Sept. 1, 1963.

Pt. II—March 20, 1950. Vol. XLV—June 7, 1965.

EGRORS ois Crane ao nie R. C. Jackson
Editorial Board ........ GerorcE Byers, Chairman

KENNETH ARMITAGE
CHARLES MICHENER
Paut Kiros
RICHARD JOHNSTON
DELBERT SHANKEL

THE UNIVERSITY OF KANSAS

SCIENCE BULLETIN

VoL. XLVI Paces 117-144 Aprit 14, 1967 No. 4

Sperm Transport Fertilization and Preimplantation Loss in
PHH and PHL Mice’
By

Joun E. Lotuers, JR.

Hall Laboratory of Mammalian Genetics, Department of Zoology

ABSTRACT

To investigate reasons for the smaller litters and higher sex ratio in litters
sired by PHH mice compared to litters sired by males of the PHL strain, sperma-
tozoa were subjected to a variety of treatments.

Single and mixed batches of spermatozoa were inseminated artificially. The
interval between injection of PHH and PHL spermatozoa varied between 0 and
96 minutes. Preimplantation and post-implantation losses were investigated by
means of ova and fetal counts along with counts of corpora lutea and resorptions.
Spermatozoa from the vasa deferentia and from ejaculates were examined for
numbers, morphology and percentage viability. Proportions of normal and
abnormal spermatozoa reaching the oviduct following copulation by an intact
male were compared with proportions found in the uterus.

Following artificial inseminations with mixtures of PHH and PHL sperma-
tozoa, there were significantly fewer litters containing progeny from PHH than
PHL spermatozoa, even when conditions were seemingly heavily weighted in favor
of PHH. The competitive disadvantage of PHH spermatozoa could be partially
compensated for by injecting 69 or more minutes before PHL. Hyaluronidase
added to the PHH extender also seemed to compensate partially for the disad-
vantage and an effect of hyaluronidase on the sex ratio is indicated. In vivo and
in vitro (B-amylase) capacitation of PHH spermatozoa before insemination
proved ineffective. Preimplantation losses as determined from counts of ova,
fetuses, resorptions, and corpora lutea tend to be higher following copulation with
PHH than PHL males. Much of the difference is attributable to a few cases in
which PHH spermatozoa fertilize only a few of the available ova. There was no
significant difference between strains in sperm number or morphology. However,
the proportion of spermatozoa from the vas deferens that were viable was higher
in PHL than PHH. There was wide fluctuation within strains in sperm counts

* Adapted from a dissertation submitted in partial fulfillment of the requirements for the
degree of Doctor of Philosophy, University of Kansas. This investigation was supported by re-
search training grant 5 T1-GM-246, and research grant GM-7786, National Institutes of Health,
USPHS.

118 Tue UnNiversiry ScrENCE BULLETIN

from ejaculates. There was no evidence for a greater proportion of morphologi-
cally abnormal PHH than PHL spermatozoa entering the oviduct. Looped
(bent) spermatozoa, found in larger proportions in PHH than in PHL ejaculates,
are able to enter the oviduct.

PHH spermatozoa are less successful than PHL particularly when the two are
competing for ova in the same female. The difference may be caused in part by
slower penetration by PHH spermatozoa. The smaller number of functional
PHH as compared with PHL spermatozoa together with the slower penetration,
and perhaps other factors, sometimes lower the fertilizing potency of a PHH
ejaculate below the threshold necessary for fertilization of most of the available
ova. Frequently none of the ova are fertilized.

I. INTRODUCTION

Parameters for the strains used in this study (Weir, 1962) include a
difference in sex ratio (PHH 52.8 + 1.00; PHL 41.8 = 0.93) and a difference
in litter size. Litters sired by PHL males averaged 0.3 to 1.5 more mice per
litter than litters sired by PHH, and this is independent of the strain of
female. The strains A/He, AKR, C3H and BALB/c and the outbred “K”
stock were used (Weir, 1960, 1962). The sex ratio difference, attributable to
the male, is not due to differential mortality in the last trimester (Beck, 1957).
Finn (1964), using randomly bred mice, found an effect of the individual
male on litter size; Krzanowska (1960) found an effect of strain of male on
litter size in the P inbred line.

The present study was undertaken to determine the nature of the physio-
logical difference between PHH and PHL spermatozoa. PHH spermatozoa
seem to be functionally deficient.

The effect on litter size that shows up statistically may be due to a num-
ber of factors having threshold effects. Litter size may be affected by in-
sufficient numbers of functional spermatozoa at the site of fertilization at the
crucial time, by loss of fertilized ova before implantation, or by loss after
implantation. The number of spermatozoa at the fertilization site could be
limited in several ways. Perhaps the most important of these would be the
number of spermatozoa in the ejaculate. The number of motile spermatozoa
in an ejaculate required to prevent reduction in fertilization rate in 129/Re,
C57BL/6 and F: hybrids of these strains is more than 500,000 (Baker 1962).
According to Chang (1958) the large number of spermatozoa ejaculated is
necessary so there will be enough strong ones to survive in the female tract,
especially if ovulation is late. Krzanowska (1964a), in explaining the longer
time required for fertilization in inbred mice as compared with outbred,
pointed out that the proportion of spermatozoa with full fertilizing capacity
is lower in inbred males. Thus the number of spermatozoa ejaculated from
an inbred male reaching the fertilization site could be a limiting factor in
litter size, particularly in a strain of low fertility.

SPERM TRANSPORT FERTILIZATION AND PREIMPLANTATION Loss 119

The number of spermatozoa that reach the fertilization site in the female
is limited by the uterotubal junction and probably by the isthmus of the
oviduct. Braden and Austin (1954a) found an average of about 17 sperma-
tozoa per tube at the fertilization site in the mouse 10 to 15 hours after
copulation. A reduced number of spermatozoa at the fertilization site at the
optimal time for fertilization may be related to male mating behavior. Weir
(1962) reported copulation by PHH males with females in late estrus, re-
sulting in reduced litter size. PHH males with continuous access to females
may also copulate before the optimal time for fertilization. Not only must
spermatozoa reach the ampulla of the oviduct, but also there must be an
adequate number with unimpaired function. Function might be impaired by
a slower than normal capacitation, by inadequate motility, or by reduced
ability to penetrate cumulus cells, zona pellucida or vitellus. Necessity of
capacitation of mouse spermatozoa has been suggested by Braden and Austin
(1954b) and by Krzanowska (1964a). According to Cross (1958) motility is
crucial only for penetration of the ovum. Krzanowska (1960, 1961, 1962)
found a decrease in number of ova fertilized in inbred compared to outbred
and crossbred mice. The greatest reduction in litter size occurred at fertiliza-
tion and from implantation to the 12th day of pregnancy (Krzanowska
1961). Outbred males improved the fertilization rate in inbred females
(both E and P lines). Fertilization rate in outbred females was reduced by
P line inbred males (Krzanowska, 1960).

Falconer (1960) and Falconer and Roberts (1960) found reduced fertility
of inbred females was due almost entirely to preimplantation loss. Differences
in male fertility contributed only 10°% or less of the litter size variation, how-
ever. McCarthy (1965) found that most of the loss due to inbreeding was
early post-implantation loss in CBA/Fa, C57/Fa and R111/Fa mice. He
found no male strain effect. Lyon (1959) also found post-implantation
mortality due to inbreeding.

The experimental procedures utilized in the study to be reported here in-
cluded: (1) artificial insemination using combinations of PHH and PHL
spermatozoa; (2) counts and classification of ova 21 to 57 hours after copula-
tion and counts of fetuses, corpora lutea and resorptions at 14 to 17 days after
copulation to determine preimplantation and post-implantation losses; (3)
determination of number, viability and morphology of spermatozoa from
the vas deferens and from ejaculates of PHH and PHL, and the proportion

of morphologically abnormal spermatozoa in the uterus and in the oviduct.

II. ARTIFICIAL INSEMINATION

The mating of 105 PHL females (treated to induce ovulation) to PHL
/males with seminal vesicles tied, followed by mating to intact PHH males,
resulted in two mixed litters and a total of only 8 litters in all. This, to-

120 Tue University ScrENCE BULLETIN

gether with results from artificial insemination of PHH and PHL females
in induced estrus indicates that females of these strains do not respond well
to PMS and HCG nor to PMS and progesterone. Some females did not
ovulate, some did not copulate, and some that copulated did not produce
litters.

In place of mixed matings, artificial insemination was employed to obtain
mixed pools of PHH and PHL spermatozoa. In a mixture spermatozoa
from one male may be superior to those from another. Beatty (1960) found
differences in success of spermatozoa from two different rabbits when they
were mixed and inseminated. Admixture of spermatozoa from more rabbits
seemed to increase the percent of inseminations yielding litters. Edwards
(1955) demonstrated an apparent superiority of spermatozoa from one strain
(C3H) when C3H, REB and G spermatozoa were used in all combinations
of two types as well as a mixture of all three types. Southard, Wolfe, and
Russell (1965), on the other hand, found that (129/Re x C57BL/6)Fi
spermatozoa from dystrophic males were of about equal potency to those
from non-dystrophic Fi males when the two were mixed and inseminated.
Data from Weir (1962) on mixed litters from PHH and PHL double mat-
ings showed that this 72 vivo mixing of the spermatozoa accentuated the
superiority of PHL spermatozoa. PHL spermatozoa showed a greater
superiority in number of progeny produced in the mixed litters than in the
single litters. Females were caged continuously with one male of each strain.

MATERIALS AND METHODS

CoMPARISON OF SPERM EXTENDERS. A pilot study was performed to compare
the following sperm extenders: Locke’s solution with extra glucose (Snell,
Hummel and Abelmann, 1944); 95°% non fat dry milk heated for ten
minutes or more and cooled prior to use (Dziuk and Runner, 1960); a 2:1
mixture of 9.5°/ non fat dry milk and 0.859% saline; 0.659% saline; the yolk
citrate, alcohol sugar extender that is used for cattle by the Kansas Artificial
Breeding Service Unit; modified Krebs Henseleit Ringer solution (Bhatta-
charya 1962) without the aureomycin-HCl1; egg yolk citrate extender num-
ber 3 of Fox and Burdick (1963) decanted from its settled solids and lacking
the glycerol; a 1:1 mixture of the egg yolk citrate with modified Krebs
Henseleit Ringer solution; and the egg yolk citrate extender with 800 units/
cc penicillin and 800 g/cc streptomycin. In the second experiment the ex-
tenders used were 0.85°% saline, the egg yolk citrate extender of Fox and
Burdick modified as described above, egg yolk citrate that had had COs
bubbled through it, and two egg yolk citrate extenders identical to the first
one except that the egg yolk was reconstituted dried yolk. Both of these were
prepared with Armour Creameries Cloverbloom powdered egg yolk. One was
standard type and the other was Type Y, which contains 10 parts corn syrup

SPERM TRANSPORT FERTILIZATION AND PREIMPLANTATION Loss 121

solids to 90 parts liquid egg yolk. The Type Y yolk was provided by the
Kansas Artificial Breeding Service Unit. The pH of all extenders was ad-
justed to within 0.2 of a pH unit of neutrality. In most cases it was adjusted
within 0.1 of a unit. Bishop and Walton (1960a) reported 7.5 as the optimum
pH for mammalian spermatozoa and metabolism, although this varies among
different mammals and the range of tolerance is quite wide. Contents of
vasa deferentia from PHH and PHL males were stripped into 0.10 cc of
0.65°% saline in separate depression slides. Two sets of extenders were used,
0.15 cc per tube, one set for each strain. After the spermatozoa were
thoroughly mixed in the 0.659% saline, an aliquot of an amount expected to
make a dilution appropriate for counting was pipetted into each extender
and mixed. An estimate of percent motility was made by haemacytometer
examination and checked several hours later. Criteria for judging the ex-
tenders were: (1) length of time motility was retained; (2) percent viability
as determined by nigrosin-eosin smears. For each haemacytometer count
notes were made on type of motility, i.e., speed of motion and whether or
not the spermatozoa were making forward progress. In the first series, using
nine extenders, the nigrosin-eosin smears were prepared within 8 hours
after spermatozoa were placed in the extender. In the second series of five
the smears were made within 4 hours after spermatozoa were put in the
extenders. Each sperm suspension was mixed with the stain for about 30
seconds prior to spreading. After nigrosin-eosin slides were prepared, all
tubes were immersed in water at room temperature and placed in the re-
frigerator, thus effecting a gradual cooling. In the first series spermatozoa
were examined via haemacytometer at two, four, six, and seven days. In
the second series they were examined at two and five days. Sperm suspen-
sions were allowed to come slowly to room temperature prior to examination.

Insemination. For artificial inseminations egg yolk citrate extender with
glycerol was used in addition to the above. The one selected in the end as
most suitable was the 9.5°% non fat dried milk with antibiotics and ascorbic
acid added. Ascorbic acid concentration ranged from 0.05 mg per ml to 0.39
mg per ml. In most cases the range was 0.20 to 0.34 mg per ml. The pro-
cedure for insemination was similar to that of Dziuk and Runner (1960).
Most of the females had been induced to ovulate by means of PMS and
HCG (2.0 or 2.5 i.u.) given intraperitoneally with an interval of about 37 to
38 hours between injections. For some earlier inseminations some of the
females in each group were given 0.5 to 1.0 mg of progesterone, whereas
others were given HCG as the second injection. In those groups that had
litters 4/8 of the HCG-injected females had litters, while 0/8 of the progester-
one-injected females had litters. The progesterone dosage for six of the eight
females was 0.5 mg. Females in post partum and cyclical estrus were used
for some inseminations. Cyclical estrus was determined by vaginal smears

122 Tue University ScreNcE BULLETIN

and used only for PHL females which showed weak response to gonado-
trophins. After trying also PHH, C3H/He, AL/N, Fi(AL/N x C57L),
C57BL/6 and BALB/c, BALB/c females became the strain of choice because
of their capacity to respond to gonadotrophins and because eye pigment
differences visible at birth allow identification of offspring sired by PHH or
PHL males. The vasectomized males, used to provide the vaginal plug, in-
cluded PHL, PHH, C3H/He and Fi(PHL x PHH). The Fi males seemed
most suitable because (1) regeneration of vasa deferentia would be revealed
by 50°% of the progeny being phenotypically distinguishable from both PHH
and PHL, (2) they copulate quickly and are likely to mate when the female is
in the optimum state of receptivity. Regeneration of vasa deferentia occurred
in one of the Fi males. Subsequently, in addition to removal of approxi-
mately one half inch from each duct, the end adjoining to the epididymis
was tied.

Vasa deferentia were stripped into the extender in a CO»-enriched atmos-
phere, PHH and PHL spermatozoa being kept in separate depression slides.
In all but the first few inseminations some epididymal spermatozoa were
introduced by making one cut through the cauda epididymis and pressing
out the contents in the depression slide adjacent to the extender. These con-
tents were then stirred into the extender. After the spermatozoa were care-
fully mixed with the extender a haemacytometer count was made on a 1:10
dilution from each strain. For some inseminations spermatozoa from the
two strains were mixed together prior to insemination. In others PHH
spermatozoa were inseminated first followed at varying time intervals by
PHL spermatozoa. The time intervals were 8 to 21 minutes, 38 to 67 minutes,
and 69 to 78 minutes. Intervals longer than these were tried, but only for a
few inseminations. When an equal number of males from the two strains
was utilized the number of motile spermatozoa from PHH was noticeably
smaller than that from PHL males. Roughly equivalent numbers could
usually be obtained by using twice as many PHH males as PHL. To in-
crease the percentage of motile PHH spermatozoa as well as to obtain sper-
matozoa that were already capacitated, or at least partially so, PHH ejaculates
that had been in the uterus for up to two hours were used to supplement the
source from vasa deferentia. To inseminate a group of three females sperma-
tozoa were used from four PHL males, six PHH males, and as many PHH
ejaculates as could be obtained from six fertile males each given a hormonally
primed female. Some females were inseminated with only PHH sperma-
tozoa. In several groups of females one or more of the females were given
only PHH spermatozoa while the other females in the group were sub-
sequently given PHL in addition to the PHH spermatozoa. In one such
group the delayed supply of PHL spermatozoa was provided by copulation
with intact PHL males. Similar experiments were performed with PHL

SPERM TRANSPORT FERTILIZATION AND PREIMPLANTATION Loss 123

spermatozoa. That is, some females were given only PHL spermatozoa and
some were given PHL followed by PHH. Spermatozoa were inseminated
by means of a blunted 22 gauge needle 11 to 14 1/6 hours after HCG injec-
tion. Most inseminations were between 12!4 and 14 hours after the HCG.

In a number of inseminations, hyaluronidase,* 8-amylase,* fructose,
bicarbonate, or an extract of PHL spermatozoa was added to the PHH
spermatozoa. Concentrations of hyaluronidase chosen for this work were 1,
2, and 4 mg per ml. At 2 mg, and especially at 4 mg per ml, there may have
been a deleterious effect on the spermatozoa as judged by examination in
the haemacytometer. More than the usual number seemed to be bent double.
Dosage of 8-amylase was selected on the basis of work on rabbit sperm by
Kirton and Hafs (1965), who used 0.1 or 1.0 mg per 100 ml of the extender.
Concentrations chosen were 0.3, 0.8, and 3.6 mg per ml. Concentrations of
fructose ranged from 2.5 to 3.0 mg per ml. The extract of PHL spermatozoa
was prepared by means of a tissue homogenizer or by alternate freezing and
thawing. Spermatozoa from vasa deferentia and epididymis of 4 to 6 PHL
males were stripped into the usual milk extender. After the cells were broken
by one of the afore-mentioned procedures, cells and cellular debris were re-
moved by centrifugation and part of the supernatant was used as the ex-
tender for the PHH spermatozoa. In all the inseminations in which an
extra ingredient was added to the extender for the PHH spermatozoa, PHH
spermatozoa were inseminated before PHL. The time interval between in-
seminations ranged from 46 to 83 minutes.

RESULTS

CoMPARISON OF SPERM EXTENDERS. There was a tendency toward higher
viability in the milk and milk saline extenders, but more active motility in
most of the others, including the egg yolk citrate extenders. The milk, milk
saline, and egg yolk citrate extenders provided longest duration of motility.
Penicillin and streptomycin added to the egg yolk citrate extender did not
affect viability or type of motility and seemed to prolong duration of motility.
The yolk citrate alcohol sugar extender, which is of proven value for bull
spermatozoa, was noticeably inferior to the other extenders. Most of the
motility had been lost one or two hours before the nigrosin-eosin slides were
prepared. The remaining motile spermatozoa were moving very slowly, so
nigrosin-eosin slides were not made.

Insemination. The sex ratios derived from inseminations, 71/202=
0.3510.035 for PHL and 19/37=0.5140.082 for PHH, are strikingly simi-
lar to the strain parameters. Comparison of PHL litters of 3 or fewer (9/29=
-0.310+0.093) with litters of 4 or more (49/144=0.340+0.042) reveals no

* Calbiochem, Los Angeles, Calif.

124 Tue University ScieNcE BULLETIN

difference in sex ratio. Results from hyaluronidase-treated inseminations are
not included in these values. The number of progeny from litters of three
or fewer is admittedly small. If X-bearing spermatozoa have an advantage,
for example, greater ability to penetrate, the excess of XX zygotes might be
accentuated when there are not enough spermatozoa to fertilize all of the
ova. The only agent that seemed to exert an affect on sex ratio was hyaluroni-
dase. Since it was added to PHH spermatozoa only, it would have to re-
main in the uterus in large enough quantities to affect the PHL spermatozoa
inseminated 46 to 83 minutes later. Sex ratio from PHL spermatozoa under
these conditions was 12/23=0.522+0.104, and for inseminations without
hyaluronidase it was 52/156=0.3350.040 (X71=3.10, 0.05<P<0.10). For
PHH spermatozoa the ratios were 5/6=0.833 with hyaluronidase, and 14/29
=0.483=0.093 without it (X71=2.46, 0.10<P<0.20).

The effectiveness of the PHH spermatozoa in the inseminations was
poor. Even when many more motile PHH than PHL spermatozoa were
inseminated there were more progeny from PHL than from PHH sperma-
tozoa. From these inseminations the percent of progeny from PHH sperma-
tozoa constituted 5/31=16.1 percent of the total. With one exception, a
case of insemination with spermatozoa from PHH only, there were no full
size litters from PHH spermatozoa. The average number of live progeny
from BALB/c females induced to ovulate with gonadotrophins and mated to
fertile males was 42/7=6.0+1.05. The mean number of ova from BALB/c
induced to ovulate with gonadotrophins was found by H. Glenn Wolfe (un-
published data) to be 204/20=10.2+1.32. The number of progeny from
inseminated PHH spermatozoa varied from one to two in inseminations in
which both strains of spermatozoa were used and from one to 14 (with 10
live) in cases in which only PHH spermatozoa were inseminated. Exclud-
ing the one large litter, the numbers varied from one to four (three live).
Number of progeny per litter from PHH and PHL spermatozoa are shown
in Table 1. Litter sizes from all double inseminations of BALB/c females,
exclusive of those on females in post partum estrus, were 30/22=1.4 for PHH
spermatozoa and 165/33—=5.0 for PHL spermatozoa. Not only were litters
from PHH spermatozoa small, but also they occurred in only a small fraction
of the females inseminated. Results are shown in Table 2. From Table 1 a
comparison can be made between the fraction of females that had progeny
when one strain of spermatozoa was injected and the fraction that had
progeny when both strains of spermatozoa were injected. The fraction that
had litters when given gonadotrophins and mated to known fertile males
was 7/10. Southard, Wolfe and Russell (1965), using hormonally primed
(129/Re x C57BL/6)F: females, found the fraction to be 127/317=43%
from natural matings. There were three groups of inseminated females in
which litters occurred both in the subgroup given only PHH spermatozoa

SPERM TRANSPORT FERTILIZATION AND PREIMPLANTATION Loss 125

Taste |. Litters from BALB/c females artificially inseminated with PHH,
PHL, or combinations of PHH and PHL spermatozoa.*

Single inseminations Double inseminations
Single Mixed Totals from
PHH PHL PHH PEE PHH PHL PHL
sperm sperm sperm sperm sperm sperm Mixed sperm
Litters/ 2 @
inseminated 5/24 3/8 0/90 13/90 — 9/90 — 9/90 22/90
Number of
mice born 25 12 eae 60 10 5) 65 115
Mice per litter 5.0 4.0 se 4.6 He 6.1 Ip Sy
Range of litter size 1-10 1-8 Pe 1-10 1-2 2-9 3-10

* Only live progeny were counted.
Inseminations using hyaluronidase and those with time interval between inseminations ex-
ceeding 67 minutes are not included in the results from double inseminations.

and in the subgroup given spermatozoa from both strains. In these insemina-
tions PHH spermatozoa from a single source in approximately equal num-
bers were inseminated into the females given only PHH spermatozoa, as were
inseminated into the females given both strains of spermatozoa. Results
were: three of the four females receiving only PHH spermatozoa had litters;
mean litter size was 2.7 with a range of 1 to 4. Three of the five females
receiving spermatozoa from both strains had litters, but only one of these
had progeny from PHH spermatozoa. This litter had one mouse from PHH
spermatozoa. The work of Weir (1962) showing an enhanced superiority
of PHL over PHH spermatozoa in mixed litters is supported. The present
data are not statistically significant, but consistent. There were more progeny
per litter from PHH spermatozoa and a larger proportion of inseminated
females having progeny from PHH spermatozoa in single than in double
inseminations. If the inferiority of PHH to PHL spermatozoa is in fact
accentuated when both strains of spermatozoa are inseminated, the difference
between the strains could be expressed between ejaculation and fertilization,
or it could result from greater loss of ova fertilized by PHH spermatozoa in
mixed litters than in litters from PHH spermatozoa alone. It might be
argued that much of this loss occurs after implantation due to crowding in
mixed litters. Table 3 shows post-implantation losses obtained by comparing
litter size with number of implantation scars. Unfortunately, counting of
implantation scars was confined to the latter part of the investigation. The
data indicate that loss of zygotes from PHH spermatozoa is no greater than
loss from PHL spermatozoa. Also, the presence of zygotes from PHH
spermatozoa in litters from double inseminations does not seem to increase
post-mplantation loss appreciably. Greater loss from crowding might be
expected in the mixed litters since litter sizes tended to be higher. There is

126 Tue UNiversiry ScIENCE BULLETIN

Tasce 2. Litters from artificial insemination of BALB/c females with spermatozoa
from PHH and PHL males.*

PHH spermatozoa

not precapacitated PHH spermatozoa precapacitated
PHH sperm
inseminated
Mixed first. 8-21 Mixed PHH sperm inseminated first
insemi- min. insemi- 8-21 min. 38-67 min. 69-78 min.
nation interval nation interval interval interval TOTAL
Litters/ 9 2
inseminated 1/14 2/8 2/16 1/4 12/31 7/16** 25/89
Litters with
mice from
PHH sperm/ 2 2
inseminated 1/14 2/8 1/16 1/4 3/31 6/16 14/89
Live progeny
from PHH
sperm/total
live progeny 1/3 Balle: 2/12 1/8 3/67 9/33 18/135

* Data shown for the 38-67 minute interval and the 69-78 minute interval were obtained using
the same extender (9.59% milk with antibiotics and ascorbic acid added).
Results from inseminations utilizing hyaluronidase are not included.
** Includes one litter consisting of progeny from PHH spermatozoa only. In every other case
tabulated litters having progeny from PHH also had progeny from PHL spermatozoa.

no compelling reason to believe that the slightly higher post-implantation loss
in mixed litters, compared to litters from single inseminations, was due to
differential loss of embryos. Alternative explanations for low effectiveness of
PHH spermatozoa include later arrival of PHH spermatozoa at the surface
of the zona pellucida or a slower maturation of PHH than PHL spermatozoa.
However, comparison of litters from inseminations with PHH spermatozoa
that were precapacitated (spent some time in a uterus prior to insemination)
with those from PHH spermatozoa not so treated showed no effect from
precapacitation (see Table 2). Insemination of PHH spermatozoa 38 to 67
minutes before injection of PHL spermatozoa actually seemed to reduce the
effectiveness of PHH spermatozoa. Possible explanations are: (1) If PHH
spermatozoa were more effective in mixed inseminations than with the 38-67
minute head start, the difference might be attributed partly to a difference
in extenders. Not many of the mixed inseminations were done with ascorbic
acid in the milk. Ascorbic acid may cause a higher percentage of conception
from PHL spermatozoa. Either it does not help PHH spermatozoa or its
influence was offset by a seasonal effect. (2) At the longer intervals the
PHH spermatozoa did not have the benefit of the copulation and plug nearly
as soon after insemination as did the PHL spermatozoa. Copulation stimu-
lates uterine contraction and probably speeds transport of spermatozoa, as
demonstrated for the cow by Van Demark and Hays (1952). Spermatozoa

SpeRM TRANSPORT FERTILIZATION AND PREIMPLANTATION Loss 127

Taste 3. Loss of implanted fetuses as determined by counts of implantation scars.

Double inseminations (both kinds of sperm) Single inseminations

Litters containing progeny from:

both PHH PHL PHH PHE
strains sperm only — sperm only sperm sperm
Mean
number last ......... ier at 2.9 0 25 13 i 7/
RITES 3 epee ee pe ee 1-6 0-6 0-5 1-2
iD top 9 2 8 41 Bie
Mean litter size
ivier-t- dead) So 6.1 1.0 ae) 5.0 6.5
Weadmatebinthy 2 cree. 0.1 0 0.1 Wed 2.0
Range of
ie? SVD pike Aes oe 2-10 - 2-7 1-14 5-8

* One female was noted pregnant but the litter was never found. She had one implantation scar.
Mean litter size, number dead at birth, and range of litter size are based on 2 rather than 3
individuals.

from the later insemination (PHL) are likely to be carried into the oviduct
soon after insemination. Also, the plug prevents loss of inseminated sperm
suspension due to seepage through the cervix. (3) For both mixed insemina-
tions and those in which the interval between injection of PHH and PHL
spermatozoa was 21 minutes or less, the PHL spermatozoa were held in
vitro for a longer time prior to insemination. For these inseminations the
vasa deferentia from the two strains were stripped alternately, whereas with
longer intervals PHL males were not opened until after PHH spermatozoa
had been inseminated. (4) If the ova were not released until some time after
PHH insemination, the interval between inseminations would merely allow
more time for PHH than PHL spermatozoa to deteriorate. The interval
would be a handicap to PHH spermatozoa rather than a help.

When f-amylase was included in the extender used for PHH sperma-
tozoa no improvement in performance of PHH spermatozoa was evident.
The time interval between insemination of PHH and PHL spermatozoa
ranged from 54 to 64 minutes. Of a total of nine females inseminated two
had litters. One of these consisted of only one mouse which was dead at
birth. Paternity was unknown since this cannot be determined accurately
in stillborn mice. None of the females had any known (live) progeny from
PHH spermatozoa. Although numbers of motile spermatozoa in counts
from PHH were low, progeny were produced in other inseminations with
estimates of number of motile PHH spermatozoa as low or lower. 8-amylase
has been shown to be the capacitating factor for rabbit spermatozoa (Kirton
and Hafs, 1965), so may perform a similar function in mice.

128 THe University ScIENCE BULLETIN

Results from precapacitation of PHH spermatozoa, from giving a head-
start of up to 67 minutes and from including -amylase in the suspension,
argue against the hypothesis that the superiority of PHL over PHH sperma-
tozoa in double inseminations is caused by more rapid capacitation of PHL
spermatozoa.

When the interval between inseminations was increased to 69-78 minutes,
PHH spermatozoa were more successful than when the interval was shorter
(see Table 2). A comparison of the 69-78 and 38-67 minute intervals gives

*=10.88 (P<0.001); comparing the 69-78 and 0-67 minute intervals, X°=
8.26 (P<0.01). Of the mixed inseminations and those with an 8-21 minute
interval few were done with the milk (with antibiotics and ascorbic acid)
extender. Only the inseminations utilizing this extender were included in
the comparison. The difference is not significant when the 69-78 minute
interval is compared with the 8-21 minute interval and mixed inseminations,
but the reasons given previously for poor performance of PHH in the 38-67
minute interval relative to shorter intervals and mixed inseminations are
applicable here also.

Hyaluronidase added to the portion of the extender used for PHH sper-
matozoa seemed to increase the number of successes of PHH relative to PHL
spermatozoa. The time interval between inseminations, 46-62 minutes, was
comparable to that of the 38-67 minute interval group. Results from those
inseminations in this group in which neither hyaluronidase nor the PHL
sperm extract were used are included for comparison in Table 4. Results
from inseminations that utilized hyaluronidase were compared with those
from all other double inseminations, from zero through a 67 minute interval.
Only results from inseminations in which the extender consisted of skim
milk with antibiotics and ascorbic acid were included (X°=3.65, 0.05<P-
<0.10).

Results from inseminating an extract of PHL spermatozoa along with
PHH spermatozoa are shown in Table 4. The effect, if any, was small.

Fructose was used in the extender for insemination of 10 females. With
one of the groups, a group of three females, the fructose was included only
in the extender used for PHH spermatozoa. In the remaining groups it was
used for both strains of spermatozoa. Time interval between inseminations
ranged from 44 to 96 minutes. Three of the 10 inseminated females had
litters. The progeny were: one from PHH spermatozoa, 5 from PHL. There
was no detectable enhancement of sperm motility except in one insemination,
and no litter was produced from it. Nevo (1965) found that for bull, ram,
and cock spermatozoa the respiratory rate and critical oxygen concentrations
were not appreciably altered by the presence of glucose or fructose. Al-
though additional testing might have shown a beneficial effect on PHH
spermatozoa, the effect of fructose was not striking.

SPERM TRANSPORT FERTILIZATION AND PREIMPLANTATION Loss 129

Tas_e 4. Effect of hyaluronidase and extract of PHL spermatozoa on function of
PHH spermatozoa in double inseminations.*

Time intervals

between Litters with Live progeny
PHH and PHL mice from from PHH
inseminations Litters/ 2 2 PHH sperm/ 2 @ sperm /total
in minutes inseminated inseminated live progeny
Hyaluronidase with
PHH spermatozoa ........ 46-62 6/13 4/13 5/24
Extract of PHL
spermatozoa
inseminated with
PHH spermatozoa ........ 59-69 3/2 2/9 2/24
Neither extract
nor hyaluronidase ........ 38-67 9/25 Dii25 2/47
With Without
hyaluronidase hyaluronidase
PHH progeny 5 2 ley eZ
PHL progeny 19 tS) | 64
24 47 sera
Xx°=4.91
0.02<P<0.05

* The extender consisted of skim milk with antibiotics and ascorbic acid.

Hamner and Williams (1964) reported stimulation of respiration in
rabbit, rooster, bull and human spermatozoa by bicarbonate. The concentra-
tion for maximal stimulation in rabbit spermatozoa was 2.36 x 10°M. Bicar-
bonate was placed in the extender used for the PHH spermatozoa in the
insemination of one group of females. The concentration used was 2.2 x
10°M. No litters were produced, but in three other groups of females bicar-
bonate was contained in the extender used for both PHH and PHL sperma-
tozoa. Three of the nine females had litters with one mouse from PHH
spermatozoa, 13 from PHL. The numbers were small, but no improvement
in success of PHH spermatozoa was observed. Stimulation of respiration
prior to insemination would be harmful if early expenditure of energy
shortens the period of motility. Examination of small samples of sperm
suspensions set aside before insemination with the remainder has confirmed
that PHH spermatozoa are likely to lose motility more quickly than PHL.

Since haemacytometer examination of spermatozoa showed that a con-
sistently higher proportion of viable PHH than viable PHL spermatozoa were
bent double, a comparison was made of ejaculates from PHH and PHL
males. Two nigrosin-eosin slides were made per ejaculate and two counts
were made per slide. Four ejaculates from 3 PHH males had a mean of 249%
of all spermatozoa viable and bent double behind the midpiece. The range
of counts was 12% to 3494. A mean of 2% of all spermatozoa were viable

130 Tue University ScrENcCE BULLETIN

and bent at the neck. The range of counts was zero to 5°/. Three ejaculates
from 2 PHL males had a mean of 5°% of all spermatozoa that were viable and
bent double behind the midpiece. The counts ranged from zero to 11%. A
mean of 6°% of all spermatozoa were viable and bent at the neck. The counts
ranged from 2 to 13°.

DISCUSSION

Some steps between insemination and fertilization at which PHH sperma-
tozoa might be retarded could be summarized as follows: Capacitation of
spermatozoa is believed to be a prerequisite for fertility. Also, spermatozoa
must penetrate the cumulus cells, the zona pellucida and the vitellus of the
ovum; penetration of the cumulus cells depends upon hyaluronidase carried
by the spermatozoa (Austin 196la). Penetration of the zona pellucida seems
to depend upon another enzyme carried by the spermatozoa, called “zona
lysin” by Austin and Bishop (1958). Sperm motility is required for at least
part of the penetration process. Penetration of the vitellus may or may not
depend upon motility. Blandau and Odor (1952) observed that motility was
involved in penetration of the vitellus by rat spermatozoa. Austin and Braden
(1956) described a more passive penetration by rodent spermatozoa. The
vitellus absorbed the spermatozoa.

Most of these factors operating prior to fertilization have been studied by
various modifications in the procedure for artificial insemination. The prin-
cipal reason for testing different extenders was to find one in which the PHH
spermatozoa would remain motile in large numbers for a reasonably long
period of time. Fructose was also tested for this purpose. Apparently the
only way to be sure of having enough functional PHH spermatozoa is to
inseminate them in much larger numbers. The required number may not
often have been attained. A factor that further decreases the number of
functional PHH spermatozoa is the bent condition shown by many of
them. Bishop and Walton (1960b) termed the bent tail condition (called
looped tail) a secondary abnormality. They stated that secondary abnormali-
ties consist of spermatozoa that appear to have been normally differentiated
but subsequently deformed while passing through the male reproductive
tract. Decapitate spermatozoa are also included in this category. PHH males
with large numbers of decapitate spermatozoa are occasionally found. This
condition rarely occurs in PHL males.

Capacitation does not explain the inferior performance of PHH sperma-
tozoa. PHH spermatozoa, given an opportunity to be capacitated prior to
insemination and then inseminated up to 67 minutes before PHL sperma-
tozoa, showed no improvement in performance. Braden and Austin (1954b)
suggest capacitation time to be an hour or more in the mouse.

SperM TRANSPORT FERTILIZATION AND PREIMPLANTATION Loss 131

Data from inseminations with hyaluronidase added must be interpreted
with caution. The effect of hyaluronidase needs to be tested by dividing
equally a dense suspension of spermatozoa between extender with and ex-
tender without hyaluronidase. Chang (1950) has stated that the hyaluronidase
in the spermatozoa is adequate for penetration of cumulus cells and that a
supplementary supply is not required. Leonard, Perlman and Kurzrok
(1947) found that hyaluronidase introduced into the rat uterus did not get
into the oviducts, at least not in large enough amounts to denude the ova.
However, spermatozoa might absorb some hyaluronidase from the extender.
Emmens and Swyer (1948) showed that rabbit spermatozoa reabsorbed a
significant amount of homologous hyaluronidase but not of heterologous
(bull) hyaluronidase. If mouse spermatozoa can absorb heterologous (bull)
hyaluronidase or if some could enter the oviduct, supplementary hyaluroni-
dase could facilitate penetration of cumulus cells by the PHH spermatozoa.
Penetration of cumulus cells in the mouse requires enough time to have
caused the extra delay between ovulation and fertilization in the inbred
strain with the greater density of cumulus cells (Braden 1958a, 1962). Thus
a difference in time required to penetrate the cumulus cells could be partly
responsible for the inferior performance of PHH spermatozoa.

An extract from PHL spermatozoa was added to PHH spermatozoa to
provide zona lysin as well as any unknown factors. Hyaluronidase would
also be provided by the extract. Failure to produce a striking improvement
might be because the needed factors were not provided by the extract or be-
cause concentrations carried into the oviduct were inadequate. Higher con-
centrations should be tested. In the race with PHL spermatozoa, once a
spermatozoan penetrates the zona and attaches to the vitellus, the block
against polyspermy and the zona reaction are elicited and the race is over
(Austin and Braden, 1956; Austin, 1959; Braden, Austin and David, 1954).
Hence poor competition of PHH with PHL spermatozoa can not be caused
by deficiency of a factor required for penetration of the vitellus.

Interpretation of the data requires information on timing. If mature ova
are not available until after the PHL spermatozoa are inseminated, putting
PHH spermatozoa in first would be a handicap rather than an aid, allowing
more time for deterioration without providing an effective headstart. Lin
and Bailey (1965), from an investigation not intended to establish exact
time of ovulation of BALB/c females, found that ovulation was completed
by 14 hours after the HCG injection. Ovulation could conceivably have been
completed earlier. Marston and Chang (1964) determined time of ovulation
in Swiss Webster mice. They found that mature females had completed
ovulation within 12 hours after HCG injection. Some females had ovulated
by the tenth hour. Immature females required 14 hours to complete ovula-
tion. Gates and Beatty (1954) reported that ovulation in immature hybrid

132 Tue University ScIENCE BULLETIN

mice occurred 11 to 13 hours after HCG injection. A preliminary study of
induced ovulation in BALB/c females has shown that ovulation has begun
in some females at 12 hours, but apparently not at 11 hours after HCG in-
jection. There were ova present at 11 hours, but they were denuded and most
likely were released somewhat earlier as a result of the PMS alone. Since
they were denuded, most of them were probably too old for normal fertiliza-
tion and development. Marston and Change (1964) showed that ova from
mature Swiss Webster mice were largely denuded (represented by 0.2 on a
scale of 0 to 6) by 30-31 hours after HCG, which would be some 18 hours
after ovulation. At this time there was a sharp drop in number of ova pene-
trated. Lewis and Wright (1935) reported ova denuded as early as 12 hours
after copulation.

Usually the PHH spermatozoa were inseminated more than 12 hours
after the HCG injection so that in many cases ova should already be present
in the oviducts. If maturation of ova membranes is necessary, this would
have a chance to occur while the spermatozoa are maturing and being
transported to the fertilization site. Data from Braden and Austin (1954b)
indicated that 114 to 2 hours might be required for egg maturation, but
Edwards and Gates (1959) reported that little time was needed for matura-
tion of egg membranes. According to Edwards and Gates, approximately
70 minutes elapsed between ovulation and penetration. The 70 minute in-
terval would include the time required for penetration of the cumulus cells
and the zona. Transport of mouse spermatozoa to the site of fertilization
requires 4 to 1% hours (Merton, 1939; Lewis and Wright, 1935; Braden and
Austin, 1954b; Braden, 1962). Fifteen minutes represents the minimum.
Braden and Austin (1954b) suggest 34 hour as the mean time required.
Yanagimachi (1963) has reported more rapid sperm transport after ovulation
than before ovulation in the hamster. This may be true in the mouse, also.
Since in artificial insemination the stimulus provided by the plug is delayed,
uterine contractions are unlikely to occur at the time the spermatozoa are
injected. However, injection of spermatozoa deep into the uterus may have
a compensatory effect on the time of transport.

CONCLUSIONS

PHH spermatozoa are inferior to PHL spermatozoa in ability to fertilize
ova. The reasons remain obscure, but the following indicate that PHH sper-
matozoa penetrate more slowly than do PHL spermatozoa: (1) A possible
accentuation of PHL superiority when both strains of spermatozoa are in-
seminated as compared with single inseminations. (2) Improvement of
PHH performance when the interval between PHH and PHL insemination
is increased to 69 minutes or more. (3) A possible beneficial effect of
hyaluronidase upon performance of PHH spermatozoa.

SPERM TRANSPORT FERTILIZATION AND PREIMPLANTATION Loss 133

The poor conception rate from PHH spermatozoa may be from failure
of enough of them to maintain vigorous motility for a long enough period
of time.

Ill. PREIMPLANTATION AND POST-IMPLANTATION LOSSES

Design of the experiment involving ova counts was similar to that of
Whitten and Dagg (1961). In addition to differences in counts and develop-
mental rates differences with respect to time of copulation during the diurnal
eyele could be detected, Fi(PEIL? x PHH ¢ ) and Fi(PHH? x PHL )
were used in addition to the two strains of inbred males. Since pH is
correlated with lactic acid levels, and lactic acid levels are inherited in a
manner that suggests sex linkage (Weir 1962), the Fi males were used to
test for possible sex-linked inheritance of the litter size effect.

MATERIALS AND METHODS

Single pair matings were made, PHH, PHL, and C3H/He females with
PEiteandePEle males, and PHEt and PEL females with fi(PHE? =x
PHH ¢ ) and Fi(PHH 2 x PHL ¢ ) males. The light cycle was adjusted to
19 hours of light and 5 hours of darkness with lights coming on at 3:00 p.m.
and going off at 10:00 am. Females were introduced within the two hour
interval preceding the dark phase. With few exceptions the females were
70 to 120 day old virgins that had been stored up to seven per cage. The
females were examined for vaginal plugs at 8:00 a.m., 3:00 p.m., and at
approximately 10:00 p.m. Ova were examined at intervals ranging from
about 21 to 57 hours after the estimated time of copulation for all four types
of pairings involving PHH and PHL males and females (two inter se and
two crossed). The time of mating was considered to be the midpoint of the
dark period if the plug was found at 3:00 p.m. (at the end of the dark period).
Otherwise, mating was considered to have occurred half way between the
time the plug was detected and the previous time the female was examined
for the presence of a plug. If the vulva was recorded as red or moist from
frequent intromissions by the male, and a plug was observed at the subsequent
examination, copulation was considered to have occurred within one hour.
Ova from C3H/He females were examined 36 to 54 hours after the estimated
time of copulation. Females paired with Fi males were examined for ova 27
to 57 hours after estimated time of copulation; most were examined 37 to
53 hours after copulation. The inter se pairings and reciprocal crosses of
PHH and PHL males and females were checked for loss of ova at the longer
intervals. Mean number ovulated from each strain of female examined up
to 3214 hours was compared with mean number from females examined 40
to 57% hours after copulation. There was no detectable loss at the longer

134 Tue University ScIENCE BULLETIN

time intervals. The ova were obtained by flushing approximately 0.02 cc of
Ringers solution through the oviduct. The cut for removal of the oviduct
was made near the ovarian end of the uterine horn. The oviduct with ac-
companying ovary was transferred to a depression slide containing Ringers
solution where the ovary, fat and mesentry were detached by means of micro
dissecting forceps. After the oviduct was blotted and transferred to clean
Ringers solution, the ovarian end was freed. Following a second blotting
and transfer, the end was slipped over a blunted 31 or 33 gauge needle and
the contents flushed out. A gentle rotary motion of the depression slide
facilitated settling of ova in the center of the slide. Corpora lutea were
counted under a dissecting microscope at 10 or 19.5 magnification. Ovaries
were measured to the nearest 4, millimeter and in some cases weighed. Ova
were examined at 19.5 X and classified as normal fertilized or as one of
several abnormal types. Normal cleavage was used as the criterion for fertili-
zation of ova when 27 hours or more had elapsed since estimated time of
copulation. In many cases normally cleaved zygotes were found at the shorter
intervals, such as 23 hours. If none were cleaved, the ova were not classified
unless they were all abnormal or unless two polar bodies could be clearly
seen on the normal one-celled ova. Occasionally a higher magnification,
60 X, was used. Zygotes with too many cells to be counted accurately were
transferred to a microscope slide, stained, and examined at 100 X.

Corpora lutea, fetuses, and resorption sites were counted at 14 to 17 days
gestation. In most cases the examination was made 15 to 16 days after forma-
tion of the plug. Biggers, Curnow, Finn and McLaren (1963) found that
corpora lutea begin to regress at about the 17th day.

C3H/He females were paired with PHL and PHH males, one per male,
and examined daily until plugs were found. Corpora lutea were examined at
19.5 X magnification.

RESULTS

MALE EFFECTS ON NUMBER OF NORMAL zZycoTEs Results from ova counts are
shown in Table 5. There was a consistent but nonsignificant difference in
number of normal zygotes attributable to the two inbred strains of males,
PHH and PHL. No such difference was detected from the two types of Fi
males. The difference can be partly accounted for by the difference in number
of fragmented ova. The difference in number of fragmented ova was largely
attributable to a few PHH males. Females with which they copulated had a
large proportion of fragmented ova and hence few normal zygotes. When
all normal zygote counts of 3 or fewer are omitted, the difference between
mean numbers of normal zygotes attributable to the strain of male is reduced
from 1.5 to 0.8 for PHH females, from 1.0 to 0.7 for PHL females, and from
1.3 to 0.6 for C3H/He females. There were fewer small zygote counts fol-

SPERM TRANSPORT FERTILIZATION AND PREIMPLANTATION Loss 135

Tase 5. Classification of ova from inter se matings and crosses.

Matings Mean ova counts per litter
Ab-
normal*
zona Normal
Kind Normal Frag- Degen- pel- unferti- Im-

Q fof Number fertilized mented erate lucida lized mature TOTAL
PHH X PHH 22 O==0F79 1.2 0.1 1.2 0.3 0.1 7 Neste)
PHH X PHL 25 10.70.57 0.1 0.5 0.9 0.1 01047 223-0140
PHD XxX PHEL 24 9.5+0.61 0.6 0.3 0.5 0.3 0.2 4037
PHL X PHL aA 10.50.49 0.2 0.2 0.4 0.1 0.1 == O23 iL
C3H/He X PHH 18 8.10.60 0.7 0.5 0.3 0.3 0.1 9.9>=0.27
C3H/He X PHL 16 9.40.37 0 0.4 0.3 0.1 0 NOLZE= O29
PHH X F,(HxL) 10 ies (OS) 0 0.2 0.5 0 0.2 12.30.45
PHH X Fi(LxH) 8 NL SEO45 0.1 0.9 0.1 0 0.2 12.60.50
PHL X F:(HxL) 12 10.60.26 0 0.7 0.2 0 0 11.50.40
PHL X F:(LxH) 11 10.6£0.45 0.2 0.3 0.2 0 0 eS == 04

* Empty zona pellucida, zygote with one or more cells outside zona pellucida, or ovum or zygote
lacking zona pellucida.

lowing copulation with PHL compared to PHH males. The counts were
not included unless there was at least one normal zygote. This precaution was
necessary since the purpose of the study was to attempt to isolate causes of
smaller litter size coming from PHH sires. If no ova were fertilized obvi-
ously there would be no litter. It was found that fragmented ova occurred
following copulation with a larger proportion of PHH than of PHL males.
The proportion of PHH males was 23/36 and of PHL males 7/29. With 10
of the PHH males fragmented ova occurred following more than one copu-
lation. Several (six) of those for which only one litter containing frag-
mented ova was recorded were checked only the one time. Fragmented
ova occurred following more than one copulation with only one of the PHL
males. Two of the other six were checked only once, so there was no op-
portunity for more than one group of ova containing fragmented ones.
Possible causes of fragmentation include late fertilization (Austin, 1961b;
Marston and Chang, 1964) and aging of unfertilized ova (Marston and
Chang, 1964). PHH males are not likely to wait until ova have aged before
copulating even though they will mate with females in late estrus (Weir,
1962). Probably most or all of the fragmented ova were aged unfertilized
ova. Spermatozoa counts were made on ejaculates from 6 PHH males that
seemed to be responsible for fragmented ova in females with which they
copulated. Two of the males, from which one ejaculate each was examined,
transferred no spermatozoa to the female. A third male had a low sperm
count (675,000). All three of these males were fertile enough to sire litters

136 Tue University SCIENCE BULLETIN

although one was kept with a hybrid female for several months and should
have sired more than one litter if fertility had been normal. The incidence
of litter sizes of 4 or less was rather high among these males (4 of 14 litters).
The data show that there can be a sharp fluctuation in spermatozoa counts
(from none to enough to sire a litter). Also, failure to fertilize all the ova
may not be caused by low spermatozoa counts alone. Two males had sperma-
tozoa counts in the upper end of the range for PHH males. (The highest
count that has been found was 11,700,000.)

If the occasional litter with several unfertilized (including fragmented)
ova and few normal fertilized zygotes is in part or largely responsible for
small litters from PHH males, one would expect to find much of the differ-
ence caused by a few small litters. Examination of Weir’s data (1960, 1962)
shows this to be the case. Omitting litters of 3 or fewer reduces the difference
in litter size attributable to the strain of male from 0.6 to 0.2 for K females,
and from 1.7 to 1.2 for C3H females. Ten of 145 litters from K females and
PHH males had 3 or fewer mice while only one of 102 litters from K
females and PHL males had 3 or fewer mice.

According to Beck (1957) more PHH than PHL progeny are lost in the
last trimester and at parturition, but these were from inter se matings, so
maternal effects cannot be separated from effects acting through the zygotes.

ADDITIONAL STRAIN COMPARISONS. There was no detectable difference due
to strain of male in developmental rates in any of the combinations (PHH &
x PHH ¢, PHH ° x PHL 6, PHL? x. PHin; PH oP ree an
fe? x PME, Cire 2. x. Pres.)

There seems to be a consistent tendency for PHH females, compared to
PHL and C3H/He, to release more ova (see Table 5), but the difference is
small. Also, PHH females compared to PHL tend to have more ova that
are broken out of the zona pellucida, but the difference is not statistically
significant.

Examination of the corpora lutea showed that there were many cases in
which corpora lutea were not distinct enough to determine the number of
ova released. Even when they were clearly distinguishable there was not
always a one-to-one relationship with ova count. Although corpora lutea
were often indistinct in females examined 29 to 57 hours after estimated time
of mating, they were even less often distinct at shorter time intervals. Con-
sidering only the counts taken during the 29 to 57 hour interval, of the 71
counts where ova were more numerous than corpora lutea 49 differed by
only one. Of the 27 cases in which the number of corpora lutea exceeded the
number of ova, 19 differed by only one. Of the 61 counts showing an equal
number of ova and corpora lutea 11 cases seemed to indicate transmigration
of blastocysts from one uterine horn to the other. McLaren and Michie

SpERM TRANSPORT FERTILIZATION AND PREIMPLANTATION Loss 137

(1954) have presented some evidence for crossing of blastocysts from one
side to the other.

Possible causes for discrepancy between counts of corpora lutea and of
ova include failure to distinguish partially confluent adjacent ones (Falconer
and Roberts, 1960), delayed luteinization, and, rarely, polyovular follicles
(Engle, 1927). Also, there may be a loss of ova (particularly at the longer
time intervals) or follicles that have no ovum. Packs of cumulus cells without
ova were quite frequently seen, especially when counts were made shortly
after copulation. Since comparison of ova counts up to 3244 hours after
copulation with those 40 to 57'4 hours after gave no indication of loss of ova
at the longer intervals, losses were small if they occurred.

A difference between PHH and PHL females with respect to the portion
of the light cycle during which they will copulate has been observed. AI-
though the X? test for independence does not show statistical significance
using light versus dark, inclusion of females that copulated shortly after
the start of the light period with those that copulated during the dark gives
a significant difference (X71=10.51, P<0.01). PHH females had a greater
tendency than PHL to copulate during and shortly after the dark period.
Females having a wet and red vulva at the end of the dark period (3:00 p.m.)
and a plug when examined at 10:00 p.m. were assumed to have copulated
shortly after the dark period ended. If only the first series of males (seven
from each strain) is considered, a significant difference at the 5°% level is
observed in time of mating of males of the two strains (X71=4.42). For the
entire group of males, however, the difference is not significant. The greater
tendency shown by PHH males to copulate during the dark may be due to
their more aggressive mating behavior. They may copulate with those females
that are still somewhat refractory during the dark period, whereas PHL males
may delay until these females are in optimum estrus.

The elapsed time preceding copulation was shorter and more regular
with PHH than with PHL females. A X° test for independence shows sta-
tistical significance when the number of copulations occurring through the
fourth day is compared with those occurring five or more days after pairing
(X*;=5.33, P<0.05). The difference between strains is even greater if
account is taken of the number of females not copulating when paired with
males for a period of time in excess of six days (X71=9.97, P<0.01). The
longest time required for a plug was 12 days for PHL females and eight days
for PHH. C3H/He females were not observed concurrently with the other
two strains, but the data available show they copulate sooner after pairing
than do PHL females (X71=5.07, P<0.05).

Feta counts. Results from fetal counts are shown in Table 6. Corpora
lutea were larger, and in most cases more distinct than at preimplantation
stages. All but one of 17 resorptions from PHH males and all but one of

138 Tue Universiry SciENCE BULLETIN

TasLe 6. Fetuses and corpora lutea from C3H/He females mated to PHH and

PHE males:
Number Corpora Preimplan-
of lutea Fetuses Resorptions tation
litters (mean) (mean) (mean) loss
PHH males 21 9.80.35 7.60.42 0.8 eat
PHL males 20 9°4==0:34 8.20.36 0.4 0.8

9 from PHL males were very early post-implantation losses—prior to forma-
tion of the placenta. In one litter from a PHH sire the preimplantation loss
was 7 out of 10. When this litter is omitted, the mean preimplantation loss
from PHH sires was 1.1. All of the eleven PHH males used were known
to be fertile. The several that did not sire litters or fertilize ova were not
used. Several of the PHL males were not tested in advance for fertility, but
only one of 20 females mated to PHL males failed to conceive from the first
copulation. Eight of 22 females with PHH males failed to conceive from
the first copulation. In two cases sterile copulations occurred twice before a
fertile mating.

DISCUSSION

The data from fetal counts and ova counts support the hypothesis that
much of the difference in litter size from PHH and PHL sires is attributable
to a few litters in which PHH spermatozoa do not fertilize all of the ova.
Also, there may be some difference in the number of resorptions.

Krzanowska (1964b) found a longer time interval between pairing and
copulation in females of two inbred strains than in females of their Fi hybrid.
The data from C3H/He, PHH and PHL females indicate that some strains
have been affected more than others by inbreeding in this respect.

IV. SPERM NUMBER, MORPHOLOGY AND MOTILITY

Differential staining of spermatozoa has been used to determine per-
centage of viability by Lasley, Easley and McKenzie (1942), Hancock (1951),
and Beatty and Sharma (1960), among others. Beatty (1961) has stated that
one of the best guides to fertility of a semen sample is the percentage of un-
stained sperm cells in a nigrosin-eosin preparation. Beatty and Sharma
(1960) reported marked differences among strains for head breadth and area
and for midpiece length. Braden (1958b) showed a strain difference in shape
at the posterior end of the sperm head, the head tending to be angular in
C57BL males, but smoothly rounded or slightly flattened in CBA males.

MATERIALS AND METHODS

Sperm from the vasa deferentia were obtained by stripping or flushing
them out. After nigrosin-eosin slides were made, samples were taken for

SPERM TRANSPORT FERTILIZATION AND PREIMPLANTATION Loss 139

haemacytometer counts. Dimensions of spermatozoa on the nigrosin-eosin
slides were obtained by means of an occular micrometer at a magnification
of 970. In addition, spermatozoa from one male of each strain were compared
by means of a microprojector. For preparation of nigrosin-eosin slides the
suspension was spread immediately after a thorough mixing which required
less than 30 seconds. The excess at one end of the slide was blotted, followed
by air drying at 24°C. Venetian turpentine and a cover glass were then
applied. Morphology of spermatozoa was studied by determining the per-
centage of spermatozoa having abnormally shaped heads and by taking
measurements of the head and midpiece. Abnormalities of the head in-
cluded missing acrosome, gross distortion of the shape, and cases in which
the tip of the acrosome was bent back toward the base of the head.

To compare normal ejaculates, containing spermatozoa from the epi-
didymis as well as from the vas deferens, ejaculated spermatozoa from the two
strains were obtained. More than one ejaculate was obtained from some of
the males to determine variation within males. There were 8 ejaculates
from 4 PHH males and 7 from 4 PHL males. A second series of counts was
made a year later at about the same time of year—November and December.
Ages ranged from 191 to 335 days for PHL and from 178 to 311 for PHH
males. Females in induced estrus were placed with males 11 hours after
the HCG injection. Within 14 to 2 hours after copulation the female was
killed, the entire reproductive tract removed, the uterus cut just above the
plug, and the uterine contents flushed out by 1.0 ml physiological saline per
uterine horn. For 16 out of 23 counts the time interval between copulation
and examination did not exceed 40 minutes. In some cases 0.5 ml of
physiological saline, followed by 0.5 ml of sodium citrate, per horn, were
used instead of saline alone. Fluid remaining after flushing was expressed by
means of forceps. After the nigrosin-eosin slides were prepared, live sper-
matozoa remaining in the solution were killed with a drop of formaldehyde.
Two samples were taken for haemacytometer counts. For each sample sper-
matozoa were counted in each of five squares. After uterine spermatozoa
were counted, oviducts were examined for spermatozoa up to 11 hours after
copulation. Oviduct contents were obtained by flushing. The procedure was
the same as that used to obtain ova. After the cumulus cells and ova had
settled in the center of the depression slide, they were transferred by means
of a narrow bore pipette to a microscope slide. A cover glass with a thin
rim of vaseline all the way around was placed over the transferred material.
Spermatozoa were examined by means of a phase contrast microscope and
classified as normal or abnormal with respect to head morphology. In a later
study of the proportion of bent (looped) spermatozoa entering the oviduct
the spermatozoa were flushed directly onto the microscope slide.

140 Tue Universiry ScIENCE BULLETIN

RESULTS

The mean sperm count from vasa deferentia of eleven PHL males was
2,296,000 and that from eight PHH males was 1,597,000. The difference is
not statistically significant. Counts of spermatozoa from the first series of
ejaculates ranged from 1,750,000 to 11,700,000 per ejaculate for PHH males
and from 250,000 to 8,850,000 for PHL males. For the second series counts
for 3 ejaculates from 2 PHL males ranged from 2,875,000 to 8,250,000. Counts
from two PHH males were 675,000 and none. One male from each strain
in the first series had no spermatozoa in the ejaculate. No significant differ-
ences between PHH and PHL were detected. There were large variations
within strains and even within the same male. Counts of the three ejaculates
from the PHL male with the lowest count were 3,100,000; 250,000; and
3,000,000. No male was used twice within the same week, and the low count
from this particular male came nearly 3 weeks after the last previous
ejaculation.

The males used for ejaculates were tested for fertility before counts were
made. In most cases fertilized ova were recovered from females following
copulation, but some males had sired litters prior to the time the counts were
made. Subsequently most males sired litters. The only exceptions were the
two males in the first series (one PHH and one PHL) that had no sperma-
tozoa in their ejaculates. The PHH males in the second series fertilized some
but not all the ova when they were tested for fertility. Both males sired
litters subsequent to the sperm counts, however.

The only measurement that showed a tendency toward difference between
strains was the percentage of live spermatozoa (PHH 3043.3, PHL 445.1
for spermatozoa from the vas deferens). Also, on numerous occasions when
percentages of motile spermatozoa were estimated from haemacytometer
counts the PHL males consistently provided a larger fraction of motile
spermatozoa than comparable samples from PHH, and motility of PHL
spermatozoa was more vigorous than that of PHH spermatozoa. Nigrosin-
eosin stains made 4 and 8 hours after removal of spermatozoa from the vasa
deferentia showed a higher percentage of viability of PHL compared to
PHH spermatozoa. The difference seemed to be accentuated at these longer
time intervals. The percentages of viable spermatozoa from the first series
of ejaculates were 70 for PHH (range 43 to 84) and 77 for PHL (range 57
to 89).

Head and midpiece dimensions were the same for the two strains. There
Was no consistent difference in shape of the posterior end of the head. Seem-
ingly normal spermatozoa from both strains have several variations in shape.

Percentages of morphologically abnormal spermatozoa for 8 ejaculates
from PHH males were 41 (range 28 to 49) from the uterus and 17 (range
10 to 43) from the oviducts. Percentages for 7 ejaculates from PHL males

SPERM TRANSPORT FERTILIZATION AND PREIMPLANTATION Loss 141

were 45 (range 35 to 53) from the uterus and 20 (range 6 to 33) from the
oviducts. Over half of the abnormal spermatozoa from some oviducts were
still motile when examined. In most instances there were a number of sper-
matozoa which could not be classified as to morphology. Either the head was
embedded in cumulus cells or it was turned sideways. Many of the sperma-
tozoa that were embedded in cumulus cells were motile and omitting them
may have led to an overestimate of percent abnormal. A lapse of time be-
tween flushing and classification might have the same effect, particularly
since absence of acrosome was considered an abnormal condition.

Bent spermatozoa were seen in oviducts from all four females in which
oviducal spermatozoa were found. In one female there were enough sper-
matozoa that a meaningful comparison could be made between percent of
bent sperm in the uterus and in the oviducts. The ejaculate came from a
PHL male. The average value for uterine spermatozoa that were bent at the
posterior end of the midpiece was 6%. An average of 99% were bent at the
neck immediately posterior to the head. These values were obtained from
two nigrosin-eosin-stained smears and assumed that no bent spermatozoa
straighten (or straight ones bend) as the smear is made. Corresponding
values for the oviducts were 15/159=99% bent at the midpiece, and 3/159=
2°%, bent at the neck.

In a study of ejaculates from males that lacked seminal vesicles sperma-
tozoa were found in the oviduct when there were only 10,000 in the uterus.
Perhaps this would be possible even with lower counts.

DISCUSSION

Krzanowska (1962) found differences in morphology and sperm counts
among inbred, crossbred, and outbred males that differed in fertility, but
no difference in percent of viable spermatozoa from the vasa deferentia.
The situation seems to be different in PHH and PHL males. PHL com-
pared to PHH males have a higher percentage of viable spermatozoa from
vasa deferentia, but apparently do not differ in percentage of abnormal sper-
matozoa or sperm counts.

A study of the effect of various “t” alleles on fertility by Braden and
Gluecksohn-Waelsch (1958) revealed no correlation between infertility and
proportion of morphologically abnormal spermatozoa. Although many
normal spermatozoa were ejaculated by infertile males, none of these sperma-
tozoa were found in the distended portion of the ampulla (site of fertiliza-
tion). Krzanowska (1962) reported that few abnormal spermatozoa traverse
the mouse uterotubal junction and the abnormality of those that do is
usually of rather mild form, affecting only the end of the acrosome. She
suggested that some morphologically normal spermatozoa are not capable
of fertilization. It seems safe to conclude from the data presented here that

142 Tue UNiversiry ScIENCE BULLETIN

reduced fertility of PHH spermatozoa as compared with PHL is not caused
by a proportionally higher number of morphologically abnormal spermatozoa
ejaculated and transported to the site of fertilization. The data show that
the uterotubal junction admits a lower proportion of morphologically ab-
normal than of normal spermatozoa. The values for percentage of abnormal
spermatozoa are useful for comparing proportion of abnormal spermatozoa
delivered to the uterus and to the oviducts, but the age range of males for
both strains included ages at which, according to Beatty and Mukherjee
(1963), there is an increase in proportion of spermatozoa with abnormal
acrosome caps. Hence, the values cannot be compared with those found for
other strains. Bent sperm seem capable of entering the oviduct. Viability of
spermatozoa from the vas deferens is lower for PHH than for PHL males.

V. SUMMARY AND CONCLUSIONS

PHH spermatozoa fertilize relatively few of the available ova following
artificial insemination. Competition with PHL spermatozoa seems to make
PHH spermatozoa even less effective. The number of fertile inseminations
and number of progeny per litter from PHH spermatozoa is higher in
single than in double inseminations. Capacitation time seems not to be
the reason. Slow penetration, particularly of the cumulus cells, may be
partly responsible because: (1) the effectiveness of PHH spermatozoa is
less when in competition with PHL; (2) PHH spermatozoa are more often
successful when the interval between PHH and PHL inseminations is 69
minutes or more; (3) the effectiveness of PHH spermatozoa is slightly
improved when hyaluronidase is injected with them. Compared with PHL
spermatozoa a lower proportion of PHH spermatozoa (from the vas de-
ferens) are viable and more of the viable spermatozoa are bent double.

Some ejaculates from PHH males fertilize only a few of the available
ova. This may account for at least half the reduction in size of litters sired
by PHH males. The effect of the low zygote counts on the mean number
of normal zygotes is equivalent to the effect of small litters on litter size
and on number of fetuses at 14 to 17 days gestation. More PHH males com-
pared with PHL males have the tendency to fertilize only a few of the ova,
and PHH males frequently fertilize none of the ova.

Failure of normal PHH spermatozoa to be transported to the oviduct does
not seem to be responsible for the low fertility of PHH spermatozoa.

Since the function of PHH spermatozoa seems to be impaired and the
percentage of normal, motile spermatozoa is low the variation among ejacu-
lates, particularly the variation in number of sperm, may be sufficiently large
to cause the occasional small litter and the more frequent sterile copulation.
The required threshold for number of functional PHH spermatozoa is not
always attained.

SPERM TRANSPORT FERTILIZATION AND PREIMPLANTATION Loss 143

LITERATURE, CITED

Austin, C. R. 1959. Fertilization and development of the egg. Reproduction in Domestic
Animals. Eds. Cole, H. H. and Cupps, P. T. Academic Press, New York, p. 399.

——. 196la. Proc. 4th Intern. Congr. Anim. Reprod. The Hague 4:723. Not seen. Quoted
in Mann, T. 1964. The Biochemistry of Semen and of the Male Reproductive Tract.
Wiley, New York, p. 138.

——. 1961b. The Mammalian Egg. Charles C. Thomas, Springfield, Ill. p. 85.

— , and M. W. H. Bisuop. 1958. Role of the rodent acrosome and perforatorium in fertiliza-
tion. Proc. Roy. Soc. London B, 149:241-248.

— , and A. W. H. Brapen. 1956. Early reactions of the rodent egg to spermatozoon pene-
tration. J. Exper. Biol. 33:358-365.

Baker, R. D. 1962. Effect of sperm numbers on fertility in two inbred strains of mice and their
F, hybrid. Summer progress report. The Jackson Laboratory Bar Harbor, Maine.

Bearry, R. A. 1960. Fertility of mixed semen from different rabbits. J. Reprod. Fert. 1:52-60.

. 1961. Genetics of mammalian gametes. Animal Breeding Abstracts, 29:243-256.

, and D. P. Muxueryee. 1963. Spermatozoan characteristics in mice of different ages.
J. Reprod. Fert. 6:261-268.

, and K. N. SHarma. 1960. Genetics of gametes. III. Strain differences in spermatozoa
from eight inbred strains of mice. Proc. Roy. Soc. Edin. B, 68:25-53.

Beck, S. L. 1957. Prenatal sex ratio and mortality in three lines of mice. M.S. Dissertation.

University of Kansas Library. Lawrence, Kansas.

Buatracuarya, B. C. 1962. Die verschieden Sedimentationsgeschwindigkeit der X- and Y-
Spermien und die Frage der willktrlichen Geschlechtsbestimmung. Zeitschrift fur
wissenschaftliche Zoologie 166:203-250.

BicceERs, J. D., R. N. Curnow, C. A. Finn and A. McLaren. 1963. Regulation of the gestation
period in mice. J. Reprod. Fert. 6:125-138.

Bishop, M. W. H. and A. Warton. 1960a. Metabolism and motility of mammalian spermatozoa.
Marshall’s Physiology of Reproduction, I1(2). Ed. A. S. Parkes. Longmans Green, Lon-
don, pp. 264-309.

.1960b. Spermatogenesis and the structure of mammalian spermatozoa. Marshall’s
Physiology of Reproduction, 1(2). Ed. A. S. Parkes. Longmans Green, London, pp.
1-129.

Bianpav, R. J. and D. L. Opor. 1952. Observations on sperm penetration into the ooplasm and

changes in the cytoplasmic components of the fertilizing spermatozoon in rat ova. Fertil.
and Steril. 3:13-26.

Brapen, A. W. H. 1958a. Variation between strains of mice in phenomena associated with
sperm penetration and fertilization. J. Genetics, 56:37-47.

.1958b. Strain differences in the morphology of the gametes of the mouse. Aust. J.
Biol. Sci. 12:65-71.

. 1962. Spermatozoon penetration and fertilization in the mouse. Symposia Genetica et
Biologica Italica, 9:94-101.

,and C. R. Austin. 1954a. The number of sperms about the eggs in mammals and its
significance for normal fertilization. Aust. J. Biol. Sci. 7:543-551.

. 1954b. Fertilization of the mouse egg and the effect of delayed coitus and of hot shock
treatment. Aust. J. Biol. Sci. 7:552-565.

,C. R. Austin and H. A. Davip. 1954. The reaction of the zona pellucida to sperm
penetration. Aust. J. Biol. Sci. 7:391-409.

,and S. GLUECKSOHN-WAELscH. 1958. Further studies of the effect of the T locus in
the house mouse on male fertility. J. Exp. Zool. 138:431-452.

Cuane, M. C. 1950. Fertilization, male infertility and hyaluronidase. Ann. N.Y. Acad. Sci.
522 1192-1195:

. 1958. Fertilizing capacity of spermatozoa. Recent Progress in the Endocrinology of
Reproduction. Ed. C. W. Lloyd. Academic Press, New York, pp. 131-165.

Cross, B. A. 1958. Hypothalamic influences on sperm transport in the male and female genital
tract. Recent Progress in the Endocrinology of Reproduction. Ed. C. W. Lloyd.
Academic Press, New York, pp. 167-177.

Dziux, P. J. and M. N. Runner. 1960. Recovery of blastocysts and induction of implantation
following artificial insemination of immature mice. J. Reprod. Fert. 1:321-331.
Epwarps, R. G. 1955. Selective fertilization following the use of sperm mixtures in the mouse.

Nature Lond. 176:215-216.

, and A. H. Gares. 1959. Timing of the stages of the maturation divisions, ovulation,

fertilization and the first cleavage of eggs of adult mice treated with gonadotrophins.

J. Endocrin. 18:292-304.

144 Tue University SciENCE BULLETIN

EmMens, C. W. and G. I. M. Swyer. 1948. Observations on the motility of rabbit spermatozoa
in dilute suspension. J. Gen. Physiol. 32:121-138.

EnotE, E. T. 1927. Polyovular follicles and polynuclear ova in the mouse. Anat. Rec. 35:341-
343.

Fatconer, D. S. 1960. The genetics of litter size in mice. J. Cellular Comp. Physiol. 56

(suppl. 1):153-167.

, and R. C. Roserrs. 1960. Effect of inbreeding on ovulation rate and foetal mortality

in mice. Genet. Res. Camb. 1:422-430.

Finn, C. A. 1964. Influence of the male on litter size in mice. J. Reprod. Fert. 7:107-111.

Fox, R. R. and J. F. Burpicx. 1963. Preservation of rabbit spermatozoa: ethylene glycol vs.
glycerol for frozen semen. Proc. Soc. Exper. Biol. Med. 113:853-856.

Gates, A. H. and R. A. Bearry. 1954. Independence of delayed fertilization and spontaneous
triploidy in mouse embryos. Nature Lond. 174:356.

Hamner, C. E. and W. L. WituiaMs. 1964. Identification of sperm stimulating factor of rabbit
oviduct fluid. Proc. Soc. Exper. Biol. Med. 117:240-243.

Hancock, J. L. 1951. A straining technique for the study of temperature shock in semen.
Nature Lond. 167:323-324.

Kirton, K. T. and H. D. Hars. 1965. Sperm capacitation by uterine fluid or beta-amylase in
vitro. Science, 150:618-619.

Krzanowska, H. 1960. Studies on heterosis II. Fertilization rate in inbred lines of mice and
their crosses. Folia Biologica, 8:269-279.

—. 1961. Zagadnienie zwiekszonej plodnosci mieszancow. Roczniki Nauk Roln. t.
78-B-1:97-114.

—. 1962. Sperm quantity and quality in inbred lines of mice and their crosses. Acta
Biologica Cracoviensia (s. Zoologia) 5:279-290.

—. 1964a. Time interval between copulation and fertilization in inbred lines of mice and
their crosses. Folia Biologica 12:23 1-244.

—. 1964b. Studies on heterosis HI. The course of the sexual cycle and the establishment of
pregnancy in mice, as affected by the type of mating. Folia Biologica, 12:415-426.

Lastey, J. F., G. T. EAstey, and F. F. McKenzie. 1942. A staining method for the differentia-
tion of live and dead spermatozoa. Anat. Rec. 82:167-174.

LronarD, S. L., P. L. PErLMAN and R. Kurzrox. 1947. Relation between time of fertilization
and follicle cell dispersal in rat ova. Proc. Soc. Exp. Biol. Med. 66:517-518.

Lewis, W. H. and E. S. Wright. 1935. On the early development of the mouse egg. Contri-
butions to Embryology. Carnegie Institution of Washington, 25 (number 144 to 151):
113-144.

Lin, T. P. and D. W. BartLey. 1965. Difference between two inbred strains of mice in ovulatory
response to repeated administration of gonadotrophins. J. Reprod. Fert. 10:253-259.

Lyon, M. F. 1959. Some evidence concerning the “mutational load” in inbred strains of mice.
Heredity, 13:341-352.

McCarthy, J. C. 1965. The effect on litter size of crossing inbred strains of mice. Genetics,
Sule ZW-22 20

McLaren, A. and D. Micure. 1954. Transmigration of unborn mice. Nature Lond. 174:844.

Marston, J. H. and M. C. Cuane. 1964. The fertilizable life of ova and their morphology fol-
lowing delayed insemination in mature and immature mice. J. Exp. Zool. 155:237-252.

Merron, H. 1939. Studies on reproduction in the albino mouse. III. The duration of life of
spermatozoa in the female reproductive tract. Proc. Roy. Soc. Edin. B, 59:207-218.

Nevo, A. C. 1965. Dependence of sperm motility and respiration on oxygen concentration. J.
Reprod. Fert. 9:103-107.

SNELL, G. D., K. P. HuMMEL and W. H. ApeLtMann. 1944. A technique for the artificial in-
semination of mice. Anat. Rec. 90:243-253.

SouTHarD, J. L., H. G. Worre and E, S. Russety. 1965. Artificial insemination of dystrophic
mice with mixtures of spermatozoa. Nature Lond. 208:1126.

Van Demark, N. L. and R. L. Hays. 1952. Uterine motility responses to mating. Am. J.
Physiol. 170:518-521.

Weir, J. A. 1960. A sex ratio factor in the house mouse that is transmitted by the male.

Genetics, 45:1539-1552.

. 1962. Hereditary and environmental influences on the sex ratio of PHH and PHL mice.

Genetics, 47 :881-897.

Wuirten, W. K. and C. P. Dace. 1961. Influence of spermatozoa on the cleavage rate of
mouse eggs. J. Exp. Zool. 148:173-183.

YANAGIMACHI, R. 1963. Sperm ascent through the oviduct of the hamster and rabbit in relation
to the time of ovulation. J. Reprod. Fert. 6:413-420.

— a

ENG

THE UNIVERSITY OF KANSAS

SCIENCE BULLETIN

BIONOMICS AND ZOOGEOGRAPHY OF TIGER
BEETLES OF SALINE HABITATS IN THE
CENTRAL UNITED STATES
(COLEOPTERA: CICINDELIDAE)

By

Harold L. Willis

Vo.. XLVII Paces 145-313 Ocroser 11, 1967 No.5

ANNOUNCEMENT

The University of Kansas Science Bulletin (continuation of the Kansas Uni-
versity Quarterly) is issued in part at irregular intervals. Each volume contains
300 to 700 pages of reading matter, with necessary illustrations. Exchanges with
other institutions and learned societies everywhere are solicited. All exchanges

should be addressed to
LIBRARY OF THE UNIVERSITY OF KANsas,
LawreENce, Kansas 66044

PUBLICATION DATES

The actual date of publication (1.e., mailing date) of many of the volumes of
the University of Kansas Science Bulletin differs so markedly from the dates on
the covers of the publication or on the covers of the separata that it seems wise to
offer a corrected list showing the mailing date. The editor has been unable to
verify mailing dates earlier than 1932. Separata were issued at the same time as
the whole volume. Beginning with Volume XLVI, publication was by separate
numbers and the date on each number is the actual publication date.

Vol. XX—October 1, 1932. Vol: XXXIV Eee I— Oct ls 1952
Vol. XXI—November 27, 1934. Pt. Ii—Feb. 15, 1952.
Vol. XXII— November 15, 1935. Vol. XXXV,Pt. I—July 1, 1952.
Vol. XXIII—August 15, 1936. Pt. 1I—Sept. 10, 1953.
Vol. XXIV—February 16, 1938. Pt. III—Nov. 20, 1953.
Vol. XXV—July 10, 1939. Vol. XXXVI,Pt. I—June 1, 1954.
Vol. © XXVI—November 27, 1940. Pt. 1I—July 15, 1954.
Vol. XXVII,Pt. I—Dec. 30, 1941. Vol. XXXVII, Pt. I—Oct. 15, 1955.
Vol. XXVIII, Pt. I—May 15, 1942. Pt. II—June 29, 1956.

Pt. II—Nov. 12, 1942. Vol. XXXVIII, Pt. I—Dec. 20, 1956.
Vol. XXIX, Pt. I—July 15, 1943. Pt. II—March 2, 1958.

Pt. 11—Oct. 15, 1943. Vol. XXXIX—Nov. 18, 1958.
Vol. XXX, Pt. I—June 12, 1944. Vol. XL—April 20, 1960.

Pt. II—June 15, 1945. Vol. XLI—Dec. 23, 1960.
Vol. XXXI, Pt. I—May 1, 1946. Vol. XLII—Dec. 29, 1961.

Pt. il—Nov. 1, 1947. Vol. XLII—Supplement to, June 28, 1962.
Vol. XXXII—Nov. 25, 1948. Vol. XLIII—Aug. 20, 1962.
Vol. XXXIII, Pt. I—April 20, 1949. Vol. XLIV—Sept. 1, 1963.

Pt. II—March 20, 1950. Vol. XLV—June 7, 1965.

Vol. XLVI— March 3, 1967

Editor’ .iciak Ramee ete R. C. Jackson

Editorial Board ........ GerorceE Byers, Chairman
KENNETH ARMITAGE
CHARLES MICHENER
Paut Kiros
RICHARD JOHNSTON
DELBERT SHANKEL

THE UNIVERSITY OF KANSAS

SCIENCE BULLETIN

Vout. XLVII Paces 145-313 OcroseEr 11, 1967 No. 5

Bionomics and Zoogeography of Tiger Beetles
of Saline Habitats in the Central United States
(Coleoptera: Cicindelidae)'

By

Harotp L. WItis

INTRODUCTION

Cicindelids of saline habitats first came to my attention when, upon visit-
ing a salt flat in Stafford County, Kansas, for the first time, in October 1961,
I was amazed to find a species of tiger beetle (Cicindela togata ) abundant on
the barren, salt-encrusted soil. Further collecting in such habitats revealed
that there are a number of species adapted for living in saline areas that are
found nowhere else. It seemed worthwhile to investigate how these insects
“make a living” in such harsh environments and how they came to be dis-
tributed and adapted to saline habitats scattered widely over the central
United States.

A survey of the literature showed that although adults of North American
cicindelids are quite well known taxonomically, relatively little work has
been done on the life history, ecology, or zoogeography of most species.
Shelford (1907, 1908, 1911, 1913d, 1917) and Criddle (1907, 1910) have done
excellent work on the life cycles and ecology of a number of species in north-
eastern North America. Hamilton (1925) described the larvae of about 25
United States species and Spangler (1955) described another. Many other
authors have made some mention of bionomics in addition to other subjects;
Ortenburger and Bird (1933) are among the few to mention cicindelids of
saline habitats in the central United States. Several general works with
zoogeographical emphasis have included some or all cicindelids of the United

‘Contribution No. 1342 from the Department of Entomology, The University of Kansas,
Lawrence, Kansas.

146 Tue UNIverSITY SCIENCE BULLETIN

States (Horn, 1908-1915; Papp, 1952; Schilder, 1953b; Rivalier, 1950, 1954,
1957, 1961, 1963). Studies by Wickham (1904a, b), Cazier (1948, 1954), and
Rumpp (1956, 1957, 1961), have included species of saline habitats of the
southwestern United States and Mexico, a few of which also occur in the
central United States. Except for brief habitat and distribution notes in many
papers, these are the only works having any direct connection with the species
in this study. Many of the minor works and others dealing with foreign
species will be mentioned later.

ACKNOWLEDGMENTS

I would like to thank Dr. George W. Byers for his guidance, aid, and sugges-
tions concerning this problem, and for reading and editing the manuscript. I am
also indebted to Drs. Henry S. Fitch, Karl A. Stockhammer, and F. James Rohlf
for reading all or part of the manuscript and offering suggestions. Thanks are also
due (in alphabetical order): Dr. Mont A. Cazier, for suggesting this sort of thesis
problem; Mr. Ronald Bartcher, for processing the geographic variation data on
the computer; Mr. Richard Freitag, for ideas concerning the phylogeny of cicin-
delids, and for carabid identification; Dr. Daniel H. Janzen, for suggestions about
the ecology of cicindelids, and for ant identification; Mr. James K. Lawton, for
unpublished information about cicindelid parasites; Dr. W. Wayne Moss, for
mite identification, and for the translation of certain papers; Mr. Charles W.
O’Brien, for curculionid identification; Dr. F. James Rohlf, for assistance and
suggestions concerning the statistical aspects of the problem; Dr. Norman L.
Rumpp, for information about populations of cicindelids in the southwestern
United States; and Dr. Ronald V. Southcott, for mite identification. Mrs. Joetta
Weaver typed portions of the manuscript.

Finally, I thank the following institutions and persons for the loan of speci-
mens or assistance in examination of specimens: Dr. R. D. Alexander, University
of Michigan; Dr. N. L. Anderson, Montana State College; Dr. W. T. Atyeo,
University of Nebraska; Dr. G. E. Ball, University of Alberta; Dr. W. F. Barr,
University of Idaho; Dr. W. W. Boyle, Pennsylvania State University; Dr. O. P.
Breland, University of Texas; Dr. W. J. Brown, Entomology Research Institute,
Canada Department of Agriculture; Dr. H. R. Burke, Texas A. and M. Univer-
sity; Dr. O. L. Cartwright, U.S. National Museum; Dr. Leland Chandler, Purdue
University; Dr. A. C. Cole, University of Tennessee; Mr. J. F. Cornell, North
Carolina State College; Dr. P. J. Darlington, Jr., Museum of Comparative Zoology,
Harvard University; Dr. H. A. Denmark, Division of Plant Industry, Florida
Department of Agriculture; Dr. W. A. Drew, Oklahoma State University; Dr.
J. G. Edwards, San Jose State College; Dr. W. R. Enns, University of Missouri;
Dr. Roland Fischer, Michigan State University; Dr. P. H. Freytag, Ohio State
University; Mr. Saul Frommer, University of California at Riverside; Mr. G. C.
Gaumer, Nacogdoches, Texas; Dr. H. J. Grant, Jr., Academy of Natural Sciences
of Philadelphia; Dr. R. C. Graves, Bowling Green, Ohio; Dr. H. M. Harris, Jowa
State College; Dr. C. E. Hopla, University of Oklahoma; Mr. Ronald L. Huber,

BIoNoMIcs AND ZOOGEOGRAPHY OF TIGER BEETLES 147

Minneapolis, Minnesota; Dr. Preston Hunter, University of Georgia; Dr. M. T.
James, Washington State University; Mr. J. B Karren, University of Kansas;
Dr. G. F. Knowlton, Utah State University; Dr. F. E. Kurczewski, Syracuse
University; Mr. R. L. Langston, University of California at Berkeley; Dr. Url
Lanham, University of Colorado; Dr. Ira LaRivers, University of Nevada; Dr.
John Lattin, Oregon State University; Mr. H. B. Leech, California Academy of
Sciences; Dr. A. T. McClay, University of California at Davis; Dr. R. E. Pfadt,
University of Wyoming; Dr. R. H. Painter, Kansas State University; Dr. L. L.
Pechuman, Cornell University; Dr. C. L. Remington, Yale University; Dr. H. H.
Ross, Illinois Natural History Survey; Mr. P. E. Slabaugh, Bottineau, North
Dakota; Dr. Marion Smith, University of Massachusetts; Dr. R. R. Snelling, Los
Angeles County Museum; Mr. A. H. Squires, University of Connecticut; Dr.
F. W. Stehr, University of Minnesota; Dr. J. E. Sublette, Eastern New Mexico
University; Dr. T. O. Thatcher, Colorado State University; Mrs. Patricia Vaurie,
American Museum of Natural History; Dr. G. E. Wallace, Carnegie Museum,
Pittsburg; Dr. L. O. Warren, University of Arkansas; Dr. Rupert Wenzel, Field
Natural History Museum, Chicago; Dr. George Wheeler, University of North
Dakota.

This research was supported by a series of National Science Foundation

Graduate Fellowships (1962-1966).

MATERIALS AND METHODS

As many saline habitats as possible in the central United States (southern
Nebraska, Kansas, western Missouri, and Oklahoma) were visited during
1963-1965 to obtain specimens and data. Many were visited repeatedly to
obtain information on seasonal distribution. When possible, large series of
specimens were collected for statistical analysis. Both larvae and adults were
collected alive and brought into the laboratory for life history studies.

In the laboratory, larvae were kept in tall jars full of soil from their na-
tural habitat and were fed either small arthropods from weed sweepings or
Mediterranean flour moths, Anagasta kuehniella, from a stock culture. Adults
and young larvae resulting from their oviposition were kept in terraria partly
filled with soil from saline habitats. A small Stender dish with water and a
shelter made from a card with its corners bent down were provided, and the
adults were given food similar to that offered the larvae. The temperature of
the laboratory was a nearly constant 245°C, although occasionally a goose-
neck lamp was placed over the terraria to provide more heat. The soil of the
larval and adult containers was moistened occasionally. Eggs, pupae, and
teneral adults were kept in Stender dishes for observation.

In addition to over 3600 specimens collected personally, over 15,400 speci-
mens were borrowed from or examined at most of the major insect collections
in the United States and Canada. More details about certain methods will be
given later.

148 Tue University ScrENcE BULLETIN

SALINE HABITATS

Saline terrestrial habitats are found throughout much of the world. The
most widely distributed are coastal, including beaches, tidal flats, and salt
marshes. Chapman (1960) has reviewed the extensive literature on coastal
saline habitats. Coastal salt marshes are generally densely vegetated and thus
unfit for cicindelids; however, tidal flats and beaches are often well populated.

Away from the coast, one may encounter inland saline habitats, which are
not subjected to periodic inundation by tides or wave action. Inland saline
and alkaline habitats are found in all continents, primarily in semiarid and
arid regions. These inland areas are associated with three types of soil; one
contains underground deposits of sodium chloride, another has excess sodium
chloride and sodium sulfate in the surface layers, and the third has sodium
sulfate, sodium carbonate, and/or magnesium sulfate in a subsurface layer.
The soils with subsurface deposits are called alkali soils (Chapman, 1960).
Richards (1954) defines a saline soil as one that contains an excess of soluble
salts (the electrical conductance of a saturated paste in the unit of measure;
the arbitrary value of 4 mmhos/cm is the lower limit of saline soils) and an
alkali soil as having an excess of exchangeable sodium (15°94 exchangeable
sodium is the lower limit of alkali soils). Soils with an excess of both soluble
salts and exchangeable sodium are called saline-alkali soils. The pH of saline
soils is ordinarily less than 8.5; that of alkali soils is above 8.5, sometimes
reaching 10. The pH of saline-alkali soils is variable, but usually below 8.5
(Richards, 1954). The commonness of saline and alkaline habitats in arid
and semiarid regions is due partly to the evaporation rate and low rainfall of
such areas, which decrease the leaching of salts into the ground water where
they can be carried away. Restricted drainage is another factor contributing
to the salinization of soils; temporary or permanent flooding, as well as irri-
gation, may raise the ground water level and cause accumulation of salts at
the surface (Richards, 1954). Alkalization of soil occurs by cation adsorp-
tion on the surfaces of soil particles as a consequence of electrical charges
(Richards, 1954).

The original source of the salts is the minerals which weather out of the
rocks of the earth’s crust; however, a more direct source is from marine
deposits of earlier geologic ages. The salts are usually moved by surface and
ground water to areas in which they are concentrated.

Most saline habitats contain characteristic vegetation which has been vari-
ously classified. Chapman (1960) used the terms halophyte and glycophyte
(mesophyte or non-halophyte), but said that one cannot always readily dis-
tinguish between them. Many halophytes exhibit characteristic physical
modifications, such as a glaucous appearance, succulence, water storage hairs,
small leaves, a glabrous surface, and salt-secreting glands (Chapman, 1960).

BronomMics AND ZOOGEOGRAPHY OF TIGER BEETLES 149

The vegetation of coastal and inland saline habitats is often very similar,
often consisting of closely related species; however, there are greater differ-
ences from one continent to another. Both coastal and inland saline habitats
often exhibit a zonation of vegetation, primarily in response to varying de-
grees of salinity, although other factors such as drainage and topography are
important. Chapman (1960) reviewed many studies of this phenomenon.
In coastal areas, a definite succession to mesic conditions has been found;
however, Chapman (1960) said that in inland areas, the zonation is usually
static and the vegetation must be considered an edaphic climax. Baalman
(1965), in a study of a salt flat in Oklahoma, decided that little change in
vegetation could be expected in highly saline areas. Ungar (1965) called the
vegetation of a Kansas salt marsh a sub-climax which would change to a
mesic climax if excess salts and water were removed.

Fewer studies have been made on the fauna of saline habitats. Davis
(1962) surveyed the seasonal abundance of insects in North Carolina coastal
salt marshes. Smalley (1960) studied the energy flow in populations of a
Georgia salt marsh grasshopper, and Teal (1962) reported on energy flow in
an entire Georgia salt marsh ecosystem. Ortenburger and Bird (1933),
Jackson and Warfel (1933), and Williams (1954, in Baalman, 1965) studied
the ecology of several Oklahoma salt flats. Lengerken (1929) did a detailed
study of halophilic beetles of the coast of Germany. Pearse, Humm and
Wharton (1942) studied the ecology of sand beaches in North Carolina.

The area here arbitrarily called the central United States (southern Ne-
braska, Kansas, western Missouri, and Oklahoma) has been variously sub-
divided by biologists and geographers. The system given in Kendall, Glen-
dinning, and MacFadden (1958) seems adequate: the western three-fourths
of Nebraska, the western two-thirds of Kansas, and the Oklahoma panhandle
are in the Great Plains; the rest of Nebraska and Kansas, the northwestern
half of Missouri, and central Oklahoma are in the Central Plains; south-
eastern Missouri and Oklahoma are in the Interior Uplands.

Satine Hasirats oF THE CENTRAL Unirep States. Saline habitats of the
central United States may be divided into two categories having ecological
significance for tiger beetles: fluvial, or salty rivers and streams, and non-
fluvial. The latter may be arbitrarily subdivided on the basis of size and
moisture into small salt patches (usually dry), salt flats (dry or moist), salt
marshes, and salt lakes. The two major categories may be in close contact
or superimposed, as when salty patches occur next to saline streams or when
drainage streams cross large salt flats.

The distribution of saline habitats in the central United States is some-
what irregular. The larger ones occur in a broad band running obliquely from
southwestern Oklahoma through central Kansas to southeastern Nebraska.
Smaller habitats occur in central and northeastern Oklahoma, southeastern

150 Tue University ScrENcE BULLETIN

Kansas, and north-central Missouri. In the central United States, such
habitats are usually far isolated from one another compared to parts of the
southwestern United States. They are of both natural and man-made origin.
In northern Kansas and southern Nebraska, the salt comes from deposits in
upper Dakota shales (Cretaceous), while in southern Kansas and northern
Oklahoma, the gypsum redbeds (Permian) and Cretaceous strata are respon-
sible (Ungar, 1965; Baalman, 1965). In oil-producing areas, small salty
patches or small salt flats frequently occur where brine has been released
during drilling. The major oil producing areas of the central United States
are found in southeastern and central Kansas and central Oklahoma (Moore

and Haynes, 1917; Redfield, 1927).

The man-made saline areas can be dated relatively well. Oil was found
in Kansas in 1860, soon after its discovery in Pennsylvania, but most drilling
was not done in Kansas and Oklahoma until the late 1800’s and early 1900's
(Moore and Haynes, 1917; Gould, 1930); thus these saline habitats are not
over 80 or 90 years old. Natural saline areas north of the limits of Pleistocene
ice (Wright and Frey, 1965) have obviously not existed in their present con-
dition before this time. Frye and Leonard (1952) stated that the present Kan-
sas landscape is a product of erosion and deposition during the Pleistocene.

In pollen analyses of sites in Meade County, Kansas, and Harper County,
Oklahoma, Kapp (1963) and Stephens (1959), respectively, concluded that
these areas were similar to the present eastern Dakotas or moderate elevations
in the southern Rockies during the Illinoian glacial period. Today these areas
are short grass prairie.

The physical and chemical conditions of saline habitats are often extreme,
particularly at the level at which insects live. Salinity varies greatly with the
season (less in spring when most rain occurs), depth (higher at surface), and
topography, but may reach as high as 3°4 (Unger, 1965); when the weather
is dry, a white encrustation of crystalline salt usually occurs on the surface.
Because of this variability and since tiger beetles occur in most parts of saline
habitats, salinity was not measured in this study.

Temperature may be extreme on the bare or sparsely vegetated surface of
saline habitats. Geiger (1965) stated that in the summer the surface tempera-
ture of bare soil may reach 60, 70, or even 80°C. Above or below the surface
the temperature drops rapidly. A study by Sinclair (*2 Geiger, 1965, and
Allee et al., 1949) in Tucson, Arizona, showed that the soil just below the
surface reached 71.5°C, was 62.5°C at a depth of 2 cm, dropped to 42.2°C at
10 cm, and was 20°C at 60 cm; meanwhile the air temperature in a standard
shelter was 425°C. Also, the daily range of temperatures was greatest near
the surface (56.5°C) and less below (40.1°C at 2 cm, 13.8°C at 10 cm, 0°C at
60 cm) or above (31.5°C) the surface. Occasionally measurements of soil and
air temperature taken in this study gave similar results; surface temperatures

BioNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 151

much greater (5-13°C) than air temperatures were frequently noted. Geiger
(1965) also stated that a vegetational cover may have a moderating effect on
temperature, but what vegetation does occur in saline habitats is usually low
and sparse and would have little effect (Geiger said that in grass under a
meter high, the maximum temperature is still at the surface).

The availability of water varies considerably with the season as well as
with the type of habitat. In rainy weather, saline habitats may be quite moist
or flooded, but during the hot, dry months of July and August the surface
may become very dry and hard. The salinity of available water varies widely,
depending on its origin, and may reach 3°%% (Ungar, 1965). Humidity also
varies greatly; Ortenburger and Bird (1933) noted that the relative humidity
ona salt flat at Cherokee, Oklahoma, varied from 80 or 90°% in the morning
to 20 or 30% at midday.

Wind velocity, usually high in prairies anyway, is especially great on the
wide, bare expanses of salt flats. In the summer months the wind normally
blows steadily from the south, and on salt flats in southern Kansas and
northern Oklahoma, I have estimated the maximum velocity to be 40 to 50
miles per hour during fair weather. Fluvial habitats are usually more pro-
tected from wind. The strong wind not only has a physical effect important
to flying insects but also has a marked desiccating effect.

Another characteristic of many saline habitats is the general lack of or
low amount of cover which would allow insects to escape many of the above
conditions as well as predators and parasites.

The vegetation of saline habitats in the central United States is primarily
affected by variations in local topography, drainage, and salinity (Ungar,
1965). In most habitats there is an area of very high salinity (2-39) where
no flowering plants occur. In areas of less salinity (up to 2.759%), Distichlis
stricta, a low, wiry grass, and Suaeda depressa, a sparse, succulent chenopod,
are the dominant plants. In more marginal and less saline areas, Atriplex
patula (Chenopodiaceae), Sporobolus airoides, Poa arida, Hordeum jubatum,
Spartina pectinata (Gramineae), and Tamarix gallica (Tamaricaceae, an
introduced woody shrub) occur along with the above-mentioned species.
Many other species are less common, although some (Salicornia, Chenopo-
diaceae) become dominant species farther west in the United States. More
complete analyses of the vegetation of two saline habitats in Kansas and
Oklahoma can be found in Ungar (1964, 1965) and Baalman (1965).

On many saline habitats, low mounds or hummocks have been formed
by the pioneering vegetation (Distichlis and Suaeda ) collecting blowing sand
or soil at their bases. These hummocks may continue to grow, reaching
heights of a meter or more in some cases, and allow normal prairie flora and
fauna to inhabit their tops (Ortenburger and Bird, 1933; Baalman, 1965).

Some typical saline habitats visited in this study are shown in Figures 1-8.

a7 aS st )y- 2 > TT be AS -

Fic. 1. Small salty patches in corner of plowed field, 11 m. north, 1 mi. east of Lincoln,
Lincoln Co., Kansas (17 June 1963). Fic. 2. Small salt flat near oil wells, 2 mi. north, % mi.
east of Chanute, Neosho Co., Kansas (18 August 1964). Fic. 3. Small salty patches near Salt
Creek, 1 mi. northwest of Fredonia, Wilson Co., Kansas (20 April 1963). Fic. 4. Small inter-
mittent creek with salty banks, 5 mi. north of Yates Center, Woodson Co., Kansas (20 Aprii
1963). Fic. 5. Large (8 miles long, 2 miles wide) salt flat on the Cimarron River, 2.5 mi.
southwest of Plainview, Woods Co., Oklahoma (29 August 1963). Fic. 6. Large hummocks on
salt flats, Great Salt Plains National Wildlife Refuge, 3 mi. east of Cherokee, Alfalfa Co., Okla-
homa (27 August 1963). Fic. 7. Salt marsh with cattails and sedges in area of greatest moisture,
VY, mi. east, 1 mi. south of Talmo, Republic Co., Kansas (18 June 1963). Fic. 8. Salt Lake,
with bare salty patches around shore, Lincoln, Lancaster Co., Nebraska (19 June 1963).

* cael .

BioNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 153

For the purposes of this paper, most saline habitats can be divided into a
central bare, nonvegetated area and a sparsely vegetated marginal area (in-
cluding the margins of the islandlike hummocks), although in some, the
marginal vegetation is tall and dense.

INTRODUCTION TO CICINDELA

The genus Cicindela® is found in terrestrial habitats throughout most of
the world from about 50° south latitude to the Arctic Circle, except for high
mountains and many midoceanic islands. Except for the closely related
monotypic genera Eurymorpha and Apteroessa, Cicindela is considered to be
the most highly evolved genus in the family Cicindelidae (Horn, 1926).
Many species of Cicindela throughout the world live in saline habitats, and
many of these are not closely related, hence the ability to live in such habitats
has apparently evolved independently a number of times. In general, how-
ever, species of saline habitats are among the more highly evolved species
of the genus.

Considering the North American fauna of Cicindela, many of the more
advanced species (the arrangement of Rivalier, 1954, is being followed, with
slight modifications) live in saline habitats exclusively, and a sprinkling of
less advanced species occur in such habitats occasionally or exclusively. The
cicindelids found in saline habitats of the central United States include some
species found only in such habitats and some found in both mesic and saline
habitats with more or less equal frequency. The species included in this
study, with brief notes on their habitat and distribution, are listed in Table 1.
As one moves outside the area here arbitrarily called the central United
States, other species may be encountered in saline habitats. Some of the
species in Table 1 that are primarily found in mesic habitats are relatively
uncommon in the habitats studied, and others have been or are being more
thoroughly studied by others. Therefore, this study is concentrated on the
following species: C. circumpicta, C. cuprascens, C. fulgida, C. macra, C.
nevadica, C. togata, and C. willistoni. In addition, certain other species which
are closely related to these will be briefly considered.

BIONOMICS OF CICINDELA

The following account is based on observations of several species. Most of
the work on the life history was done with C. togata, but C. circumpicta, C.

* A number of workers have suggested splitting Cicindela into a variable number of genera.
The most recent and best arrangement to date is in a series of papers by Rivalier (1950-1963)
based primarily on the male genitalia. This proposal has met some opposition; many workers
feel that Rivalier’s “genera” should be treated as subgenera. I shall follow the latter viewpoint
in this work.

154 Tue University SciENCE BULLETIN

Tasre 1. Species of Cicindela included in this study, types of habitats in which
they occur, and their general distribution (F—fluvial, N=nonfluvial).

C. circumpicta La Ferté; N (sometimes F) saline habitats; N. Dak., Neb., Mo., Kans., Colo.,
Okla., N. Mex., Tex., Tamaulipas (Mexico).
C. cuprascens LeConte; F mesic and saline habitats; Manitoba (Canada), Mont., Wyo., Colo.,

N. Mex., Tex., La., Miss., Ala., Tenn., Ky., Ohio, Ind., Ill., Ia., Minn., and areas enclosed
within this circle.

C. duodecimguttata Dejean; F mesic and saline habitats; Northwest Terr. Alberta, Sask.,
Manitoba, Ontario, Quebec, Labrador, Newfoundland, Nova Scotia, and N. Brunswick
(Canada), south to Ga., Ala., Miss., Ark., Tex., Colo., Wyo., Mont., and areas enclosed
within this circle.

C. fulgida Say; N saline habitats; Alberta, Sask., and Manitoba (Canada), N. Dak., Mont.,
S. Dak., Wyo., Neb., Colo., Kans., N. Mex., Okla., Tex., Ariz.
C. hirticollis Say; F mesic and saline habitats; Brit. Col., Alberta, Sask., Manitoba, Ontario,

Quebec, Newfoundland, Prince Edw. Is., and N. Brunswick (Canada), most of continental
United States, Baja Calif., Chihuahua, and Vera Cruz (Mexico).

C. macra LeConte; F mesic and saline habitats; Ohio, Ky., Tenn., La., Tex., Colo., Wyo.,
Neb., S. Dak., Minn., Wisc., Mich., and areas enclosed within this circle.

C. nevadica LeConte; F and N saline (sometimes mesic) habitats; Sonora and Coahuila (Mex-
ico), Calif., Nev., Ariz., Ut., N. Mex., Tex., Okla., Colo., Kans., Neb., Wyo., S. Dak.,
Mont., N. Dak., Sask. and Manitoba (Canada).

C. punctulata Olivier; N (sometimes F) mesic and saline habitats; Alberta, Sask., Manitoba, and
Ontario (Canada), most of continental United States except Pacific Northwest and Calif.,
Sonora, Chihuahua, Coahuila, Durango, Zacatecas, Hidalgo, Puebla, Distrito Federal, and
Mexico (Mexico).

C. repanda Dejean; F and N mesic and saline habitats; Brit. Col., Alberta, Sask., Manitoba,
Ontario, Quebec, Labrador, Newfoundland, N. Brunswick, and Nova Scotia (Canada), most
of continental United States except Southwest.

C. schauppit G. Horn; N saline and mesic habitats; Okla., Tex., Nuevo Leon (Mexico).

C. togata La Ferté; N (sometimes F) saline habitats; Neb., Kans., Colo., Okla., N. Mex., Tex.,
La., Miss., Ala., Fla., S. Car., Tamaulipas (Mexico).

C. tranquebarica Herbst; N mesic and saline habitats; Northwest Terr., Brit. Col., Alberta, Sask.,
Manitoba, Ontario, Quebec, N. Brunswick, Nova Scotia, and Prince Edw. Is. (Canada), most
of continental United States.

C. willistoni LeConte; N (sometimes F) saline habitats; Calif., Nev., Ore., Ut., Ariz., Wyo.,
N. Mex., Tex., Okla., Kans.

fulgida, and C. nevadica were also studied. Nearly all the species in Table 1
were considered from the ecological viewpoint. This account will consider
each stage of the life cycle separately. The following brief summary, taken
mostly from Balduf (1935), will serve as an introduction to the bionomics of
Cicindela in general.

The eggs are laid in the soil; the newly hatched first instar larva enlarges
the hole produced by the adult’s ovipositor into a burrow. Burrows are
usually perpendicular to the surface, more or less straight, and with the open-

BioNomMics AND ZOOGEOGRAPHY OF TIGER BEETLES ib»)

ing flush with the surface. The three larval instars lie in wait at the entrance
to their burrows for prey, the head and pronotum forming a camouflaged
“plug” to the burrow. Most small arthropods are accepted as food and are
taken to the bottom of the burrow to be consumed unless they are too large
for the diameter of the burrow. The last instar larva digs a special pupal cell
in which transformation to the adult occurs. The adult digs its way out and
spends most of its active hours hunting prey or reproducing. Small] arthro-
pods are the usual food. Adults usually prefer open, sparsely vegetated areas
and are most active during warm, sunny weather. The female carefully
selects the type and moisture of the soil used for oviposition. The length and
type of life cycle varies with the species. In some, the adults emerge in the
fall, hibernate, and then mate and oviposit in the spring, dying during the
summer; in others, the adults emerge during the summer and die before
winter. The larval stage occupies most of the life cycle, which may take one
to four years to complete.

Tue Eco. Since most adult Cicindela are about the same size, the eggs of
most species are probably very similar. Shelford (1908) said that the eggs of
C. purpurea and C. repanda are about 2 mm long and 1-1.5 mm wide, a
translucent creamy yellow, and larger at the anterior end. Moore (1906) gave
similar sizes for two eggs of C. repanda, but judging from his descriptions,
they were apparently atypical and shrunken. Huie (1915) said that the eggs
of the European C. campestris are 2 mm long, oval, smooth, and yellowish
when laid. Ponselle (1900) found the eggs of C. flexuosa to be 2 mm long
and 1 mm wide.

I found the eggs of C. circumpicta, C. duodecimguttata, C. nevadica, and
C. togata to be similar. The mean length and width of 38 eggs of C. togata
measured with an eyepiece micrometer was 2.08 x 1.01 mm; the ranges were
1.85 - 2.43 x 0.92-1.12 mm. The average size of four eggs of C. circumpicta
was 2.21 x 0.99 mm. One egg of C. duodecimguttata was 1.80 x 1.14 mm,
and a desiccated C. nevadica egg was 1.70 x 0.74 mm. Eggs will absorb water
and swell slightly if placed in a moist environment. The eggs examined were
ovoid and not much larger at the anterior end than the posterior. There is
sometimes a slightly concavity on the ventral side. The chorion is shiny, but
under high magnification a fine reticulate pattern can be seen. The chorion
is not very strong, and the egg is easily ruptured. When first laid, the egg is
a creamy or light straw yellow and filled with yolk granules. In C. togata,
the posterior end of the egg is attached to the soil at the bottom of the hole
made by the ovipositor by a short stalk of sticky material (Fig. 13). In C.
circum picta, no definite stalk was seen, but the egg adhered to the soil because
its posterior end was sticky. Moore (1906), Huie (1915), and Zikan (1929)
also noted that Cicindela eggs are sticky or fastened by a stalk.

Histological sections of the ovaries of C. togata were made and stained

Fics. 9-12, ovary of C. togata, longitudinal section; Fic. 9, stained according to the periodic
acid—Schiff procedure; inset: enlargement of the peripheral cytoplasm of the terminal oocyte;
areas stained are shaded. Fic. 10, stained according to the azo-coupling protein method; inset:
enlargement of the peripheral cytoplasm of the terminal oocyte; areas stained are shaded.
Fic. 11, stained according to the Sudan black B method for lipids; inset: enlargement of the
peripheral cytoplasm and female pronucleus of the terminal oocyte; areas stained are shaded.
Fic. 12, stained according to the methyl green-pyronin Y method for nucleoproteins; areas
stained are shaded: black=green, stipple=purple-red.

BroNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 157

according to the following methods: the periodic acid-Schiff (PAS) pro-
cedure (Barka and Anderson, 1963) for gylcogen, muco- and glycoproteins,
glycolipids, mucopolysaccharides, and simple proteins; the azo-coupling pro-
tein method (Barka and Anderson, 1963) for proteins in general; the Sudan
black B method (Pearse, 1960) for lipids; and the methyl green-pyronin Y
method (Pearse, 1960) for nucleoproteins. The results are shown in Figures
9-12. The ovaries of cicindelids are of the polytrophic meroistic type, as are
those of all Adephaga (with follicles of nurse cells, or trophocytes, alternating
with follicles of oocytes). In Figures 11 and 12, nutrient material can be seen
entering oocytes from trophocytes.

In Figure 9, it can be seen that PAS-positive nutrients do not enter the
oocyte until very late, since only the terminal oocyte is stained. An enlarge-
ment of it shows a layer of cytoplasm near the vitelline membrane with few
stained granules. All cells are lightly stained by the azo-coupling reaction
(Fig. 10), which is not surprising, since all cells contain proteins. In the
terminal oocyte, relatively few granules contain protein. In the Sudan black
B lipid reaction (Fig. 11), the cytoplasm of all cells is stained, and the nuclei
are only very lightly stained. In the oocyte of intermediate development, a
dense-staining central core is present, indicating that most of the lipids have
entered the oocyte or have been synthesized within it by this time. This may
be the same as the corelike “polar vitelline granules” of Hirschler (1932). In
the terminal oocyte, different sized granules are stained in varying degrees,
and the nucleus is stained to a greater degree than in other oocytes. In the
methyl green-pyronin Y reaction, DNA stains green and RNA stains purple-
red. In Figure 12, the nuclei of all cells stained green and the cytoplasm
purple-red, as expected.

EmpryotocicaL DrvetopMent. The gross embryology of C. togata was
studied. Fragmentary observations on C. circumpicta and C. duodecimgut-
tata were similar to those made on C. togata and will not be discussed.
Almost no work has been done on the embryology of cicindelids. Shelford
(1908) published a small drawing of an embryo of C. purpurea within the
egg and stated that the egg hatches in about two weeks. Huie (1915) men-
tioned that more mature eggs of C. campestris exhibit two pairs of eyes
through the chorion. Zikan (1929) said that embryological development
takes 9-29 days, depending on the species and temperature (he studied other
genera of cicindelids as well as Cicindela; the genera Ctenostoma and Odon-
tochila take about one month).

Eggs of C. togata were recovered from terrarium soil in which adults had
oviposited. When possible, eggs were dug up just after oviposition. The eggs
were kept individually in small covered Stender dishes in which a small
amount of water was placed to avoid desiccation. They were examined under
a dissecting microscope with transmitted light twice a day. The laboratory

IS

0.5 mm | mm l

scale,Figs.|8 and 21 scale, other Figs.

Fic. 13. Newly laid egg of C. togata, showing stalk at its posterior end attaching it to the
substrate. Fics. 14-21, embryos of C. togata; Fic. 14, early embryo (about | day old), lateral
aspect; Fic. 15, same, ventral aspect; Fic. 16, about 2 days old, lateral aspect; Fic. 17, same,
ventral aspect; Fic. 18, same, enlargement of anterior portion; Fic. 19, about 2.5 days old,
lateral aspect; Fic. 20, same, ventral aspect; Fic. 21, same, enlargement of anterior portion.

BIonoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 159

temperature was 24.5°C. The transparent chorion makes eggs of Cicindela
ideal for such observations. In addition, the embryos of fixed eggs were
dissected out and examined.

The gross embryological development of C. togata is shown in Figures
13-34. The newly laid egg (Fig. 13) is filled with homogeneous appearing
yolk granules. Such features as the cleavage center and cleavage nuclei could
not be seen. As can be seen in the histological sections (Fig. 11), the female
pronucleus is located near the periphery on one side. After about one day,
the yolk in the anterioventral portion of the egg appears less dense than the
rest (in live eggs). Presumably the germ band and germ layers are being
formed or have been formed by this time. Then follow several stages that
have been observed only in fixed eggs. Figures 14 and 15 show an early
embryo. Head and thoracic segmentation are well developed, and abdominal
segmentation is nearly complete. Only four segments can be seen in the head
region, the apical one being composed of the paired lateral lobes. A median
line, or primitive groove, is just beginning to develop midventrally. At a
later stage (Figs. 16-18), the lateral lobes of the head are much enlarged;
four pairs of head appendages (antennae, mandibles, maxillae, and labium)
plus a median bilobed labral bud are clearly visible; the maxillae and labium
are beginning to segment; what is probably the stomodeum can be seen as a
depression at the base of the labral bud. The thoracic and first abdominal
appendages are quite long and are beginning to segment; other abdominal
appendages are mere bumps. The median line is clearly evident. At a
slightly later stage (Figs. 19-21), the head appendages have enlarged and
thickened; the maxillae are beginning to become bilobed; the labial ap-
pendages have moved closer together. The thoracic and first abdominal
appendages are clearly segmented, and the other abdominal appendages are
beginning to segment. The legs are longer. The fifth abdominal segment
is slightly larger in diameter than those surrounding it. Slightly later (Figs.
22-24), the head appendages have elongated; the mandibles begin to assume
their future scimitar shape; the outer lobes of the maxillae have elongated
considerably; what may be the anterior tentorial pits can be seen at the
bases of the mandibles when the antennae are straightened out. The
first abdominal appendage is three-segmented, and the other abdominal
appendages are slightly two-segmented. At this stage the early embryo
has reached maximum elongation. Later, when the embryo is about three
to four days old (Figs. 25, 26), the lateral lobes of the head have enlarged
greatly; the maxillae have assumed a characteristic L shape. The legs are
longer, but the body is shorter and wider. The abdominal appendages have
begun to regress. Returning to observations of living eggs, Figure 27 shows
a stage slightly more advanced (4 to 5.5 days old) than that in Figures 25
and 26. The amount of yolk has decreased considerably; the embryo has

28

0.5 mm | | mm

scale, Fig.24 scale, other Figs.

Fics. 22-30, embryos of C. togata; Fic. 22, about 3 days old, lateral aspect; Fic. 23, same,
ventral aspect; Fic. 24, same, enlargement of anterior portion; Fic. 25, about 3.5 days old,
lateral aspect; Fic. 26, same, ventral aspect; Fic. 27, about 4-5.5 days old, lateral aspect;

Fic. 28, about 5-6.5 days old, lateral aspect; Fic. 29, about 7-9 days old, lateral aspect;
Fic. 30, same, ventrolateral aspect.

BroNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 161

moved dorsally; and the abdomen is beginning to flex ventrally. The man-
dibles are much elongated and clearly sickle-shaped, and the L-shaped,
bilobed maxillae are much longer. Dark segmental “spots” can be seen
internally in the thorax and abdomen. A paired pulsating organ appears in
the anterior region of the prothorax. During the next day, the body continues
to elongate, forcing the head and abdomen closer together (Fig. 28). The
abdomen may be seen moving slightly. At an age of 7 to 9 days, the body is
slightly longer, and two pairs of large developing stemmata become visible
as very faint orange ovals on either side of the head (Figs. 29, 30). A tubular
pulsating area can be seen extending posteriorly through the thoracic region
from the paired organ near the head. The internal segmental “spots” are no
longer visible. Probably the dorsal closure is completed by this stage. In
another day (8 to 10 days days after oviposition), the large ovals representing
developing stemmata have become thicker and dark orange; three additional
pairs of smaller stemmata become visible as faint gray dots, two pairs mesad
of the large pairs and one pair laterad of the most posterior large pair; the
large pairs of developing stemmata are surrounded by transparent circles.
The dorsal segmentation can be clearly seen. What appears to be the labium
becomes light orange (Fig. 31). The entire embryo may move slightly in
this and subsequent stages. About one day later (Figs. 32, 33), the small
pairs of stemmata are darker gray, and another pair is beginning to develop
laterad of the posterior large pair; the median pairs of small stemmata have
transparent circles around them. The labium is dark orange; the tips of the
mandibles (but not the extreme tips) are orange; the tarsal claws of the first
legs are orange and those of the second legs light orange. A few hours to
one-half day later (Fig. 34), the stemmata mentioned above are darker
orange (large ones) or gray (small ones); the two pairs of medial stemmata
have begun to fuse (the anterior ones with the posterior) and each fused pair
is surrounded by a common transparent circle; another medial pair of stem-
mata is visible posterior to the fused pairs as very light gray dots. The
orange of the mandibles has spread slightly; the tarsal claws of the second
legs are orange, and those of the third legs are light orange. Sclerites of the
legs, thorax, and abdomen are very light gray; the setae at the apical ends of
the legs are dark. When development reaches this stage, the larva bursts the
chorion at the anterior end and in about five minutes, wiggles out. The
duration of embryological development, measured in four individuals, is
from 10.5 to 11.25 days under laboratory conditions (temperature 24.5°C).

Tue Larva—Descriptions. The larvae of cicindelids are relatively little
known, either biologically or taxonomically. A number of workers have de-
scribed many of the palearctic species, and the larvae of some tropical genera
are known. Among the more important papers for these regions are van

Emden (1935, 1943), Zikan (1929), Friederichs (1931), Blair (1920), Hamil-

162 Tue UNiversiry ScrENcE BULLETIN

Fics. 31-34, embryos of C. togata; Fic. 31, about 8-10 days old, lateral aspect; Fic. 32,
about 9-11 days old, lateral aspect; Fic. 33, same, ventral aspect; Fic. 34, just before eclosion,
lateral aspect.

ton (1925), and Gilyarov and Sharova (1954). In North America, G. Horn
(1878) described one species in each of the four North American genera;
Schaupp (1879a) listed the species that had been described at that time;
Shelford (1908) described rather superficially the larvae of 12 species; Hamil-
ton (1925) described in detail 28 North American and six palearctic species
of Cicindela, three species of Megacephala, three species of Omus, one species
of Amblychila, and two species of tropical genera, and standardized the
morphological terminology; Ortenburger and Bird (1933) published crude
drawings of the larva of C. willistoni and the fifth abdominal segment of the
larvae of C. cuprascens and C. togata; Spangler (1955) described the larva of
C. circumpicta, but his drawings do not show some important details and
are incorrect in others.

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 163

In this study, the larvae of six species were collected; four were identified
by rearing and two (C. fulgida and C. willistoni) by the process of elimina-
tion. One of these, C. duodecimguttata, has been adequately described by
Hamilton (1925) and will not be described here; the larvae of C. togata,
C. nevadica, C. fulgida, and C. willistoni are described for the first time, and
C. circumpicta is redescribed. See Hamilton (1925) for an explanation of
morphological terminology. Descriptions of larval cicindelids have tradi-
tionally been based on last instar larvae, no doubt partly because of a scarcity
of specimens of younger instars. I shall follow this convention, although I
have also collected or reared first and second instar larvae of most of the
species described here.

Larvae will be deposited in the following institutions: U.S. National
Museum (all five species), American Museum of Natural History (all species
except C. fulgida), and The University of Kansas (all species except C. ful-
gida and C. willistont).

C. circumpicta johnsoni Fitch—third instar larva (Figs. 35-40)

Color. Head cupreous, cupreous-green, brassy green, or blue, with green,
blue-green, blue, or purple reflections; labrum red-brown with black margin;
pronotum with most of disk chestnut brown in a pattern which may be more
or less developed (Fig. 35); cephalolateral angles yellow-brown to yellow;
mesonotum dark brown anteriorly, yellow-brown posteriorly; metanotum
yellow-brown; basal two antennal segments yellow-brown, distal segment
reddish brown, penultimate segment intermediate; mandibles reddish brown
basally with apices and retinaculum black; maxillae and labium yellow-
brown. Dorsal cephalic and pronotal setae transparent, other setae yellow-
brown.

Head. Setae on dorsum medium in length and prominent; diameter of
stemma IT subequal to that of stemma I and slightly greater than distance
between I and II; fronto-clypeo-labral area slightly wider than long; U-shaped
ridge on caudal part of frons with 2 setae; antennae with distal segment 0.85
as long as penultimate, proximal segment slightly longer than second seg-
ment, proximal segment with 6-7 setae, second with 7-9, third with 3-4, and
distal with 3-5; maxillae with 3 setae on mesal margin of proximal segment of
galea and 4-5 on distal segment; maxillary palpus 3-segmented, palpifer with
7 setae, penultimate segment with 2 setae; distal segment of labial palpus
with 1 ventral seta, penultimate with 3 spurs and 2 setae on either side of
spurs; ligula with 4 setae.

Thorax. Pronotum with cephalolateral angles extending as far cephalad
as mesal portion; lateral margins slightly carinate; primary setae medium in
length; secondary setae few, 5 or 6 pairs.

. é7\% S
)

‘ 4 L
Ul

|
1 \
we Qari

Fics. 35-40, C. circumpicta johnsoni, third instar larva; specimen is from Geuda Springs,
Sumner Co., Kansas; Fic. 35, head and pronotum, dorsal aspect; left half of pronotum shaded
to show the pattern; Fic. 36, left antenna, ventral aspect; Fic. 37, third abdominal segment,
lateral aspect of left half, with the middorsal line at top and midyentral line at bottom; ventro-
lateral suture dotted; Fic. 38, dorsum of fifth abdominal segment, dorsal aspect; stippled area
is weakly sclerotized; Fic. 39, ninth abdominal sternum, ventral aspect; Fic. 40, pygopod,
dorsal aspect.

BioNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 165

Abdomen. Sclerotized areas distinct, supplementary lateral sclerotized
areas variable in number and shape; secondary setae numerous, some long
and slender, some short and fine; eusternum of ninth segment bearing 2
groups of 3 long setae caudally; pygopod usually bearing 14 setae, 7 on a side;
median hooks of fifth segment with 4-5 (rarely 6) setae, the distal one stout
and directed mesad, the others more slender and directed laterad; inner
hooks with 6-8 setae, 3-5 of which are notably stouter than the others; spine
of inner hooks minute to obsolete.

Measurements. Total length of larva, 16-21 mm; width at third abdomi-
nal segment, about 2 mm; diameter of stemma I, 0.34-0.38 mm; diameter of
stemma II, 0.27-0.37 mm; distance between stemmata I and II, 0.21-0.30 mm;
length of fronto-clypeo-labral area, 1.5-1.7 mm; width of fronto-clypeo-labral
area, 1.56-1.80 mm; length of pronotum, 1.93-2.19 mm; width of pronotum,
2.93-3.25 mm.

C. fulgida fulgida Say—third instar larva (Figs. 41-46)

Color. Head red-brown with brassy green and cupreous-purple reflections;
labrum red-brown with black margin; pronotum brown or dark red-brown
with cupreous-purple reflections; cephalolateral angles lighter brown; meso-
notum dark brown anteriorly, yellow-brown posteriorly; metanotum yellow-
brown; antennae red-brown; mandibles red-brown basally with apices and
retinaculum black; maxillae and labium yellow-brown. Dorsal cephalic and
pronotal setae white, other setae yellow-brown.

Head. Setae on dorsum medium in length; diameter of stemma II notice-
ably smaller than that of stemma I and smaller than distance between I and
II; fronto-clypeo-labral area slightly longer than wide; U-shaped ridge on
caudal part of frons with 2 setae; antennae with distal segment 0.7 as long as
penultimate, proximal segment shorter than second segment; proximal seg-
ment with 5-6 setae, second with 9, third with 2, and distal with 3-4; maxillae
with 3 setae on mesal margin of proximal segment of galea and 5 on distal
segment; maxillary palpus 3-segmented, palpifer with 7 setae, penultimate
segment of palpus with 2 setae; distal segment of labial palpus with 1 ventral
seta, penultimate with 3 spurs and 2 setae on either side of spurs; ligula with
4 setae.

Thorax. Pronotum with cephalolateral angles extending cephalad slightly
more than mesal portion; lateral angles carinate; primary setae medium in
length; secondary setae few, 7 or 8 pairs.

Abdomen. Sclerotized areas distinct; secondary setae numerous, most
long and slender, a few short; eusternum of ninth segment bearing 2 groups
of 3 long and 1 shorter seta caudally; pygopod usually bearing 18 setae, 9 on
a side; median hooks of fifth segment with 2 setae; inner hooks with 2 setae
on a shoulder and long spine over one-third the length of the hook.

Fics. 41-46, C. fulgida fulgida, third instar larva; specimen is from 11 mi. northeast of
Hudson, Stafford Co., Kansas; Fic. 41, head and pronotum, dorsal aspect; Fic. 42, left antenna,
ventral aspect; Fic. 43, third abdominal segment, lateral aspect of left half; ventrolateral suture
dotted; Fic. 44, dorsum of fifth abdominal segment, dorsal aspect; stippled area is weakly
sclerotized; Fic. 45, ninth abdominal sternum, ventral aspect; Fic. 46, pygopod, dorsal aspect.

Fics. 47-52, C. nevadica knaust, third instar larva; specimen is from 11 mi. northeast of
Hudson, Stafford Co., Kansas; Fic. 47, head and pronotum, dorsal aspect; Fic. 48, left antenna,
ventral aspect; Fic. 49, third abdominal segment, lateral aspect of left half; ventrolateral suture
dotted; Fic. 50, dorsum of fifth abdominal segment, dorsal aspect; stippled area is weakly
sclerotized; Fic. 51, ninth abdominal sternum, ventral aspect; Fic. 52, pygopod, dorsal aspect.

168 Tue UNIversiry SCIENCE BULLETIN

Measurements. Total length of larva, 14-18 mm; width at third abdominal
segment, 1.7 mm; diameter of stemma I, 0.30 mm; diameter of stemma II,
0.26 mm; distance between stemmata I and II, 0.28 mm; length of fronto-
clypeo-labral area, 1.48 mm; length of pronotum, 1.83 mm; width of prono-
tum, 2.95 mm.

C. nevadica knausi Leng—third instar larva (Figs. 47-52)

Color. Head black with metallic blue-green, green, purplish, or some-
times bronze reflections; labrum red-brown; pronotum with cephalolateral
angles red-brown and disk black with purple, blue-green, brassy, and cupre-
ous reflections; mesonotum dark brown anteriorly, yellow-brown posteriorly;
metanotum yellow-brown; antennae reddish brown; mandibles reddish
brown basally with apices and retinaculum black; maxillae and labium
yellow-brown. Dorsal cephalic and pronotal setae transparent, other setae
yellow-brown.

Head. Setae on dorsum prominent, long to short; diameter of stemma II
subequal to that of stemma I and greater than distance between I and II;
fronto-clypeo-labral area slightly wider than long; U-shaped ridge on caudal
part of frons with 2 setae; antennae with distal segment 9.6 as long as penul-
timate; proximal and second segments about equal in length; proximal seg-
ment with 6-9 setae, second with 8-9, third with 2, and distal with 3; maxil-
lae with 3 setae on mesal margin of proximal segment of galea and 5 on
distal segment; maxillary palpus 3-segmented, palpifer with 7 setae, penulti-
mate segment of palpus with 2 setae; distal segment of labial palpus with 1
ventral seta, penultimate segment with 2 large and one reduced spur and
two setae on either side of spurs; ligula with 3 setae.

Thorax. Pronotum with cephalolateral angles not extending as far cepha-
lad as mesal portion; lateral margins slightly carinate; primary setae long to
short; secondary setae few, 4 to 6 pairs.

Abdomen. Sclerotized areas distinct; secondary setae few, some very long
and slender, some medium in length; eusternum of ninth segment bearing
2 groups of 3 long and 1 shorter seta caudally; pygopod usually bearing 14
setae, 7 on a side; median hooks of fifth segment with 3-4 setae, all of about
the same diameter; inner hooks with 4-5 setae, 3 of which are notably stouter
than the others; spine of inner hooks minute to obsolete.

Measurements. Total length of larva, 18-20 mm; width at third abdominal
segment, about 2 mm; diameter of stemma I, 0.31-0.33 mm; diameter of
stemma II, 0.27-0.33 mm; distance between stemmata I and II, 0.22-0.26 mm;
length of fronto-clypeo-labral area, 1.42 mm; width of fronto-clypeo-labral

area, 1.44-1.48 mm; length of pronotum, 1.57-1.72 mm; width of pronotum,
2.43-2.63 mm.

0.5 mm |

Fics. 53-58, C. togata globicollis, third instar larva; specimen is from 11 mi. northeast of
Hudson, Stafford Co., Kansas; Fic. 53, head and pronotum, dorsal aspect; left half of pronotum
shaded to show pattern; Fic. 54, left antenna, ventral aspect; Fic. 55, third abdominal seg-
ment, lateral aspect of left half; ventrolateral suture dotted; Fic. 56, dorsum of fifth abdominal
segment, dorsal aspect; stippled area is weakly sclerotized; Fic. 57, ninth abdominal sternum,
ventral aspect; Fic. 58, pygopod, dorsal aspect.

170 Tue Universiry Science BULLETIN

C. togata globicollis Casey—third instar larva (Figs, 53-58)

Color, Head cupreous and green with green reflections; labrum red-
brown with 2 opalescent white spots and black margin; angulus frontalia
opalescent white; pronotum with white margin, disk with pattern of red-
brown to yellow-brown on yellow-brown to yellow background, which may
be more or less developed (Fig. 53); mesonotum dark brown anteriorly,
yellow-brown posteriorly; metanotum yellow-brown; basal two antennal seg-
ments opalescent white, distal two segments yellow-brown; mandibles red-
dish brown basally, with apices and retinaculum black; maxillae and labium
yellow-brown, Dorsal cephalic and pronotal setae transparent, other setae
yellow-brown.

Head, Setae on dorsum short and prominent; diameter of stemma IT sub-
equal to that of stemma Land slightly greater than distance between I and 1;
fronto-clypeo-labral area wider than long; U-shaped ridge on caudal part of
frons with 2 setae; antennae with distal segment 0.57 as long as penultimate,
proximal segment equal in length to second segment; proximal segment with
6 setae, second with 8-9, third with 3, and distal with 3; maxillae with 3 setae
on mesal margin of proximal segment of galea and 5 on distal segment;
maxillary palpus 3-segmented, palpifer with 7 setae, penultimae segment of
palpus with 2 setae; distal segment of labial palpus with 1 ventral seta,
penultimate with 3 spurs and 3 setae on either side of spurs; ligula with
4 setae.

Thorax, Pronotum with cephalolateral angles extending cephalad as far
as or slightly beyond mesal portion; lateral margins slightly carinate; pri-
mary setae short; secondary setae few, 2 to 4 pairs.

Abdomen, Sclerotized areas distinct, supplementary lateral sclerotized
areas variable in number and shape; secondary setae fairly numerous, some
long and slender, some short and fine; eusternum of ninth segment bearing
2 groups of 3 long setae caudally; pygopod usually bearing 12 setae, 6 on a
side; median hooks of fifth segment with 4-5 setae, the distal one stout and
directed mesad, the others more slender and directed laterad; inner hooks
with 5-8 setae, 3-5 of which are notably stouter than the others; spine of inner
hooks minute to obsolete.

Measurements, Total length of larva, 17-19 mm; width at third abdomi-
nal segment, about 1.9 mm; diameter of stemma I, 0.32-0.35 mm; diameter
of stemma IL, 0.27-0.34 mm; distance between stemmata I and I, 0.23-0.27
mm; length of fronto-clypeo-labral area, 1.28-1.35 mm; width of fronto-
clypeo-labral area, 142-150 mm; length of pronotum, 1.83-2.08 mm; width
of pronotum, 2.85-3.20 mm.

. PFiANhE ke, ; So}
7 iwi

i / —

ed.

0.5 mm

65

Fics. 59-64, C. willistoni, new subspecies, third instar larva; specimen is from Il mi. north-
east of Hudson, Stafford Co., Kansas; Fic. 59, head and pronotum, dorsal aspect; Fic. 60, left
antenna, ventral aspect; Vic, 61, third abdominal segment, lateral aspect of left half; ventrolateral
suture dotted; Fic, 62, dorsum of fifth abdominal segment, dorsal aspect; stippled area is
weakly sclerotized; Fic. 63, ninth abdominal sternum, ventral aspect; Fic. 63, pygopod, dorsal
aspect,

172 THe University ScriENcCE BULLETIN

C. willistoni, new subspecies, described later—third instar larva (Figs. 59-64)

Color. Head dark brown to red-brown with brassy, green, blue-green, or
purplish reflections; labrum red-brown; pronotum with disk usually dark
red-brown (rarely red-brown in a pattern), with cephalolateral angles red-
brown (rarely yellow-brown); mesonotum dark brown anteriorly, yellow-
brown posteriorly; metanotum yellow-brown; basal two antenna segments
red-brown to yellow-brown, distal two segments red-brown; mandibles red-
dish brown basally, with apices and retinaculum black; maxillae and labium
yellow-brown. Dorsal cephalic and pronotal setae white, other setae yellow-
brown.

Head. Setae on dorsum medium to short and prominent; diameter of
stemma II subequal to that of stemma I and slightly greater than distance
between I and II; fronto-clypeo-labral area about as wide as long; U-shaped
ridge on caudal part of frons with 2 setae; antennae with distal segment 0.6
to 0.7 as long as penultimae; proximal segment shorter than second segment;
proximal segment with 5-6 setae, second with 9-10, third with 2, and distal
with 3; maxillae with 3 setae on mesal margin of proximal segment of galea
and 5 on distal segment; maxillary palpus 3-segmented, palpifer with 7 setae,
penultimate segment of palpus with 1 seta; distal segment of labial palpus
with 1 ventral seta, penultimate segment with 3 spurs and 2 setae on either
side of spurs; ligula with 4 (occasionally 6) setae.

Thorax. Pronotum with cephalolateral angles not extending as far cepha-
lad as mesal portion; lateral margins slightly carinate; primary setae medium
to short, secondary setae few, 3 to 4 pairs.

Abdomen. Sclerotized areas distinct, secondary lateral sclerotized areas
variable in shape; secondary setae few, some long, some short; eusternum of
ninth segment bearing 2 groups of 3 long and 1 shorter seta caudally; pygo-
pod usually bearing 16 setae, 8 on a side; median hooks of fifth segment
with 3-4 setae; inner hooks with 2 setae on a shoulder, spine long, over one-
third the length of the hook.

Measurements. Total length of larva, 18 mm; width at third abdominal
segment, about 2.3 mm; diameter of stemma I, 0.30-0.35 mm; diameter of
stemma II, 0.27-0.34 mm; distance between stemmata I and II, 0.24-0.35 mm;
length of fronto-clypeo-labral area, 1.42-1.70 mm; width of fronto-clypeo-
labral area, 1.42-1.76 mm; length of pronotum, 1.66-2.02 mm; width of
pronotum, 2.56-3.03 mm.

The larvae of C. fulgida and C. willistoni are segregated to couplet 7 in
Hamilton’s (1925) key; they can be separated from the species in that couplet
by the following key:

1. Setae of dorsum of head and pronotum brown ...........2-2....-2-2:2-000-0--- sex guttata
Setae of dorsum of head: and pronotum: white-.. = e 2

BIoNoMIcs AND ZOOGEOGRAPHY OF TIGER BEETLES 173

2. Median hooks of fifth abdominal segment with two setae fulgida
Median hooks of fifth abdominal segment with more than three setae .......... 3
Saemtenmal: scape shorter thaw) pedicel). c:s2..aoce- nse -ceeececseacececcenencceceeansiezecise willistoni
Scape and pedicel of antenna subequal in length ~.....-.200- cam pestris

The larvae of C. circumpicta, C. nevadica, and C. togata are segregated to
couplet 24 in Hamilton’s (1925) key; they can be separated from the species
in that couplet by the following key:

1. Inner hooks of fifth abdominal segment with nine or ten setae —...... mar ginata
Inner hooks of fifth abdominal segment with fewer than nine setae .............. 2
ye scape and pedicel of attennavopalescent White <2... ccseeesccacseecsee cease! togata
Scape and ipedicel ob antenna brOwmnre.: 422. ee 3
BaP airennali pedicels with tem, to [2 setae. ooo eee abdominalis
PMtcninalypediceh with: fewer than tem setae ...2 2 4
4. Cephalolateral angles of pronotum not extending as far cephalad as
mesal portion; pronotum less than 2.7 mm wide .............222-2...20.--2---.--- nevadica
Cephalolateral angles of pronotum extending as far cephalad as mesal
portion; pronetum more than: 2.8 mim wide -.0222.)-200-2o. circum picta

Tue Larva—Bionomics. Many brief notes and papers have appeared con-
cerning the bionomics of cicindelid larvae, some of which will be mentioned
later. Among the more important are Zikan’s (1929) large paper on South
American species, Shelford’s (1908) paper, Huie’s (1915) work on C.
campestris, Friederich’s (1931) detailed study of eyes, and two papers by
Criddle (1907, 1910).

Sclerotization. The first instar larva of C. togata, upon hatching, is about
3 mm long. The body is light straw yellow except for the tips of the man-
dibles and the large stemmata, which are orange, and the meso- and meta-
thoracic, abdominal, and leg sclerites and the small stemmata, which are
light gray. The two hind pairs of legs are slightly darker gray than the
front legs. After about four hours the above mentioned sclerites are darker
gray, and the dorsum of the head and parts of the pronotum are gray. The
gray of the mandibles has spread basally to about half their length. In about
six more hours, the body sclerites are gray-brown, and the top of the head is
dark and iridescent cupreous and green. The venter of the head is light
brown. The mandibles are almost entirely black. In about 15-24 hours after
eclosion, the larva is complete sclerotized. The abdominal sclerites are light
gray-brown; the thoracic and leg sclerites (except the pronotum) are dark
gray-brown; the pronotum is light yellow-brown with a darker brown pat-
tern; the dorsum of the head is dark brown with cupreous and green reflec-
tions; the venter of the head is orange-brown; the mandibles are black; and
the other head appendages are light-brown.

Burrows and digging. In nature, after the first instar larva is sclerotized,
and if the soil is moist enough, it enlarges the cell that contained the egg into

174 Tue Universiry ScIENCE BULLETIN

a burrow. Shelford (1908) said that the larva of C. purpurea first digs the
burrow from the depth of the egg to the surface, then digs beneath this to a
depth of 10-15 cm. The exact method of digging the first burrow was not
determined in this study. The effect of soil moisture was noted in the
laboratory terraria. Soil that had begun to dry out was watered, and shortly
afterward numerous first instar larval burrows began to appear. If the soil
becomes too dry again, the larva plugs the burrow with soil, apparently to
conserve moisture.

In digging a burrow, the larva (of all instars; third instars are described
here), head downward, loosens some soil with its mandibles, using its legs
for support. Then the anterior end of the body is bent around in the other
direction, and the head and pronotum are placed shovellike under the loose
soil. The larva finishes turning right-side-up in the burrow and elevates the
soil up the burrow on top of its head and pronotum. Upon reaching the sur-
face, the larva may flip its head and pronotum backward, throwing the soil
several centimeters away (if the soil is moist and sticky, it is deposited in the
form of small pellets). Some species pack the soil around the entrance of the
burrow by turning the head upside down and pushing with the legs. When
the larva is beginning a burrow from the surface, as when it is introduced
into a jar of soil in the laboratory, slightly different tactics are used. The
thorax is humped, the larva supported by its front and especially hind legs
(the middle legs are normally held horizontally from the body to help sup-
port the larva in the center of the burrow and are useless for walking when
it is outside the burrow), in order to allow the mandibles, which slant upward
from the head, to dig into the soil. When the hole is about 1 cm deep, the
larvae may hold the abdomen in the air while using the legs to gain leverage.
The larva uses the method of digging upside down and backing out the hole
to flip the soil away until the hole is about 2 cm deep. At depths below that,
the method of turning around in the burrow, described above, is used. The
temporary bottom of the burrow is made slightly larger than the finished
diameter; soil is later “plastered” on the walls as the burrow descends. When
the burrow is completed, the larva clears the soil around the entrance of all
movable obstructions within a distance of about half its body length, forming
a slightly concave smooth area. Some of the same observations on digging
have been made by Criddle (1907), Fackler (1918), Enock (1903), and
Macnamara (1922). Shelford (1908) noted an exception: the burrow en-
trance of C. macra is ragged at the edge, rather than smooth. Some authors
(Macnamara, 1922; Bryson, 1939) have noted an increase in burrowing
activity after rains, and Criddle (1907, 1910) noted that most digging is done
at night except late in the season when the nights are cold.

The depth of the burrow varies with many factors, including the instar,
species, weather and climate, season, soil moisture, and possibly type of soil.

BroNomics AND ZOOGEOGRAPHY OF TIGER BEETLES 175

Shelford (1908) gave depths of (presumably) third instar burrows of eight
species ranging from 5-90 cm, depending on the species, temperature, and
possibly soil moisture. Criddle (1907, 1910) gave depths for six species rang-
ing from 15-200 cm, depending on the instar, species and season (larvae
deepen their burrows before hibernation). Zikan (1929) showed burrow
depths of the species he studied. The depths of burrows measured in this
study are shown in Table 2.

Tasce 2. Depths of larval burrows measured (in cm.) in the field (F) and lab-
oratory (L).

Instar
Species First Second Third
Gs GRALDTITIE oe A ee ree 6.5-9 (F) 13.5-16.5 (F) 11-29.5 (F)
6-14 (L)
Gmduodecameutiaia pe 1.5-3.5 (F) 3.5-4.5 (F)
2-2.6 (L) 4 (Z) 6-11 (L)
(Cs: LEGION, oe Oe oR a 13 (F)
Co ERLE A 8 Se a eee 18-28 (F) 22-35 (F)
(Cp FEET ee Se ne eee 2.5-4 (F) 6.5-10 (F) 10-18 (F)
5) (GL) 10-35 (L)
GCL TITS EO TD ee nee ee AS 7-13 (F) TZ (GE) 16-35 (F)

The diameter of most larval burrows is slightly greater than the diameter
of the head and pronotum, although Zikan (1929) and Williams and Hun-
gerford (1914) showed the terminal part of the burrows of some species of
other genera as being enlarged, and Shelford (1911) showed a similar burrow
of C. limbalis. The larva can easily turn around inside the burrow. In doing
this it bends the anterior end of the body dorsally using the legs, forcing the
head past the dorsum of the abdomen (Shelford, 1911).

The burrows of most species are approximately straight and perpendicular
to the soil surface, which may be vertical, horizontal, or oblique. However,
there are many exceptions and much intraspecific variation. Burrows that I
have dug up in the field or laboratory are aften curved, oblique to the surface,
or both. Such variations have also been shown by some of the above authors
as well as Hood (1903). Reineck (1923) noted that the larvae of C. silvicola
will dig around large obstructions in their path.

The burrows of many species open flush with the surface or with a slight
depression as noted above, but others are quite different. Shelford (1908)
noted that the burrow of C. lepida, which is found in dry sand, has a funnel
at the entrance formed by the action of gravity on the sand. Ortenburger
and Bird (1933) noted similar craterlike entrances to burrows of C. cupra-

scens. Shelford (1908) and Criddle (1910) found that C. formosa builds a

176 Tue Universiry ScrENCE BULLETIN

pit about 4 cm wide and 2.5 cm deep. From about half way up one side, the
burrow begins horizontally, then curves downward. Similar burrows were
noted by Dow (1916) for four species, including C. lepida and C. scutellaris.
However, he did not collect larvae for identification, but used the uncertain
method of placing wire screen cages over the burrows to catch the emerging
adults (uncertain because the adult does not necessarily dig its way out along
the old larval burrow; larval burrows of different species may be close
together, and the adult from one may emerge near another). Dow’s de-
terminations are thus in doubt, since Shelford (1908) and I have noted differ-
ent burrow entrances for C. lepida and C. scutellaris, respectively.

Lesne (1897, 1921) and Reineck (1923) reported that C. Aybrida and C.
silvicola, which burrow in sloping areas, build a semicircular lip above the
entrance (apparently to deflect rain) and a pit below the entrance.

Shelford (1908) said that C. /imbalis builds a chimneylike structure about
6 mm high at the entrance. Macnamara (1922) stated that the larvae of C.
tranquebarica build a similar structure when necessary, as when an immov-
able obstruction is present. Hamilton (1925) said that an unidentified species
from Colorado builds a chimney about 2.5 cm high in its early instars. Zikan
(1929) showed a similar structure on a burrow of Megacephala brasiliensis.
A second instar burrow of C. fulgida that I found in northern Kansas was
situated among dead Distichlis stems; the larva had built a chimney about
0.5 cm high to elevate the entrance above these obstructions (other burrows
of the same species had no such structure). This chimney-building habit is
best developed in C. willistoni. Ortenburger and Bird (1933) first noticed
this phenomenon in Oklahoma, but did not know which species is involved;
I have studied it in Oklahoma and Kansas (in a new subspecies). In this
species the larva always builds a chimney (called a turret by Ortenburger and
Bird) relatively much higher than any other Cicindela known and adds two
projections at the top, giving the apex a saddlelike appearance (Figs. 65-67).
First instar turrets are usually 1-3 cm high; second and third instar turrets
are 1.5-4 cm high (one second instar turret was 5.5 cm high). The apical
projections have no special orientation. In the laboratory, larvae build shorter
turrets and never add the projections. The function of these turrets is uncer-
tain. They do not seem to serve for flood protection since they crumble in a
heavy rain. Another possibility is to elevate the larva above the surface,
which is the hottest part of the environment; however, when conditions
become very hot and dry, the larvae usually plug their burrows and remain
underground. A third possibility is that some insects that could serve as prey
may be attracted to such projections as landing places.

At various times (after feeding, in unfavorable weather, before hiberna-
tion or estivation, before molting or pupation) the larva may close the burrow
with a plug of soil. This is done by scooping a small amount of soil from the

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 177

Fic. 65. Turret of third instar larva of C. willistoni, 2.5 mi. southwest of Plainview, Woods
Co., Oklahoma. Fic. 66. Turret of second instar larva of C. willistoni, 11 mi. northeast of
Hudson, Stafford Co., Kansas. Fic. 67. View of a number of turrets of larvae of C. willistoni
on salt flat 2.5 mi. southwest of Plainview, Woods Co., Oklahoma; insect net is about | m long.

wall or bottom of the burrow onto the head and applying it to the entrance
repeatedly. The burrow is unplugged in reverse fashion, the soil from the
plug being plastered onto the walls. The thickness of the plug varies from
less than one to several centimeters; it is thicker if the larva spends long
periods underground (hibernation, pupation, etc.). I have noticed that the
larvae in terraria often plugged their burrows temporarily soon after the soil
was moistened. In nature this reaction probably saves their burrows from
being flooded during rains.

Because the burrow diameter is nearly the same size as the head width,
one can easily tell the instar of the occupant if one knows what species is
involved. The latter reservation is necessary because the first instar burrow
of a large species may be as large as the second or third instar burrow of a
small species.

Food and feeding. After the burrow has been dug, the larva assumes a
position at the entrance to lie in wait for food. The head and pronotum

178 Tue University SciENcE BULLETIN

together form a circular lid or plug to the burrow and are usually colored so
that the larva is very well camouflaged. Surprisingly, only one species of larva
known from saline habitats, C. togata, has a light coloration of these parts of
the body. However, some species, C. nevadica in this study, C. silvicola
(Reineck, 1923), and C. sexguttata, accumulate a thin layer of soil on the
pronotum and thus are even better camouflaged.

In maintaining its position at any point in the burrow, the larva uses its
legs and abdomen. The front and hind legs are directed ventrad and the
middle legs dorsad. The abdomen assumes a sigmoid position; the first five
segments are bent ventrad, giving the larva a swayback appearance; the
spines and large setae on the fifth abdominal tergum dig into one burrow
wall; the rest of the abdomen is directed forward perpendicular to the axis of
the burrow; and the spines on the tenth abdominal segment dig into the
opposite burrow wall. Thus the larva is supported at eight points: the six
legs and the fifth and tenth abdominal segments. To move down the burrow,
the larva straightens its abdomen, flexes its legs, and if the burrow is vertical,
falls with the aid of gravity (if the burrow is not vertical, the legs are used) ;
to move up the burrow, the legs alone are used, although Shelford (1911)
said that the abdomen assists also. Larvae are easily frightened and will drop
down their burrows at the slightest footstep or movement of a human
observer.

The larva usually lies in wait at the burrow entrance continually during
favorable weather, by night as well as by day. Sometimes however, larvae
close their burrows at night, and they frequently do so after feeding. If
suitable prey alights or crawls over the burrow, it is very quickly seized by
the larva’s mandibles. A click is often heard when the prey is seized, appar-
ently caused by the mandibles striking together. Wigglesworth (1929), in
experiments on unidentified African species (probably of several genera),
said that the larvae will not strike unless certain tactile setae on the back of
the head and front of the pronotum are stimulated. Enock (1903), however,
stated that the larva of C. campestris strikes when the prey is within 1.5 cm.
He also gave a good description of how the larva strikes: the larva throws
its body backward half out of the burrow; the median hooks of the fifth
abdominal tergum dig into the edge of the burrow to anchor the larva.
Friederichs (1931) noted the importance of vision in catching prey and said
that the larva strikes in the above manner when the prey is between 3 and 6
cm away. Of course, if the prey should land very close to or directly on the
head of the larva, it does not strike in the above manner, but simply snaps its
mandibles shut on the prey. Probably vision is most important in the day
and tactile senses at night for prey capture. The hooks of the fifth abdominal
tergum, directed anteriad, serve to help prevent the larva from being pulled
out of the burrow by strong prey.

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 179

Those species that build pits below the burrow entrance appear to use
them as traps. Criddle (1910) and Shelford (1908) reported this for the
larva of C. formosa; throwing small ants in the pit resulted in their immedi-
ate capture, according to Criddle.

The prey, if small enough, is usually dragged down the burrow to be
eaten; if it is too large, it is held at the top of the burrow, but large prey items
may be taken down the burrow after they have been partially eaten. A third
instar larva of C. circumpicta that was observed in the laboratory sometimes
quickly and sometimes slowly (in one case not for one hour) dragged the
prey part way or all the way down the burrow; then it often turned around
in the burrow before eating. Sometimes (in the laboratory) a larva, instead
of eating the prey, would return to the entrance within a short time and lie
in wait again. Two larvae of C. willistoni were thus “fed” repeatedly; one
dragged 21 adult Anagasta kuehniella moths down its burrow, and the other
30, within one hour (three days later the first larva had thrown 13 moths out
of its burrow, uneaten, and the other did the same with 15 moths three days
after that; other uneaten moths were found when the burrows were dug up
a month later after the larvae had died). Such behavior is probably not
normal. In eating, the larva manipulates its prey slightly with its mouth-
parts. Wigglesworth (1929) said that larvae eject fluid (with a pH of 6.26.4
and containing trypsin) from the mid-intestine which predigests the prey;
the larva then consumes the liquified tissues, straining out solid particles
with setae on the labium. The hard cuticular portions of the prey are not
eaten and are disposed of, either by tossing them out of the burrow or keep-
ing them in the end of the burrow. The latter method has only been found
to be used by C. silvicola (Reineck, 1923), Megacephala brasiliensis (Zikan,
1929), and Amblychila cylindriformis (Williams and Hungerford, 1914).
Unacceptable prey is tossed away.

Larvae will eat nearly any small arthropod they can catch. Literature
references indicate that food eaten in nature includes caterpillars and other
insect larvae, butterflies, moths, flies, beetles, dragonflies, ants, spiders, centi-
pedes, and land crustaceans. In captivity, larvae have been fed houseflies,
ants, sowbugs, small beetles, decapitated woodboring larvae, ant pupae, thy-
sanurans, caterpillars, small pieces of raw meat, and apple (it is unlikely that
apple was eaten). Criddle (1910) said that larvae of C. formosa do not accept
Hemiptera. I have only once found a larva eating in nature, a first instar
larva of C. willistoni that had a salticid spider at the top of its turret. Dr.
F. E. Kurczewski (personal commun.) has seen larvae of an unidentified
species eat a wasp, Tachysphex terminatus. In the laboratory I have fed
larvae small spiders and phalangids (a large phalangid was refused), may-
flies, nymphal short- and longhorned grasshoppers, nymphal cockroaches
(Supella supellectilium ), mirids (Adelphocoris rapidus and others), nabids,

180 Tue Universiry SciENcCE BULLETIN

cicadellids (nymphs and adults), an immature fulgorid, aphids, chrysopids,
a small cantharid, trichopterans, various caterpillars (including arctiids,
geometrids, and a noctuid), small adult moths (including Anagasta kueh-
niella), small adult flies (including a culicid, a syrphid, a bombyliid, a
trypetid, a calliphorid, and other muscoid species), and ants. An immature
cercopid was not readily accepted. Silphid, coccinellid, and chrysomelid
larvae were rejected (one coccinellid larva was manipulated by the larva’s
mouth-parts for 10-15 seconds, then forcibly flipped out of the burrow un-
harmed). Sawfly larvae (Tenthredinidae ?) were sometimes refused and
sometimes eaten. A small bee was refused by one larva but accepted by
another. Goldsmith (1916), Macnamara (1922), Reineck (1923), Schaupp
(1879b), and Shelford (1908, 1911) mention that larvae may eat each other
in captivity. Some authors attribute cannibalism to crowding and others to
hunger. I have noted cannibalism once; a beheaded larva was found outside
its burrow in a jar that contained three larvae. Dr. F. E. Kurczewski (per-
sonal commun.) has seen on three occasions an unidentified species of larva
in Groton, Tompkins County, New York, eating adult C. formosa, starting
at the abdomen. As will be mentioned later, I once found a C. circumpicta
with its abdomen missing. We may thus conclude that nearly all small arthro-
pods are acceptable as food to larval Cicindela. The time since the last meal
probably also affects the acceptability of food, but has not been investigated.

Macnamara (1922) was surprised to find how seldom larvae catch prey in
nature; in over ten hours of watching a group of C. tranquebarica larvae, he
only saw one small ant eaten. He also said that the larvae throw their sem)-
liquid excrement away from the burrow.

The time interval between meals depends on the individual, the instar,
the size of the meal, and probably also on the species and temperature. Some
larvae in the laboratory ate Anagasta kuehniella moths about every other
day, while others ate very irregularly and often fasted for a number of weeks.
The rough treatment of being caught and transported and the artificial con-
ditions in the laboratory apparently made some larvae refuse to come to the
tops of their burrows for food, with the result that they starved to death. In
nature, of course, nothing is eaten during estivation or hibernation.

Molting and stadia. First instar larvae only need one meal (if it is large
enough) to store enough energy to molt; second and third instar larvae need
several meals. As mentioned above, the larva usually closes its burrow while
it molts. Shelford (1908) stated that the larvae he studied take about five to
seven days to molt; i.e., the burrow is closed that long (as far as is known, no
one has seen the actual molting process). Zikan (1929) found that some
tropical genera have five instars and close their burrows for two to four weeks
during molts. Huie (1915) reported that C. campestris larvae closed their
burrows about ten days while molting at the start of the second instar. I have

BIoNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 181

Tasce 3. Length of stadia in days of larvae reared in the laboratory. Numbers

between dots are numbers of days spent between the two events in the respective

columns; numbers in parentheses are number of Anagasta kuehniella adults (A)
or larvae (L) eaten.

First Instar Second Instar Third Instar

Species burrow burrow burrow burrow burrow burrow
open closed open closed open closed
GANCiiCnIn picid ye e— |2——e—_ 1/2 —_e—_ 34 _@—___ 1 (5
8(A)
C. duodecimguttata _.. e— 47—e
Gap cuglica eee SS]. Ui

o——____—_ 148 ——

GORA eee ee ee FV 2 1 SS HSS
(8A) (14A)
C= 3 SO 2 ISO
(SA) (15A, IL)
C= 10-541 SC
(7L) (13A)
20
31—e
(4A)
a
(7A, 2L) (9A)

SS — eo
Sh

193 —@

(13A)

C. willistoni eo—_—————__ | ———_e—___ | > ———_e—_ 12

eo—— 9——e—17
(9A)
118
341——e
(40A)
@e— 8——e—__ 7/—_-®
(5A)

found that a larva of C. duodecimguttata in the laboratory closed its burrow
four days to molt from first to second instar, while a C. willistoni took seven
days to make the same molt (after eating one Anagasta kuehniella three days
before closing its burrow). These seem to be minimal times; other larvae
closed their burrows for many weeks or months, then appeared as the next
instar. The enlargement of the burrow to fit the new instar occurs after
molting.

182 Tue UNniversiry ScrENCE BULLETIN

The interval between molts (stadium) varies with the individual, the
species, the instar, abundance of food, amount of favorable weather, and
temperature. Shelford (1908) stated that the first instar larva of C. limbalis
molts about three to four weeks after hatching; the other stadia are much
more variable, the second being about five weeks and the third about ten to
eleven months under favorable conditions. Huie (1915) said that the first
stadium of C. campestris is about six weeks. Zikan (1929) found that the
stadia lasted from one to four months, the first being the shortest. Stadium
lengths and molting intervals found in this study are shown in Table 3.
As can be seen, there is much individual variation, much is probably due to
the artificial laboratory conditions. There seems to be much less variation
in the amount of food consumed in a stadium than in the length of stadia.
The number of moths eaten in normal second stadia ranged from 5-9, and
9-15 for third stadia; the first instar probably can molt after eating one moth.
The average weight of an Anagasta kuehniella adult is about 0.01 g; thus
the amount of whole moths needed for the stadia is: first stadium, 0.01 g;
second stadium, 0.05-0.09 g; third stadium, 0.09-0.15 g. Because of the in-
dividual variation and the small number of larvae reared in this study, few
definite conclusions can be drawn about the lengths of stadia. For C. togata,
the minimum time for the second stadium under laboratory conditions is
about four to five weeks. As will be seen later, hibernation and estivation
greatly lengthen the stadia in which they occur.

Activity. Field observations indicate that some larvae of most species are
active throughout the warm months (in Kansas, from about March through
October). This is partly because of the overlap of generations and long life
cycle of cicindelids. If conditions become severe (high temperature or dry-
ing out of the soil), estivation usually occurs. As mentioned above, larvae
are active day and night, but probably not on cold nights.

Those species that live so near the margins of saline habitats that their
burrows are near vegetation or on the side of a bank or hummock probably
are able to be active for a longer time than species living on bare flats; the
vegetation probably reduces the extremes of temperature and temperature
fluctuation found on bare flats. However, there is the possibility that larvae
of bare flats compensate for this by digging deeper burrows.

Microhabitats. Many authors have noted that the larvae of one species or
another occur only in limited or characteristic areas. Shelford (1911) made
a detailed study of C. limbalis, C. tranquebarica, and C. sexguttata, and
found that the larvae of these species are found in quite restricted areas char-
acterized by vegetation, exposure, slope, and kind and moisture of soil. In
laboratory experiments, he found that the adults choose optimum micro-
habitats for oviposition (see more complete discussion under the adult).
Microhabitats of larvae in this study are shown in Table 4.

BIoNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 183

Tase 4. Microhabitats in which larvae have been found. The greater the num-
ber of X’s the more frequent the occurrence of larvae.

Near Sloping Moist Dry Small Near hum-

Species water, bank, bare salt bare salt flats, near mocks, among
sand bar creek bank flats flats margin vegetation

C. circum picta XXX XX XXX

C. duodecimguttata XX XXX

C. fulgida* XX XX XXX

C. nevadica X XX XXX

C. togata Xx XXX XX XX XX

C. willistoni XXX XX

* Few larvae have been collected of this species.

As noted above, larvae usually dig their burrows where their eggs were
oviposited; however, a number of workers have found that the larvae of some
species will leave their burrows if conditions are unsuitable. Shelford (1908,
1911) stated that the larvae of C. hirticollis and C. repanda often leave their
burrows if the soil becomes too dry or too wet; under extreme conditions,
such as flooding of the habitat, a small percentage of larvae of the other
species he studied left their burrows. Montgomery and Montgomery (1930)
also noted larvae of C. Airticollis leaving their burrows, and Hefley (1937)
saw larvae of C. cuprascens moving to a cooler, moister place.

The question of how species of fluvial habitats, whose larvae inhabit sand
bars and stream banks, survive flooding is intriguing. Criddle (1907) thought
that both larvae and adults of C. duodecimguttata must often die in hiberna-
tion when their habitat is flooded. Hamilton (1885) noted that hibernating
adult C. repanda survived flooding for seven days. I have seen active larvae
and swarms of adults in areas that had been recently flooded, so cicindelids
most certainly do survive flooding. Possibly air trapped in their closed bur-
rows is instrumental in their survival. Floods that cause much erosion prob-
ably wash out and kill many beetles.

Larvae can tolerate a certain amount of disturbance of their habitat, such
as cattle walking and disfiguring the soil or plowing by man, but cicindelids
are generally absent from areas where such disturbance is frequent or
prolonged.

Predators and parasites. Predators of larvae are few; Grandi (1951) stated
that histerid beetles of the genus Saprinus penetrate larval burrows and eat
the occupants; Zikan (1929) said that birds occasionally eat larvae, and that
ants (Solenopsis geminata) frequently dig into the larval burrow, sting the
larva, and eat it; Shelford (1911) reported that ants may overcome a C. lim-
balis larva, and after chewing off its antennae and tarsi, drag it from the
burrow. No evidence of predation on larvae was seen in this study.

184 Tue Universiry SciENcE BULLETIN

A number of larval parasites are known. Reineck (1923) reported that
C. silvicola larvae are occasionally found dead and covered by a moldlike
substance; however, he did not mention that the larvae might have died from
other causes and their dead bodies later attacked by fungi. Shelford (1913a)
studied the life history of a bombyliid fly, Anthrax analis, which he found
parasitizing C. scutellaris and C. hirticollis. The adult fly oviposits in larval
burrows; the fly larvae attach to the thorax or abdomen of the cicindelid
larva and feed from the outside. After the cicindelid larva has built its pupal
cell (see next section), the last instar fly larva pupates. The fly pupa digs its
way to the surface and the adult emerges. Shelford also said that the larvae
of C. limbalis were parasitized by a larva somewhat different from Anthrax
analis. Hamilton (1925) stated that four larvae (out of 34 collected) of C.
obsoleta (?) were parasitized by a total of seven dipterous larvae, which he
assumed to be Anthrax analis. Frick (1957) reported that he saw a small
black bombyliid (not collected) oviposit in three larval burrows of either
C. haemorrhagica or C. pusilla; the burrows were soon closed by blowing
sand, and in a few weeks no larvae could be found. Williams (1916) found
some larvae of C. punctulata (?) parasitized by orange maggots, probably
Anthrax. Batra (1965) saw Anthrax analis apparently ovipositing in burrows
of cicindelid larvae. James K. Lawton (in litt.) has found larvae of C. ¢ran-
quebarica in Wisconsin parasitized by Anthrax analis; adults were seen ovi-
positing one to three times in larval burrows. Dr. F. E. Kurczewski (per-
sonal commun.) has seen Anthrax albofasciatus ovipositing in Cicindela
larval burrows in New York. I have often seen adults of Anthrax analis in
saline habitats and once saw one oviposit repeatedly in two burrows of second
instar C. togata (?) before being collected. The fly lands beside a burrow,
stands high on its legs, and bends the tip of its abdomen forward, flipping
eggs at the entrance. I have found no parasites on any larvae that I have
collected throughout the central United States.

Criddle (1919) stated that the chalcid wasp, Tetrastichus microrhopalae
(Eulophidae), was reared from C. limbalis larvae. Williams (1928) reported
that two species of tiphiid wasps of the genus Pterombrus (misspelled
“Pterombus’’) parasitize larvae of Cicindela sp. and Megacephala affinis in
Brazil. A number of species of the tiphiid genus Methocha have been found
to parasitize cicindelid larvae: M. tchneumonides in Europe (Bouwman,
1909; Champion and Champion, 1914; Champion, 1915; Pagden, 1925; and
others), M. striatella and M. punctata in the Philippines (Williams, 1919),
M. japonica and M. yasumatsui in Japan (Iwata, 1936), M. sp. in Brazil
(Zikan, 1929), M. californicus in California (Bridwell, 1912; Burdick and
Wasbauer, 1959), and M. stygia in Massachusetts (Williams, 1916) and
Wisconsin (James K. Lawton, zn litt.). The antlike female wasp approaches
a burrow with the larva lying in wait at the entrance and induces the larva

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 185

to seize it (in other cases the wasp avoids the mandibles of the larva). Then
the wasp curls its abdomen under the heavily sclerotized head and stings the
larva in the tender, membranous neck region, paralyzing it. After secondary
stingings, the wasp lays an egg on the venter of the thorax or abdomen.
Then the burrow is closed partially or level with the surface by the wasp.
The wasp larva feeds from the outside and may change positions. Pupation
occurs off the remains of the host. No parasitism by Hymenoptera was noted
in saline habitats; however, a Pterombrus rufiventris was collected on salt
flats in northern Kansas. The life cycle of this species is unknown.

The pupal cell. Before pupation the third instar larva closes its burrow
(normally) and digs a chamber for pupation, the pupal cell. The diameter of
the pupal cell is about twice that of the larval burrow. It may simply be an
enlargement of the larval burrow or adjoin the larval burrow directly (C.
campestris, Enock, 1903; C. limbalis, C. lepida, C. punctulata, Shelford, 1908).
In other species, a tunnel the same diameter as the larval burrow and branch-
ing off the latter is built between the pupal cell and the larval burrow (C.
limbata, Criddle, 1910; C. scutellaris, Shelford, 1908). The soil from these
new cavities is used to plug much of the larval burrow. The walls of the
pupal cell are made smooth by the larva with the mandibles and ventral side
of the head. The shape of the pupal cell varies in different species; in some
it is only about twice as long as wide and oval; in others it is very elongate
and slightly bowed. There is some intraspecific variation in the shape and
position of the pupal cell relative to the larval burrow. The pupal cell is built
relatively close to the surface, often as close as 2.5 cm. Shelford (1908) found
in experiments that the depth of the pupal cell is influenced by soil tempera-
ture, being greater the warmer the soil. Some typical pupal cells that I found
in laboratory-reared individuals are shown in Figures 68-71.

Tue Pura. Having completed construction of the pupal cell, the larva
becomes quiescent, head up, with the thorax and abdominal hump resting
against the bottom or side of the cell. At the end of one to three weeks, it is
not able to move its legs. The abdomen gradually thickens and turns a trans-
lucent cream color, indicating internal changes. The tubercles of the first
five abdominal segments of the pupa (see below) can be seen through the
larval skin folded toward the midline. A few days after these changes, the
larval cuticle splits along the frontal suture of the head and the dorsal thoracic
midline. Contractions of the ventral muscles flex the body slightly, and the
head is gradually withdrawn. The abdomen is freed by later movements.
Ecdysis is accomplished in only a few minutes. Just after emergence, the
pupa is only slightly shorter than the larva, but it soon contracts to its normal
size. The above events have not been seen personally and have been taken
from Shelford (1908) and Enock (1903). However, one larva of C. circum-
picta became quiescent after digging the pupal cell the previous two days;

186 Tue University ScrENCE BULLETIN

ES

68 69 70

b
(e
—|
il | 5 cm

scale, Figs.68 -7|

0.5 mm =
WS ry 74 | 1mm

Fics. 68-69. Pupal cells and larval burrows (partly plugged) of C. circumpicta, from 1 mi.
northwest of Fredonia, Wilson Co., Kansas, shown in laboratory rearing jars. Fic. 70. Pupal
cell, adult escape burrow (a), and portions of larval burrow (b), partly plugged, of C. togata,
from 3 mi. west, 2 mi. south of Barnard, Lincoln Co., Kansas, shown in laboratory rearing jar.
Fic. 71. Pupal cell with newly emerged adult (a), pupal skin (b), and larval skin (c) of C.
nevadica, from 11 mi. northeast of Hudson, Stafford Co., Kansas, shown in laboratory rearing
jar. Fics. 72-74, C. cimcumpicta pupa; Fic. 72, newly emerged pupa, from 11 mi. northeast of
Hudson, Stafford Co., Kansas, ventrolateral aspect; Fic. 73, right eye, caudal aspect; Fic. 74,
labrum and mandibles of pupa, from 1 mi. northwest of Fredonia, Wilson Co., Kansas, ventral
aspect; pupa 1s about 18 days old.

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 187

the period of quiescence lasted 28 days. A freshly emerged pupa of another
individual of the same species was dug up 16 days after the larva was last seen.

The newly emerged pupa of C. circumpicta (Fig. 72) is creamy white;
the thoracic region is darker yellow, and part of the eyes contains light tan
pigment arranged in columns (Fig. 73). The first four abdominal segments
bear paired dorsal tubercles, each with an apical ring of setae. Those of the
fifth segment are larger. The setae and tubercles function to hold the pupa
off the substrate. Shelford (1908) gave a brief summary of pupal color
changes in C. purpurea: in about ten days the eyes have become completely
dark brown; about two days after that, the tips of the mandibles and man-
dibular teeth begin to darken; the darkening of the mandibles is complete in
one to two days; about 13 days after pupation, the tarsal claws begin to
darken; and 14 to 15 days after pupation, the proximal portion of the tibiae
and the outer margins of the trochanters begin to darken. In 1917, Shelford
gave a slightly more detailed schedule for C. tranquebarica: initial stages are
about the same as for C. purpurea; the darkening of the tibiae, which spreads
from proximal to distal parts, takes two to three days; about the time that
this occurs, the middle of the folded adult hind wings (appear as the tips of
the pupal wings) begin to darken; color centers on the last two abdominal
segments may develop just before emergence of the adult.

Four pupae of C. circumpicta, kept in Stender dishes, were observed dur-
ing all or part of their development. In the following schedule, the numbers
indicate the approximate number of days after emergence:

0-12 or 16: the eye pigment gradually becomes diffused and darker red-
dish brown until the eyes are uniformly dark brown.

14-18: tips of the labral and mandibular teeth and tarsal claws begin to
turn brown (the latter two usually start slightly before the labrum), a process
that takes about 1-1.5 days (Fig. 74); soon after this, the apical tibial spines
also begin to turn brown.

16-19: a metallic purplish and green iridescence begins to appear on the
frons, first antennal segment, and tibiae, later spreading over the head and
legs.

17-19: parts of the genitalia begin to turn brown (gonapophyses in @
| Fig. 75], aedeagus in ¢), the posterior margin of the eighth abdominal
sternum (of 2?) begins to turn brown, the proximal ends of the tibiae and
the margins of the trochanters become light brown, and the tips of the pupal
hind wings (=middle of folded adult wings) turn light gray; soon after this,
spots of light brown appear on the labrum at the bases of the submarginal
setae, the apices of the tarsal segments and the tips of the terminal maxillary
palp segments become light brown; the terminal few segments of the an-
tennae become light gray, the laciniae of the maxillae become brown, and
the clypeal region turns light brown. By now, light brown areas have

188 Tue Universiry ScteNcE BULLETIN

Cee

Fics. 75-76, C. circumpicta pupa; Fic. 75, Eighth, ninth, and tenth abdominal segments and
apices of hind tarsi of female, ventral aspect; pupa is about 19 days old; Fic. 76, metathoracic
(a) and second abdominal (b) terga of same, dorsal aspect; the middorsal position of the heart
is shown by dotted lines. Fics. 77-78, C. circumpicta adult, from 11 mi. northeast of Hudson,
Stafford Co., Kansas; Fic. 77, venation of right elytron (recently emerged), dorsal aspect;
C=costa, Sc=subcosta, R=radius,; M=media, Cu=cubitus; Fic. 78, mesonotum, metanotum

and first three abdominal terga, 15 hours after emergence, dorsal aspect; pn=pronotum, el=
elytron, s=scutellum.

appeared on the dorsum of the thorax and abdomen (Fig. 76); the beating
heart can easily be seen through the dorsal cuticle, its rate being somewhat
variable, with occasional stops, and averaging about 37 beats per minute.
The gray of the pupal hind wings spreads to about half their visible surface
and becomes darker; about half the tibiae become brown, and the tips of the
terminal maxillary palpal segments become dark brown; also the tarsal tips
may begin to twitch at or before this point. The gray of the antennae spreads
proximad; the pupal skin begins to shrivel, and the dorsal abdominal tuber-
cles become soft.

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 189

18-21: the tibiae and tarsi are completely brown (the proximal parts of
the tibiae are darkest), and the proximal ends of the femora begin to turn
brown. The lateral and posterior margins of the labrum begin to turn
brown, the proximal halves of the pupal hind wings begin to turn light gray
(the distal halves are very dark gray), the genitalia become darker, and the
last three abdominal sternal margins (2 ) are light brown. The rest of the
maxillary palpal segments become light brown, the bases of the first antennal
segments begin to turn brown, the terminal antennal segments are dark gray,
and the distal tips of the first, second, and third segments are light brown;
the trochanters are dark brown, the distal ends of the coxae and margins of
the coxal cavities are brown, and iridescent reflections are visible on all parts
of the head and eyes. Movements of the legs, maxillary palps, mandibles, and
genitalia may be seen within the pupal skin, and the abdomen may swell and
contract slightly. An hour or so later, the movements become more exten-
sive: the head and prothorax are flexed backward; the whole body may be
moved slightly (straightened); the mandibles, labrum, maxillary palps,
genitalia, and legs are flexed; abdominal movements continue; finally, the
legs and maxillary palps are moved freely. The adult may emerge at this
point or such movements may continue for several hours. After a flexing of
the head and prothorax, the pupal skin is apparently broken dorsally, because
several seconds later air bubbles can be seen in the fluid beneath the pupal
skin, and the nonbrown portions of the labrum and mandibles, which were
translucent, quickly become opaque creamy white. Movements of the abdo-
men, head, thorax, and appendages assist in extricating the adult from the
pupal skin. After about an hour, the head, antennae, front and middle legs,
and entire dorsum are free. The body setae are wet and matted. After about
two hours the adult has emerged completely. As implied above, the time
spent in the pupal stage varies from 18 to 21 or 22 days at a temperature
of 24.5°C.

Tue Aputt—Post Emergence Changes. Following emergence, hardening
and darkening of the adult cuticle is completed. Shelford (1917) described
the process in detail for C. tranquebarica and less completely for several other
species. At the time of emergence, the stage of sclerotization of C. circum-
picta is more advanced than in C. tranquebarica (see description of pupa just
before emergence). The newly emerged adult is creamy white except for the
above noted parts. The elytra are expanded before the pupal skin is com-
pletely off. About 1.5 hours after emergence, the elytral pattern becomes
visible because their future pigmented parts have a faint green metallic color.
The tracheae of the elytra are easily visible at this time; the typical venation
is shown in Figure 77. About 1.5 to 2 hours after emergence, the hind wings
are fully expanded and begin to change from translucent to transparent; the
veins begin to darken. About 3 to 4 hours after emergence, the distal wing

190 Tue UNIversiry ScIENCE BULLETIN

veins are noticeably darker, and about 4 hours after emergence, the hind
wings are folded. About 15 hours after emergence the head, prothorax, and
femora are light brown, and the elytra are very light brown except where the
white pattern will be. The meso- and metanota are creamy white with a few
gray spots, and the abdominal dorsum is creamy laterally and dark gray
mesally (Fig. 78). The antennal scape and pedicel are brown, the distal
ends of the third and fourth antennal segments are brown, and the sixth to
eleventh segments are dark gray. In about 3 to 4 hours, the head and pro-
thorax are dark brown and the elytra are brown except the future white pat-
tern, which is translucent creamy white. At about this time, if the individual
is a female, the terminal abdominal segments (eighth, ninth, and tenth)
begin to retract, a process that is completed about 40 hours after emergence.
About 24 hours after emergence, the elytral pattern begins to become more
opaque white. About 40 hours after emergence, the beetle can support its
own weight and stand. About 68 hours after emergence, the venter of the
abdomen is dark, and the antennae, which were held close to the body over
the back, are held out in the normal position perpendicular to the long axis
of the body. The beetle now becomes quite active, and in nature, adults
probably begin digging their way out of the soil at this stage. In captivity,
beetles will accept food four to seven days after emergence. About six days
after emergence, the elytral pattern becomes opaque white. In the field,
adults have been collected with the elytra soft and the pattern still opaque,
supporting the conclusion that they dig out of the soil about three days after
emergence.

Besides the above changes, a series of color changes, begun in the pupal
stage, occurs before the final adult color is acquired. Shelford (1917) studied
this in detail, and Huie (1915) noted that on the third day after emergence,
C. campestris is bluer than when more mature. In C. circumpicta, a number
of color forms occur, most individuals being either reddish, green, or blue. In
the early stages, all individuals show about the same colors: green, blue, and
purple on the head, thorax, proximal antennal segments, and legs in the
pupal stage and early post-emergence period. In a future green individual,
the changes are as follows: from about 4 to 17 hours after emergence, the
elytra become almost completely purplish; about 18 to 20 hours after emer-
gence, the head and pronotum, which were purplish, have become brassy
green, and the elytra have become purplish and blue; about 40 hours after
emergence, the blue of the elytra has increased, a process that continues for
over a day; about 70 hours after emergence, the elytra begin to acquire a
turquoise color; and about four days later there is more green than blue in
the elytral color, but the margins still have a purplish tinge; there is little
color change after this. An individual that finally had blue elytra and a
green-blue head and pronotum, first had purple-blue elytra and a blue-green

BroNomics AND ZOOGEOGRAPHY OF TIGER BEETLES 191

Taste 5. Percentage of adult C. f. fulgida in two color classes (dorsal color)
collected at four times of the year in the central United States, and sample size.

Time of coll. °% Bright red °% Purplish red N
Atnrileear lye May esc eee eet ees 72.5 27.5 40
Lats IMiewyeceidhy JhOU 22) ee 59.0 41.0 464
Hate gUNe=) lyst ese he eee 20.5 795 39
PATI SIS t= Septet bein eee eee eee eee 50.0 50.0 128

Probable scheme of color change in C. f. fulgida. The width of the band indicates
the abundance of specimens; O=bright red, X=purplish red.

OFOLOLOVORXSX
(0) 3€ 107 (©) ©) 2:€ ©) (O) (0) 2: () ) 2«
OQLOTOFOLO OLOVOVOTOVO ORG O ORG OKA
OFOVORS OL OLS OF OF ORS Oe OO UX OX OFX OO KORO XK XE OX
OCIOTORS C1010 OFOLOVORS OOO XX ORG ORK Xx
OCLORSOLO XS OLOZGOR OX
ORS OLORXEXGO OO
OO0O0000
OQLOLOLOLOIOEG OLOLOLOLO
OLOLOFOVORS
(oe)

April-ear. May | It. May-ear. June | It. June-July | August-Sept.

head and pronotum. An individual that finally had dark cupreous elytra and
a brownish head and pronotum was mostly green and cupreous three hours
after emergence; the green then increased for about a day, then a reddish
wine predominated for about two days before the final color appeared.

As Shelford (1917) noted, individuals of some species continue to change
color long after emergence, sometimes until death. I have found that this
occurs in C. fzlgida, at least in its southern subspecies, C. f. fulgida. I noticed
that most adults collected from northern Kansas in September 1963, and kept
in the laboratory over winter were dark purplish red (dorsally) in the spring
instead of bright red as they were when captured. When about 270 specimens
in my collection and 400 in the Snow Entomological Museum were categor-
ized according to color and time of year collected (Table 5), a trend was
noted for most specimens collected in the spring to be bright red and late
summer specimens to be dark. A chi square test of independence in a 4 x 2
table was performed on the original data (not °/%), with the null hypothesis
(Ho) being that the time of collection and color are independent. A highly
significant X* of 27.7 was obtained, rejecting the Ho and indicating that the
color is dependent on the time of collection. Three chi square tests with one

192 THe University ScrENcCE BULLETIN

degree of freedom were then performed: no significant difference was found
between the April-early May and the May-early June groups (X?=2.9); the
late June-July and August-September groups were significantly different
(X*°=10.6); the two early groups combined and the late June-July groups
were significantly different (X°=21.9). The bottom of Table 5 shows quali-
tatively a possible scheme for such data. As will be discussed in more detail
later, C. fulgida is called “double brooded”; that is, adults emerge from the
pupae in the late summer and fall, hibernate, emerge the next spring, and
gradually die out in the summer. Freshly emerged individuals are bright red,
gradually becoming darker.

Tue Aputt—Mating. Many species begin to reproduce soon after emer-
gence as adults, but, as will be discussed later, others hibernate as adults and
do not become sexually mature until the spring after they emerge. When
sexually mature, and if the weather is suitable, the adults do little else than
eat and reproduce. Mating has been observed by a number of workers in
the past and seems to be similar in all the genera (Mitchell, 1902, for Mega-
cephala carolina and C. ocellata;’ Shelford, 1908, for C. purpurea; Moore,
1906, for C. repanda; Lengerken, 1916, 1929, for C. hybrida and C. martima;
Goldsmith, 1916, for C. punctulata; Fackler, 1918, for C. cuprascens, C. hirti-
collis, and C. repanda; Zikan, 1929, for Cicindela, Prepusa, Euprosopus,
Tresia, Oxychila, Odontochila, and Megacephala; Lesne, 1921, for C. silvicola;
and Pratt, 1939, for Omus). I have closely observed mating in the laboratory
in C. circumpicta, C. fulgida, C. nevadica, and C. togata; and mating pairs
of many other species were frequently seen in the field.

The males “take the initiative” and usually approach a female in short
runs. When several centimeters away, the male makes a final quick dash and
leaps on the dorsum of the female, grasping her between the prothorax and
elytra with his mandibles, and frequently around her abdomen with his first
and sometimes second pair of legs. Males have a dense brush of setae on the
venters of the first four front tarsal segments that are apparently an adapta-
tion for grasping the female. The male supports himself with the last and
sometimes second and first pairs of legs (Fig. 79). Males are quite aggressive
and have been seen trying to mount other males (specimens pinned on the
same pin as “mating pairs” have been seen in museum collections that con-
sisted of two males of the same or different species) or mounted pairs. The
female may unseat the male as soon as he mounts and drive him away. If
not, the two beetles may remain in the mounted position for some time before
or after copulation (a half hour or more), and often the female “goes about
her business” of eating, drinking, or ovipositing with a male riding along

* The species which students of American cicindelids have been calling C. flavopunctata Chev.
should be called C. ocellata Klug because the former name is a junior homonym (Schilder,
1953a).

Bionomics AND ZOOGEOGRAPHY OF TIGER BEETLES 193

- fakes aes
Fic. 79. Mounted pair of C. togata,
from 3 mi. west, 2 mi. south of Barnard,
Lincoln Co., Kansas. Note male’s use of
mandibles in grasping female. Fic. 80.
Ovipositing C. togata, from same locality.

(thus, because specimens collected in the field were mounted, this does
not mean that they were actually mating). Eventually, the male everts his
aedeagus, which normally is retracted within the abdomen, and brings it into
a ventral position pointing forward, attempting to insert it into the female’s
genital opening. The female may make this difficult by turning the end of
the abdomen down and may even drive the male away. If the male is suc-
cessful, the aedeagus is inserted one or several times for one to several minutes
each. At this time, the pair may touch and quiver their antennae. Lengerken
(1929) said that the male strokes the female’s elytra with the palpi of the
mouthparts, but I have not seen this. Following couplation, the male may
remain mounted or be driven off by the female. Males and females may mate
repeatedly with the same or different partners.

194 Tue Universiry ScrENcE BULLETIN

The literature mentions mating in the daytime only, and Lengerken
(1929) said that it only occurs during hot sunshine; however, I have seen a
number of mounted pairs of C. circumpicta near Roswell, New Mexico, after
dark in the early evening (the soil was still warm from the day). Shelford
(1911) mentioned seeing C. tranquebarica mating and ovipositing in the
laboratory on damp, cloudy days.

Tue Aputt—Oviposition. Oviposition has occasionally been seen in
nature. Ponselle (1900) mentioned it for C. flexuosa; Mitchell (1902) saw it
in Megacephala carolina and C. ocellata; Moore (1906) twice saw C. repanda
ovipositing; Shelford (1908) reported on C. purpurea; Huie (1915) noted
that C. campestris fills the oviposition hole and may eat its own egg if dis-
turbed during oviposition; Goldsmith (1916) said that C. punctulata may
oviposit in cracks if the soil is hard; Fackler (1918) briefly commented on
C. repanda; Lesne (1921) reported on C. silvicola; Lengerken (1929) men-
tioned that eggs are apparently laid singly in the soil; Zikan (1929) said that
all the cicindelids he observed close their oviposition holes with material
from a second hole made at the same site but slanting obliquely.

I have observed oviposition in the laboratory in C. circumpicta and C.
togata. The ovipositor consists of the eversible terminal abdominal segments
(eighth, ninth, and tenth) and the sclerotized gonapaphyses of the eighth
and ninth segments, and has been morphologically studied by Shelford
(1908), Tanner (1927), and Zikan (1929). In seeking an oviposition site, a
female C. togata was seen touching her antennae to the soil and occasionally
biting the soil with her mandibles. Occasionally, she dug for a short time
with her ovipositor. Females often dig holes into which no egg is laid. These
have been called “exploratory” or “test” holes, in the literature. When a suit-
able site has been found, the female digs a true oviposition hole. In digging,
the ovipositor is everted and the body is inclined at a steep angle by the front
and middle legs (steeper in C. togata than in C. circumpicta). The hind legs
are spread wide for support (Fig. 80). The gonapophyses are the primary
digging tools, and the abdomen makes assisting thrusting motions. Some-
times some soil is removed from the hole by an upward motion of the whole
body. The oviposition holes takes about five to ten minutes to dig and is
from 0.5 to 1 cm deep. The female then remains quiet for several seconds
while the egg is laid. Then the hole is usually filled, apparently sometimes
with soil from an oblique hole at the same site, as reported by Zikan (1929),
and also using the loose soil around the hole that was thrown out in digging.
The soil is tamped with the end of the ovipositor, the whole body assisting in
the motions, and finally the soil is raked with the gonapophyses, leaving
little or no trace of its having been disturbed. The filling of the hole takes
one or two minutes; the entire oviposition process usually takes eight to

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 195

twelve minutes. Occasionally, C. togata have been seen not to fill the oviposi-
tion hole, and Shelford (1908) said that C. purpurea does not close its holes.
This would seem to be very harmful to the egg by permitting desiccation.
One egg is laid in each hole. I have only once seen a cicindelid oviposit in
nature, a C. circumpicta near a saline pool near Roswell, New Mexico, on a
hot, sunny day.

Shelford (1908) reported watching a C. purpurea lay about 50 eggs, and
was uncertain as to whether more might be laid. This is the only known
estimate of fecundity.

Shelford (1911) did interesting experiments on the selection of the ovi-
position site, placing adults in cages containing different types of soil and
different degrees of slope and moisture. He found that C. limbalis chooses
only clayey soil and prefers steep slopes; C. tranquebarica oviposits in many
kinds of soil, but prefers sandy, moist soils; C. sexwguttata prefers sand with
a small amount of humus and usually oviposits near or under twigs or leaves.
He also noted that oviposition sites are different from or more restricted than
the habitats in which the adults hunt for food. Thus it seems that the adults
select optimal microhabitats for the larvae.

Tue Aputt—Food and feeding. There have been many notes in the
literature concerning the food of adult cicindelids; those for the genus Czcin-
dela are summarized in Table 6. In addition, past workers have fed the
following to adults in captivity: decapitated woodboring beetle larvae, meal-
worms (Tenebrio larvae), caterpillars, a tipulid fly, a large tabanid fly, calli-
phorid larvae and adults, freshly killed house flies, and ants. Arthropods that
I have seen eaten or attacked by Cicindela in the field and in the laboratory
are listed in Table 7. From these lists, one may conclude that adults of
Cicindela eat nearly any arthropod that they can subdue and which occurs in
their microhabitat. Occasionally, a beetle will reject an insect that possibly
may possess distasteful chemicals, but at other times such insects are eaten.
Probably the degree of hunger of the predator affects the acceptability of
distasteful prey.

Balduf (1925) observed a C. punctulata catch and eat nine large nymphs
and one adult chinch bug, Blissus leucopterus, in 26 minutes. The beetle saw
the bugs from 5 to 8 cm away (in all cases they were moving), and after
catching them in its mandibles, struck them against the ground several times
in rapid succession before eating them. The beetle returned to a certain
elevated spot to eat; the exoskelton of the prey was discarded in the form of
a small pellet. Lesne (1921) also noted that the hard parts of the prey are
discarded. Moore (1906) noted that C. purpurea may return to its adult
burrow to eat its prey, that it could seen ants 10 to 13 cm away, and that it
rushed up to an ant, bit it once, and gave it a toss, repeating this behavior
until the ant showed no signs of life. He also noted that the vision of C.

196 Tue UNiversiry ScrENcE BULLETIN

Tase 6. Food of Cicindela eaten in nature that has been reported in the literature.

Crustacea—fiddler crabs (young) and marine “fleas” (C. ocellata)
Arachnida—unidentified species of spiders (C. repanda, C. sexguttata)
Insecta
Orthoptera
Acrididae—Melanoplus spretus (C. circumpicta, C. formosa, C. fulgida, C. purchra,
C. punctulata, C. scutellaris, C. sexguttata, C. tranquebarica) ; Stenobothrus sp.
(nymph) (C. silvicola)
Dermaptera—(C. campestris )
Hemiptera
Lygaeidae—Blissus leucopterus (C. punctulata)
Coreidae—Leptocoris trivittatus (C. splendida)
Homoptera
Aphididae—(C. tranquebarica)
Coleoptera
Carabidae—Harpalus pennsylvanicus (C. sexguttata, was larger than cicindelid and
escaped); Bembidion sp. (C. cuprascens, C. hirticollis)
Heteroceridae—(C. cuprascens, C. hirticollis)
Elateridae—Horistonotus uhleri (C. rufiventris)
Coccinellidae—(C. Aybrida)
Tenebrionidae—Tenebrio molitor (C. hybrida)

Chrysomelidae—Disonycha quinquevittata (C. formosa)
Leptinotarsa decimlineata (small larvae) (C. punctulata)
Curculionidae—Phytonomus punctatus ? (C. repanda ?)
Small, unidentified adults—(C. tranquebarica)
Lepidoptera
Phalaenidae—Pseudaletia unipuncta ? (C. sexguttata); unidentified larvae (“cutworms”’)
(C. scutellaris, C. tranquebarica)
Notodontidae—“‘puss moth” (C. campestris)
Galleriidae—Galleria melonella (larva) (C. hybrida)
Pieridae—‘white butterfly” (C. campestris)
Unidentified larvae (C. campestris)
Diptera
Calliphoridae—Lucilia caesar (C. hybrida); Calliphora volitoria (C. hybrida)
Sarcophagidae—Sarcophaga camaea (C. hybrida)
Unidentified muscoid larvae and adults (C. repanda)
“Gnats” (C. sexguttata)
Unidentified larvae (C. cuprascens, C. hirticollis)
Hymenoptera
Formicidae—Pogonomyrmex occidentalis (C. fulgida); Formica pallidefulva (C. formosa,
was unsuccessful); unidentified adults (C. ocellata, C. formosa, C. longilabris,
C. punctulata, C. purpurea, C. repanda, C. scutellaris, C. sexguttata, C. tranquebarica)
“Bees” —(C. tranquebarica)
Halictidae—Lasioglossum zephyrum (dead adults) (C. repanda); Nomia melanderi
(dead adults) (C. haemorrhagica, C. pusilla)
Other
Freshly dead carcasses (fish, rabbits, etc.) (C. ocellata—this is doubtful; perhaps they were
eating carcass-feeding insects)

BioNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 197

Tasre 7. Arthropods that adults of Cicindela have eaten or attacked in the field
(F) and in a laboratory terrarium.

Arachnida
Salticidae (F)
Lycosidae ?
Small, unidentified species
Insecta
Orthoptera
Gryllidae—Oecantha sp. (nymph)
Tettigoniidae—(nymph over | cm long)
Hemiptera
Gelastocoridae—Gelastocoris sp. (F) (attacked, but not eaten)
Miridae
Nabidae—Nabis ferus
Lygaeidae—Ischnoderus falicus; nymph of another species
Cydnidae (F)
Pentatomidae (nymph)
Homoptera
Cicadellidae
Aphididae (F)
Neuroptera
Chrysopidae
Coleoptera
Cicindelidae—Cicindela togata
Caradibae—A nisodactylus sp. (F)
Coccinellidae (could not get a grip on it, unsuccessful)
Chrysomelidae—Ceratoma trifurcata (F—unsuccessful, apparently distasteful) ;
Diabrotica undecimpunctata; Halticinae (unident.); Monoxia puncticollis (F—
unsuccessful, apparently distasteful)
Lepidoptera
Pyralidae—A nagasta kuehniella
Geometridae (larvae)
Small, unidentified caterpillars
Diptera
Stratiomyidae
Tachinidae
Muscidae
Hymenoptera
Formicidae—Pogonomyrmex occidentalis (F—ant heads found with mandibles clamped
on cicindelid antennae or palps); Crematogaster sp. (dealate queen) (F)

repanda seems to be limited to 8 to 13 cm, that it takes C. repanda four or
five minutes to eat a housefly, and that this species seems to be afraid of ants.
Huie (1915) reported that a C. campestris in captivity ate freshly emerged
adults of the same species. Goldsmith (1916) observed that C. sexguttata
would give up an attack if the prey offered any resistance. He also watched
a group of 27 C. repanda feeding on a colony of “small red ants.” Another

198 Tue Universiry ScriENcE BULLETIN

time he saw two beetles of the same species devour a whole colony of ants by
alternating eating each ant that came to the entrance of the hole; when no
more ants appeared, one beetle dug half the length of its body into the
mound in search of more food. Swiecimski (1957) studied the role of sight
and memory in food capture by C. Aybrida in terraria, using various types
of live and dead insect bait. He found that the beetles obtain food by random
search (only immobile prey) or deliberate attack caused by perception from
a distance (up to 25 cm); they do not react to dead prey moved artificially, or
are frightened. In random search, vision does not play an important part,
the beetle often trying to eat pebbles or other inanimate objects. Apparently
chemoreceptors do not function until the prey is very close. Naturally moy-
ing prey evokes attack, which may be divided into several stages: 1) prepara-
tory attitude, consisting of elevating the front of the body and turning toward
the prey, 2) actual attack, consisting of a quick, interrupted run toward the
prey, 3) capture, 4) eating the prey. In some cases, certain stages are omitted,
usually caused by variations in the behavior or nature of the prey. Memory
of the shape and location of the prey appears to play a part occasionally,
because when prey was taken from the beetles, they searched the area where
it was, or if it was moved, they reattacked it.

Friederichs (1931) found that European species (C. campestris, C. hy-
brida, C. silvicola, C. silvatica) have binocular vision for about 90° of their
forward field of vision, and that they react only to movement. This is
probably why beetles turn toward their prey before attacking it : to locate
it more accurately.

Evans (1965) gave a detailed account of how the food is eaten by C.
hybrida. The prey is seized with the long distal teeth of the mandibles;
pieces of the cuticle may be cut away to reach the soft inner parts. The food
is then raked back into the preoral cavity by rotary movements of the laciniae
and accumulates in the form of a bolus that may be chewed for some time
by the proximal molar portions of the mandibles. The maxillae move only
slightly and, together with the labial palps, help to hold the bolus in place.
When enough food is accumulated, the mandibles begin to move the food
in a rotary motion in the preoral chamber from the mandibles upward and
backward, across the cibarial opening, and down between the labial palps.
The labial palps may then push the food up to the maxillae to be recirculated.
Rows of setae on the hypopharynx and epipharynx strain out solid particles
and allow only fluid and very small solid particles to enter the cibarium.
Evans found some evidence of extraoral digestion, that is, regurgitation of
enzymes from the gut. Lengerken (1929) also supposed that this occurs.

Observations that I have made on the feeding behavior of C. circumpicta,
C. duodecimguttata, C. fulgida, C. nevadica, and C. togata in the laboratory
confirm many of the above reports. C. togata saw the movements of small

BIoNoMiIcs AND ZOOGEOGRAPHY OF TIGER BEETLES 199

spiders (about 2 mm long) from a distance of 2 to 8 cm, and C. fulgida saw
adult mites (Androlaelaps casalis) about 1 mm in diameter from 2 to 3 cm
away (the mites are scavengers that probably were carried into the laboratory
with the soil). The prey, if it is distant, is approached by a series of short
runs; the beetle then lunges at it with the mandibles open. If the prey is
small or stops moving, the beetle may miss it and lunge repeatedly, often
merely biting the soil. The beetle seems to rely almost entirely on sight; if
the prey eludes it in the above manner, the beetle searches “blindly” the
immediate area and does not recognize immobile prey until its head is almost
directly over it. Once the prey is seized with the distal part of the mandibles,
the beetle may become quite excited, running about, holding the prey with
its mandibles. Its excitement is often transferred to other beetles in the
terrarium, which also run about searching for food or fighting with the suc-
cessful individual for the prey. Usually the prey is first masticated thoroughly
with the distal parts of the mandibles and maxillae. These organs move
laterally alternately and gradually work the prey back and forth. This opera-
tion takes about two minutes for a small catepillar about 5 mm long and
probably serves to break up large sclerites of the prey. Then the food is
moved farther back into the preoral cavity for mastication and circulation as
described by Evans (1965). This process may take five to ten minutes, and
is occasionally interrupted by the mouthparts becoming still and the head
and prothorax being protruded slightly. This is interpreted as swallowing.
During mastication, the mandibles are moved laterally quite regularly and
mechanically at a rate of 80 to 140 times per minute. Finally, a small, com-
pact pellet about 1 mm in diameter is ejected and laid on the substrate by
the beetle. After a beetle had eaten an Anagasta kuehniella caterpillar, the
resulting pellet was put into water and teased apart. It contained the cater-
pillar’s mandibles, masticated head capsule (broken into irregular pieces
ranging from 0.05 to 0.25 mm in diameter), and the thin skin, relatively
intact and including setae and proleg crotchets. A medium sized prey (about
5 to 8 mm long) takes about five to 20 minutes to eat. Once a C. togata was
observed to lay down part of a hemipteran, chew on the other part, then pick
up and eat the first part. As noted in Table 7, cannibalism occasionally
occurred. One C. togata was found without a head, the elytra of another
was found, and a third was seen being eaten by an individual of the same
species. Probably only weak or sick individuals are cannibalized. In cap-
tivity, adults will eat an insect the size of Anagasta kuehniella once every
one, two, or three days.

Defecation consists of deposition on the substrate of a drop of opaque,
pinkish or brownish fluid, which eventually dries into a spot of the same
color.

200 Tue University SciENCE BULLETIN

Tue Aputt—Drinking. A number of workers have reported tiger beetles
drinking in captivity: Moore (1906), Williams and Hungerford (1914),
Huie (1915), Lengerken (1929), and Zikan (1929). Apparently only Mitchell
(1902) has seen drinking in nature.

I have occasionally seen C. duodecimguttata, C. fulgida and C. togata
drink in the laboratory. Moore (1906) reported that although he provided
C. repanda with a drinking container, the beetles did not use it, drinking
instead from moist sand. My beetles drank both from a Stender dish and
from moist soil. In drinking, the mouthparts, mandibles agape, or much of
the head is thrust into the water or moist soil for one-half to several minutes.
Sometimes soil is bitten to bring the moist soil closer to the mouth. Presum-
ably the water is sucked up by the toregut.

Balduf (1935) thought that cicindelids require water as often as food;
however, cicindelids have been seen drinking in nature so infrequently that
this is questionable. Probably much water is obtained from the food, and
many species frequent the moister parts of their habitats, which probably
reduces their rate of water loss.

Mitchell (1902) stated that C. ocellata eats algae and fine moss near
springs, but it is more likely that the beetles were sucking water from these
plants.

Tue Apvutt—Burrows and digging. Many species dig burrows as adults
for various purposes. Some hibernate as adults, usually in deep burrows (5-
122 cm deep in Canada; Criddle, 1907). During the warm part of the year,
many species make shallower burrows in which to spend the night (Davis,
1921; Rau, 1938), and hot or dry weather (Wallis, 1961). Reineck (1923)
rainy or cloudy weather (Moore, 1906; Wille and Michener, 1962; Blanchard,
1921; Dengerken, 1916, 1929; Moore, 1906; Mitchell, 1902; Blanchard, 1921),
stated that C. silvicola does not dig adult burrows, hiding in natural crevices
in cold or rainy weather. I have noted, as has Graves (1963), that C. sexgut-
tata may take shelter under loose bark. Mitchell (1902) reported that only
females of Megacephala carolina dig burrows to spend the day (this species is
nocturnal), while males hide under logs, trash piles, dead leaves, or bunches
of grass. Usually, however, both sexes burrow. Good descriptions of the
digging process are given by Moore (1906), Criddle (1907), and Lengerken
(1929).

I have noted adult burrows in the field for C. circumpicta, C. duodecim-
guttata, C. fulgida, C. hirticollis, C. nevadica, C. repanda, and C. willistoni,
and have made more detailed observations of digging by C. fulgida in the
laboratory.

In digging, the beetle loosens bits of soil with its mandibles; the soil is
then kicked backward under the body by the first, middle, and hind pairs of

legs in succession. Soon a pile of loose soil accumulates at the entrance of

BIoNOMIcs AND ZOOGEOGRAPHY OF TiGER BEETLES 201

the burrow, often plugging it. As the beetle goes deeper, it accumulates a
small pile of soil directly behind it, then backs up the burrow to push it out
the entrance. The beetle may stop occasionally to rest. Most of the burrow
is slightly larger than the body width, but the end is larger, to allow the
beetle to turn around (it rests facing outward). The entrance is oval and
rather ragged in contrast to the round, smooth holes of most larvae. For
short stays, the burrow is usually plugged loosely or with a thin plug; hiber-
nation burrows are entirely plugged except for the bottom 10 to 25 cm
(Criddle, 1907). The speed of digging depends on the type of soil. In sand,
a beetle can dig its body length in a minute (Lengerken, 1929). A C. fulgida
dug 3 cm in moderately compact clay in one hour. Usually non-hibernation
burrows slant gently downward, are more or less straight, and range from
2 to 10 cm deep. Hibernation burrows are more nearly vertical and often
crooked, and when in hard soil are shallower than those of the same species
in soft soil (Criddle, 1907). In warmer areas, hibernation burrows are not as
deep as reported by Criddle; Blaisdell (1912) found 64 C. senilis hibernating
under three rocks in relatively short, often interconnected burrows. From
one to five individuals were in each chamber. More than one individual in
the same burrow is uncommon. Criddle (1907) found both C. duodecim-
guttata and C. tranquebarica occasionally in the same burrow as C. repanda.
Moore (1906) sometimes found two or more C. repanda in the same burrow.
In digging 67 C. willistoni from temporary burrows about 3-5 cm long, |
found two beetles in the same burrow in only three cases. The burrows were
usually plugged except for the bottom 2 cm and had a mound of loose soil
over the hole.

C. togata was never seen to burrow in captivity or in the field, although
one once spent the night in a C. fulgida burrow. They readily take shelter
under available objects. Their long legs do not seem to be well adapted for
digging.

Tue Aputt—Cleaning. Adults frequently clean themselves, as after dig-
ging, eating, or drinking. Lengerken (1929) stated that beetles often stop in
the midst of digging to clean themselves of sand, and then continue. The
front tarsi are rubbed over the front and upper surfaces of the head (simul-
taneously or one at a time), and the antennae are pulled between the apical
tibial spurs and the tarsi of the front legs. The front legs are used to clean
the mouthparts and the mouthparts to clean the front legs. Primarily the
middle tibiae are used to clean the elytra, although the hind tibiae some-
times assist, by being rubbed quickly over the dorsum. In addition, the distal
parts of the legs are cleaned by being rubbed together, two at a time, the first
with the middle or the hind with the middle. The front and particularly
the middle tibiae bear on their inner apical portions, areas of short, closely
set setae that seem to serve for cleaning. When an apparently distasteful

202 Tue University ScrENCE BULLETIN

insect is seized, a beetle will often dig its mandibles into the soil repeatedly,
probably in an attempt to clean them of the offensive substance.

Tue Aputt—Sleep. Behavior resembling sleep is known in many insects.
No specific mention of sleep in cicindelids is known, although some authors
implied that sleep occurs in the burrow or other retreat (Rau, 1938; Davis,
1921; Moore, 1906). In the laboratory, I have on several occasions found
adults apparently asleep (C. fulgida, C. nevadica, and C. togata). The beetles
were motionless in a normal standing position or with the venter resting on
the substrate and acted half dead when touched. After several seconds of
being prodded, they seemed to “wake up” and became active.

Tue Aputt—Activity. Many workers have noted that Cicindela tends to
be most active on hot, sunny days. Lengerken (1916, 1929) and Shelford
(1908) noted that mating occurs only in such conditions (see above section
on mating). Reineck (1923), Moore (1906), Huie (1915), and others have
noted that beetles remain hidden or in burrows on cool and cloudy or rainy
days. Lengerken (1916) reported that even a cloud passing before the sun
curtailed the activity of C. maritima and C. hybrida. Davis (1921) watched
a C. tranquebarica dig at about 4:45 PM a burrow in which to spend the
night. Moore (1906) said that C. repanda retires at about 5 PM on ordinary
days and about 7 PM if the weather is very warm, and that C. purpurea
remained active until late in the evening if the day was hot. He also said that
C. repanda becomes active at about 8 or 9 AM or earlier on very warm days.
Remmert (1960) found that C. campestris alternates variable periods of rest
and activity during the day; hungry individuals have longer periods of
activity and shorter periods of rest (and vice versa for full individuals); at
higher temperatures or in stronger light, the periods of rest are shorter.

I have noted that on hot days, Cicindela is active in spite of clouds; in
fact, if the habitat is very hot, activity decreases, the bettles remaining in the
shade of vegetation or other objects, or in cracks until clouds come and lower
the temperature. More will be said about the effects of physical conditions
on activity in a later section.

Tue Anutt—Fighting. Huie (1915) noticed that female C. campestris in
captivity were frequently disturbed by other beetles running into them, pull-
ing their legs, or seizing them by the body (probably mating attempts).
Lengerken (1916, 1929) often noted fighting among captive beetles. Some-
times when two individuals met they would stop and elevate the front of
their bodies, mandibles open wide; often this apparent threatening behavior
averted actual combat. If two individuals ran into one another from opposite
directions, they tumbled around together briefly before continuing on their
way. Moore (1906) observed fighting in captive and free C. repanda. One
beetle rushed at another, snapping at it with its mandibles or merely bump-
ing into it and then running away. The other beetle then chased the first for

BioNomics AND ZOOGEOGRAPHY OF TIGER BEETLES 203

a while. Legs and antennae were sometimes lost as a result of these en-
counters in captive individuals.

I frequently observed fighting in captive C. togata and C. fulgida, often
for food. Once two C. fulgida were seen fighting for a small lycosid spider,
one riding on the others back, both chewing on the spider. Another time a
female C. togata rushed up to a male that had just caught a small spider; they
faced one another and fought with their mandibles for about ten seconds;
then the female rushed at the male twice before giving up. In the spring
when mating occurs, C. fulgida are quite aggressive; one often seized
another’s leg, and they tumbled around for a few seconds. Once a C. togata
that was in a C. fulgida burrow was quickly approached by a C. fulgida;
they fought briefly with their mandibles, and the C. togata remained in the
burrow. A C. nevadica was found with a broken left hind tibia that it
apparently received in a fight, and specimens with tarsi or antennae missing
have been seen. In the field, I once saw a number of C. circumpicta rushing
at one another in the manner described by Moore (1906).

Tue Avutt—Flying and escape. Davis (1921) watched a C. tranquebarica
for an hour and saw it fly only once; and Moore (1906) watched a C. pur-
purea traverse about 40 meters of a road in an hour without flying. These
notes confirm the impression that I have received from casual observations
that although most species of Cicindela have the power of flight, they rarely
fly. When going about their usualy activity of preying, Cicindela typically
runs in short bursts, often in a zig zag course. Usually only when disturbed
by a larger animal do they fly, and even then some species (or under certain
conditions) fly only as a last resort.

As Moore (1906) noted, before flying, a beetle squats close to the ground.
If further frightened, the beetle jumps into the air and takes wing (Moore
remarked that deformed beetles unable to fly could jump about 3 cm high).
Several authors have noticed differences in the flying abilities and habits of
different species. Most species of Cicindela fly in a low (1 to 2 m), level path
and land 5 to 15 meters from the source of disturbance. Some early authors
thought that beetles always landed facing the source of danger, but later
observations showed that they land facing the wind (Moore, 1906). Often a
beetle will circle and land behind the disturbance, particularly if flushed
repeatedly. More intricate midflight maneuvers may be made, if the beetle
“sees” it is about to land in unsuitable territory, to bring it to a “desired”
landing place.

As implied above, escape behavior varies with the species, the external
conditions, and the degree of danger. Moore (1906) noted that C. purpurea
usually flew only a short distance down a road unless persistently pursued;
then they flew into the nearby grass, ran to a clear space, and either flew
again or squatted ready to fly. C. togata, a species with long legs, often tries

204 Tue University SciENCE BULLETIN

to outrun the danger. If hard pressed, it usually runs in a zig zag course and
may run into short, sparse Distichlis grass rather than fly. When it does fly,
it frequently flies into dense grass. The same zig zag running behavior has
also been noticed in C. circumpicta, and this species was also seen to fly or
run into vegetation to escape. C. fulgida, which is normally found among
sparse vegetation, usually flies at the slightest danger, often into dense vege-
tation, but sometimes out onto a bare salt flat. Sometimes species that are
surprised in short vegetation have difficulty flying because of bumping into
the plants. C. Airticollis is another very wary species that is difficult to
capture. String wind and low temperature generally deter or inhibit flying.

Tue Aputt—Injuries and deformities. As mentioned above, injuries are
sometimes inflicted by other cicindelids. Townsend (1884) mentioned find-
ing C. tranquebarica with antennae, legs, or elytra injured or missing. I have
collected specimens with injuries that probably were not caused by fighting.
A C. fulgida was collected with both hind tibiae and tarsi missing, a C. togata
was found with the tip of one elytron missing, and a C. circumpicta was
found on its back, legs kicking, with its entire abdomen missing. Probably
predators inflicted these injuries.

Several authors have reported deformities in cicindelids. Moore (1906)
noted that some C. purpurea have fused elytra and cannot fly. Townsend
(1884) found a C. tranquebarica with a wrinkle across one elytron and noted
(as I have) that some individuals of this species cannot fly. Horn (1927)
reported atypical elytral markings, short elytra, deformed leg segments, and
a deformed head and prothorax in Cicindela and several other genera. Shel-
ford (1913c) showed atypical patterns of elytral tracheation in Cicindela.
Shelford (1915) noted that leg and antennal abnormalities are rare, while
elytral and labral abnormalities are more common in nature. He also experi-
mentally produced deformities by injuring larvae and pupae of C. punctulata;
injuring the pupal labrum produced adult deformities, but injuries to the
pupal legs or larval labrum had little or no effect on the adult. He thought
that elytral deformities such as an abnormally short elytron (often accom-
panied by reduced markings) or holes in the elytron were caused by injuries
to the pupal elytra. I have occasionally found labral and elytral deformities
in Cicindela; some of these are shown in Figures 81-84.

Wood (1965) found a C. scutellaris with a trifurcate ninth antennal seg-
ment (each fork having two additional segments), and noted that similar
abnormalities have been reported in a cerambycid. Park (1931) and Lavigne
(1965) found them in a tenebrionid and cerambycids. I collected a C.
nevadica with a similar deformity, Figures 85 and 86. The third segment of
the left antenna is bifurcate at the apex, and segments four to eleven are
duplicated.

Bionomics AND ZOOGEOGRAPHY OF ‘TIGER BEETLES 205

Sea eee

Baerga

85

Fic. 81. Deformed labrum of male C. nevadica nevadica, from Saltdale, Kern Co., Cali-
fornia, frontal aspect. Fic. 82. Deformed labrum of male C. nevadica knausi, from 2.5 mi.
southwest of Plainview, Woods Co., Oklahoma, frontal aspect. Fic. 83. Normal (left) and
deformed (right) elytra of female C. nevadica olmosa (paratype), from 25 mi. west of Tularosa,
Sierra Co., New Mexico, dorsal aspect; a crease runs transversely across the right elytron, inter-
rupted by a small hole near the margin; stippled areas are discolored (light brown), normally
white maculation. Fic. 84. Anterior half of right elytron of female C. nevadica knausi, from
1 mi. northeast of Tucumcari, Quay Co., New Mexico, dorsal aspect; shaded areas represent
holes; the posterior portion of the humeral lunule (caudad from largest hole) is atypical (see
Fig. 83 for typical humeral lunule). Fic. 85. Head and deformed left antenna of male C.
nevadica knaust, from 3 mi. east of Cherokee, Alfalfa Co., Oklahoma, cephalic aspect. Fic. 86.
Enlargement of third and duplicated fourth segments of same.

Tue Aputt—Qdor. It has often been mentioned in the literature (e.g.,
Leng, 1902; Eckhoff, 1939; Graves, 1963) that C. punctulata emits a fruity or
applelike odor when handled. Leng (1902) also reported that C. sexguttata
emits a fragrant odor and C. longilabris a musky odor when captured. Leng-

206 Tue University SctENcE BULLETIN

erken (1929) noted a pleasant odor produced by C. maritima in hot weather
and when the beetles were alarmed, as when captured. I have noticed that
C. ocellata and C. tranquebarica, when handled, emit an odor similar to
that of C. punctulata, but weaker, and that C. duodecimguttata produces a
musky odor.

These odors are produced by the anal, or pygidial glands. Dierckx (1899,
1901) discussed the anatomy of these glands in C. hybrida and C. campestris,
noting that they are similar to those of carabids, with an oval, cuticular,
sparsely muscled reservoir; a collecting canal; and a long, cylindrical gland
proper. Brandt (1888) remarked that the anal glands of cicindelids (no
species given) are more poorly developed than in carabids because the cicin-
delids have other well developed means of escape (running, flying) and
defense (mandibles). Although the subject has not been studied further,
Brandt’s explanation is probably correct.

Tue Aputt—Sound production. Certain cicindelid genera, Mantica, Man-
tichora, Oxychila, and Chiloxia, possess stridulatory organs, ridges on the
margins of the elytra and hind tibiae (Horn, 1908-15). I have noticed that
captive Cicindela (C. fulgida, C. nevadica) sometimes raise their elytra syn-
chronously very slightly once or several times. Sometimes when this is done,
faint but clearly audible sounds are produced, best described as short buzzes.
Captive C. repanda have been heard making a continuous buzz lasting about
one second while the elytra were raised slightly. Apparently the sound is
produced by slight irregularities on the elytra where they join, although none
are readily visible under magnifications of about 80X. The sound is probably
purely accidental.

Sound is also produced in flying species when the wings beat. It is usually
not audible to humans except in the larger species, in which a low buzz can
be heard as the beetle takes off.

Tue Aputt—Predators and parasites. Relatively few records of enemies
of adult cicindelids have appeared in the literature. Ingram (1934) found a
mite (unidentified) parasitic on the thorax and legs of C. haemorrhagica.
Graves (1962) watched a dragonfly, Aeshna interrupta, catch a C. repanda in
flight and carry it away. Asilid flies seem to be the most common predator:
Fox (1910) reported them catching and eating C. dorsalis; Fackler (1918)
saw a Proctacanthus (near rufus) catch a C. repanda, inserting the proboscis
between the elytra to feed; Bromley (1914) saw Proctacanthus philadelphicus
eat C. punctulata, on three different occasions, and once saw Promachus fitchi
catch the same species; Davis (1910) saw a C. sexguttata caught and eaten by
a Laphria sp., about as large as the beetle; Wallis (1913, 1961) reported
Proctacanthus milberti catching a C. purpurea and inserting the proboscis
between the elytra to feed, as well as asilids twice catching beetles in flight;
Stevenin (1948) saw an asilid repeatedly attack a C. apiata in Uruguay.

BronNomics AND ZOOGEOGRAPHY OF TIGER BEETLES 207

Blaisdell (1912) found evidence that lizards eat C. senilis. Zikan (1929) said
that chickens and a “wren” may eat adults. Fackler (1918) reported that
the remains of Amblychila had been found in a hawk’s stomach; Fitch
(1963) found remains of Cicindela in pellets of the Mississippi kite. Crid-
dle (1907) said that badgers sometimes destroy large numbers of adults;
Snow (1877) found a freshly eaten Amblychila in the stomach of a skunk;
Stains (1956) reported fragments of Cicindela in scats of the raccoon.

Many of the C. circumpicta 1 collected on 28 August 1963 from near
Drummond, Garfield County, Oklahoma, were heavily infested with larval
mites of an undescribed species of Eutrombidium (‘Trombidiidae). Most of
the mites were under the elytra, on top of the abdomen, and on the hind
wings; a few were on the venter of the abdomen. The infestation was ap-
parently only “accidental” because these mites have not been found on the
same species from that locality in later years, nor from beetles from other
localities. This genus of mites lives in the soil as nymphs and adults, eating
orthopteran eggs; larvae have been found parasitizing several genera of
orthopterans (Evans et al., 1961). I have also found a number of mites
(Uropodidae) attached to the thorax and legs of two museum specimens of
C. sexguttata. This family of mites is not actually parasitic, but phoretic,
attaching to insects in the deutonymph stage for transportation (Evans et al.,
1961).

I have twice caught asilids eating Cicindela: Diogmites symmachus with
a C. togata, and Proctacanthus milberti with a C. formosa. Asilids are often
common in and near saline habitats.

On a small salt flat in north central Kansas, I found two regurgitated
pellets about 1 cm in diameter, which an ornithologist, Dr. Richard J.
Johnston, thought were produced by a sparrow hawk. They contained
remains of insects: a bee, a carabid, several chrysomelids, and three species
of cicindelids, C. circumpicta, C. togata, and C. punctulata. Other birds
which eat insects, such as killdeers and snowy plovers, are often common in
saline habitats and likely eat some Cicindela.

Tue Aputt—Ecological relationships. Certain of the relationships of
cicindelids with other animals have been discussed in the sections on food
and predators and parasites. Other arthropods, besides Cicindela, that I have
seen or collected in saline habitats are listed in Table 8. These relationships
can be summarized in a diagram (Fig. 87), showing the interactions of the
major organisms in a saline habitat. Saline habitats and their assemblage of
organisms could be considered ecological communities, since their organisms
show a certain amount of interdependence and function somewhat as a unit.
However, as can be seen from Figure 87, there is also an intimate connection
between the saline habitat and the surrounding prairie. Because of this, the
saline habitat should more properly be called a minor community.

208 Tue Universiry ScrENcE BULLETIN

Tasie 8. Arthropods, other than Cicindela, seen and collected in saline habitats
in the central United States.

Crustacea
Isopoda
Diplopoda
Arachnida
Acarina (infesting Cicindela)
Araneida
Insecta
Odonata
Libellulidae, Gomphidae, Aeshnidae, Coenagionidae
Orthoptera
Acrididae, Trimerotropis sp.
Tetrigidae, Paratettix sp.
Tridactylidae, Tridactylus minutus
Hemiptera
Gelastocoridae, Gelastocoris oculatus
Saldidae, Pentacora signoreti
Cydnidae, Schirus cinctus
Pentatomidae, Rhytidolomia sp.
Homoptera
Fulgoridae, Scolops sp.
Coleoptera
Cicindelidae, Megacephala virginica
Carabidae, Agonoderus lineola, A. obliqulus, Anisodactylus sp., Aspidoglossa subangulata,
Bembidion coxendix, Calosoma sp., Clivina dentipes, Cratacanthus dubius,
Diplocheila assimilis, Dyschirius criddlet, Elaphrus ruscarius, Geopinus incrassatus,
Harpalus amputatus, H. pennsylvanicus, Pogonistes planatus, Pterostichus sayt,
Scarites substriatus, Selenophorus sp., Tachys sp.
Omophronidae, Omophron nitidus
Staphylinidae, Bledius sp.
Histeridae, Hister biplagiatus
Meloidae, Epicauta conferta, E. segmentata
Tenebrionidae, Eleodes hispilabris, E. opaca, Lobometapon sp.
Scarabaeidae, Cotalpa subcribrata
Chrysomelidae, Calligrapha sp., Leptinotarsa decimlineata, Monoxia puncticollis
Curculionidae, Cleonts angularis, Hypera punctata, Lixus sp., Ophryastes vittatus,
Pantomorus pallidus, Phytonomus nigrirostris, Sphenophorus aequalis, S. australis,
S. callosus, S. destructor, S. germari, S. parvulus, S. scoparius, S. venatus
Diptera
Chironomidae
Ceratopogonidae
Culicidae
Tabanidae, Chrysops vittatus
Stratiomyidae, Eulalia communis
Bombylhiidae, Anthrax analis, Exoprosopa dodrans, E. sordida
Asilidae, Diogmites symmachus, Laphystia sp., Nerax sp., Proctacanthus milberti
Dolichopodidae
Syrphidae, Evistalis aeneus, Eupeodes sp.

BroNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 209

Otitidae, Cheroxys latiuscula
Ephydridae, Ephydra sp.
Sarcophagidae
Muscidae
Anthomyiidae
Hymenoptera
Tiphiidae, Pterombrus rufiventris, Myzinum quinquecinctum
Mutillidae, Dasymutilla leda, D. occidentalis, D. quadriguttata, D. vesta, D. waco
Formicidae, Crematogaster sp., Pogonomyrmex occidentalis, Prenolepis imparis
Sphecidae, Ammophila varipes, Cerceris sp., Prionyx atratus, Stigoides unicinctus
Pompilidae, Anoplius sp., Poecilopompilus interruptus
Andrenidae, Calliopsis sp.
Anthophoridae, Eucerinae

BARE FLATS MARGIN PRAIRIE

snowy plover coyote

t ae eee killdeer
Cicindela a

Va ils

carabids % |

ys

Cicindela

EAidee (eclediuse a
Male
7

insectivorous
vertebrates *

predaceous
insects

herbivorous

Jt
flies Monoxia —> ants
] 4 vie
algae Sugeda, Soy ae Ba z

S/

Fic. 87. Interrelations of the major organisms in a salt flat community (subdivided into bare
flats and margin) and the surrounding prairie. The organisms that the arrows point to use those
at the other ends for food.

*toads, lizards, sparrow hawk, raccoon, badger

Another sort of relationship between cicindelids and other animals is
found in the closely related phenomena of mimicry and cryptic coloration, in
which a species evolves to look or act like something inedible (or at least not
high on the list of “preferred” foods) to a predator species. R. Shelford (1902)
and Robinson (1903) discussed some instances of mimicry of tropical Asian
cicindelids (Tricondyla, Collyris, Cicindela) by locustids, cerambycids, a scio-
myzid fly, and wasps. Blickle (1958) reported that a species of tabanid in
Florida resembles C. dorsalis in color and flight habits.

210 Tue University ScIENCE BULLETIN

Townsend (1886) noted that the species that live in wooded areas, such as
C. sexguttata, are often green and difficult to see, even when sitting on a log;
while others that frequent bare soil (C. macra, C. cuprascens, C. repanda, C.
tranquebarica, and others) are often the same color as their background. He
also noted that the white elytral markings, rather than being conspicuous,
break up the outline of the insect and make it even harder to see; and that the
ventral coloration, not visible from above, is often more brilliant than the
dorsum. Wallis (1961) also mentioned matching of the background color,
citing the nearly white species, C. lepida, which inhabits light sand and
whose shadow is often more conspicuous than itself. Fox (1910) noted the
close resemblance of some species on the New Jersey coast to their back-
ground, but also noted that in two species, individuals sometimes occurred
on the “wrong” background. N. L. Rumpp (7 /itt.) said that two subspecies
of C. willistoni (pseudosenilis and praedicta ) are not at all well camouflaged,
being dark blue-green forms and occurring on white salt pans in the Mojave
Desert.

I too have noted that individuals of many species are very difficult to
detect against their natural background, particularly when they do not move.
Of the approximately 80 species of Cicindela in the United States, over 50
have a rather dull (brownish, blackish, dark green) dorsal color, over 20 are
dull in some parts of their ranges and bright in others, and only about five
are consistently bright (of these, two are green woodland forms). Of about
35 species that I have collected, about 25 are well camouflaged, five are well
camouflaged in some parts of their ranges or in certain polymorphic forms,
and only a few do not match their background well. It is interesting that in
some species which appear dull to the naked eye (e.g., C. macra, which
appears brownish), the elytral color, when viewed through a microscope, is
actually made up of spots of bright blue or green on a red background; if
the blue or green spots are large, the beetle appears dark brown, and if they
are small, the color is bright reddish brown.

It seems likely that many of the predators of Cicindela exert a selection
pressure that results in the beetles resembling their background by the killing
of ill-matched individuals. Many birds and insects are known to have color
vision. Evidence that this occurs is found in the sparrow hawk pellets that I
examined (see section on predators and parasites). Remains of C. circum-
picta, which occurs in reddish, green, and blue color forms in Kansas, were
present; all were of green individuals, even though over 60% of the in-
dividuals are reddish in that area.

The pattern of geographic variation of color in some species (particularly
C. togata) has led me to hypothesize that certain species of saline habitats
respond to two selection pressures: one, to match the color of the soil in their
local area; and two, to match the white of the salt. As noted in the section on

BroNomics AND ZOOGEOGRAPHY OF TIGER BEETLES 211

saline habitats, in dry weather saline habitats become covered by a crust of
crystalline salt; however, in wet weather the salt dissolves and the color of
the underlying soil appears. Thus, the Cicindela of these habitats live on a
substrate that frequently changes color. In many of the species in this study
(C. circumpicta, C. cuprascens, C. hirticellis, C. macra, C. nevadica, C. togata,
and C. willistoni), | have noticed a definite correspondence between the
dorsal coloration and the color of the soil in the locality. This is investigated
more thoroughly in later sections. In the most nonfluvial species, C. togata,
there is a decrease in the amount of white on the elytra in the eastern parts of
the range, where the climate is more humid; while the southwestern part
of the range, where the soil is more often dry, the white of the elytra increases.
In one locality in western Texas, where the soil is nearly white, the elytra are
almost completely white. This great tendency of some species to match their
background can sometimes be used as a clue to past dispersals or ranges of
these species.

In most species, the dorsum of the abdomen (as well as the venter of the
body) is a bright metallic color, usually green or blue, even if the rest of the
dorsum is dull. When the beetles fly, this bright area of the body is suddenly
visible as the elytra are raised. It is possible that this acts as a flash or startle
coloration to predators.

The species in this study exhibit certain morphological, physiological, and
behavioral traits, or adaptations, for living in saline habitats, listed in Table
9. Some species have more or different adaptations than others. Since few

Taste 9. Important morphological and behavioral traits for living in saline
habitats (L—found in larvae also). Not all traits are necessarily found in all
species studied.

Those shared with species not found in saline habitats;
1. Inactivity during the least favorable parts of the year (winter, summer); L.
2. Dig burrows to withstand temporary harsh conditions (storms, daily temperature
fluctuations).
3. Hide under vegetation, in cracks, etc., to escape heat.
4. Protective coloration (resemble color of substrate, white markings act as disruptive
coloration).
5. Cuticle of adult and setae of larvae protect from salt.
6. Reduction of competition (spatial and temporal segregation), L.
7. Able to be active at high temperatures; L.
Those found primarily in species of saline habitats:
8. Long legs:
a. For rapid running to catch prey and escape enemies.
b. To elevate body higher above hot substrate.
9. Increased white markings and body setae; L.
10. Tend not to fly because of high winds.

212 Tue University ScriENcCE BULLETIN

of the adaptations are restricted to species of saline habitats (even increased
white on the body is found in some species living on light sand), probably
few evolutionary changes were necessary for these species to become adapted
to saline habitats. Those modifications that are necessary apparently have
developed repeatedly, for many relatively unrelated species or species groups
throughout the world are found in saline habitats. The first-listed adapta-
tion, inactivity during the least favorable parts of the year, could be either an
advantage or a disadvantage; if a species could become adapted to being
active during the hot, dry summer, it could avoid competition from other
cicindelids, provided there was sufficient food to make this a “worthwhile”
expenditure of energy.

The literature is replete with notes about habitats in which adult cicin-
delids are found, and no attempt will be made to review them all. Blanchard
(1921) found definite habitat preferences for 11 species in Michigan, and
Fox (1910) did the same for 11 species in New Jersey. Sherman (1908) noted
different species occurring at different elevations in western North Carolina.
Vaurie (1950) gave brief habitat notes for 27 species in north central North
America. In general, some species are found in quite restricted habitats,
while others frequent many types of habitats. Such factors as the type and
moisture of the soil, amount of vegetation, disturbance of the habitat, and
climate are important in determining whether a species will inhabit an area.

One notable characteristic of many saline habitats in the central United
States is that they support numerous species and individuals of Cicindela.
On one June morning I collected 11 species (some being abundant) within
an area of about one acre on a large salt flat in Woods County, Oklahoma.
This is one extreme, but many habitats have five or more species. According
to the competitive displacement principle (Gause’s law), different species
having identical ecological niches cannot coexist for long in the same habitat
(DeBach, 1966); therefore, one naturally wonders how all these species can
exist together, since all are general predators of about the same size, and since
food appears to be scarce in saline habitats. By collecting throughout the
warm months and noting the distribution of species in a particular saline
habitat, I have found that there is a tendency for the species to be separated
both spatially and temporally.

Spatial segregation is effected by the preference of various species for
different microhabitats. This phenomenon is shown graphically in Figure 88,
where species most likely to be active at the same time of year are grouped
together. Clearly, some species, (C. cuprascens, C. duodecimguttata, C. hirtt-
collis, C. macra, C. nevadica, and C. repanda) “prefer” moist conditions, one
(C. togata) is most common on bare salt flats, one (C. fulgida) is common in
dry, vegetated areas, and others (C. circumpicta, C. punctulata, and C. wil-
listoni) are found in a variety of habitats. Another sort of spatial segregation

BionNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 213

A

i, C.fulgida
‘
50 i
D
i
40 j
% !
30 ;
7
u
20 Pe, ots,
*, oe Sexe
sy yee 7 ~Sn
lO Bo *© C.willistoni

C.repanda
C.duodecimguttata

C.cuprascens

C.punctulata

OC togata

© C.circumpicta

88

Fic. 88. Per cent of instances in which the species were noted in various microhabitats,
arranged from wet to dry and vegetated; A=spring and fall species; B=summer species; sample
sizes range from 10 to 80, most being above 15; microhabitats: a=near water, or sand bar,
b=low wet area, or moist ditch, c—near creek, or creek bank, d=moist bare salt flat, e=dry
bare salt flat, f=small flats, or near margin, g=near hummocks, or among vegetation.

214 Tue UNIversiIty SCIENCE BULLETIN

is found in differential soil preference. One species (C. duodecimguttata ) is
nearly always found on clayey soil, while some (C. cuprascens, C. hirticollis,
and C. macra) occur on sandy soil. The others seem to have no “preference”
and are found on both types of soil.

Temporal segregation is shown in Figure 89, where nonfluvial and fluvial
species are grouped together. Some species (C. duodecimguttata, C. fulgida,
C. repanda, and C. willistoni) emerge in the spring after hibernating as
adults, mate, oviposit, and die out during the summer; in the fall a new
brood of adults emerges. In other species (C. circumpicta, C. cuprascens, C.
hirticollis, C. macra, C. nevadica, C. punctulata, and C. togata), the adults
emerge during the summer, often nearly all at once (C. nevadica), and some-
times show a lull during the hottest months and a smaller emergence in the
fall (C. circumpicta).

No diurnal temporal segregation was observed. Adults seem to be active
throughout the day in fair weather when the air temperature is above 20° C
and below 37° C. One would expect to find from a careful study that the
spring and fall species are active at lower temperatures than summer species
(see below). This could produce some diurnal temporal segregation during
parts of the year when these two types of species occur together.

Although the temporal and spatial segregation are only partial, the com-
bination of the two provides nearly complete separation of some species and

C. cuprascens C.macra A
Chirticollis

C.repanda C.duodecimguttata

C. duodecimgut tata

Mar. Apr. May June July Aug. ' Sept. Oct.

C.nevadica
C.willistoni —¢ fylgida B

Mar. Apr. May June July Aug. Sept. Oct.

89

Fic. 89. Relative abundance of the species throughout the year, generalized from data from
localities in southern Kansas and northern Oklahoma; A=fluvial species, B—nonfluvial species;
the exact positions of the peaks and ends of the curves, and their heights vary from year to
year and at different localities and latitudes; C. willistoni has not been collected in the fall,
but should be active.

Bionomics AND ZOOGEOGRAPHY OF TIGER BEETLES Z15

partial separation of others, thus considerably reducing the possibility of com-
petition. No organized data were collected on the subject, but casual observa-
tions indicated that food is the most important resource of the environment
that is in short supply. Small arthropods suitable for food are usually very
uncommon in saline habitats, and only very rarely was an adult or larva seen
eating. Since the larva is the primary feeding stage and lives much longer
than the adult, no doubt competition for food is even more severe among
larvae than adults; furthermore, adults often occur in the same microhabitats
as larvae and thus compete with them. Perhaps the long life of larvae and
their habitat of lying in wait for prey at night are evolutionary “efforts” to
reduce larval-adult competition. Insufficient data are available for the larvae,
but it appears that those of most species are active at the same times of year
as the adults; however, in species whose adults emerge only in the summer,
the larvae are active in the spring as well. Many larvae of C. togata are active
throughout the hottest part of the summer, when the larvae of most species
are inactive (i.e., have their burrows plugged).

Another resource that could be in short supply for species frequenting
marginal microhabitats is space, initially for oviposition sites, but ultimately
space for larval burrows. Many times I have seen favorable larval sites literally
riddled with larval burrows.

If competition among species of Cicindela is important, one would expect
to find fewer species in habitats with fewer or less abundant resources. In
Table 10, the number and abundance of species in 14 saline habitats with
varying types and abundance of microhabitats are compared. Those habitats
with many microhabitats, abundant water, and a large area have more species
which are more abundant than the more impoverished habitats. There are
other possible reasons why certain of the habitats in Table 10 have fewer
species than others: some are fluvial habitats and would have few if any
nonfluvial species; and some of the habitats are outside the ranges of some
species. However, it is likely that the “preference” of different species for
different microhabitats and the differing ranges of species evolved at least
partly because of competition.

Some of the effects on Cicindela of physicochemical factors of the environ-
ment have been alluded to above. The primary factors that govern the
activity of adults seem to be temperature, humidity (actually evaporation),
probably light, and wind. Shelford (1913b) found that C. scutellaris reacted
negatively to dry air and positively to moister air, and that beetles moved
against a stream of warm air (in the laboratory). Chapman et al. (1926), in
a study of sand dune insects in Minnesota, found that C. formosa and C.
lepida become active at 15-20° C, and that C. lepida dies at 45-50° C and
C. formosa at 50-55° C. They emphasized temperature and noted that the
“successful” sand dune insects are able either to endure great extremes or

Tue University ScrENcE BULLETIN

216

e I I oO + -+ O *“SUBY “YOIMUIII5f
b I if G O alr + ao “suey ‘opeioq [q
b ¢ I O + + + “suey ‘ATR
€ I Z oO O O ede "PPO “VIUZIO
b I I G =F O O gs ea "ePAO ‘tyINH
li € & . + + fo) +++ “E[¥O ‘22940
9 € I Z tr at oO fe tues “FIO ‘PPYyArW
9 Z Z ie + _ Oo + ‘suey ‘o1oplAjog
C Z ¢ + + O + ‘e]YO ‘puowumiqg
l b ¢ + + O ++ + ‘suey “ssdg epnay
9 I ¢ Z + + + + + + + “suey ‘U0}[9ZeF]
8 Z I I b ss a act ap ect: 4) Gis ogee ieee alee are cing ie "EEIO 23497949
6 Z I Po allay gente ae te aise ey wine ee | Gabe salted santa aly aio oigy Steoads ‘suey “Uospn}y
II b I I € eat ae alc serie ie te aac Gtr ae [ate stealer "PLYO ‘Motautelq
N oR Y uourWoOsUyF) ~UOWWWIOT, JURepUNGY UONLIISIA syep [[euus sey JOVEM (UMo} JsoIvoU )
satgads jo yoquiny y0ys Jo ulsieut AIq uado ‘adie JUSUPUIOg sont[eo0T
SJEUQEYOINY

*(sarsads Jo Jaquinu [e10I=N) syeuqey
asayi We so.uULpuNqe sANLJaI May pue vjepu11y Jo saisads Jo Joquinu pue ‘(aUepuNgE aaNe[er aIvoIpur sasnjd “uUasqe—Q)
BWOYL]YO UsJoYyIOU puUe sesURY UJoYINOs UT syeUqey IUT][es fF] I sieWqeyosUW snoreA Jo suLpuNqe pue 29dUISITq “()[ ATV],

BIoNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 217.

avoid them by being active at other times. Rensch (1957) found similar
ranges of activity for C. hybrida in Europe (25-45° C), C. bicolor in India
(25-43° C), and C. nilotica in Egypt (27-37.5° C). He noted a few individuals
of the latter species active on moist sand at 46° C, but never any on dry sand
at 43-46.5° C. Remmert (1960) studied daily changes of light and tempera-
ture preference in C. campestris in the laboratory, finding that during the
day the preferred temperature is 34.6° C, while at night it is 26.0° C; how-
ever, the temperature preference depends on the physiological state of the
beetle: hungry and thirsty individuals have a lower preference than fed ones.
His experiments indicated that the beetles are more positively phototactic
during the day, but he did not keep temperature and humidity constant, and
the results are not conclusive. Payne (1964) studied temperature preferences
(humidity not controlled) of C. repanda and C. rufiventrts in the laboratory.
C. repanda, a spring and fall species, preferred 25-32° C, and C. rufiventrts,
a summer species, preferred 25-38° C.

I have made a few incidental measurements of temperatures at which
Cicindela become active or cease activity in the field: C. repanda was seen
“sunning” but not running about when the air temperature was 15° C; C.
duodecimguttata became active at about 25° C; C. willistont became active at
about 18-19° C; C. circumpicta was active at about 21° C, and became in-
active (in shade of vegetation) at about 36° C; C. togata was sluggish and
unable to fly at 22° C; C. punctulata, C. circumpicta, C. togata, and C. repan-
da were found near water, in grass, in burrows, or in cracks in the soil at
about 38-39° C. As mentioned above, I found C. circum picta mating at night,
probably because the habitat was too hot during the day; thus temperature
seems to be more important than light in influencing activity. Adults of
some species (C. circumpicta, C. cuprascens, C. macra, C. nevadica, C. punc-
tulata, C. schauppi and C. togata in this study) also are attracted to artificial
lights, making this a profitable means of collecting.

Another effect of temperature (as well as humidity) on adults of certain
species was investigated by Shelford (1917). He found that the color and
elytral pattern of the adult are affected by the temperature and humidity at
which the prepupal and pupal stages develop. Individuals reared at high
temperatures had more reduced markings (cover a smaller portion of the
elytra) than those reared at low temperatures, Similarly, colors were brighter
in dry conditions and darker in moist conditions. He noted that in some
species, forms with bright colors and reduced markings occur in hot, dry
areas, while forms with bright colors and expanded markings occur in areas
of hot, moist climate. However, he noted the importance of microhabitat
conditions, which still have not been studied or mapped well.

As mentioned above, strong wind reduces the tendency of some species to
fly; when they do, they are often carried to unsuitable habitats. The other

218 Tue University ScrENCE BULLETIN

effects of wind have not been studied, but no doubt one is increase in rate of
water loss, probably causing many species to seek protected microhabitats.

Cicindelids do not seem to be adversely affected by the high concentration
of salts in saline habitats. Adults, as well as larvae, of at least some species are
found in all parts of saline habitats that are not well vegetated or flooded.
Apparently the thick exoskeleton of adults and the long spines and setae of
larvae and pupae serve as a mechanical protection from salts.

Tue Aputt—Dispersal. Besides being the reproductive stage, adult Cicin-
dela are the dispersal stage of the species. The fluvial species, none of which
is restricted to saline habitats, almost certainly disperse along streams. How-
ever, nonfluvial saline habitats are today separated by many miles of land that
is unsuitable for species of saline habitats. Yet these species are found at most
of them, including man-made ones near oil wells that could not have been in
existence for more than 80 or 90 years. How has this dispersal taken place?
Although most species of Cicindela are agile fliers, they rarely fly in nature
unless disturbed by an animal larger than themselves. I have noted that in
strong wind, when the beetles do fly, they sometimes lose control and go
sailing out of sight over the prairie. Quite probably, strong winds are a factor
in dispersal. Storms may or may not be important; adults usually take cover
and become inactive in bad weather, although Woodruff and Graves (1963)
thought hurricanes might have been responsible for the introduction of a
Cuban species into Florida. Between the major saline habitats are scattered
many small salty patches, not ideal habitats for large populations of Cicindela,
but able to support small numbers. Such areas no doubt form “stepping
stones” for dispersal between major saline habitats. For example, I have
found small populations of C. circumpicta, C. fulgida, C. nevadica, and C.
togata in small salty spots in pastures and fields. Another factor in some
cases is that some nonfluvial saline habitats are interconnected by streams.
Species that “prefer” salt flats have occasionally been found on sand bars of
streams and evidently disperse along them.

After some beetles have dispersed to a saline habitat, what are the factors
that determine whether they can survive there, and if so, how numerous can
they become? One of the most important factors is that the soil be of the
proper type and moisture content for larval development. Many small salty
patches in pastures and roadside ditches become too dry in the summer for
the larvae of most species. In some parts of the country (the desert South-
west), extremes in temperature might be a limiting factor, but in the central
United States, most species escape unfavorable temperatures by becoming
inactive or seeking a microhabitat that is more equable. However, there are
some species in the northern United States and western mountains (e.g.,
C. pusilla) which probably do not occur as far south or east as Kansas at
least partly because of high temperatures. The availability and abundance of

BIoNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 219

food are obvious and important limiting factors, as is the amount of vegeta-
tion. Competition, discussed above, is another. No doubt a combination of
these factors is responsible for the decrease in abundance and number of
species on impoverished habitats noted above (Table 9). Salinity seems not
to affect the beetles at all adversely, since they are found (larvae and adults)
in all bare parts of saline habitats.

Man has no doubt had a great effect on the distribution and abundance of
cicindelids. By altering or destroying habitats, such as cutting forests or
plowing prairies, he has restricted the distribution of some species and made
it possible for others to expand. Dirt roads, paths, road cuts, eroded gullies,
vacant lots, field edges, etc., provide favorable habitats for colonization by
some species, such as C. punctulata. Similarly, the construction of farm
ponds, lakes, and irrigation canals has probably aided species that live near
water. On the other hand, increased use of insecticides, housing develop-
ments, polluted waterways, and the damming of rivers has undoubtedly
adversely affected some species.

Man has also affected many saline habitats. Agricultural activities may
lead to silting in of saline habitats in natural depressions. Schaffner (1898)
said that a salt marsh in Republic County, Kansas, had suffered much in this
manner already by 1897, being much smaller than it formerly was and having
only several hundred acres without vegetation. Today, nearly the entire area
is vegetated, some is being used for raising crops, and during the course of
this study, a marshy area (Fig. 7) was drained and an unsuccessful attempt
made to grow crops on it. As mentioned above, man may also create saline
habitats by irrigation or drilling for oil. Fender (1945) reported finding one
specimen of C. cuprascens (dead) and several of C. punctulata (three alive)
in McMinnville, Oregon, in mail sacks from Council Bluffs, lowa. These
species are attracted to lights and must have fallen into the mail sacks at
night. Thus, any species coming to lights may be transported considerable
distances by man under favorable circumstances. Also, the larvae of any
species could conceivably be transported by man in loads of soil.

Tue Lire Cycie. Only a few of the world’s species of cicindelids have had
their life cycles worked out completely or even in part. Development takes
one to several years, and some species do poorly under laboratory conditions,
making them difficult to rear. Enock (1903) and Huie (1915) worked out
the four-year cycle of C. campestris. Criddle (1910) studied the cycles of
C. limbata, C. limbalis, C. formosa, and C. lengi in the field. Shelford (1908)
worked out the cycles of C. punctulata, C. purpurea, C. lepida, C. limbalts,
C. formosa, and C. sexguttata from laboratory and field observations and
gave partial cycles for other species. Zikan (1929) reared many Brazilian
species in several genera partly through their cycles. Shelford (1908, 1911)
noted that temperature, moisture, climate, and food influence the length of

220 Tue University ScrENcE BULLETIN

Taste 11. Life cycles of Cicindela worked out by previous authors (Criddle, 1910;
Shelford, 1908; Huie, 1915). Only one possible cycle is shown for each species;
larval stages may be lengthened. The exact times of appearance of the stages
varies geographically and from year to year. A=adult; O=oviposition; 1L, 2L,
3L—first, second, and third larval instars; P—pupa; ....—hibernation.

1 year cycle 2 year cycle 2 year cycle 3 vear cycle 4 year cycle
(C. punctulata) (C.purpurea) (C.lepida) (C.lengi) (C. campestris)
' tance el Li tld eee ie ae
'
Pe apes: eae: eee
April SIL ; 3L Sb, AL ‘ JIG SIL IN WAL SIL SNL
|
0 0 0
May IL, P IL
P IL.
June P
A 2L b\. Sal,
|
0 0 PAL
A P
July Ie Bibe
IL 3L IL
ZL,
August 2L A A
2L
Sie,
Sept. 3L

com me
ee——=

Oct.

BronomMics AND ZOOGEOGRAPHY OF TIGER BEETLES IAI

the different stages of the life cycle, particularly the larval stages; thus some
of the species that Shelford studied at Chicago were found by Criddle to
have longer life cycles in Manitoba, Canada. This, plus the facts that life
cycles are long and adults oviposit for a number of weeks or months, often
gives rise to great overlaps and a heterogeneity of cycles for a species. In
general, two major types of life cycles can be delimited. In the first, adults
emerge from the pupa in the fall, feed, hibernate, become active in the spring,
and oviposit; variable larval stadia may produce a two to four year cycle. In
the second type, adults emerge in the early summer and oviposit soon there-
after, dying off before winter; the total life cycle may last one to two years.
The life cycles worked out by the above authors are shown diagrammatically

in Table 11.

I succeeded in rearing only one individual of C. togata completely through
its life cycle. Adults of C. togata were collected from Lincoln County, Kan-
sas, on 7 September 1963, and kept in a laboratory terrarium. Several days
later they began ovipositing. First instar larvae were first seen on 24 Septem-
ber 1963; on 28 January 1964, the larva that was eventually reared became a
second instar larva; and on 20 April 1964, it became a third instar larva. The
third instar burrow was last closed on 20 July 1965. An adult female, with
elytra still soft, dug its way out of the soil on 6 September 1965, two years
after the adults were collected. This evidence, plus observations on larvae in

C. circumpicta

C. duodecimguttata

C. togata

C.willistoni

90

Fic. 90. Temporal distribution of larvae of different instars (1=first, 2=second, 3=third)
and adults (height of curve indicates relative abundance) of four species of saline habitats. A
boldface number indicates that the instar was very abundant.

222 Tue Universiry SciENcE BULLETIN

the field, leads to the conclusion that the life cycle takes two years (but may
be lengthened to three), and is similar to that of C. lepida in Table 11.
Larvae apparently hibernate in any of the three instars (Fig. 90).

In Figure 90 are shown times of the year that I have seen or collected
larvae of different instars of four species. From these data and the times of
adult activity, certain tentative conclusions can be drawn about the life cycles:
C. circumpicta appears to have a life cycle similar to that of C. togata. C. duo-
decimguttata and C. willistoni probably have cycles similar to that of C.
purpurea in Table 11; however, from the temporal distribution of larval
instars, their cycles may only take one year.

ZOOGEOGRAPHY OF CICINDELA OF SALINE HABITATS

Zoogeography is a very broad field of study, drawing its data from the
disciplines of ecology, systematics, phylogeny, paleontology, paleoclimatol-
ogy, physical geology, pedology, and geography. Zoogeography may be de-
fined as the study of the distribution of animals in space and time, how and
why this distribution came about, and prospects for future changes. Discus-
sions of the principles and methods of zoogeography can be found in Darling-
ton (1957) and Munroe (1963).

Many papers have appeared recently on the zoogeography of North
American insects; e.g., Gressitt (1958), Howden (1963), Linsley (1939,
1958), Miskimen (1961), Rhen (1958), Ross (1953, 1958), and Ross and
King (1952). The results of some of these can be applied to cicindelids.
Works dealing with the zoogeography of cicindelids are fairly numerous.
Horn (1908-1915) made speculations on the phylogeny and past dispersal of
the family, but he was hampered by the relatively incomplete geological
knowledge of the time. Rapp (1946) listed the distribution of the cicindelid
genera on the seven major land masses of the world and hypothesized about
their origin and dispersal. Crowson (1946) quickly criticized some of Rapp’s
conclusions. Kolbe (1935) expounded his theory of “morphological progres-
sive animal dispersal,” using the palearctic species C. Junulata as an example;
the theory proposes that the most primitive forms occur at the place of origin
and the most advanced at the periphery of the range. This is similar to the
“age and area” hypothesis of J. C. Willis (see Darlington, 1957). Mandl
(1954), studying the male genitalia of many of the cicindelid genera, hypothe-
sized about their evolution and dispersal. Papp (1952), in a study of the
male genitalia of 33 North American species of Cicindela, placed the species
into groups of close relatives, noted the existence of close relatives in the pale-
arctic region, and drew general conclusions about the evolution and dispersal
of the North American cicindelid fauna. Schilder (1953b) subdivided the

classical Cicindela into 18 genera and 29 subgenera, based at least partly on

BIioNoMIcs AND ZOOGEOGRAPHY OF TIGER BEETLES 223

distribution, and speculated about their evolution and dispersal. Rivalier
(1950, 1954, 1957, 1961, 1963) published a much more “natural” classification
of the classical Cicindela (also subdividing it into genera) based primarily on
the male genitalia. Ihering (1926) speculated about the dispersal of Mega-
cephala carolina. Van Dyke (1929, 1939), in more general papers on North
American Coleoptera, mentioned the faunal affinities of Megacephala, Ambly-
chila, and Omus. Leconte (1875a, 1875b) proposed that the occurrence of
C. hirticollis and C. lepida in similar habitats along the Atlantic coast and in
the interior of North America could be explained by assuming that the inland
populations are remnants of populations that lived on the shores of Creta-
ceous seas. Wickham (1904a, b) speculated on the evolution of C. willistoni
and correlated its supposed evolution and dispersal with geological events.
Shelford (1907) noted the preference of C. sexguttata for oak-hickory forests
and predicted that as most forests of northeastern North America changed to
a beech-maple climax (man eliminated), this species would be driven out of
the area. Leng (1912) discussed factors controlling the distribution of the
species of Cicindela of eastern North America: temperature, mountain bar-
riers, local environment, behavior and adaptability of the species, accidents of
climate and geological changes, and place of origin. Cazier (1948) treated
the origin and dispersal of the cicindelid fauna of Baja California, Mexico,
as it was known then. In a later publication (1954), he grouped the Mexican
species of Cicindela according to their faunal relationships. Rumpp (1956,
1957, 1961) studied a number of species in the southwestern United States,
including C. willistoni, C. nevadica, C. fulgida, C. circumpicta, and C. togata,
and correlated their possible evolution and dispersal with geological events.
Freitag (1965) postulated phylogenetic and zoogeographic schemes to ex-
plain the distribution, geographic variation, and systematics of nine North
American species, particularly C. duodecimguttata, C. oregona, and C. de-
pressula.

In this work, seven species were studied in detail zoogeographically: C.
circumpicta, C. cuprascens, C. fulgida, C. macra, C. nevadica, C. togata, and
C. willistoni. Brief mention will be made of other species. Each species will
be discussed individually.

Mertnops. Data from several sources were used. The present known dis-
tribution of the species was determined by personal collecting and from
records from the literature and museum collections. Geographic variation
was studied statistically. Samples of specimens from various localities,
assumed to have been collected at random, were measured. External morpho-
logical characters were used, such as lengths and widths of body parts, shapes
of the white elytral maculation, and color. Males and females were studied
separately. Size measurements were taken using an eyepiece micrometer in
a binocular dissecting microscope and were later converted to millimeters.

224 Tue University ScrENCE BULLETIN

Fic. 91. Left elytron of C. cuprascens (dorsal aspect), illustrating terminology of the
maculation: al=apical lunule; b=basal dat, hl=humeral lunule, mb=middle band, ml=
marginal line. Fic. 92. Color wheel used to study geographic variation of elytral color in five
species of Cicindela. The brightest (most saturated) colors are around the rim; dark colors are
toward the center, with black being at the center. The numbers are used for later reference. The
marks outside the rim divide the wheel into six sectors for more simplified representations.

BrioNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 225

The possible shapes of the elytral maculation were subdivided into several
arbitrary states, which were given numbers; specimens were then scored in
decimal fractions of these states. The terminology used with the maculation
is as follows (Fig. 91). The maculation is quite variable; however, a typical
pattern consists of a humeral lunule (with or without a basal dot), middle
band, and apical lunule. These markings may be connected at the margin
by a marginal line. From this pattern the markings may be reduced to dots
or complete absence, or they may be fused until the elytra are white or
nearly so.

In the case of C. togata, variations of elytral color could be arranged in a
linear series and represented by a scale of numbers; color was then treated as
any other character. However, in the other species (except C. fulgida, for
which color was not measured because of the temporal change of color noted
in the section on adult post-emergence changes), the range of colors is more
complex and had to be studied separately from the other characters. The
possible colors, with but few exceptions, could be arranged in a color wheel
(Fig. 92), with bright green, blue, and reddish at equidistant points at the
edge, intermediate colors between them (blue-green, purple, red-green), and
darker shades of the same colors toward the center (black being at the
center). The color wheel was subdivided into 42 sections (numbers in
Figure 92), and specimens were found which fit into all of them. The names
of these colors, in the Ridgway (1912) and Inter-Society Color Council-
National Bureau of Standards (Kelly and Judd, 1955) systems are given in
Table 12. Specimens were examined against a white background with the
naked eye, using overhead lighting from Sylvania cool white fluorescent
bulbs. With the eye perpendicularly above the elytra, the position the speci-
men occupied on the color wheel was determined and a dot made on a data
sheet with an outline of the wheel and its sections. Viewing such a sheet,
with many dots representing a sample from a given locality, gives one an
idea of the “color structure” of that population. For the purpose of compar-
ing many populations on a map, a simplified, less detailed version of the
color wheel was used because of space limitations on the maps.

The measurements, except those involving the color wheel, were analyzed,
using an IBM 7040 computer, by the method of multivariate generalized
discriminant functions. The computer program, available at the Computa-
tion Center of The University of Kansas, was written by Dr. F. James Rohlf,
University of Kansas. It is called MULDIS, short for multigroup discrimi-
nant functions. The theory of generalized functions is explained in Jolicoeur
(1959) and Seal (1964), where it is referred to as “canonical analysis.” In
brief, the method consists of a simultaneous analysis of the characters of
specimens drawn from a number of localities. Any differences that exist
among the groups of specimens can be displayed in the most efficient man-

226

Tue UNiversiry ScIENCE BULLETIN

Tasie 12. Color names of the sections of the color wheel (first column) in the

Ridgway (1912) and Inter-Society Color Council—National Bureau of Standards

systems. The names and numbers of the latter system often have greater latitude
than those of the Ridgway system, accounting for duplication.

Section

Ridgway

ISCC-NBS

BRO ee
SCO WN NUMA WNRK OW ONDA UB WN

ia)
Noe

ND MH NH NH
NU BW

Nm hw NY
\o co NI

30

WWW WW W W
NID UW WN

bb BW W
N= © oO: 6

Brick red

Hay’s brown

Clove brown
Chaetura drab
Russet

Prout’s brown

Natal brown
Cinnamon brown
Saccardo’s olive
Dark olive

Dresden brown
Forest green

Roman green

Dark dull yellow green
Cossack green
Varley’s green
Danube green
Dusky olive green
Meadow green
Bottle green

Duck green

Dark viridian green
Invisible green
Dusky dull green
Wall green

Myrtle green

Dusky green-blue (1)
Dusky dull bluish green
Alizarine blue
Dusky orient blue
Dark delft blue
Bluish slate-black
Dark aniline blue
Dull violet-black (3)
Dull purplish black
Burnt lake

Dusky auricula purple
Blackish violet-gray
Vandyke red
Madder brown
Diamine brown
Dark mineral red

strong reddish brown (40)
moderate reddish brown (43)
dark grayish reddish brown (47)
brownish black (65)

strong brown (55)

strong brown (55)

grayish brown (61)
moderate brown (58)

dark olive brown (96)

dark olive brown (96)

light olive brown (94)

deep yellow green (118)
moderate olive green (125)
moderate olive green (125)
deep yellowish green (132)
dark yellowish green (137)
dark green (146)

very dark yellowish green (138)
strong green (141)

deep green (142)

very dark green (147)
strong green (141)

dark bluish green (165)
very dark bluish green (166)
deep bluish green (161)
deep bluish green (161)
dark greenish blue (174)
very dark greenish blue (175)
deep blue (179)

deep blue (179)

dark blue (183)

dark blue (183)

deep purplish blue (197)
dark purplish blue (201)
dark violet (212)

very deep purplish red (257)
very dark red (17)

blackish purple (230)

dark red (16)

dark red (16)

very deep red (14)

very deep red (14)

BroNomMics AND ZOOGEOGRAPHY OF TIGER BEETLES 227

ner. One means for doing so is to transform the locality means into specially
standardized units, so that the means of the localities can be plotted as points
in an n-dimensional hyperspace, where n=number of characters, or number
of localities —1, whichever is the smaller. The coordinate axes (generalized
discriminant function) of these points are constructed so that the greatest
amount of variance among localities (relative to that within localities) is
explained by the first discriminant function, the next greatest amount by the
second function, etc. Chi square tests are performed on each function; non-
significant functions are ignored. A matrix of generalized distances is pro-
duced, giving the distance from the mean of one locality to that of any other;
the greater the distance, the more different the specimens from the compared
localities in the characters measured. One can also perform analyses of vari-
ance (anova) on each character with intermediate output of the program.
A test is also made for homogeneity of the variances of the groups.

The significance of the differences in position of the means of the localities
in n-dimensional hyperspace was tested using an unpublished method of
Dr. K. R. Gabriel, called “likelihood ratio manova simultaneous testing
procedure” (STP). Program 6, option D, available at the Computation
Center of The University of Kansas, was used. In this method, the prob-
ability of making a type I error (that is, rejecting a true hypothesis) is known;
a value of 5°% was used. This is an experiment-wise error rate; a type I error
will be made in 5% of the studies, or experiments. Another characteristic of
this method is that if a certain set of means is found to be not significantly
different, no subset within it will be significantly different.

Inferences about the phylogeny of the species studied were drawn from
the literature, particularly Rivalier (1954), and from the results of this study.
The grouping of species by Rivalier (1954) was used as a basis for the system-
atic arrangement of the North American fauna of Cicindela; however, it was
found that certain minor corrections were necessary, on the basis of examina-
tion of the male genitalia of certain species. Mandl (1954) gives instructions
for the preparation and observation of the male genitalia; however, in his
method the inner sac is everted, making it difficult to see the interrelation-
ships of the various internal parts. Freitag (1965) gave very brief instruc-
tions. The method I have developed is as follows:

1) If working with dried specimens, relax them for at least a day, or use a relaxing fluid.

2) Holding the beetle upside down, reach inside the genital opening with fine-pointed forceps
and gently remove the aedeagus (it is best to use a low power dissecting microscope). Choosing
specimens with the aedeagus partly extruded makes this easier. It may be necessary first to
separate the aedeagus from the sclerites around the anus, to which it often adheres because of
dried body fluids. Be careful that the curved proximal part does not break or that the basal
piece and lateral lobes do not become separated.

3) Place the aedeagus in a 10% potassium hydroxide (KOH) solution for about 5-10 minutes
to dissolve the tissue that usually adheres to the proximal part. Transfer it to a small Stender

dish or microscope slide with a small amount of water and remove any remaining tissue with
forceps. A weak acetic acid bath may be used to stop the action of the KOH.

228 Tue University ScrENcE BULLETIN

I |_mm
35

| 0.5 mm

2/5)

Fic. 93. Aedeagus of C. cuprascens (Douglas County, Kansas), dorsal aspect. Abbreviations:
b=“beak,” bp=basal piece, c=concavity, d=denticles, ed=ejaculatory duct, ll=lateral lobes
(only one is drawn), o=orifice, r=ridge. Fic. 94. Inner sac of aedeagus of C. fulgida (Stafford
County, Kansas); a, dorsal aspect; b, ventral aspect. Abbreviations: ap=arciform piece, cp=
central plate (partly weakly sclerotized), dm=denticulate membrane, ed=ejaculatory duct,
f=flagellum, pa=paired areas with large denticles, r—=rod, ssr—small stiffening rod, t=tooth.
Fic. 95. Inner sac of aedeagus of C. cuprascens (Douglas County, Kansas); a, dorsal aspect;
b, ventral aspect. Abbreviations same as in Fig. 94.

4) Place the aedeagus on its “dorsal side” (as it lies at rest in the beetle) and make a drawing
of it (see Fig. 93). This is best done with a camera lucida or ocular grid and graph paper. Note
ridges, concavities, and denticles near the distal end. Also note the shape (curvature, thickness,
etc.) of the entire aedeagus and of the heavily sclerotized part (often beak-shaped) at the apex,
as well as the relative length of the lateral lobes. Sometimes a useful character is found by
viewing the apex from the “side”; i.e., from the direction of the orifice. Add a millimeter scale
beside the drawing.

5) Slit the aedeagus longitudinally from about the middle to the apex. A scalpel or razor
blade might work for this, but I used an insect pin with the point bent into a tiny hook, the pin

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 229

mounted in a wooden handle. Hold the proximal part of the aedeague with forceps and push
the insect pin into it proximal to the inner sac to make a starting hole. Then pull the tool
toward the apex, being careful not to poke it deep inside, using the hook to tear a slit. If the
aedeagus is heavily sclerotized, this operation may be difficult.

6) Holding the proximal part of the aedeagus with forceps, reach inside the slit near the
middle of the aedeagus with a fine teasing needle or another forceps, grasp the ejaculatory duct
proximal to the inner sac, and gently pull the inner sac out of the slit. It will be connected to
the aedeagus at the orifice by membranous cuticle; the membrane should be severed with a
scalpel or teasing needle, being careful not to tear it where it continues over the inner sac.

7) Place the inner sac in a 10% KOH solution for 5-10 minutes, immerse it briefly in an
acetic acid bath, remove it to a slide, and remove excess tissue with forceps, exposing the
sclerotized membranes and pieces that compose the inner sac. The great complexity of these parts
makes them difficult to draw. Make drawings from the dorsal and ventral aspects and any
others that clarify the shapes and interrelationships of the parts. The sclerotized pieces are
covered by membranes, often produced into complex folds; some parts of the membrane are
transparent and some are covered by denticles of various sizes. The size and distribution of
denticles is often an important character, usually more so among species groups than among
closely related species. The sclerotized pieces show the most important characters; they vary
widely in shape within the genus Cicindela, but they can usually be homologized from one
species to another (see Rivalier, 1950-1963). Two examples, one of a more primitive species
and one of a more advanced species, are shown in Figures 94 and 95. One can often see the
parts of the inner sac more clearly if it is examined (by transmitted light) immersed in glycerin
rather than water.

8) Keep all parts in glycerin in a corked microvial on the pin of the specimen.

9) Examine the genitalia of at least several specimens of a form to determine the range of
variation of the characters.

10) Be certain that the aedeagi and inner sacs are oriented the same way for drawings; a
slight rotation can change the appearance of many parts.

PHYLOGENY

Unfortunately, few fossils of cicindelids are known. Horn (1906) dis-
cussed a Megacephala from the Baltic amber, which he said was conspecific
with the modern American M. carolina, and hypothesized about how it
happened to get there; however, Cockerell (1920) though that this specimen
was a fake. Cockerell (1920) described a fossil, represented only by an ely-
tron, from White River, Colorado (Green River age), as Czcindelopsis
eophilus; however, as he pointed out, it does not have an inner apical elytral
angle as do known cicindelids, and it may be from a cerambycid. G. Horn
(1876) discussed two species of fossil Cicindela from a post-Pliocene cave at
Port Kennedy, Pennsylvania. He said one is C. haemorrhagica, a species
now restricted to the southwestern United States and northwestern Mexico;
he did not name the other species. A close relative of C. haemorrhagica,
C. rufiventris now occurs east of the Rockies; Horn could have misidentified
the specimen.

Consequently, conclusions about the phylogeny of cicindelids must be
drawn almost entirely from the present fauna. Earlier attempts to make
phylogenies, e.g., Horn (1908-1915) and Schilder (1953b), were based on
external morphological characters, particularly the elytral maculation in
Cicindela. Mandl (1954) produced a much more satisfactory scheme (al-
though fragmentary) for the family, using the male genitalia in addition to
external characters. Papp (1952) arranged the species that she studied into

:
230 Tue Universiry Science BULLETIN

Habroscelimorpha Microthylax

Brasiella , Eunota f- Opilidia

e
°
.?
a2 oe
<
% *
\

. ~~ e a
i * Ag eo
ici idi Seglene” llipsopt
Cicindelidia “tafe? Ellipsoptera
|
: Dromochorus
: Cylindera
Cicindela ;
~
(

PALEARCTIC REGION

Cicindela Habroscelimorpha
C.hirticollis C.circumpicta

C.duodecimguttata
Eunota

C. repanda
C.togata

C.fulgida

C.tranquebarica Ellipsoptera

C. nevadica
Cicindelidia
C.cuprascens

C.willistoni
C.macra

C.punctulata

96 C. schauppi

Fic. 96. Proposed scheme of evolution of the North American fauna of Cicindela. Names
are the “genera” of Rivalier (1954). The species in this study are listed under their respective
“genera” at the bottom.

BIioNomMics AND ZOOGEOGRAPHY OF TIGER BEETLES 231

groups progressing from phylogenetically old to young, as did Rivalier (1950-
1963) in his more complete work. Rivalier made few statements about the
actual phylogeny of his groups except for some remarks in his last paper
(1963). From the meager comments of Mandl (1954), Papp (1952), and
Rivalier (1963), and from the excellent systematic arrangement of Rivalier,
it is possible to speculate on the phylogeny of the Cicindela of North America.

The family Cicindelidae and the genus Cicindela probably arose in Africa
(Mandl, 1954). From there, secondary centers of evolution in the Old World
tropics were colonized. Population of the New World by Cicindela probably
occurred only from the north, via the Bering land bridge (although Mandl,
1954, thought that other genera dispersed to the New World at an earlier
time via a southern land bridge). Much radiation occurred in the New
World, producing many indigenous groups (“genera” of Rivalier). A hypo-
thetical scheme of evolution, using the names of Rivalier’s “genera” is shown
in Figure 96. At least two major ancestral stems crossed the Bering land
bridge. One was Cicindela s. s., a group found throughout the Old World as
well. This “genus” gave rise to Cicindelidia, which is restricted to the New
World. The other major stem was Cylindera, a group also found throughout
the Old World. It gave rise to Ellipsoptera and Dromochorus, which are
restricted to the New World. The remaining five “genera” are an endemic,
heterogeneous group whose origin is more uncertain. Probably at least some
of them evolved from ancestors that crossed the Bering bridge; others may
have split off the Cylindera stem.

Because of the paucity of cicindelid fossils, it is difficult to date the evolu-
tion of Cicindela. According to Carpenter (1953), the earliest known beetle
fossils are from late Permian strata; cicindelids probably arose at about this
time. The evolution of most of the cicindelid genera probably occurred
during the Mesozoic. Horn (1908-1915) thought that the genus Cicindela
became differentiated in the early Tertiary and that the ancestors of the
American fauna crossed the Bering bridge in late Tertiary. Rumpp (1961)
considered that the ancestors of C. willistonit had already reached North
America by the beginning of the Tertiary, and that there has been little
evolution of this species since Miocene time. Freitag (1965) thought that the
ancestral stock of Cicindela s. s. may have been in existence in early Tertiary,
and that living species may have evolved during late Tertiary or early Pleisto-
cene. Thus it is possible that the evolution shown in Figure 96 took place in

“The question of land bridges is a touchy one. The presence of the Bering bridge during
certain periods of geological time has been well established. Some early biogeographers were
prone to “build” bridges where there was not the slightest shred of geological evidence to support
them. For many years the trend has been away from this extreme. A related and even more
volatile subject is continental drift. See Darlington (1957, 1965) and Simpson (1965) for recent
discussions of these topics from the zoogeographical viewpoint.

232 Tue University SciENCE BULLETIN

late Mesozoic and early Cenozoic times. The Bering land bridge was uplifted
during most of this time, and the climate at that latitude was mild (Miski-
men, 1961).

SUBSPECIES

There has been a recent revival of interest in the question of whether sub-
species have any reality in nature and, if so, whether they should be named in
the formal system of Linnean nomenclature. Of particular importance is a
series of papers in the journal, Systematic Zoology. Wilson and Brown (1953)
condemned the arbitrariness and subjectivity of naming subspecies and
pointed out several difficulties: 1) the tendency for independent characters
to show independent geographic variation: 2) the ability of characters to
appear in more than one geographic area, producing “polytopic” races; 3) the
occurrence of microgeographic races; and 4) the necessary arbitrary lower
limit of distinction of the subspecies. They further stated that subspecies as
currently used are not units of evolution and that naming them conceals
much variation. Other papers followed, supporting (Mayr, 1954; Parkes,
1955; Durrant, 1955; Smith and White, 1956) or refuting (Hubbell, 1959;
Edwards, 1954; Peters et al., 1954; Gosline, 1954; Gillham, 1956; Hagmeier,
1958; Christiansen, 1958; Pimentel, 1959; Owen, 1963) the naming of sub-

species.

My feelings on the subject, tempered by many of the above papers, as well
as others, are as follows. Most of the difficulties mentioned by the opponents
of subspecies are due to the extreme variability of animals. Populations of
animals exist today in all stages of evolution, from a small, specifically distinct
population inhabiting a single island, mountaintop, or valley, to widely dis-
tributed species, some of whose populations are not capable of interbreeding
(see Mayr, 1963, for examples). In some species, one can find populations or
groups of populations which are quite “distinct” in one or more characters
from other populations of the species, while in other species this cannot be
done. Thus, “distinct” intraspecific groups may exist within species. In the
past, many such “distinct” groups (and some not so distinct) have been
named as subspecies. At least some such groups (but by no means all) seem
worthy of being named, if for no other reason than convenience in referring
to them. In groups in which formal names have already been applied, one
might just as well retain them. If no names exist, one may wish instead to
use locality names or symbols to refer to them, as suggested by Hubbell
(1954); although this may lead to confusion, as pointed out by Smith and
White (1956). In some cases, some subspecies do appear to be units of evolu-
tion, as in geographically isolated populations, while in other cases they are
not (at present), as in “distinct” groups among which gene flow occurs or is

ie)
Wo
Us

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES

possible. There are all degrees of “distinctness” among intraspecific groups,
making it impossible to set a nonarbitrary “lower limit” for subspecies.
Statistical methods are useful in defining distinctness, but they can be mis-
used. It is possible to find statistically “significant” differences between almost
any two conspecific populations by using a large enough sample size, but
these differences may have no biological significance. The many arbitrary
limits, such as the “75°/ rule,” the “84°/ rule,” or the “959% rule,” have been
used with only limited success, and any one “rule” does not work well with
all phyla of animals. Sokal and Rinkel (1963) pointed out the inadvisability
of using such rules. I think that statistical methods, plus the opinion of a
systematist familiar with the group of animals, are the only practical way to
decide if a species should be subdivided and how, at our present state of
knowledge. The recent development of methods such as numerical taxon-
omy may change this situation in the future.

Usually the naming of subspecies does conceal much variation; the study
of species should not stop here. The variation of species, geographic and
temporal, should be measured, studied, and shown by maps, graphs, etc. The
use of a system of categories with a “higher degree of resolution,” such as
natio, subspecies, prospecies, species, and superspecies (see Schilder, 1953b, for
an example using Cicindela), has not been popular, and in many cases the
assignment of a form to a particular category is quite arbitrary. Subspecies
are necessarily an oversimplification of the “true” situation. Nevertheless, if
subspecies reflect to some degree the actual pattern of variation, as well as
being convenient “handles” for reference, their value seems sufficient to
justify their recognition.

In the genus Cicindela many of the species exhibit pronounced geographic
variation, particularly of color and elytral maculation. For example, of the
approximately 80 species occurring in North America north of Mexico, about
45 exhibit pronounced variation. Past workers have named many forms,
which are now recognized as subspecies, varieties, intergrades, melanic forms,
seasonal forms, and individual variants. Often, forms were described before
the total distribution of variation of the species was known, or were described
from inadequate series. Thus, one task of the present-day student of North
American cicindelids is to make thorough studies of the variation of species
and decide whether subspecies or formerly unrecognized sibling species exist
(and should be named) within what have been called species. This will be
done with the seven species studied thoroughly here.

Being familiar with the variation of many of the Cicindelas of North
America, I think a subspecies in this group should have the following charac-
teristics: (1) occupy a well defined geographic area or ecological habitat,
separate from that of other subspecies within the same species (zones of
intergradation may occur between neighboring subspecies); (2) exhibit a

234 Tue UNiversity ScrENcE BULLETIN

time scale

Postglacial
Wisconsin
glacial

Sangamon
interglacial

Illinoian
glacial

Yarmouth
interglacial

Quaternary period
Pleistocene epoch

. s ¢ ¢ Q -
(Tote
°o

Y
Kansan Cascadiar4

glacial Revolution

cor PS |

Nebraskan
glacial

<t
ar
ud
<
ro)
N
e)
Fie
Ld
So)

Oligocene

Paleocene < yaa

Cretaceous
period

Tertiary period

o ™M o

~y) oO Oo << a
oo ro) oS 3
ey re) O° °
zs 2 = 2 =

@
ao)
(2)
(2)
=P

MESOZOIC

Fic. 97. Partial geological time scale and important events in North America. From
Dunbar (1960) and Dorf (1960).

BIoNomMics AND ZOOGEOGRAPHY OF TIGER BEETLES 233

relatively uniform expression of characters within itself; (3) be readily sepa-
rable from other subspecies by one and preferably more characters.

These characteristics are obviously vague and contain no minimum degree
of difference for a form to be a subspecies. I think each case should be con-
sidered individually (but comparison with the variational pattern of other
species may be helpful) because cases will no doubt be found in which two
forms may be “on the verge” of becoming species. Any clues on the past
distribution or dispersal of the species (e.g., ancient river courses for fluvial
species) should be considered. The problem is further complicated by the
existence in many species of some forms which are much more distinct (and
obviously subspecies) than others. Clearly there must be some subjectivity
in the naming of subspecies.

CENOZOIC GEOLOGY AND CLIMATE

Before discussing the zoogeography of individual species, it would be well
to review the major geological changes and paleoclimates that occurred in
North America during the Cenozoic Era, when most of the evolution and
dispersal of the species in this study probably took place. This amount is
drawn from a number of sources. See Miller (1952), Dunbar (1960), King
(1958), MacGinitie (1958), Dorf (1960), Braun (1947, 1955), Axlerod (1950),
Chaney (1947), Deevey (1949), Dillon (1956), Thornbury (1965), and
Wright and Frey (1965) for more details. See also Figure 97 for a geologic
time scale.

The Cenozoic Era was marked by two great periods of orogeny, or moun-
tain building activity. Beginning in the Cretaceous and continuing until
early Eocene, the Laramide Revolution thrust up the initial Rocky Mountains
in western North America. These mountains were eroded almost to a level
peneplain by Miocene times, when the Cascadian or Cordilleran Revolution
began, uplifting the western mountains again to their present height. During
the Laramide Revolution, numerous basins were formed in the western
United States, and in the Miocene, the Basin and Range province, which had
a high mountainous surface and exterior drainage, began to assume its
present character. As the initial Rockies were eroded, these basins were filled
and sediment was carried eastward across the interior of the continent.

During most of the Tertiary, eastern North America, from central Ala-
bama to the Gulf of St. Lawrence, which had been uplifted slightly at the
close of the Cretaceous, was eroded to a nearly flat peneplain. Near the end
of the Tertiary the area was again uplifted unevenly, and erosion increased
to produce the present topography. The Atlantic and Gulf coastal plains
were partially submerged from Paleocene to Oligocene or even Pliocene in
some areas, but in general they gradually grew in size throughout the

236 Tue Universiry SciENcE BULLETIN

Tertiary. A remnant from Cretaceous times, the Mississippi embayment, an
arm of the Gulf of Mexico extending up the present Mississippi valley to
southern Illinois, was gradually filled with sediment in the late Tertiary.

In late Cretaceous times, the climate was mild throughout most of the
earth. Plants such as figs, cycads, palms, and tree ferns grew as far north as
central Greenland and Alaska. The climate was nevertheless zoned as it is
today. The tropics extended northward to about 35 or 40° latitude. From
here to about 55 or 60° N latitude, subtropical conditions prevailed. Temper-
ate climates extended to about 70° N latitude, north of which subarctic
conditions occurred. ;

Conditions during the early Tertiary did not change greatly. The western
mountains were mostly of moderate elevation and did not affect climate
appreciably. By mid-Eocene times there were at least three botanical provinces
in North America. In the far western states a subtropical forest extended
along the coast and inland as far as northwestern Wyoming (Neotropical-
Tertiary flora). The low-lying shores of the Mississippi embayment were
occupied by a tropical flora (Wilcox flora). Far to the north extended the
hardwood deciduous and coniferous Arcto-Tertiary forests. Arid conditions
began to appear locally in northwestern Colorado and southern Wyoming,
although the modern desert vegetation had not yet evolved.

Beginning in the Oligocene, a trend of gradual cooling and drying climates
began. The northern Arcto-Tertiary forests began to move southward
through the western United States, displacing tropical and subtropical floras.
A climatic barrier of reduced rainfall prevented any of these species from
entering the Appalachian region or Mexico. The Miocene saw the evolution
of the semiarid Madro-Tertiary flora in northern Mexico and southern
California, while the Arcto-Tertiary flora became more restricted. An ecotone
existed between them in southern Nevada. The Arcto-Tertiary flora evi-
dently still had a dispersal path to the eastern United States through southern
Canada. In the vicinity of Washington, D.C., a low coast existed, lined with
cypress swamps and coastal sand dunes.

Beginning in mid-Miocene and especially during the Pliocene, the pres-
ent grasslands developed in the western two-thirds of the continent, replacing
subtropical scrub in the region between the Rocky Mountains and the
Mississippi embayment. Farther east, the mixed deciduous forests retreated
and were replaced by oak forests. Temperatures cooled, rainfall decreased,
and seasonal fluctuations increased, until in late Pliocene conditions were
essentially like they are today. The uplift of the western mountains and their
resultant rain-shadow effect played a major role in the development of the
prairies and deserts.

The climatic changes begun in the Tertiary culminated in the Pleistocene
epoch. A cyclic climatic pattern developed, producing alternating glacial

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 257A

and nonglacial stages. Four major glacial periods (the Wisconsin is sub-
divided by some authors) alternated with interglacials (Fig. 97). During the
glacials, huge masses of ice moved southward, and mountain glaciers de-
veloped and increased in size. South of the glaciers rainfall increased, creat-
ing many large lakes in the Great Basin (the glacials are also called pluvial
periods). The continental glaciers pushed farthest south in the central
United States, reaching the present Ohio River and northeastern Kansas.
Climatic zones were greatly compressed near the glaciers and shifted south
(or down mountains) a certain amount; exactly how far is a point of dis-
agreement among authors. Early authors thought that glaciation pushed
the flora and fauna far south into Mexico and Central America. More recent
authors, e.g., Dillon (1956), Dorf (1960), felt that climatic zones were more
compressed than shifted south, and that the Gulf coast was still subtropical.
Graham (1964) thought the glaciers had little effect on the biota of the south-
eastern United States. Probably a narrow band of tundra existed immediate-
ly south of the glaciers, followed by bands of subarctic, temperate, and sub-
tropical climate as one moved south. Ranges of mountains in the West
would produce a more complicated pattern, greatly influenced by elevation.
During the interglacial periods the climatic zones and biota moved north-
ward (and up mountains) to or slightly beyond their present positions.

The glaciers had profound effects on the sea level. During glacials the
sea level was about 100 meters lower than at present, exposing much of the
continental shelves and allowing the Bering land bridge to connect Eurasia
and North America. The glaciers also had great effects on many North
American streams. As mentioned, during glacials rainfall increased near the
glaciers. Many streams in the Great Basin which now are intermittent or dry
could have been important in the dispersal of certain organisms. Also, many
changes in the courses of streams occurred and may have affected the distri-
bution of organisms.

At present we appear to be in an interglacial period, with the next glacial
period predicted to occur in 10,000 to 15,000 years. During the last several
hundred years the mean world temperatures have been rising, glaciers are
melting, and northward movements of animals such as seals, codfish, and
armadillos have been noted.

MINOR SPECIES

Miscellaneous distribution records on the species not studied intensely
follow.
C. duodecimguttata
The systematics and zoogeography of this species were studied thoroughly
by Freitag (1965). Besides what has been mentioned in earlier sections, I
offer the following personal collection records:

238 Tue University ScriENCE BULLETIN

KANSAS: Republic Co., % mi. east, 1 mi. south of Talmo, 6 Sept. 1963; 4 mi. northwest
of Jamestown, 14 Sept. 1964; Lincoln Co., 3 mi. west, 2 mi. south of Barnard, 14 Apr., 7 Sept.
1963; Greenwood Co., 2 mi. west, 1 mi. north of Severy, 11 May 1963; 1.5 mi. west of Severy,
11 May 1963; Wilson Co., west edge of Fredonia (city park), 11 May 1963, 6 Apr., 20 June
1964; Woodson Co., 5 mi. north of Yates Center, 20 Apr., 10 May 1963, 6 Apr. 1964; Butler
Co., west edge of El Dorado, 11 July, 9 Sept. 1964, 21 May 1965; Sumner Co., 4.5 mi. west of
Geuda Springs, 21 Apr. 1963; MISSOURI: Howard Co., 1 mi. north of Rocheport, 24 June
1963; TEXAS: Rockwall Co., 2.5 mi. southwest of Royse City, 30 June 1965.

C. hirticollis

This species is currently being studied by Dr. R. C. Graves. Collection
records from saline habitats of the central United States are as follows:

KANSAS: Lincoln Co., 11 mi. north, % mi. east of Lincoln, 14 June 1964; Stafford Co.,
11 mi. northeast of Hudson, 7 Apr. 1965; Barber Co., 3 mi. southeast of Hazelton, 27 Aug.
1963; 17.5 mi. west, 4 mi. north of Hardtner, 30 Aug. 1963; Comanche Co., 12 mi. south of
Protection, 29 Aug. 1963; OKLAHOMA: Hughes Co., 5 mi. north of Holdenville, 29 June
1965; Seminole Co., 12.5 mi. south of Seminole, 29 June 1965; McClain Co., south edge of
Purcell, 1 July 1965; Logan Co., 3 mi. north of Guthrie, 1 July 1965; Creek Co., just north of
Oilton, 19 Aug. 1964; Woods Co., 2.5 mi. southwest of Plainview, 3 June 1963; Woods-Harper
Co. line, 6 mi. west-northwest of Plainview, 29 Aug. 1963; Grant Co., just east of Pondcreek,
10 Sept. 1964; Major Co., 2 mi. northeast of Orienta, 12 July 1964; Kingfisher Co., 2 mi. south
of Dover, 10 Sept. 1964; Blaine Co., 7 mi. south of Okeene, 21 June, 10 Sept. 1964; TEXAS:
Cooke Co., 1 mi. northeast of Rosston, 30 June 1965.

Populations of C. Azrticollis in the central United States have traditionally
been called the subspecies C. 4. ponderosa Thomson, but this is not satisfac-
tory because the type locality is Veracruz, Mexico, and specimens from the

two areas are not the same. The study by Graves should clarify the matter.

C. punctulata

This is an extremely common species which is found from central Mexico
to southern Canada, and from Utah and Arizona to Maine and Florida. Most
of the Mexican populations belong to the subspecies C. p. catharinae Chevy-
rolat (see Cazier, 1954). In the southwestern United States is found the
bright green, blue, or purple form, C. p. chihuahuae Bates. In general, the
form east of the Rocky Mountains is the dark C. p. punctulata Olivier, al-
though populations from Colorado, northeastern Utah, western Kansas, the
Oklahoma and Texas panhandles, and eastern New Mexico seem to be inter-
grades between C. p. punctulata and C. p. chithuahuae. Collection records
from this study (county only) are as follows:

NEBRASKA: Lancaster, Saunders; KANSAS: Republic, Lincoln, Mitchell, Russell, Neosho,
Greenwood, Woodson, Wilson, Montgomery, Butler, Sedgwick, Sumner, Stafford, Kingman,
Kiowa, Barber, Comanche, Clark; MISSOURI: Howard; OKLAHOMA: Seminole, Logan,

Creek, Osage-Pawnee Co. line, Woods, Alfalfa, Grant, Major, Garfield, Blaine, Kingfisher,
Beckham, Harmon; TEXAS: Fannin, Hunt, Collin, Montague.

C. repanda

This widely distributed species occurs from Texas (and possibly Mexico,
near El Paso, Texas) to about 58° N latitude in Canada, and from Washing-

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 239

ton to Nova Scotia (it is absent from the Great Basin but occurs on the
Colorado Plateau). It exhibits little geographic variation. Populations in
Nova Scotia and vicinity have reduced elytral maculation and have been
designated C. r. novascotiae Vaurie. Individuals in east central Utah are
small, with expanded elytral maculation, and are called C. r. tanneri Knaus.
The remaining populations may be called C. r. repanda Dejean, although
other names have been applied to certain forms. Collection records from
saline habitats are as follows (county only):

NEBRASKA: Saunders; KANSAS: Republic, Lincoln, Russell, Wilson, Sumner, Kiowa,

Barber, Comanche; MISSOURI: Howard; OKLAHOMA: Hughes, Seminole, McClain, Creek,
Woods, Alfalfa, Grant, Garfield, Kingfisher; TEXAS: Cooke.

C. schauppi

This is a species of southern distribution which reaches its northern limit
in the central United States. North of Texas it is uncommon. I have collected
it only once, in Okfuskee Co., Oklahoma, just northwest of Pharoah (19 Aug.
1964), on a bare, slightly saline area caused by oil drilling. This and other
localities from the literature and museum collections are shown in Figure 98.
C. schauppi exhibits some variation, specimens from northern localities hav-
ing a longer, thinner, and more oblique middle band than do those from

Fic. 98. Known distribution of C. schauppi. Fic. 99. Known distribution of C. circumpicta;
triangle=C. c. circumpicta, dot=C. c. johnsont.

240 Tue Universiry ScrENcE BULLETIN

southern and western Texas, but so few large series of specimens exist that a
statistical study of variation would not be very meaningful. Further study
may show that subspecies are worthwhile naming. Locality records are as
follows (county or state only):

OKLAHOMA: Okfuskee, Coal, Hughes, Jefferson, Comanche; TEXAS: Wichita, Hunt,
Wilson, Atacosa, Brazos, Frio, Uvalde, Limestone, King, Dimmet, Cameron, Kleberg, Burnet,
Dallas, Childress, Kimble, Zavala, Bexar, Mason, Runnels, Duval, Denton, Nevarro, Nueces,

Howard, Carson, Harris; MEXICO: Nuevo Leon. Doubtful locality: KANSAS: Parsons, Labette
Co.; this specimen more resembles those from southern Texas than those from Oklahoma.

C. tranquebarica

This is a very widely distributed species, found from the Pacific coast to
Newfoundland and from the Gulf coast to 60° N latitude in Canada. Popu-
lations east of the Rocky Mountains exhibit little variation and are called
C. t. tranquebarica Herbst. West of the Rocky Mountains there occurs a
multitude of subspecies (at least 11), which have been insufficiently studied.
Collection records for saline habitats are the following:

KANSAS: Republic Co., 4 mi. west, 1 mi. south of Kackley, 6 Sept. 1963; Stafford Co., 11

mi. northeast of Hudson, 9 Apr. 1964; OKLAHOMA: Woods Co., 2.5 mi. southwest of Plain-
view, 3 June 1963.

MAJOR SPECIES

In this section, those species that were studied most thoroughly are dis-
cussed. For each species, I will consider: (1) present distribution (as well as
it is known), geographic variation, and the presence or absence of subspecies;
(2) why it is found where it is, including adaptations, evolution, and dispersal.

C. circumpicta

This species is of primarily southern distribution, being most common
along the western Gulf coast, along the Pecos River, and in Oklahoma and
Kansas. It also occurs in Colorado, Missouri, Nebraska, and has recently
been collected in eastern North Dakota (Fig. 99). Doubtful records exist
(Lyons, Cook Co., Illinois; Riley Co., Kansas; Patagonia, Santa Cruz Co.,
Arizona) ; and the town of Manzenda, Colorado (—Manzanola, Otero Co. ?)
could not be located. Locality records are as follows (county or state only):

NORTH DAKOTA: Grand Forks; NEBRASKA: Lancaster, Nuckolls, Saunders; MIS-
SOURI: Boone, Howard, Saline; KANSAS: Barber, Butler, Clark, Clay, Gove, Kiowa, Lincoln,
McPherson, Meade, Mitchell, Montgomery, Neosho, Reno, Republic, Sedgwick, Seward, Stafford,
Sumner, Wallace, Wilson, Woodson; COLORADO: Bent, Otero; OKLAHOMA: Alfalfa,
Beaver, Beckham, Blaine, Cleveland, Creek, Garfield, Grant, Harmon, Harper, Jackson, King-
fisher, McClain, Noble, Okfuskee, Oklahoma, Osage, Pawnee, Payne, Tillman, Tulsa, Washing-
ton, Woods; NEW MEXICO: Chaves, Eddy, Guadalupe, Quay, Roosevelt, Santa Fe, Torrance;
TEXAS: Andrews, Cameron, Childress, Colorado, Dallas, Dimmet, Galveston, Goliad, Hall,
Hardeman, Howard, Hunt, Jackson, Jefferson, Kenedy, Kleberg, Loving, Nueces, Palo Pinto,
Pecos, Randall, Reeves, Victoria, Webb, Wichita, Wilbarger; TAMAULIPAS.

BIoNoMIcs AND ZOOGEOGRAPHY OF TIGER BEETLES 241

The described forms within this species are the following:

Cicindela circumpicta LaFerté, 1841. Type locality: Texas.

Cicindela collaris LaFerté, 1841. Type locality: Texas.

Cicindela Johnson Fitch, 1856. Type locality: prairies west of Arkansas.

Cicindela circumpicta ambiens Casey, 1913. Type locality: Kansas.

Cicindela circumpicta inspiciens Casey, 1913. Type locality: Point Isabel, Texas.

Cicindela circumpicta salinae Vaurie, 1951. Type locality: Lincoln, Lancaster County, Nebraska.

In recent years, three subspecies have been recognized: C. c. circumpicta,
the dark, mostly coastal form; C. c. johnsoni, the bright inland form occur-
ring in three color phases; and C. c. salinae, the dull reddish or brown form
from the Lincoln, Nebraska, vicinity.

In the study of geographic variation, the following characters were meas-
ured: (1) width of head at widest point (the eyes); (2) width of pronotum
at widest point; (3) width of left elytron at widest point when viewed from
directly above, not at an oblique lateral angle; (4) width of the white elytral
maculation at a specified place (Fig. 100a); (5) length of left elytron from
the level of the anterior end of the scutellum to the most apical part; (6) num-

i)
as
bo

Tue University ScrENCE BULLETIN

aoe eee

lOl

Ni
\ fo” "3g Gales

lOO [ey

Fics. 100-102. Illustration of certain characters measured on C. circumpicta; Fic. 100,
a=width of elytral maculation, b=number of punctures in 0.45 mm* area at this position;
Fic. 101, a=width of labrum, b=length of labrum, c=length of labrum at one lateral edge;
Fic. 102, arbitrary units for shapes of the middle bands.

ber of punctures in a 0.45 mm? square on a specified part of the left elytron
(Fig. 100b); (7) width of labrum at widest point (Fig. 101a); (8) length of
labrum at midline, including tooth (Fig. 101b); (9) length of labrum at one
lateral edge (Fig. 101c); (10) length of hind tarsus, excluding claws; (11)
length of hind tibia; (12) shape of middle band in arbitrary units (Fig. 102) ;
(13) color of elytra, using color wheel.

Specimens from 23 localities were measured for characters 1-12; the locali-
ties and sample sizes are given in Table 13. Specimens from 21 of these
localities and ten additional localities were included in the color analysis
(Table 13). When possible, samples of 20 specimens of each sex were used,
but in a few cases only small samples were available or specimens from two
or more nearby localities were lumped into a larger sample. The values
obtained from these less desirable samples are thus less reliable than those of
samples from one locality.

The means for the characters and localities are given in Tables 14 and 15.
Analyses of variance of the individual characters showed that there are
significant differences (p < 0.01) among the means of all characters for both
sexes. Pooled within-locality variance-covariance matrices are given in Tables
16 and 17. The sums of the among-locality variance components are 100.907
for males and 127.336 for females. The first six components for males and
the first seven for females are highly significant (p < 0.01), and the seventh
is significant at the 5°% level for males; however, the first three functions
account for 77.67% of the variance (among localities relative to that within)
in males and 78.1994 in females. The first six functions account for 93.96%

BIoNoMIcs AND ZOOGEOGRAPHY OF TIGER BEETLES

243

Taste 13. Localities and samples sizes of the specimens of C. circumpicta

measured.
N

Locality bs Q

1. NORTH DAKOTA, Grand Forks Co., 1.5 mi. n.e. of Emerado _...................... 20 20
2. NEBRASKA, Lancaster Co., Lincoln (west edge) -.................. DL ee ee ee) 20
Ce MAINSAS@ Republic Con zumiiaess i s.\Of Malmo eee 9 14
4. KANSAS, Republic Co., 4 mi. w., 1 s. of Kackley —...-0.-00-..0...... ee 20 15
DRLCAIN SAG alLincolnnGomponmninsw asec ass Obi baniande ssa ae ee 20 20
GHELCANSAS = Stators Comm llimuemses) Of eidsom eee) 20 20
je MISSOURIA Howard: Cos lime ns 2)we of Boonesboro) 0-2) 20 20
8. KANSAS, Montgomery Co., 3 mi. s. of Elk City 2... oe Ea rss eet, 20 20
OR OKMALIOMAC Creek Com suimin's:e: Of Sapulpales = se ee 20 19
HOS OKEAROMAS Allfalta (Cos, 5) mine, of (Cherokee 2. 2 20 20
MM ROKEALONA™ Woods Go. 2-5) mis Saw- lof Plainview 2.02 20 20
IV AISCAUN SA Sa Glanks GorepringlewOOd! 0.2 erences 5 A eee eee 20 20
SO KUNEONA Blaines Come mis ssf Okeene ses ee ee 7 10
(RO KEATIONAY BeckhamiCo:, 6) mis swe ol Mayfield. sess 14 11
15. OKLAHOMA, Jackson Co., 3.5 mi. s. of Eldorado ................. ihe ee ee ah 20 17
GleNEAW We Mbp COM Chaves Col! Omit es) se Of Roswell) see eee ee 20 20
a NE WeMb xd COMEddyaCosouminves 2 ny of Moving p22) ee 11 8
18. TEXAS, Reeves Co., vicinity of Pecos, Lake Balmorhea, Toyah; Pecos, Co.,

Bitar S tocktomism leavin oa © Obes eae ae oar Ree ee ee core Ui Mette er esas ce ee 13 20
ESAS Webbs Complanedos: Dimmet Col. Carrizo) Sprinos) 22 ee 6 6
20. TEXAS, Cameron Co., Port Isabel, Boca Chica, Brownsville ...................-...--.-- 12 19
AP EDCNS sal eberoa Conmivieras beach semester ee ee ee 20 20
22. TEXAS, Galveston Co., Seabrook, Galveston, Dickinson* .....................---......--- 1 2
23), WEDNS. IDbillllas (Co, IDEWEISS Tshuiate (Co, WiVolbie (Cie fo ec ee 2 3

Localities included in color analysis only:

24. KANSAS, Lincoln Co., 11 mi. n. of Lincoln

25. KANSAS, Wilson Co., 1 mi. n., % w. of Fredonia
26. KANSAS, Neosho Co., 2 mi. n., % e. of Chanute
27. KANSAS, Sumner Co., just n. of Geuda Springs
28. KANSAS, Barber Co., 3 mi. s.e. of Hazelton

29. KANSAS, Kiowa Co., 1 mi. n. of Belvidere

30. OKLAHOMA, Tulsa Co., 1.5 mi. s. of Skiatook
31. OKLAHOMA, Garfield Co., % mi. n.w. of Drummond
32. OKLAHOMA, Beckham Co., 3 mi. s. of Carter

33. TEXAS, Wichita Co., 2 mi. n.e. of Burkburnett
34. TEXAS, Hardin Co., Sour Lake

* Excluded from color analysis.

244 Tue University ScrENCE BULLETIN

Tasie 14. Means of 23 localities (see Table 13) and 12 characters for males of
C. circumpicta. Values for characters 1-5, 7-11 are in mm.

Character

Locality 1 2 3 4 5 6 a 8 9 10 il 12
1 3:36 2:57 221 0:6” “8.67 93045 COP e075 0:47) 9 eS ome
2 333° 254 215 i041 ‘7.97 Sait We54 L019 044s C468 4 /lleZeo6
3 3.46 2.73) 225 044 851 933is) 1620 0) OAC orl ool
4 353) 2:81 234 7046) (8:62.55 S09) S6> ONS 0S 2 om meee OD)
5 3.26 256 2.13) (0:41 (7288) 33:8) eS OWS) 0M ate OE OO
6 353) 2276 233) 052) 854 28:8) kos) SO OM ciel 2 eet)
7 351 2:73 229 OAl 8:42 295. 162) 079) 1050489 oS eal
8 350) 2672 23 044 8520 2956) PLC ON 0276 0 47 eo ee ee
9 3:39 259) 2220) 10:50) 809 3.08) 155 02/6.) SOG 4S ao Ceeesee
10 Bra 2.69) 223) 0460) 8230 e297 SO 0S 08 EO
11 3520 2.72 2341050) 85l)) 28:8) ole 1079 0498 Salsa ome)
12 B25 a 2eote eZ 0:49) 7.90 Ss Eto 07045 eo Oe SSeS
13 3-405 2:47 220) (043) 7.91) 29:0) 14S 072) S04 oe A/S eee eee
14 3145) 2263) 2230) Oe 85525 SEG 0 0216) OA Sai Zee
15 36 22788 243) 054 1881 29S E65) 0:81 0 Sl O ena
16 347 2:67 230 10.57. 8:39) 325) 160) 0/7 049 Za oe
17 331 248" 2216" 057) 789) 346) EAS 072) 0G) 05 Ole
18 354 273 235 0:64 855) 29!6) 163) 10380) 050 S12 > ee
19 3245) 2250 2221 0428892910) 15S 0! SSO 8 Cue lye BOS
20 39/9) 2:87) 2238) 0:40 8/8 22.0) Os OS'S in 052 een 2s mmm ei
21 3:92— 2:97 2-48) 10F42)) 8:99) 22-4 E78 10895 054 StS oe: Comme S
22 S37 225 2.05! 046) 77, 3320) ES C069 0396 ete Ome)
23 303 ASS) All) Os) cI PSK) Ss OOS SE) GIS) TLS

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES

Tase 15. Means of 23 localities (see Table 13) and 12 characters for females of

C. circumpicta. Values for characters 1-5, 7-11 are in mm.

Character
Locality 1 2 3 4 5 6 8 9 10 11 12
1 BEGOmE2 SOL d en ODD S02 mace = eZ O80) 0:47 4:28 4:73 14102
2 BED 9 eA OA eeise72 Olea 1269) 30;80) “047, 4.377 4279-298
3 B22 OO O47 ool Sal) = 1270) 079 = 08 “4:49 4288 2789
4; Bee St 2255) O49 9104 286) = 174 90282) 048) 4 4872297
5 BES eS tO lec tle SOM kG) VOS/5)) (OAGH = 4228" 4:66) 93.09
6 265 3.02 B50 (Oe hile WES ie ~ (ee) (esr Gis SHI SEIN
7 BY SA OO) Set = Chl le/0) 10:82) OSI 4:26" 4275) 3-13
8 Bem 9 Ae OLAS SECC Ore le7le) OL800 049 S44 4:85) eS .'6
9 35 DS Den) Sil By wees sy OI 7 esa adele) SL
10 Ben Se ORAS) e045 2838 le 70) SOLS 0F49S Aa Ge 93335
11 BH DOP DEG ORG BO Pre Al (Ww aes SB Sy
12 Bue DAO ODO 842 2810) EG OG OA 84:58) 4578593338
13 Be/Ae2e 8S DOO 4D ee otO 08 29:85 0 10:83 OMS ARS S289
14 275 BO) PS Of7 BOR PAS ie Mss WS) ais <4) eyuil
15 ZO B05 AS O57 O32 Aye Uns WeX5i Oss Silt Sail) Se
16 310 Diy 2253 O56 87 308 as O40 OO) Hs) SO S74
ily 353 26s 23 OWS2 BE5 se ss) (ey by ais Ais) Siefeis)
18 339 299 De) Wes CS BO ins WYH5 (Ss) 4h So siete
19 35 D2 D237 O89 G65) 262 USA Oss WS) Ghee aif — ileils
20 Anil 308 253 O30 Gul 207 il) OO 053 48 Sa Isis
21 12) 2320 265 O39 O46 iS Is O85 W577 SHA S70) ies
22 35 Dl Bs OAs Cay Bos iW Wkw OA asi aire IL oh0)
23 310), 273 B34 O43 bess Bao” aS Wey Wa aes Gye; Ilolls)

Taste 16. Pooled within-locality variance-covariance matrix for males of C.
circumpicta. Variances are along the diagonal, and covariances compose the rest
of the matrix. Values have been multiplied by 104 to conserve space; thus

2550010253:

Character

ConNT ON UB WwW NH

= _—
NK oOo OO

bo
VI
Ww

2

240
394

3
184

304

284

4

41
62
64
143

5

647
1074
908
190
3503

Character
7 8 9 10 11 12
—1049 134 82 50 351 370 —38
846 216 ily 70 593 623 —86
983 176 97 60 504 528 —73
267 36 24 13 104 ig) wala
2670 634 346 213 1864 1953 —259
141199 509 221 156 2148 2068 670
137 70 43 350 367 —33
49 2 191 198 2
19 120 125 —5
1256 1178 —69
1301 —49
4403

246 Tue University SclENCE BULLETIN

Taste 17. Pooled within-locality variance-covariance matrix for females of C.
circumpicta. Variances are along the diagonal, and covariances compose the rest
of the matrix. Values have been multiplied by 10°-* to conserve space; thus

“304”—0.0304.
Character

1 2 3 4 5 6 7 8 9 10 ee

1 304 251 204 45 635 —2078 130° W730 Waa Wes eeesiomeenS

2 258 187 40 567 —1950 J17 66 39 44 s7ay memes

3 184 41 490 —1585 96 53 32 203 23QNmemas

4 63 <8 266 21 0 5 39 S600

oh bs 1767 —5664 308 179 104 663) SSIs ueeren
= 6 100873 —1039 —620 —343 —2390 2543) oz
E 7 12 37 22 133 Sey
S 8 34 14 184 eee
9 12 647 5Gn ets

10 461 410 268
11 Aye
12 3047

in males and 92.10°% in females. The distributions of the means in the first
three discriminant functions are shown in Figures 103 and 104. The localities
fall into two loose clusters in both sexes, with localities 19-23 in one and all
other localities in the other. The two clusters are not as widely separated in
males as they are in females; however, there is still a distinct break between
them because localities 19, 22, and 23 have high values in Ks, while the nearest
localities in the other cluster have lower values. The results of the simultane-
ous testing procedure show that all combinations of pairs of means are
significantly different at the 5°/ level except 6 vs. 11, 10 vs. 11, 19 vs. 22, and
22 vs. 23 for both sexes; 4 vs. 6, 11 vs. 14, 11 vs. 15, and 16 vs. 18 for males;
and 2 vs. 3, 8 vs. 10; 11 vs. 13, 3 vs. 14, 6rvs. 14, 8 vs. 14, 10°vs. 14; 1vselas
and 13 vs. 14 for females. Thus, several pairs within the clusters are not
significantly different, but the nearest intercluster pairs are significantly dif-
ferent. Localities 16, 17, and 18 could be considered a subcluster; however,
including samples from geographic areas between them and their nearest
neighbors would probably destroy this appearance.

Sets of vectors (Figs. 105 and 106), plotted for the first two discriminant
functions, show the geographic trends of the 12 characters when compared
with Figures 103 and 104. Thus, specimens with high values in Ki (the
right-hand cluster) have high values for characters 1, 8, 10, and 11 and low
values for characters 2, 3, 5, and 7; while specimens with low values in Ki
show the opposite trend (when all characters are considered simultaneously).
Similarly, specimens high in K, are also high in characters 4 and 9, and low
in 7 and 8,

P N
I7 6 6 2 -8
6) \ “b ©
i _ O \ 9
Wine renin Vay ce
q VF L : a

ai
vai
eae

103

K,
13

lO4

O 6

ane

-13

-I4

B( -I5

ial (eee Ne
ot eS Ky
1s ae a
a aes r

2

Fics. 103-104. Distributions of the means of 23 localities for males (Fig. 103) and females
(Fig. 104) of C. circumpicta in the first three discriminant functions (Ki, Ke, Ks), drawn as
three-dimensional models, with numbered balls representing the means and vertical supports
arising from the Ki, Ke surface. Different sized balls indicate different distances from the viewer.

248 Tue University ScIENCE BULLETIN

lO6

Fics. 105-106. Vectors for the 12 characters of males (Fig. 105) and females (Fig. 106) of
C. circumpicta for the first two discriminant functions. Each vector shows the change in the
discriminant function that the corresponding character would produce if it varied independently.
Units are same as in Figs. 103-104.

The results of the color analysis are shown in Figure 107. Populations
from near the Gulf Coast and lower Rio Grande valley are nearly uniformly
dark purplish to dark olive green, although in the vicinity of Corpus Christi,
Texas, a small percentage of bright green, blue-green, and blue individuals
occur. Most other populations contain individuals in bright reddish, green,
and blue morphs in varying proportions. In general, reddish individuals are
most common, followed by green and blue. In several scattered localities,
reddish individuals compose nearly the entire population (southern New
Mexico, southwestern and west-central Oklahoma, north-central Kansas,
Nebraska, and North Dakota). In Missouri only blue to green individuals
occur, while populations in southeastern Kansas and northeastern Oklahoma
contain a higher percentage of these morphs than western populations. The
southern-most sample in eastern Oklahoma contains an unusually high num-
ber of dark individuals. The North Dakota population is unique in contain-
ing about 15°% black (or at least very dark purple) individuals. Thus, we

BroNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 249

HEN
PAW)

400 MILES
——

1
600 KILOMETERS

Fic. 107. Results of color analysis of C. circumpicta using the color wheel. Different degrees
of shading indicate different percentages of the sample (upper left). The positions of the three
major colors are shown at upper right. Numbers show the sample size for each locality.

250 Tue University ScrENCE BULLETIN

may divide the samples into two major groups: the rather uniform dark
coastal populations and the quite heterogeneous, usually bright inland popu-
lations, which exhibit much intra- and interpopulation variation.

Considering the pattern of geographic variation shown by this species, I
think it is best to recognize only two subspecies (Fig. 99): 1) C. ¢. circum-
picta, characterized by narrow maculation, particularly a thin middle band,
a relatively long labrum and wide head, and usually dark color, ranging from
dark purplish to dark olive green, with occasional bright blue or green in-
dividuals; 2) C. c. johnsont, having wider maculation, particularly a wider,
often broadly rounded middle band; relatively short labrum and narrow
head, and usually bright color, ranging from reddish to green to blue, with
occasional brown or black individuals.

The recently described C. c. salinae was separated because of its small
size, dull (not glossy) elytra, and brown or dark red color, with no blue or
green individuals. The present analysis shows that populations of quite small
individuals occur in several scattered localities. Although it was not meas-
ured, relatively dull elytra seem to be characteristic of populations from
northern Kansas, Missouri, Nebraska, and North Dakota. The tendency for
local populations to contain only one color morph seems not to be uncommon
in this species. Thus, C. c. salinae does not seem sufficiently distinct to war-
rent its continued recognition.

C. circumpicta probably arose within its present range. Its closest rela-
tives, C. praetextata and C. californica, occur in the southwestern United
States and western Mexico. It is proposed that a common ancestor to the
three became widespread from Texas to California during the late Tertiary.
The rising mountains of the Cascadian Revolution and the drier, cooler
climate of this time separated the ancestral species into at least three groups,
which evolved into the modern species. Probably while this was occurring,
proto-circumpicta, had already begun to become differentiated into coastal
and inland forms, possibly in the form of a cline along the Rio Grande valley.
Drier climates of the Pleistocene then extinguished intermediate populations
in the upper Rio Grande valley. The fluctuating sea level of the Pleistocene
probably “encouraged” C. c. circumpicta to disperse up many of the Texas
rivers; today relict populations exist in the Dallas vicinity and the lower Rio
Grande valley. C. c. johnsoni had not dispersed to Nebraska or central Mis-
souri by mid-Pleistocene because the Nebraskan and Kansan glaciers covered
these areas (or if it had reached these areas, it was driven back or extermi-
nated). As it dispersed northeastward from its place of origin, C. c. johnsoni
became more and more unlike the original stock, producing the entirely blue
or green populations of Missouri and the entirely reddish populations near
Lincoln, Nebraska. Central Missouri was apparently reached via southeast-
ern Kansas.

BIioNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES Wp) |

The occurrence of reddish, green, and blue color morphs seems to have
been a characteristic of the ancestral species of C. circumpicta, C. praetextata,
and C. californica, because they occur in all three modern species. Thus, it is
proposed that the coastal C. c. circumpicta has nearly lost this primitive char-
acter, while most populations of C. c. johnsoni have retained it. The bright
green and blue morphs would seem to be at a great disadvantage in being
camouflaged, and evidence presented earlier supports this (see section on
ecological relationships of the adult). The pattern of color variation (Fig.
107) shows that, except for populations in northern Kansas, those popula-
tions with the highest percentage of bright reddish individuals occur in those
regions that possess very red soil derived from the Permian redbeds (Fig.
108). This suggests that the bright green and blue morphs are gradually
being lost in these areas. As populations from these areas dispersed north-
ward into Kansas, Nebraska, and North Dakota, they encounted darker
soils. The high frequency of darker red and purplish individuals from these
localities indicates that evolution has occurred to produce better camouflaged
individuals. The high percentage of rather bright blue and green individuals
in southeastern Kansas and Missouri is difficult to explain; soils in these areas
are generally dark. Perhaps dark green and dark purplish of C. c. cireum-
picta match very well the dark soils prevalent in the areas where it occurs.

Fic. 108. Distribution of soils derived from the Permian redbeds. From map: Origin and
distribution of United States soils, prepared by the Technical Development Service, Civil Aero-
nautics Administration and the Engineering Experimental Station, Purdue University (1946).

iS)
WI
SS)

Tue University SciENCE BULLETIN

The wider markings of populations in southwestern Kansas, western
Oklahoma, western Texas, and New Mexico seems to be an indirect result of
the drier climate of these areas, which causes a white crust of salt to be
present on saline habitats more of the time than in moister areas (see section
on ecological relationships of the adult). The occurrence of widely maculate
individuals in North Dakota seems to be a convergence that has taken place
recently.

C. cuprascens

This species is most common between the Rocky and Appalachian Moun-
tains (Fig. 109). The distribution by counties or states is the following:

GEORGIA; ALABAMA: ‘Tuscaloosa; MISSISSIPPI: Warren; TENNESSEE: Shelby;
KENTUCKY: Campbell, Fulton, Henderson, Kenton; OHIO; INDIANA: Posey, Putnam;

Fic. 109. Known distribution of C. cuprascens; triangle=state record.

BronoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 253

ILLINOIS: Coles, Hardin, Jackson, Johnson, Masac, Morgan, Pope, Clair; WISCONSIN: Dane;
MINNESOTA; IOWA: Johnson, Monona, Pottawatamie, Story, Woodbury; MISSOURI: Boone,
Cooper, Franklin, Gasconade, Holt, Jefferson, Livingston, St. Charies, St. Louis; ARKANSAS:
Arkansas, Clay, Craighead, Crawford, Desha, Jefferson, Lincoln, Miller, Sebastian; LOUISIANA;
TEXAS: Childress, Cooke, Hall, Hardeman, Hartley, Hemphill, Hutchinson, Montague, Potter,
Randall, Wichita, Wilbarger; OKLAHOMA: Alfalfa, Beaver, Beckham, Canadian, Cherokee,
Cimarron, Cleveland, Comanche, Cotton, Custer, Dewey, Greer, Harper, Hughes, Jackson,
Jefferson, Johnson, Kiowa, Logan, Love, Major, Marshall, McCurtain, Murray, Okfuskee, Payne,
Pontotoc, Roger Mills, Sequoyah, Texas, Tillman, Tulsa, Woods, Woodward; KANSAS: Atchi-
son, Barton, Clark, Clay, Douglas, Ellsworth, Finney, Ford, Gray, Hamilton, Johnson, Kearney,
Kiowa, Leavenworth, Logan, McPherson, Meade, Pottawatomie, Reno, Rice, Riley, Sedgwick,
Shawnee, Sumner, Wallace, Wyandotte; NEBRASKA: Buftalo, Cass, Dakota, Dixon, Douglas,
Dundy, Knox, Otoe, Platte, Richardson, Scotts Bluff, Thomas; SOUTH DAKOTA: Brule, Clay,
Fall River, Shannon, Spink, Union; NORTH DAKOTA: Billings, Burleigh, Emmons, Mc Ken-
zie, Mc Lean, Mercer, Morton; MANITOBA; MONTANA: Custer, Dawson; WYOMING:
Niobrara; COLORADO: Bent, Denver, Otero, Prowers, Pueblo, Yuma; NEW MEXICO:
Chaves, Colfax, Quay. Towns that could not be located: Eastport and Eastbrook, Iowa; Wicks,
Missouri. Doubtful records: Moscow, Latah Co., Idaho; Logan Canyon, Cache Co.?, Utah.

A closely related form, which has been considered a subspecies of C.
cuprascens (Horn, 1930; Leng, 1902), a subspecies of C. macra (Vaurie,
1951), or a separate species (G. Horn, 1876; Schaupp, 1883-1884), is puritana
G. Horn. It is found along the Connecticut River in New Hampshire (Sul-
livan Co.), Massachusetts (Hampden and Hampshire Cos.), and Connecticut
(Hartford Co.), and around Chesapeake Bay in Maryland (Calvert and St.
Marys Cos.). State records exist for New York and Virginia. Some speci-
mens appear to be labelled “Windsor, Can.,” which is in southern Ontario.
Wallis (1961) does not mention this record from Canada, and the labels
probably should read “Windsor, Connecticut,” where this form has been col-
lected. After examining the adult morphology, including the male genitalia,
I think puritana should be considered a separate species, one that is more
closely related to C. cuprascens than to C. macra. The three species are com-
pared in Table 18. In certain characters, C. puritana is somewhat inter-
mediate between C. cuprascens and C. macra. Because of its distinct com-

Taste 18. Comparison of seven characters in C. cuprascens, C. puritana, and

C. macra.
Character C. cuprascens C. puritana C. macra

Shape of posterio-lateral

emargination of @ elytra —....... Acute Acute Rectangular
Shape of 2 elytral apices) Rounded Acute or Acute

occas. truncate

Depth of elytral punctation _......... Deep Deep to shallow Shallow
Bil ytiell es Uisha Gemeees eee eens Shiny Shiny, occas. dull Dull
Typical shape of Globose or Globose or Recurved or

apextor middle bande sees eee not enlarged not enlarged triangular
Sha penoina CC caAglisy = seen More slender Thicker Thicker

(Fig. 93) (Fig. 110) (Fig. 112)

Shape of tooth of inner sac ............ Long and acute Long and acute Shorter and blunt

(Fig. 95) (Fig. 111) (Fig. 113)

254 Tue Universtry SciENCE BULLETIN

MAD

¥ 4
i rk
iy

13

Fic. 110. Aedeagus of C. puritana (Windsor, Connecticut), dorsal aspect. Fic. 111. Inner
sac of aedeagus of C. puritana, ventral aspect. Fic. 112. Aedeagus of C. macra (Ness County,
Kansas), dorsal aspect. Fic. 113. Inner sac of aedeagus of C. macra, ventral aspect.

bination of characters, and because it is geographically isolated from C.
cuprascens and C. macra, I have separated it taxonomically. As a test, it was
included in the statistical analysis along with C. cuprascens.
In recent years, no subspecies have been recognized. The following forms
have been described within the species C. cuprascens:
Cicindela cuprascens Leconte, 1852. Type locality: Arkansas River (types bear green paper
circles, signifying ‘Kansas, Nebraska, and westward’’)
Cicindela cuprascens amnicola Casey, 1913. Type locality: Kentucky, Illinois, and Missouri.
Cicindela mundula Casey, 1913. Type locality: Vicksburg, Mississippi.
In the study of geographic variation, the following characters were meas-
ured : (1) length of left elytron; (2) width of left elytron; (3) width of

BronomMics AND ZOOGEOGRAPHY OF TIGER BEETLES 255

COb 5G

114 HIS

Fics. 114-115. Illustrations of certain characters measured on elytra of C. cuprascens, C.
puritana, and C. macra; Fic. 114, width of middle band; Fic. 115, arbitrary units for shape of
apex of middle band.

labrum; (4) length of labrum, including tooth; (5) width of middle band at
a specified place (Fig. 114); (6) width of head; (7) shape of apex of middle
band in arbitrary units (Fig. 115); (8) color of elytra, using color wheel.
Specimens from nine localities were measured; the localities and sample
sizes are given in Table 19. Sample sizes were adequate from all the localities;
however, specimens from several localities in Alabama and Missouri were

lumped.

Tasie 19. Localities and sample sizes of the specimens of Cu puritana and C.
cuprascens measured.

N
Locality 3 2
C. puritana:
i, (CCOININIEXCHOMCIUME. ISbraaioysl (Clo, Vivtaeoyr 12 12
C cuprascens:
2x INEINGYNMUN, Annee lKorare (Clo, Geyceiilll hore tiers 2 2, 7)
Se MISSOURIAN St. Louis! Co. St. louis: St, Charles) Cos, St. Charles 12 12
ARO ACH Rottawattamies Corn Gounelly Blt tts meee eee ee LZ 12
SED LCATN SANS a) O11] asp CO sg ete ice eran eee ees es ee ee 12 12
GUIKCAINSAS Clarke Come Guinn SyrOliS tana. 2: eeepc ee ee eee 12 12
7 OKEAHOMA Alfalfa Gol, 3 mi. m:, Die, of Cherokee 12 12
SSORIEARIOMA Cleveland (Gos cee sushi wie Ee oe eee AD 12
OM COLORADO mBent Cornias Amina: snes a aoe ewe cc ececspe ee 12 12

The means for the characters and localities are given in Tables 20 and 21.
Analyses of variance of the individual characters showed that there are sig-
nificant differences (p < 0.01) among the means of all characters for both
sexes except character 2 for females, which is significant at the 5° level.
Pooled within-locality variance-covariance matrices are given in Tables 22
and 23. The sums of the among-locality variance components are 82.768 for

256 Tue Universiry ScIENcE BULLETIN

Tas Le 20. Means of nine localities (see Table 19) and seven characters for males
of C. puritana and C. cuprascens. Values for characters 1-6 are in mm.

Character
Locality 1 2 3 4 5 6 7
1 7.32 2.01 1.39 0.64 0.31 2.84 1.54
2 6.73 1.89 1.34 0.59 0.27 2.62 2.67
3 7.16 1.90 VA 0.63 0.30 2.90 2.06
4 7.46 2.00 1.47 0.63 0.36 2.91 2.43
5 7.38 1.92 1.44 0.63 0.34 2.91 2.22
6 7.79 2.04 156 0.70 0.46 3.15 2.50
7 7.46 2.02 RDS: 0.68 0.43 3.08 DP} |
8 7.46 2.02 Sil 0.66 0.43 3.05 254
9 Wes 2.03 1.54 0.66 0.45 3.00 2.96

Taste 21. Means of nine localities (see Table 19) and seven characters for
females of C. puritana and C. cuprascens. Values for characters 1-6 are in mm.

Character
Locality 1 D 3 4 5 6 7
] 7.67 Delite} 1.43 0.64 0.34 3.03 1.60
2 7.26 2.06 1.40 0.57 0.28 2.84 2.10
3 7.96 DMS 1S 0.66 0.32 3.19 DSS
4 7.99 2.18 1.59 0.65 0.38 BLS 2.20
5 7.97 2.14 1.59 0.67 0.37 S22 2.43
6 8.09 BAG 1.60 0.68 0.46 33141 232
7 TOM DAVY) 1.61 0.69 0.43 3.29 232
8 7.96 2.18 1.61 0.66 0.41 3224 2.26
9 8.04 2.16 1.60 0.67 0.43 303} 2.63

Tasie 22. Pooled within-locality variance-covariance matrix for males of C.

puritana and C. cuprascens. Variances are along the diagonal, and covariances

compose the rest of the matrix. Values have been multiplied by 10-4; thus
“940” 0.0940.

Character

1 2 3 4 5 6 7

l 940 222 182 85 16 267 383

2 79 51 25 5 83 74

z 3 50 21 = 64 70
aoe 20 =5 29 21
See 67 =) =30
6 128 117

7 4332

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 257

Taste 23. Pooled within-locality variance-covariance matrix for females of C.
puritana and C. cuprascens. Variances are along the diagonal, and covariances
compose the rest of the matrix. Values have been multiplied by 10-4; thus

io lo 0 5h
Character =
1 2 3 5 6 7
1 71 137 108 71 39 254 —161
2 48 27 16 8 53 2298
g 3 37 14 8 45 —54
a 4 16 2 28 —5
eee 56 7 9
6 129 == 128
7

males and 55.300 for females. The first three components for males and the
first two for females are highly significant (p < 0.01), and the fourth com-
ponent for males is significant at the 5% level. The third component for
females is on the borderline of being significant at the 5°% level. The first
three functions account for 83.47°/% of the variance (among localities relative
to that within) in males and 93.13% in females. The distributions of the
means in the first three discriminant functions are shown in Figures 116
and 117. Localities 1 and 2 are rather distantly separated from the others,
which form a loose cluster in males and a tighter one in females.

The results of the simultaneous testing procedure show that all combina-
tions of pairs of means are significantly different at the 5% level except
3 vs. 5 for both sexes, and 3 vs. 4, 4 vs. 5, 4 vs. 8, 4 vs. 9, 5 vs. 8, 5 vs. 9, 6 vs. 7,
and 8 vs. 9 for females. Thus, the samples of C. puritana and C. cuprascens
from Alabama are quite different from each other and from the other
samples, which are more similar to each other.

Sets of vectors (Figs. 118 and 119), plotted for the first two functions,
show the geographic trends of the seven characters when compared with
Figures 116 and 117. Specimens from localities 1 and 2 are characterized by
having relatively wider elytra and narrower heads. Specimens from western
Kansas, Oklahoma, and Colorado (localities 6-9) have wider labra and
middle bands and narrower elytra.

The results of the color analysis are shown in Figure 120. Populations
northeast of eastern Kansas (including C. puritana) consist mostly of dark
red-green, green, or blue-green individuals, with occasional blue individuals.
Southwest of eastern Kansas, many individuals are reddish, often bright red.

On the basis of the statistical analysis, the specimens from Alabama seem
distinct enough to separate as a subspecies; however, this is probably not wise
at present because no samples were measured from areas between there and

258 Tue University SctENcE BULLETIN

Fics. 116-117. Distributions of the means of the nine localities for males (Fig. 116) and
females (Fig. 117) of C. puritana and C. cuprascens in the first three discriminant functions
(Ki, Ke, Ks), drawn as three-dimensional models, with numbered balls representing the means
and vertical supports arising from the Ki, Ke surface. Different sized balls indicate different dis-
tances from the viewer. The models are viewed from opposite directions, relative to the clusters.

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 259

18 : II9

Fics. 118-119. Vectors for the seven characters of males (Fig. 118) and females (Fig. 119)
of C. puritana and C. cuprascens for the first two discriminant functions. Each vector shows the
change in the discriminant function that the corresponding character would produce if it varied
independently. Units are same as in Figs. 116-117.

ey oa
“i

\

Fic. 120. Results of color analysis of C. puritana and C. cuprascens using the color wheel.
Different degrees of shading indicate different percentages of the sample. The position of the
three major colors are shown at right center. Numbers show the sample size for each locality.

260 Tue Universiry ScriENcCE BULLETIN

Oklahoma or Missouri. The few specimens I have examined from these
intervening areas, however, appear more like those from Missouri or Okla-
homa than from Alabama. I have not seen specimens from Georgia. The
color analysis suggests the populations in southern Kansas, Colorado, and
Oklahoma, with many red individuals, might be worth naming, but the
statistical analysis shows that the females are not distinct enough, although
males are quite distinct. Thus, I recognize no subspecies within C. cupra-
scens. As the statistical analysis showed, C. puritana is quite different from
C. cuprascens, even though only one of the characters in Table 18 was
measured.

The closest relatives to C. cuprascens are C. puritana and C. macra. The
former is restricted to the northern Atlantic Coastal Plain, while the latter is
sympatric with C. cuprascens throughout most of their ranges. However,
C. macra does not occur east of the Mississippi River in the South, as C.
cuprascens does. This could indicate that C. cuprascens and C. macra evolved
on opposite sides of the Mississippi valley, but not necessarily so. No other
explanation is readily apparent. It does appear, at least, that C. cuprascens
was once more widespread along the southeastern coastal plain than it now 1s,
since the Alabama populations are so far separated from other known popula-
tions. As its coastal plain populations were exterminated by climatic changes
during the Pleistocene, the more northeastern ones apparently became iso-
lated and evolved into C. puritana. The spread of C. cuprascens northwest-
ward into the range of C. macra has probably occurred rather recently (late
Pleistocene), since populations in this area are rather similar. Its spread into
the Pecos River system probably occurred via the Canadian River through
northwestern Texas (the headwaters of the two river systems are very close
in New Mexico).

As is the case in other species, the predominance of red color in popula-
tions from southern Kansas, Oklahoma, and Colorado is almost certainly
due to their living on red soils (or having recently dispersed from areas of
red soil). Red individuals also occur in New Mexico, Texas, Arkansas,
Wyoming, and Montana.

C. fulgida

This is primarily a northern species, which is most common in the Great
Plains north of Texas (Fig. 121). Its distribution by counties or states is
as follows:

MANITOBA; SASKATCHEWAN; ALBERTA; MONTANA: Gallatin, Prairie, Roosevelt,
Sheridan; NORTH DAKOTA: Benson, Bottineau, Burke, Burleigh, Dickey, Divide, Dunn,
Grand Forks, McLean, McHenry, Mercer, Montrail, Oliver, Pierce, Roulette, Slope, Stutsman;
MINNESOTA; SOUTH DAKOTA: Beadle, Brookings, Edmund, Fall River, Kingsbury;
WYOMING: Albany, Carbon, Goshen, Weston; COLORADO: Arapahoe, Bent, Conejos,
Crowley, El Paso, Fremont, Huerfano, Larimer, Logan, Otero, Prowers, Sedgwick, Summit,

261

BIioNoMIcs AND ZOOGEOGRAPHY OF TIGER BEETLES

.
.
*°
. 3
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

ij
i

:
r :
° :
*Peeee, .
fees :
e fees
TY Tey 3
fees
° oopeee
gs ory
-
:
:

Fic. 121. Known distribution of C. fulgida. Open circle=C. f. westbourne1, shaded circle=
C. f. fulgida, half-shaded circle=intergrade population, triangle=state record, circles with “X”

are localities included in the statistical analysis.

262 Tue University ScrENcE BULLETIN

Weld, Yuma; NEBRASKA: Dawson, Douglas, Dundy, Lancaster, Morrill, Nuckolls, Saunders;
KANSAS: Barber, Cheyenne, Clark, Clay, Cloud, Ford, Hamilton, Kearney, Kiowa, Lincoln,
Reno, Republic, Sedgwick, Stafford, Wallace; OKLAHOMA; Alfalfa, Beckham, Blaine, Har-
mon, Jackson, Woods; TEXAS: Bailey, Donley, Hemphil, Hutchinson, Knox; NEW MEX-
ICO; Bernalillo, Colfax, Guadalupe, Sandoval, San Juan, Torrance, Valencia; ARIZONA:
Navajo. Doubtful localities: “Fla.” and Woodward (La Salle Co.?), Texas.

The following forms have been described within this species:

Cicindela fulgida Say, 1823. Type locality: Missouri Territory, near the mountains on the Platte
and Arkansas Rivers.

Cicindela fulgida elegans Calder, 1922. Type locality: Westbourne, Manitoba, Canada. Pre-
occupied.

Cicindela fulgida subnitens Calder, 1922. Type locality: Lincoln, Nebraska.

Cicindela fulgida westbourne: Calder, 1922. New name for C. f. elegans.

Cicindela fulgida pseudowillistont W. Horn, 1938. Type locality: Lake Como, southern Wyo-
ming.

In recent years, two subspecies have been recognized: C. f. westbournet,
the small, dark colored northern form, and C. f. falgida, the bright, usually
red southern form.

In studying geographic variation, the following characters were measured:
(1) length of left elytron; (2) width of left elytron; (3) width of labrum;
(4) length of labrum, including tooth; (5) shape of base of middle band in
arbitrary units (Fig. 122); (6) smallest width of transverse portion of middle
band (Fig. 123). Specimens from 14 localities were measured; the localities
and sample sizes are given in Table 24. Sample sizes were adequate except
for two localities each in Canada and New Mexico. Specimens from two
localities in Saskatchewan and two in Colorado were lumped into one
sample each.

l23

Fics. 122-123. Illustrations of certain characters measured on elytra of C. fulgida; Fic. 122,
arbitrary units for shapes of base of middle band; Fic. 123, smallest width of transverse
portion of middle band.

BIoNoMIcs AND ZOOGEOGRAPHY OF TIGER BEETLES 263

Tasie 24. Localities and sample sizes of the specimens of C. fulgida measured.

Locality } Q

1. MANITOBA, Westbourne Me Deced 0 2 08th, Sete? oe 12 12

2. SASKATCHEWAN, Route 14, s.e. a Blucher; peeaash Clayet and Elston - 6 2

Dee ABB Ra Ags @Onefoun fas a1 at hi re ete eee - a ES 2 5 6
4. NORTH DAKOTA, Bottineau Co., near Bottineau —.......... 12 12

5. NORTH DAKOTA, Oliver Co., Sect. 35, Twp. 144, Rg. 83. 12 12

6. COLORADO, Fremont Co., 3 mi. s. of Penrose, near Portland 9 12

(ENE BRAS Am eancasters Gos wleimco lin (wv.este el oe) essere nee eee ee 12 11

8. KANSAS, Republic Co., % mi. e., 1 s. of Talmo ......... ne ALAA 10 12

9. KANSAS, Republic Co., 4 mi. w., 1 s. of Kackley —.......... Se Wate en Oe 12 12

10. KANSAS, Lincoln Co., 3 mi. w., 2 s. of Barnard _.......... Se eaAD 5. ae ee eee I 12
II KCAINSA'S: Stafford: Cor, il mis nes of lidsom 2222222 teense peers ee eons AND 12
IPE ROAINI SAS an Clarkes @ony Emel ewOOd ene ee ee es os eee 12 1
13. OKLAHOMA, Woods Co., 2.5 mi. s.w. of Plainview —_..... saa. - eer eevee 1 8
14. NEW MEXICO, Roosevelt Co., 4 mi. e., 1 s. of Arch —............. Bat | ea eae a 8G 3
IDS NEWaMExdCOl Torrance Con) 4eminsies oi Wallarc) eee peas 2 8

The means for the characters and localities are given in Tables 25 and 26.
Analyses of variance of the individual characters showed that there are sig-
nificant differences (p < 0.01) among the means of all characters for both
sexes. Pooled within-locality variance-covariance matrices are given in Tables
27 and 28. The sums of the among-locality variance components are 83.504
for males and 69.975 for females. The first two components for males and
the first three for females are highly significant (p < 0.01), and the third
component for males is significant at the 5°% level. The first three functions
account for 96.50°/ of the variance (among localities relative to that within)
in males and 94.63°/ in females. The distributions of the means in the first
three discriminant functions are shown in Figures 124 and 125. Localities in
the central United States and eastern New Mexico form a tight cluster, while
the other localities are rather widely separated from one another and from
the cluster.

The results of the simultaneous testing procedure show that all combina-
tions of pairs of means are significantly different at the 5°% level except
Faysol0s9 yvsotle 9 vs. 135 Movs 12,11 ys: 13, 12 vs. 13,8 vs. 14, and It vs: 14
for both sexes; 8 vs. 10, 9 vs. 10, 9 vs. 12, 10 vs. 11, 10 vs. 12, 10 vs. 13, 9 vs. 14,
and 10 vs. 14 for males; and 7 vs. 13, 8 vs. 9, 12 vs. 14, and 13 vs. 14 for
females. Thus, the members of the cluster are generally not significantly
different from one another, while the isolated samples are different from one
another and the cluster.

Sets of vectors (Figs. 126 and 127), plotted for the first two functions,
show geographic trends in the six characters when compared with Figures

264 Tue University ScrENcE BULLETIN

Tarre 25. Means of 15 localities (see Table 24) and six characters for males of
C. fulgida. Values for characters 1-4 and 6 are in mm.

Character
Locality 1 2 4 5 6
1 6.56 2.01 1.62 0.78 4.46 0.90
2 6.64 2.07 1.62 0.74 4.05 1.05
3 7.38 2.19 1.81 0.87 4.88 1.06
4 6.91 2.09 ef? 0.83 3.87 0.85
5 6.66 2.02 1.63 0.76 3.34 0.78
6 7.30 2.20 1.81 0.90 4.11 0.92
I 6.83 2.06 1.63 0.77 2.60 0.70
8 7.05 Pall 1.69 0.81 2.62 0.83
9 7.04 2.07 1.68 0.80 DB 0.71
10 6.92 2.07 1.67 0.78 2.39 0.73
11 6.98 2.08 1.70 0.80 Doi 0.72
12 6.92 2.07 1.68 0.77 2a. 0.63
13 7.01 2.11 1.69 0.80 2.10 0.67
14 7.26 2.16 aH 0.83 2AD 0.83
15 6.73 2.10 1.65 0.80 5.60 1.34

Taste 26. Means of 15 localities (see Table 24) and six characters for females
of C. fulgida. Values for characters 1-4 and 6 are in mm.

Character
Locality | 2 3 4 5 6
1 6.94 2D || edz 0.86 3.84 0.81
2 7.04 22> 1.73 0.84 3.40 0.88
3 7.40 2.39 1.87 0.94 4.28 1.02
4 e222 DB 1.84 0.93 3.48 0.82
5 6.95 Dal) aS 0.85 2297, 0.74
6 7.69 2.42 1.95 1.00 3.26 0.93
7 7.04 2.19 ez 0.83 2.27 0.61
8 7.20 2.18 1.76 0.87 2.20 0.76
9 7.26 2.21 1.76 0.88 2s 0.71
10 6.98 PMs} 1.70 0.81 2.18 0.67
1] 7.26 2.24 1.79 0.86 D7) 0.70
12 7.26 225 1.78 0.84 2.02 0.66
3 7.25 2.26 1.76 0.87 2.08 0.64
14 7.39 2.29 1.84 0.89 2.10 0.81
15 7.00 2.18 1.73 0.86 5.03 1.30

124 and 125. Specimens with high values in Ki have longer labra, wider
markings, and relatively shorter elytra. Specimens with high values in
Ke have wide labra, narrow elytra (in males; wide in females), and narrower

middle bands.

The results of this analysis show that, except for locality 15, there are no

BronomMics AND ZOOGEOGRAPHY OF TIGER BEETLES 265

Taste 27. Pooled within-locality variance-covariance matrix for males of C.
fulgida. Variances are along the diagonal, and covariances compose the rest of
the matrix. Values have been multiplied by 10-4; thus “690”—0.0690.

Character
1 2 3 4 5 6
1 690 140 137 90 274 129
' 2 49 36 23 67 29
we 23 42 24 64 30
se 4 23 33 18
O85 1549 228
6 119

Tasre 28. Pooled within-locality variance-covariance matrix for females of C.
fulgida. Variances are along the diagonal, and covariances compose the rest of
the matrix. Values have been multiplied by 10-4; thus “691”—0.0691.

Character

1 2 3 4 5 6

1 691 160 141 103 114 77

5 2 62 38 27 16 19
3s) 42 27 23 15
ert 26 5 1
ee) 2326 231
6 153

clear groupings that could be called subspecies. The tight cluster that is
separated from other samples is probably an artifact caused by the nonran-
dom geographical distribution of the samples; if samples had been included
from South Dakota, Wyoming, and Montana, they would probably fill in
the gap. There appears to be a gradual cline running from larger, narrowly
marked populations in the south to smaller, widely marked northern popula-
tions. Thus, on the basis of this analysis alone, the subspecies C. f. westbour-
nei, representing one end of a cline, cannot be recognized. However, a
character that was not measured sets it apart from all other populations: the
shape of the apex of the aedeagus. In specimens from Westbourne, Manitoba,
the apex of the aedeagus is rather evenly tapered and comparatively blunt;
in specimens from all other localities, it is more prolonged and slender (Figs.
128-136). Another characteristic of specimens from Westbourne is their dark
dorsal color, ranging from dark purplish red to black, often with a bluish
sheen. Only three out of 39 specimens before me, or about 7.5°%, have bright-
er colors (two are reddish, one is yellow-green). Specimens from other
Canadian localities, northeastern Montana, and northern North Dakota
exhibit a great variety of colors, ranging from dark purple to red to red-green
to green to blue-green to blue to purple-blue. Green-red specimens are also

266 Tue University SciENCE BULLETIN
erat
: r |
e NYE
Gay | ae, 22
@ 10/| | a(2)
all4 @) Th ‘
wal Ae 6 | eS 2
| ie \ \ Ks
\9 | \
Kp vy | 20
ig 4 i oo
/ /
vA—_+4 I9
/ is
s fs
-6 -5 oa -3 -2 K -| O 2 oS 4
124 |
x
4 \ »:
oO Ninn
Ora oe @ ‘ -
op + NON Nee
[oe , || \ \ A\ IN N
alll Z ORR es -8
216// 1 \ NTN \
| Viele) Nh lis
25 | — \ 1 N
Ko i \ -9
24 - (15) iN
23, \ es -lO
/
a 3 4 5) 6 i 8 g 10
125 .
Fics. 124-125. Distributions of the means of the 15 localities for males (Fig. 124) and

females (Fig. 125) of C. fulgida in the first three discriminant functions (Ki, Ke, Ks), drawn as
three-dimensional models, with numbered balls representing the means and vertical supports
arising from the Ky, Ke surface. Different sized balls indicate different distances from the viewer.

known from southwestern Montana, southern Wyoming, and central Colo-
rado. The population at Grants, New Mexico, is said to contain many blue
individuals (Rumpp, in itt.). The population at Lincoln, Nebraska, contains
about 40°% dark purple or black individuals. Specimens examined from other

BronoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 267

Ze 2® I27

Fics. 126-127. Vectors for the six characters of males (Fig. 126) and females (Fig. 127) of
C. fulgida for the first two discriminant functions. Each vector shows the change in the dis-
criminant function that the ‘corresponding character would produce if it varied independently.
Units are same as in Figs. 124-125.

localities are bright red to red-purple, with occasional red individuals having
a green sheen. It is not known whether northern or western populations
exhibit a seasonal change of color as was noted for those from the central
United States (see section on post-emergence changes of the adult).

Because of its genitalic difference, nearly uniform dark color, and its
isolated geographic location, C. f. westbournei will be retained as a poorly
differentiated subspecies, at least for the present. The populations of small
individuals with quite variable color in Saskatchewan, northeastern Montana,
and northern North Dakota could be considered as intermediates (Fig. 121).

Specimens with the markings wide and connected (particularly the apex
of the humeral lunule and transverse portion of the middle band) are found
in scattered localities in the northern and western parts of the range of the
species: all Canadian localities, eastern Montana, northern North Dakota,
Wyoming, south-central Colorado, and central New Mexico. Especially
widely maculate individuals are found in populations from Carbon County,
Wyoming, and Torrance County, New Mexico (locality 15 in the statistical
analysis). This form was named pseudowillistoni by W. Horn, but because
of its scattered and nonuniform distribution (it could be called polytopic), it
is best not recognized as a subspecies.

Tue University SctENcE BULLETIN

AAs
Aff

S| I32

oe

34 I35 ie
ks

Fics. 128-136. Apex of aedeagus (dorsal aspect) of C. fulgida from various localities: West-
bourne, Manitoba (Fic. 128); near Blucher, Saskatchewan (Fic. 129); Onefour, Alberta (Fic.
Oliver County, North Dakota (Fic. 132);

130);

near Bottineau, North Dakota (Fic. 131);
Rawlins, Wyoming (Fic. 133); 4 miles northwest of Jamestown, Kansas (Fic. 134); 2.5 miles
southwest of Plainview, Oklahoma 4 miles southeast of Willard, New Mexico

(Fic. 135);

(Fic. 136).

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 269

The closest relative of C. fulgida is C. parowana, which is found in the
Great Basin, western Oregon, western Washington, and southern British
Columbia. Except for a longer labrum and several differences of the male
genitalia, it is quite similar to C. fulgida in general appearance and range of
variation of color and maculation. Quite clearly, they evolved from a com-
mon ancestor that was probably widespread in the western United States and
Canada in late Tertiary times. The rising mountains of the Cascadian Revo-
lution apparently separated it into populations which became differentiated
into the two modern species. C. fulgida is basically a species of cool climates;
however, Pleistocene glaciers and the accompanying shift of climatic zones
no doubt drove it south. As climates warmed and it moved northward again,
relict populations were left in Arizona, New Mexico, and Texas (the locality
of Woodward, Texas, listed as a doubtful locality earlier, may actually be a
relict population in south-central Texas). The occurrence of dark individuals
at Lincoln, Nebraska, a characteristic of more northern populations, may be
explained by assuming that the species was not exterminated from there by
the later glaciations and became dark due to the cool climate of that time.

C. macra

This species occurs between the Rocky and Appalachian Mountains (Fig.
137). Its distribution by counties or states is as follows:

MICHIGAN: Berrien, Emmet, Leelanau; WISCONSIN: Dane, St. Croix, Waushara, Wood;
MINNESOTA: Ramsey, Scott, Wabasha, Washington; OHIO; INDIANA: Greene, Lake,
Monroe, Porter, Posey, Putnam, Vigo; KENTUCKY: Henderson; TENNESSEE; JLLINOIS:
Calhoun, Carroll, Cook, Mason, Morgan, Ogle, Whiteside; IOWA: Alexander, Benton, Black-
hawk, Boone, Clinton, Des Moines, Johnson, Lee, Linn, Louisa, Monona, Pottawatamie, Van
Buren, Woodbury; MISSOURI: Atchison, Clay or Jackson, Holt; ARKANSAS: Craighead,
Crawford; LOUISIANA; TEXAS: Childress, Dallas, Denton, Eastland, Grayson, Hall, Harde-
man, Hemphill, Kaufman, Montague, Potter, Randall, Stonewall, Wichita, Wilbarger; OKLA-
HOMA: Alfalfa, Beaver, Beckham, Caddo, Choctaw, Cimarron, Cleveland, Comanche, Cotton,
Custer, Dewey, Ellis, Greer, Harmon, Harper, Hughes, Jackson, Johnston, Kingfisher, Kiowa,
Logan, Love, Major, Marshall, McClain, Osage-Payne Co. line, Payne, Pontotoc, Roger Mills,
Seminole, Texas, Woods, Woodward; KANSAS: Barber, Barton, Cheyenne, Clark, Clay,
Comanche, Cowley, Douglas, Ellsworth, Kearney, Kiowa, Leavenworth, McPherson, Meade,
Mitchell, Ness, Osbourne, Phillips, Pottawatomie, Reno, Riley, Rooks, Russell, Sedgwick, Shaw-
nee, Stafford, Sumner, Trego; COLORADO: Denver, Larimer; WYOMING: Carbon; NE-
BRASKA: Brown, Buffalo, Cass, Chase, Cherry, Dakota, Dixon, Douglas, Dundy, Franklin,
Hamilton, Lancaster, Madison, Merrick, Otoe, Platte, Saunders, Thomas; SOUTH DAKOTA:
Brookings, Clay, Union. Towns that could not be located: Pine, Indiana; Adams and Herrold,
Iowa. Doubtful records: Shoshone, Inyo County, California; Las Cruces, Dona Ana County;
Socorro, Socorro County; and Albuquerque, Bernalillo County, New Mexico.

This species could be considered a sibling to C. cuprascens and C. puritana
(see Table 18 for a comparison of the three), and they have been confused in
the past, leading to erroneous records in the literature and misidentified speci-
mens in museum collections. The ranges of C. macra and C. cuprascens

overlap broadly, but there are notable areas of nonoverlaping; for example, C.
macra is rare northwest of South Dakota and Wyoming, but C. cuprascens

270 Tue Universiry ScrENcE BULLETIN

Fic. 137. Known distribution of C. macra. Shaded circle=C. m. macra, open one m.
fluviatilis, half-shaded circle=intergrade population between C. m. macra and C. m. fluviatilis,
shaded triangle=C. m. amplicata, open triangle=intergrade population between C.” m. fluviatilis
and C. m. amplicata, square=state record.

occurs as far north as Manitoba and Montana; C. cuprascens is absent from
Michigan, where C. macra occurs; finally, C. macra does not occur in Ala-
bama or Georgia, where C. cuprascens has been found. These two species
also occupy the same ecological microhabitat; I have often seen them running
together on the same sandbar. Nevertheless, out of dozens of mating pairs
that I have collected, none involved two individuals of the wrong species.
Also, no specimens that appear to be hybrids have been seen. Thus, I have
no doubt that these forms are specifically distinct and genetically separate.

The following forms have been described within this species:

Cicindela macra Leconte, 1860. Type locality: Wisconsin and Minnesota (types bear yellow
paper circles, signifying “Illinois, Missouri, and Central Valley’).

Cicindela macra mercurialis Casey, 1913. Type locality: Iowa.

Cicindela macra topeka Casey, 1916. Type locality: Mt. Hope, Kansas.

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES a7

Cicindela macra fluviatilis Vaurie, 1951. Type locality: Red River, north of Quanah, Hardeman

County, Texas.

Cicindela macra amplicata Vaurie, 1951. Type locality: Denton County, Texas.

In recent years, three subspecies have been recognized: C. m. macra, the
smallest dark green or green-brown form found north and east of Oklahoma;
C. m. fluviatilis, the large, reddish, widely maculate form in Oklahoma and
northwestern Texas; and C. m. amplicata, the large, dark green, narrowly
maculate form in north-central Texas.

The following characters were measured in the study of geographic varia-
tion: (1) length of left elytron; (2) width of left elytron; (3) width of
labrum; (4) length of labrum, including tooth; (5) width of middle band
at a specified place (Fig. 114); (6) width of head; (7) shape of apex of
middle band in arbitrary units (Fig. 115); (8) color of elytra, using color
wheel.

Tasie 29. Localities and sample sizes of the specimens of C. macra measured.

N

Locality 3 2

eee LINID LAINVAS Monroe s@ors Bl ommm i git tie ees a ee nee oe 5 9
2. MINNESOTA, Wabasha Co., Dumfries; Scott Co., Jordan; Washington Co.,

Gray Cloud Island; WISCONSIN, St. Croix Co., North Hudson ..._......._.. 12 10

3, WONWA, Whoodlomiay Clay, Semeqeeine 840i ee eee ee ee 12 11

4. KANSAS, Ellsworth Co., Kanopolis Lake, s.e. Ellsworth Co, -.............-.........------ 12 11

DRIKCAIN SAS Sumner: Gos usta menos G etic ae Spo ulin is eae ee sere neem eee 12 12

HO KO AITO MAW Al fal tan Cowes amisensspomes Obs @ hero kee essen ere ee ee 12 12

The, NOUSICI NTS (OWS VANS oC OME Ge Een I Sl Of ne epee ee re 9 12 12

SOKEAL OMA MarshalliG@o.,, Lake mexomal State Parks 2220s eee eee 12 9

OMBAISEDXOAS se Montage GPR Ores tl Ur: peers ee eee eee ae ee) meet nee 12 8

Specimens from nine localities were measured; the localities and sample
sizes are given in Table 29. Sample sizes were adequate from nearly all
localities; however, specimens from Minnesota and Wisconsin were lumped
into one sample.

The means for the characters and localities are given in Tables 30 and 31.
Analyses of variance of the individual characters showed that there are sig-
nificant differences (p < 0.01) among the means of all characters for both
sexes (except characters 1 and 7 for males, which are significant at the 5%
level). Pooled within-locality variance-covariance matrices are given in
Tables 32 and 33. The sums of the among-locality variance components are
54.793 for males and 62.362 for females. The first two components for males
and the first three for females are highly significant (p < 0.01), and the
fourth for females is significant at the 5°% level. The first three functions
account for 95.31°% of the variance (among localities relative to that within)

272 Tue Universiry SciENCE BULLETIN

Tape 30. Means of nine localities (see Table 29) and seven characters for males
of C. macra. Values for characters 1-6 are in mm.

Character
Locality 1 2 3 4 5 6 7
] 8.09 2.04 1.60 0.71 0.45 3.04 4.60
2 7.89 2.01 1.56 0.67 0.33 3.03 52
3 8.08 2.02 1.61 0.69 0.39 3.09 Saye
4 8.09 2.09 1.63 0.70 0.39 3.14 4.08
5 8.27 Bei? 1.67 0.74 0.47 3.20 3.86
6 8.18 2.11 1.69 0.74 E55) 3.26 3.92
I 8.17 PMI 1.69 0.72 0.71 3.29 3.92
8 8.12 2.09 1.69 0.73. 0.70 3.23 3.64
9 8.42 222i, e/5 0.78 0.42 3)5)// 3341

Tasie 31. Means of nine localities (see Table 29) and seven characters for females
of C. macra. Values for characters 1-6 are in mm.

Character

Locality 1 2 3 43 5 6 7
1 8.20 2.04 1.64 0.66 0.37 SigIl7/ 3.76
2 7.79 112)5) 1.54 0.60 0.30 3.03 Sh)
3 8.33 2.05 1.64 0.67 0.35 3.24 3.93
4 8.36 Mell il 1.68 0.70 0.46 3.35 B75.
5 8.54 Dale) sq/5) 0.73 0.41 SAW 351
6 8.78 2.31 1.78 0.77 0.60 SS 3.60
7 8.66 2.25 Le, 0.73 0.59 3.54 4.18
8 8.77 2.30 1.81 0.74 0.58 3554 4.10
9

9.04 DNS 1.83 0.77 0.41 3.65 Sts

Taste 32. Pooled within-locality variance-covariance matrix for males of C.
macra. Variances are along the diagonal, and covariances compose the rest of
the matrix. Values have been multiplied by 10-4; thus “1104”°—0.1104.

Character

1 2 3 5 6 i

| 1104 230 171 75 40 280 412

2 71 40 21 3 66 117
ae 46 20 2 61 59
s 4 24 2 33 39
= 5 124 6 79
6 120 144

BronomMics AND ZOOGEOGRAPHY OF T1GER BEETLES Da es

Tasie 33. Pooled within-locality variance covariance matrix for females of C.
macra. Variances are along the diagonal, and covariances compose the rest of
the matrix. Values have been multiplied by 10-4; thus “1076”—0.1076.

Character

1 2 3 4 5 6 7

1 1076 226 187 104 72 365 217

2 43 29 16 82 17

5S 3 55 23 16 76 36

Ss 4 26 () 43 66
igs]

5

a 5 102 11 64

6 174 7()

j/ 4406

in males and 92.86% in females. The distributions of the means in the first
three discriminant functions are shown in Figures 138 and 139 (however, the
third function is not significant in males and could be ignored). Locality 9
is widely separated from the others, which form an elongate, loose cluster in
males and three more compact clusters in females.

The results of the simultaneous testing procedure show that all combina.
tions of pairs of means are significantly different at the 5° level except 1 vs. 3,
4 vs. 5, and 7 vs. 8 in both sexes; 2 vs. 3 in males; and 1 vs. 2, 6 vs. 7, and
6 vs. 8 in females.

Sets of vectors (Figs. 140 and 141), plotted in the first two functions, show
geographic trends of the seven characters when compared with Figures 135
and 139. Specimens from locality 9 have relatively wider and shorter elytra.
Specimens from the northeastern localities have relatively long, narrow elytra
and long labra. Specimens from Oklahoma have wide heads and broad

middle bands.

The results of the color analysis are shown in Fig. 142. Populations from
north and east of Oklahoma consist mostly of dark green to dark red-green
to dark reddish (actually brown) individuals. Populations from Oklahoma
consist of mostly brighter red individuals, with fewer red-green and green
individuals. The Texas population is similar to the northeastern populations
in color.

These analyses confirm the existence of the three recognized subspecies.
Locality 9, representing C. m. amplicata, is greatly different from all others.
Localities 1-5 (C. m. macra) are significantly different from localities 6-8
(C. m. fluviatilis). The inclusion of samples from other areas would no
doubt fill in the gaps between the clusters in Figures 138 and 139. However,
study of specimens from most of the localities in Figure 137 indicates that
the variation within this species is in the form of a stepped cline, with narrow
zones of intergradation between the three subspecies. In fact. three of the

| ea es = ee ae
I39 ,

Fics. 138-139. Distributions of the means of the nine localities for males (Fig 138) and
females (Fig. 139) of C. macra in the first three discriminant functions (Ki, Ke, Ks), drawn as
three-dimensional models, with numbered balls representing the means and vertical supports
arising from the Ki, Ke surface. Different sized balls indicate different distances from the viewer.
The models are viewed from opposite directions, relative to the clusters.

BIoNoMIcs AND ZOOGEOGRAPHY OF TIGER BEETLES 275

40 aia

Fics. 140-141. Vectors for the seven characters of males (Fig. 140) and females (Fig. 141)
of C. macra for the first two discriminant functions. Each vector shows the change in the
discriminant function that the corresponding character would produce if it varied independently.
Units are same as in Figs. 138-139.

localities analyzed (5, 6, and 9) are in intergradation zones. Intergrade popu-
lations are characterized by much variability in color and maculation.

The fact that C. macra shows a greater differentiation into geographical
races than does C. cuprascens might indicate that it has been in its present
range (at least in the southern parts) for a longer time than C. cuprascens.
If true, this also supports the idea that the two evolved on opposite sides of
the Mississippi valley, as mentioned under C. cuprascens. The pattern of
color variation of C. macra shows very clearly the effect of selection to match
the substrate color. The limits of the reddish C. m. fluviatilis coincide almost
exactly with the northeastern limits of red soil (Figs. 137 and 108). Popula-
tions intermediate between it and C. m. macra in southern Kansas, north-
eastern Oklahoma, and western Arkansas contain some reddish, some brown,
and some green individuals. The area of occurrence of “true” C. m. ampli-
cata possesses dark soil; however, in the northern and western parts of the
range of this form, tan or reddish soils begin to appear, and populations
contain many reddish and brown individuals.

This species is reported in the literature from three localities along the Rio
Grande in New Mexico. No specimens have been seen from these localities,
and they have been listed as doubtful. It is hard to imagine how the species
could have gotten there.

276 Tue University SciENcE BULLETIN

a

igen Or ees ZA

green _..:

Fic. 142. Results of color analysis of C. macra using the color wheel. Different degrees of
shading indicate different percentages of the sample. The positions of the three major colors
are shown at top center. Numbers show the sample size for each locality.

BIoNoMICcs AND ZOOGEOGRAPHY OF TIGER BEETLES De,

C. nevadica

This is a western species, occurring from the Mojave Desert in California
and Sonora, Mexico, to the western edge of the Central Plains (Fig. 143).
Its distribution by counties or states is as follows:

ALBERTA; SASKATCHEWAN; MANITOBA; MONTANA: Hill, Prairie, Roosevelt,
Sheridan, Yellowstone; NORTH DAKOTA: Pierce, Ramsey; SOUTH DAKOTA: Hand;

Fic. 143. Known distribution of C. nevadica. Triangle=C. n. nevadica, open circle=C. n.
tubensis, shaded circle=C. n. knausi, square=C. n. olmosa, diamond=C. n. lincolniana.

278 Tue Universiry ScrENCE BULLETIN

WYOMING: Goshen, Weston; NEBRASKA: Dundy, Hitchcock, Lancaster, Nuckolls, Scotts
Bluff, Sheridan; COLORADO: Bent, Chaffee, Prowers, Otero; KANSAS: Barber, Cheyenne,
Clark, Cloud, Ellsworth, Ford, Gove, Hamilton, Kearney, Kiowa, Lincoln, McPherson, Meade,
Mitchell, Reno, Republic, Russell, Sedgwick, Stafford, Sumner; OKLAHOMA: Alfalfa, Beaver,
Beckham, Custer, Jackson, Logan, Major, Oklahoma, Payne, Roger Hills, Tulsa, Woods;
TEXAS: Eastland, Hudspeth, Hutchinson, Kenedy, Randall; NEW MEXICO: Bernalillo,
Dona Ana, Guadalupe, Quay, Rio Arriba, San Doval, San Juan, Santa Fe, Sierra, Taos, Torrance;
UTAH: Duchesne, Emery, Moab, Wayne; ARIZONA: Apache, Cochise, Coconino, Navajo;
NEVADA: Nye; CALIFORNIA: Inyo, Kern, San Bernadino; SONORA; COAHUILA. The
exact location of Acnegas, Coahuila, Mexico, could not be found.

The following forms have been described within this species:

Cicindela nevadica Leconte, 1875c. Type locality: Nevada.
Cicindela Knausii Leng, 1902. Type locality: Kackley, Belvidere, and Great Spirit Springs,
Kansas.

Cicindela lincolniana Casey, 1916. Type locality: Lincoln, Nebraska.
Cicindela nevadica tubensis Cazier, 1939. Type locality: Tuba City, Coconino County, Arizona.
Cicindela nenadica olmosa Vaurie, 1951. Type locality: Los Olmos, Kenedy County, Texas.

In recent years, all five of the above forms have been recognized as sub-
species, C. m. nevadica being characterized by dark brown or green-brown
color and often a reduced marginal line; C. n. tubensis by reddish color and
often expanded markings; C. n. olmosa by dark greenish brown color and
expanded markings; C. n. knaust by reddish brown, greenish brown, green,
or blue color and average markings; and C. n. lincolniana by dark greenish
brown color and reduced markings.

In the study of acne: variation, the following characters were meas-
ured: (1) length of left elytron; (2) width of left elytron; (3) width of
labrum; (4) length of labrum, including tooth; (5) shape of base of middle
band in arbitrary units (Fig. 144); (6) width of middle band in arbitrary
units (Fig. 145); (7) width of apical lunule in arbitrary units (Fig. 146);
(8) color of elytra, using color wheel.

Specimens from 18 localities were measured; the localities and sample
sizes are given in Table 34. Sample sizes were adequate for nearly all locali-
ties, except some in New Mexico, Utah, and Kansas.

The means for the characters and localities are given in Tables 35 and 36.
Analyses of variance of the individual characters showed that there are sig-
nificant differences among the means of characters 3-7 (p < 0.01), and
character 1 (p < 0.05); the means of character 2 are nonsignificant for both
sexes. Pooled within-locality variance-covariance matrices are given in Tables
37 and 38. The sums of the among-locality variance components are 150.835
for males and 129.146 for females. The first four components are highly
significant (p < 0.01); and the fifth is significant at the 5°% level. The first
three functions account for 95.67°% of the variance (among localities relative
to that within) in males and 93.1594 in females. The distributions of the
means in the first three discriminant functions are shown in Figures 147 and
148. Localities 1, 3-11, and 18 form a fairly compact cluster, while the other
localities are scattered about to varying degrees.

BronomMics AND ZOOGEOGRAPHY OF TIGER BEETLES 279

<——— — — — Ve
nee Foe
Sone re 2

l 2 3 a 5 6

Led fe eR

46

Fics. 144-146. Illustrations of certain characters measured on elytra of C. nevadicu; Fic. 144,
arbitrary units for shapes of base of middle band; Fic. 145, arbitrary units for width of middle
band; Fic. 146, arbitrary units for width of apical lunule.

The results of the simultaneous testing procedure show that all combina-
tions of pairs of means are significantly different at the 59% level except 3 vs.
8, 4 vs. 7, 4 vs. 8, 4 vs. 9, 4 vs. 10, 4 vs. 11, 5 vs. 8, 8 vs. 10, 9 vs. 11, 10 vs. 11,
and 12 vs. 13 for both sexes; 4 vs. 10, 5 vs. 7, 5 vs. 10, 7 vs. 10, 9 vs. 10, and
12 vs. 15 for males; and 1 vs. 9, 3 vs. 4, 3 vs. 5, 3 vs. 7, 3 vs. 8, 3 vs. 9, 3 vs. 10,
3 vs. 11, 3 vs. 13, 7 vs. 8, 7 vs. 9, and 8 vs. 9 for females. Thus, most members
of the cluster are not different from one another, but the scattered means are
usually significantly different from one another and from the cluster.

Sets of vectors (Figs. 149 and 150), plotted for the first two functions,
show geographic trends of the seven characters when compared with Figures
147 and 148. Specimens with high values in Ki have a wide base of the
middle band, long labrum (in males), and narrow elytra (in females); and
vice versa for specimens low in K4.

The results of the color analysis are shown in Figure 151. Specimens from
California and Nevada are predominantly dark reddish (appear brown) to

280

Tue University ScrENcE BULLETIN

Taste 34. Localities and sample sizes of the specimens of C. nevadica measured.

N
Locality ey Q
I OMA NTTOBA:. (Gy amiics., cote bal to tay ee es ce ce ee ee 10 10
2. INEBRASK Avwizancaster Gow Mincolimy Gwiestyec vic) mse eee eee eee 2 12
3. RANSAS® lincoln(Go.. lin ni 2A mi. e.. of Lincoln eee 9 5
ae SRAINSAIS] McPherson; 15 "mites (Of (Galival ieee ee ee ree ee 12 12
5s RAINSAS) Statrord: Coz, 1 litinis mie of sr luc soi ee ee ee ee eee 12 12
GS SIKAINS AS: BClank® (Gos sBimp lew God (isc ee cess ca eeeae tre e 12 12
7 AGOLORADO: Bent Cot. Tas Animas ae ee 12 12
ST LOKLAROMA, Alltaltas@oee Sumin enor uCherokeems eee 11 10
SO OKUAELIOMAS, Woodse@on 2-5 mileesswaor Planmvicyy cs ee eee 12 12
LOR RELA? Bastlandi(Gomtainkawaro tn G15 copes ease es ee ee 11 10
pe NE VWiViEd €@h Ouay Cos alnmipen ea oO feeli Gunn carl gene een een eee 12 12
[QE INE WaME x COs lacs Cosy ©) om Gallien tepe eee oe ee ae 3 4
is) NEW MEXICO s San oval Com Simiieswe Ole Samy SIG ro) aeeeee eee 4 9
[a ARIZONAS Navayjon@or. lor mirsemems yw rO se Keely em tees eeee meen nee ene 11 3
LSE UEA Eis merye Cone oa eee. 2 es ee ee ee 5 iL
[Gee INE WiMExXd CO Sienran@os a> amminmwa © lawl lc osc) een em 12 8
Wet ENE WAD ASeNyex Cora Aish: IM eal ovis) cee eer ee 12 8
8k (CALIFORINTAS Wer Cox, Saltdale: <2. o5 ae es eee 12 11
Taste 35. Means of 18 localities (see Table 34) and seven characters for males

of C. nevadica. Values for characters 1-4 are in mm.
Character
Locality 1 2 3 4 5) 6 7

1 Hols 2.04 1-53} 0.65 4.16 4.25 2.86

2 7.07 2.02 1.45 0.63 SAL 2.50 1.93

3 Holl? 2.07 1.49 0.63 4.09 3.42 2.20

4 6.94 1.98 1.43 0.61 4.04 4.24 2.99

5 7.09 2.03 1.45 0.63 4.05 4.07 2.84

6 6.82 1.94 1.43 0.61 4.08 4.66 3.94

ih 6.98 2.05 1.43 0.62 4.03 4.08 3.34

8 6.92 2.01 1.44 0.61 4.05 4.08 2.40

9 7.06 2.02 1.47 0.60 4.08 4.57 3.18

10 6.94 2.03 1.44 0.61 4.14 4.32 2.95

11 6.96 2.01 1.45 0.60 4.12 456 3.41

12 7.19 2.08 1.49 0.71 4.23 4.60 Sh 5y/

13 7.07 2.01 1.45 0.69 4.10 4.38 3.28

14 6.57 1.92 1.39 0.64 5.19 5.88 5.66

15 6.74 1.99 1.43 0.66 432 5.14 4.42

16 6.94 1.99 1.40 0.62 4.35 5.48 5.28

7 7.19 2.03 150 0.60 Ae, 4.23 1.94

18 6.86 1.98 1.41 0.59 3.98 5.18 3.89

BroNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 281

Tasie 36. Means of 18 localities (see Table 34) and seven characters for females
of C. nevadica. Values for characters 1-4 are in mm.

Character
Locality l 2 3 4 5 6 7
1 7.86 2.32 1.65 0.68 4.11 3.89 2.69
2 HEY) 225 sul 0.59 3.39 2.60 1.68
3 7.63 2.27 Lol 0.64 4.12 4.18 2.90
4 7.63 2.27 1.54 0.64 4.10 3.97 2.93
5 7.45 2.30 1.56 0.66 4.00 3.61 2.81
6 7.48 2.21 E58 0.64 4.10 4.53 3.62
7 7.34 DeP)i\ 1.50 0.62 4.09 3.88 Pei
8 7.61 DDB, 154 0.62 4.10 3.96 2.79
9 US 2.22 1.54 0.64 4.12 4.03 neil |
10 7.50 PQS 1.50 0.61 4.13 4.63 3.06
11 7.74 2.26 i555) 0.64 4.13 4.13 2.76
12 7.97 2.28 lay 0.71 4.25 4.48 3.90
13 7.40 2.14 {L5yI! 0.65 411 4.22 S)9}5)
14 7.30 2.20 eZ, 0.66 DEY) 6.20 6.10
15 TAD Deo) SA 0.68 Ai) 5.05 3.81
16 7.65 22S 1-49 0.65 4.44 5.18 5.49
7 7.60 DDS 157 0.62 1.41 1.00 1-91
18 7.38 223 1.50 0.59 3.98 4.88 3.56

Taste 37. Pooled within-locality variance-covariance matrix for males of C.
nevadica. Variances are along the diagonal, and covariances compose the rest of
the matrix. Values have been multiplied by 10-*; thus “1187”=0.1187.

Character

1 2 3 4 5 6 7

1 1187 311 220 132 135 45 203

2 108 64 39 38 29 60
oe: 60 30 33 41 38
a4 24 15 22 34
Be 717 504 659
6 2771 799

7 3503

dark green. The Utah and Arizona samples are mostly reddish, and the
northern New Mexico samples are similar, but contain many red-green in-
dividuals. The southern New Mexico population is dark red (brown) to
dark-green. Other populations are quite variable in color and contain many
individuals with brighter colors (in the south; northern populations are
darker).

These analyses, in general, support the existence of the five subspecies.
Samples representing C. n. lincolniana (2) and C. n. olmosa (16) are well

282 Tue UNiversiry ScIENCE BULLETIN

Taste 38. Pooled within-locality variance-covariance matrix for females of C.
nevadica. Variances are along the diagonal, and covariances compose the rest of
the matrix. Values have been multiplied by 10-4; thus “1198”=0.1198.

Character

l 2 3 4 5 6 7

1 1198 316 215 117 60 167 2G

2 121 65 35 32 61 =

2 3 61 24 14 33 27
a4 23 15 25 9
aS 478 321 245
6 2101 1040

il 2876

separated from the others in the statistical analysis. The means of C. n.
knausi (1, 3-13) form a compact cluster. The means of C. n. tubensis (14
and 15) are widely separated, apparently because sample 14 contained many
small, very widely maculate individuals. Localities 12 and 13 could be called
intermediates between C. n. tubensis and C. n. knausi on the basis of their
color. The means of C. n. nevadica (17 and 18) are also widely separated
because most individuals in sample 17 have no marginal line, while those in
18 do, causing them to resemble C. n. knausi (still, 18 is significantly different
from all others). However, the geographical isolation of C. n. nevadica and
its characteristic dark color and wide transverse portion of the middle band
justify its recognition. Some of the Utah specimens of C. n. tubensts lack the
red color typical of Arizona populations, being greenish brown and thus
resembling C. n. knausi; however, their markings average much wider than
those of C. n. knausi. The status of the population in southeastern Arizona
(if it is a valid record) is in doubt (Fig. 143). The eight specimens from
there are dark green to green-brown, and have average markings with a
marginal line. At present, they are being called C. n. olmosa. The popula-
tion of “C. 2. olmosa” from central New Mexico may be intergrades between
this subspecies and C. n. knausi; I have seen no specimens from there.

C. nevadica almost certainly evolved in the western United States, but its
range is so large, it is impossible to guess where. Its closest relatives are
C. sperata and C. marutha, found in the Great Basin, northern Mexico,
southern Great Plains, and (C. sperata only) southern Texas. The three
probably evolved as the rising western mountains in late Tertiary times iso-
lated populations of a common ancestor. C. nevadica then probably became
widespread in the Southwest during the early Pleistocene pluvial periods,
when moist climates permitted easy dispersal. During drier interglacial
periods, many populations were apparently extinguished and resulting com-
plete or partial isolation favored the development of geographic races: C. n.

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 283

I48

Fics. 147-148. Distributions of the means of the 18 localities for males (Fig. 147) and
females (Fig. 148) of C. mevadica in the first three discriminant functions (Ki, Kz, Ks), drawn
as three-dimensional models, with numbered balls representing the means and vertical supports
arising from the Ky, Ke surface. Different sized balls indicate different distances from the viewer.

nevadica in the western Great Basin and C. n. tubensis on the Colorado
Plateau, where it developed a red dorsal color because of the red soils in that
region (Fig. 108). A third form (or ancestors of C. n. tubensis) spread

284 THe University ScrENCE BULLETIN

I49

Fics. 149-150. Vectors for the seven characters of males (Fig. 149) and females (Fig. 150)
of C. nevadica for the first two discriminant functions. Each vector shows the change in the
discriminant function that the corresponding character would produce if it varied independently.
Units are same as in Figs. 147-148.

throughout the region drained by the Rio Grande and Pecos River; those
along the lower Rio Grande became C. n. olmosa, while the others (C. 2.
knaust) dispersed northeastward. Glaciations evidently pushed C. n. knaust
as far south as central Texas, where a relict population still exists in Eastland
County. C.n. knaust probably reached the Lincoln, Nebraska, vicinity some-
time after the Kansan glaciation, probably during the Yarmouth. From then
on, it was apparently isolated from the rest of the gene pool and evolved into
C. n. lincolniana. After the Wisconsin glaciation, C. n. knausi quickly dis-
persed northward into the Dakotas, Montana, and southern Canada. The
fact that C. n. knausi has not developed completely reddish populations in
western Oklahoma indicates that it has not been in this area as long as C. 7.
tubensis has been on the Colorado Plateau; however, the Oklahoma popula-

BIoNoOMIcs AND ZOOGEOGRAPHY OF TIGER BEETLES 285

blue er"

red

Fic. 151. Results of color analysis of C. nevadica using the color wheel. Different degrees of
shading indicate different percentages of the sample. The positions of the three majors colors
are shown at top center. Numbers show the sample size for each locality.

tions seem to be evolving in that direction, because they contain many bright
red individuals, compared to populations from north of there (Fig. 151).

C. togata

This is a southern species found along the Gulf and Atlantic coasts from
South Carolina to northern Mexico, and inland from western Texas to
Nebraska (Fig. 152). Its distribution by counties or states is as follows:

NEBRASKA: Dodge, Fillmore, Lancaster, Nuckolls, Saunders; COLORADO: Chaffee,
Otero; KANSAS: Barber, Butler, Clark, Cloud, Kiowa, Lincoln, McPherson, Mitchell, Reno,
Republic, Sedgwick, Stafford; OKLAHOMA: Alfalfa, Beckham, Blaine, Garfield, Grant, Har-
mon, Harper, Jackson, Logan, Muskogee, Woods; NEW MEXICO: Chaves, Eddy, Guadalupe,
Torrance; TEXAS: Andrews, Aransas, Bailey, Brazoria, Calhoun, Cameron, Childress, Dallas,
Dawson, Dimmet, El Paso, Gaines, Galveston, Harris, Hidalgo, Hudspeth, Jackson, Jefferson,
Kenedy, Kleberg, Live Oak, Matagorda, Mitchell, Nueces, Reeves, San Patricio, Val Verde,
Wichita, Wilbarger; TAMAULIPAS; LOUISIANA: Cameron; ALABAMA: Mobile; FLOR-

286 Tue UNIversiry SCIENCE BULLETIN

Fic. 152. Known distribution of C. togata. Triangle=C. t¢. togata, circle=C. +t. globicollis,
diamond=intergrade population, square=state record.

IDA; SOUTH CAROLINA: Beaufort, Charleston. The town of Meredith, South Carolina,
could not be located. Saltair, Utah, is a doubtful record.

The following forms have been described within this species:

Cicindela togata LaFerté, 1841. Type locality: Texas.

Cicindela togata latilabris, new name for Cicindela togata apicalis W. Horn, 1897 (not apicalis
Chaudoir, 1843). Type locality: Nebraska, Kackley (misspelled “Kakley”), Kansas. Pre-
occupied.

Cicindela globicollis Casey, 1913. Type locality: Clark County, Kansas.

Cicindela fascinans Casey, 1914. Type locality: Santa Rosa, New Mexico.

In recent years, three subspecies have been recognized: C. ¢. togata, the
rather small coastal form with the apical elytral spine of the female greatly
retracted; C. ¢. globicollis, the large northern form with the apical elytral
spine of the female slightly retracted; and C. t. fascinans, the small south-
western form with expanded markings and slightly retracted apical elytral
spine in the female.

The following characters were measured in the study of geographic varia-

tion: (1) width of head; (2) width of pronotum at widest point; (3)

BrioNomMics AND ZOOGEOGRAPHY OF TIGER BEETLES 287

po é

Fics. 153-154. Illustrations of certain characters measured on C. togata; Fic. 153: a=pro-
notal bulge, b=width of maculation, c=retraction of elytral spine of female, d—=number of
punctures in 0.45 m?® area at this position; Fic. 154, arbitrary units for shape of middle band.

“pronotal bulge,” the difference between character 2 and the width of the
pronotum at its anterior margin (Fig. 153a); (4) width of left elytron;
(5) length of left elytron; (6) width of maculation at a specified place
(Fig. 153b); (7) retraction of left apical elytral spine (distance from its
base to apex of elytron) of female (Fig. 153c); not measurable in males
because the spine is truly apical; (8) number of punctures in a 0.45 mm?
square on a specified part of the left elytron (Fig. 153d); (9) average width
of setae on center of pronotum; (10) shape of middle band in arbitrary units
(Fig. 154); (11) color of elytra in arbitrary units: 1=dark green-brown,
2=dark green-cupreous, 3=green-cupreous, 4=cupreous, 5=dark purplish
cupreous, 6=dark green-purple; (12) length of labrum, excluding tooth;
(13) width of labrum.

Specimens from 19 localities were measured; the localities and sample
sizes are given in Table 39. Sample sizes were adequate except for localities
in the southwestern part of the range and from Dallas, Texas. Specimens
from New Mexico were lumped into one sample.

The means for the characters and localities are given in Tables 40 and 41.
Analyses of variance of the individual characters showed that there are
significant differences (p < 0.01) among the means of all characters. Pooled
within-locality variance-covariance matrices are given in Tables 42 and 43.
The sums of the among-locality variance components are 258.487 for males
and 284.879 for females. The first eight components for males and the first
five for females are highly significant (p < 0.01), and the sixth component

Tasre 39. Localities and sample sizes of the specimens of C. togata measured.

Locality 3
le NEBRASKA Pancastern Caran leimncolnin (yes tac cl cc) eee sean eee eee 10
P. URINNISINGS ARGiolollite (Clog, Ye soak, ks ILS in Aellonvey 2 ee 10
Ss IKAINSASS eincolin’ Com Simi wee ee Sono tel argh cl eee eee ee 10
4. KANSAS, Butler Gon sBl Dorado (westued cc) ks ae eee ee ee 10
SA PICAINSAG= Stattord: Conelilemniee mes Ole tT G sores ne 10
GeIWAINSA'S GlarkaGo. vEnplewoods 22.2e2 ees nee ne eee eee ee ee 10
/. WOKE ATIOMA. AllfaltaiConesi mie. of Cherokees eee ee 10
8) OKLAROMAS Garfield(Con 47) mie niw. of Drummonds ee 10
Oo ORTALOMAtackson Coms:5-5 missssOt 6) dorad ope ane ee 10
10. TEXAS, Bailey Co., 9 mi. s., 9 w. of Muleshoe (Coyote L.);
INE Wr MEXICOh Roosevelt Conn ante ces ill sen@hey Airc iia sea een ee a 4
11. TEXAS, Andrews Co., 8 mi. n.w., 3.5 mi. s.w. of Andrews (Shafter L.) ...........- 9
12. NEW MEXICO, Chaves Co., 8 mi. n.e. of Roswell;
Biddy Gown6 nn. es Zamacob eon cela 6
13) DEXAS, Eludspeth Cor 90\imin es0f Ele Rason (neaneSaltellat)) sass een 7
4 “TEMAS Val Verde (Cox; Del "Rigas. 228s as ee 10
(5'y “TEXAS -DallastCo:.cD alls oe os as ee 1
GH EAS, Kileberou Com Riviera cach eee mene eeeeees ete eae enn En een ee 10
4 SE EXAS Galvestomm Go... Gallvestom) sete et eee ee ee 10
LS CALABAMA. Mobile. Co:uGoden <8 eec ee e ey, 10
OS SOUME GARONA Chanlestons Comp Fo) eB ea chien ee 10

N

Taste 40. Means of 19 localities (see Table 39) and 12 characters for males of

C. togata. Values for characters 1-6, 9, 12, and 13 are in mm.

Character

1 2 3 4 5 6 8 9 10 11 12 13

1 2104 2240) 044 220 7AG 40545916320 OLOLS OT E29 eS Oe ommmaleia

2 32 ea OFS 204 ole OS Oe CON OL OMS Gi ele 2 les eee

3 292, 223, 1037 N92) F025 L649 16324) 0.08 ON 223 tea Oey

4 3109) 24 10l44) 22027 739) 10166) 5728) OL0NG 4S ESI eS: Gp Os eeemleey

5 Bl 2240) 038)" 22038 7A OS78e 64.0 20020002 7S 0 GO cee,

6 293 225, 040) ESO) 678) O77 60s 00197 S205 SS Sn 2 ee

7 2.97 2:28 038 1290) 7:07" (0272 628) 00190 2535) 2 19S 069 ee!

8 3:09) 2:36 10943" 2:00" 7.377 0M 56:2. 0019 24 oa Ole

9 320 2241 042) 2200) 7:21) 089) 562) 00210 N 335244 0 OSS alr

10 3.10) 242 1038 196" 7-26) 10:83 a2) 010235) 9250 94S oOo mele)
11 3109) 241 1038) 2201 e733 8 TOI Seo OO 2277 SFOS On Oona)
12 272) 2200 032) 69) G40) SOB 6323 00255. S257 O02 Oop emelees
13 2103) 217 034 188) 6166 1288) 5353) 00259 G00 ara Ol79 esr,
14 295 27 036) SG 677 10162 555010200) e225 54277 Ovals
15 2182, 1195 021) 74 6:25) 0%6I G10) (010220) Si60 4:60 065 EZ
16 309 92.29) 10405 9 1295)96:86) 02725) Sas 0L0225 328 426 SES
17, 3:06 2:20) 10321093) 16.84 0!67 955225 OL0S 2 778 0 ee)
18 2.83 1.96 O27- 1:81) 6:38) 10:70 54:7e LOL0NSO 27 4078 Oo ale
19 SHO) AIC) Oya TW Oi A S527 (OMNIS Sil et) O72 NS

BronomMics AND ZOOGEOGRAPHY OF TIGER BEETLES

289

Tass 41. Means of 19 localities (see Table 39) and 13 characters for females of
C. togata. Values for characters 1-7, 9, 12, and 13 are in mm.

Locality 1

2

3

4

5

6

Characte
7

r

8

9

3.35
3.49
Sly
3.24
3.26
3.07
3.19

D2.
3.36

3.38
3.10
3.29
323
3.30
3:32
3.08

ee a
CONIDMNBPWBNK TCU WWYWA UAWNeE

3.43

3-7}

3.22.

2.63
2.63
2239.
DS
Ys
Visi
2.40

2.59)

2.34
2.61
2.48
Assil
2.10
2.3/7
DDT,
Desay
2.28
PS)
224

0.41
0.38
0.34
0.38
0.35
0.36
0.33
0.40
0.34
0.41
0.34
0.36
0.28
0.33
0.32
0.34
0.29
0.27
0.30

2.31
2a
2.06
2.12
2.21
2.05
7M
2.24
2.07
2533
2.18
2.20
1.97
21h
2.07
2.08
2.11
2.01
1.98

8.23
8.25
7.48
7 Bl
7.83
7.08
7.70
8.01
7.32
7.99
7.65
7.89
6.86
7.31
WPS
7.33
Tes
6.80
7.02

0.66
0.59
0.57
0.62
0.76
0.75
0.71
0.70
0.74
0.87
0.92
1.18
1.97
0.68
0.77
0.74
0.69
0.75
0.67

0.14
0.16
0.10
0.13
0.11
0.11
0.11
0.18
0.14
0.14
0.12
0.10
OL
0.39
0.36
0.42
0.44
0.30
0.30

56.1
55.4
2829
53.3
DHE:
IU
Shs
50.9
55.5
50.8
Sl5)
54.0
55/55)
51.8
48.7
49.2
51.8
50.5
54.0

0.0188
0.0201
0.0172
0.0178
0.0208
0.0198
0.0193
0.0205
0.0210
0.0252
0.0213
0.0230
0.0240
0.0211
0.0207
0.0239
0.0197
0.0176
0.0176

1.17 1.53 0:80
1.23, 1.44 Q8T
4 232) Or71
Z09 3:82, 0374
7238) 2.89) 0:78
ROT Sil O74
i.97 3.61 0.76
1.80: 3.52. 0:80
2.07 4.40 O75
2.28 3.88 0.86
3.16 3.88 0.82
3.00 3.90 0.86
6.00 4.20 0.83
3.01 4.82 0.77
3.07 4.70 0.73
2.82 4.71 0.76
£70 4:96 0.77
25: 47 0k70
L137 4.87 O75

1.60
1.60
44
1.46
1.55
1.39
as,
1.57
1.45
1.59
1.58
SY,
1.35
Ay,
1.38
1.43
gs
E35
1.39

Taste 42. Pooled within-locality variance-covariance matrix for males of C.
togata. Wariances are along the diagonal, and covariances compose the rest of
the matrix. Values have been multiplied by 10-4; thus “186”—=0.0186.

Character

Character
3 5 6 8 9 10 11 12 13
14 102 407 72 —2690 1 89 13 52 81
32 91 375 70 —2773 0 129 18 47 72
27 10 36 9 —344 0 28 —13 6 7
93 294 53 —1957 0 91 31 36 51
1345 193 —7617 1 315 100 141 213
104 —1594 0 229 29 22 34
228105 —9 —1602 —571 —1000 —1442
0 1 —0 0 0
3449 225 34 52
1684 —3 22
24 26
47

for females is significant at the 5°% level. The first three functions account
for 87.959, of the variance (among localities relative to that within) in males
and 88.9694 in females. The distributions of the means in the first three
discriminant functions are shown in Figures 155 and 156. Localities 14-19

290 Tue University ScrENcCE BULLETIN

form a fairly compact cluster, while the others are scattered in a loose cluster,
except 13, which is widely separated.

The results of the simultaneous testing procedure show that all combina-
tions of pairs of means are significantly different at the 5% level except 15
vs. 16 and 15 vs. 18 for both sexes; 15 vs. 6, 15 vs. 9, 15 vs. 10, 15 vs. 12, 15 vs.
14; 15°vs. 17, and 15 vse 19 for males;“and4vs.8;5 vs: 7,6 vs. 12,9 va
10 vs. 12, 11 vs. 12, and 14 vs. 15 for females. The low sample size of one
individual for males from locality 15 and of females from locality 12 accounts
for the many nonsignificant pairs involving those localities.

Sets of vectors (Figs. 157 and 158), plotted for the first two functions,
show geographic trends of the characters when compared with Figures 155
and 156. Specimens with high values in Ki (southwestern localities) have
wide maculation, a long labrum, a greater pronotal bulge, and (in males)
wider pronotal setae. Specimens low in Ki and Ke (northern localities) have
a wide labrum, narrow head, and low arbitrary color values. Specimens low
in Ki and high in Ke (coastal localities) have a wide head, a narrow labrum,
high color values, and (in females) wide pronotal setae and a more greatly
retracted elytral spine.

The pattern of variation of four of these characters is shown in Figure 159.
The width of maculation shows a gradual increase from northeast to south-
west, while all coastal and eastern Texas samples have narrow maculation.
The same general pattern is shown by the width of pronotal setae, although
some eastern Texas samples have very high values. Arbitrary color values
increase from northeast to southwest, thence to the southeast. The retraction

Tape 43. Pooled within-locality variance-covariance matrix for females of C.
togata. Variances are along the diagonal, and covariances compose the rest of
the matrix. Values have been multiplied by 10-4; thus “268”=0.0268.

Character

1 2 3 4 a pe 8 > 10°) Ai) oie

1 268 181 « 30° 152) 571. 74 «21> Soe 0) Gat pee

2 202°. <40'e 20:0 AsO" bor! “Bo g5ss a) ee = es 90

3 38. “e2ip O29. S61 =A! Sag 9 7 14

4 132° %379' 54 ais? —=D0iSer 0 91 28 40 76
5 1590 211 53 —7603° 1 344. 295° 154m

oS 6 94 12 910), ore as ee 39
8.7 18 97 0 D6 ie v8 7 9
G 8 226118 —0 —2788 1153-—840=1483
9 0 1. =0 0 0
10 2310): 13° Sar 66

—
—"

2379 —14 25
A) 31

a
Ww dO
fon'
Ww

BIoNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 291

|
i

Wy :
1p oy f + fae ‘e
\
= Cae \ \

6 4D

|
ig H
: | ©. @ oy

3 if
ath
SSS eae
SSE ROSSER EO
Seeee So Soe eee See
SSS
eS Sas AS Sa i "K,
: ;

| 10
r 9

cloaee

Fics. 155-156. Distributions of the means of the 19 localities for males (Fig. 155) and
females (Fig. 156) of C. togata in the first three discriminant functions (Ki, Ke, Ks), drawn as
three-dimensional models, with numbered balls representing the means and _ vertical supports
arising from the Ki, Ke surface. Different sized balls indicate different distances from the viewer.
The models are viewed from opposite directions, relative to the clusters.

292 Tue Universiry ScrENcE BULLETIN

-62

IS7

;
vA lge
Fics. 157-158. Vectors for the 12 characters of males (Fig. 157) and 13 of females (Fig. 158)
of C. togata for the first two discriminant functions. Each vector shows the change in the dis-

criminant function that the corresponding character would produce if it varied independently.
Units are same as in Figs. 155-156.

I58

of the female elytral spine is small in all northwestern samples and great in
southeastern samples.

The pattern of variation is complicated by specimens not measured from
localities in west-central Texas. Three out of four specimens before me from
Sand, Dawsen County, Texas, are small and with completely white elytra
(like specimens from locality 13 in Hudspeth County, Texas); the fourth is
larger and similar to specimens from east-central New Mexico or south-
western Oklahoma. Of six specimens from Potash Lake, Gaines County,
Texas, one has very wide markings, with only a narrow sutural band of the
elytra not white; the others are more “normal.”

Another characteristic of many southwestern populations of this species is
the occurrence of occasional individuals with elytral basal dots (Fig. 91).
Sometimes the dots are quite large and connect to the marginal maculation,
but usually they are poorly developed and isolated. This has been noted in
most samples from New Mexico, western Texas, and southwestern Okla-
homa; and it even occurs as far northeast as southwestern Kansas and
El Dorado, in east-central Kansas.

A clear division of the localities into two groups is shown in the above
analysis, a coastal and eastern Texas form, C. t. togata, best characterized by
the great retraction of the female elytral spine; and the northwestern popula-
tions, in which this spine is only slightly retracted. Within the latter group,

BroNoMIcs AND ZOOGEOGRAPHY OF TIGER BEETLES 293

w. Sipe setae
:

color
retr. 2 spine OB

w.macul. w. setae
50-74 0.0170-0.0184
75-99 0.0185- 0.0199
Dy : R=! 100-124 0.0200-0.0214
Dy 125-149 0.0215- 0.0229
I

150+ 0.0230-0.0244

ALBERS PROJECTION

Fic. 159. Geographic variation of four characters of C. togata: width of the maculation,
width of pronotal setae, color, and retraction of the elytral spine of the female. Values for the
first three characters have been pooled for the sexes and represented by different degrees of
shading (upper right). Values for the fourth character are beneath each circle. The positions
of the characters in the circles are shown at upper left.

a gradual cline exists in most characters measured, running from northeast
to southwest, with no clear breaks or steps (or at least none in which there
is any sort of character concordance). The type localities of the two valid
named forms in this group are Clark County, Kansas, and Santa Rosa, New
Mexico. Specimens from these areas are more similar to one another than
are those from northern Kansas and Clark County or those from western
Texas and Santa Rosa. Thus, only the earlier of these two names, C. ¢.
globicollis, should be used. The form with completely white elytra from
Hudspeth County, Texas, is certainly distinct enough to be called a sub-
species, but as noted above, some specimens from west-central Texas are
identical to or closely approach it; and intervening populations are more
nearly “normal.” Therefore, it does not seem advisable to recognize more
than two subspecies within this species (Fig. 152). The two (female) speci-
mens that I have seen from Pecos, Reeves County, Texas, appear to be inter-
mediate between the two subspecies. One has a greatly retracted elytral spine
and is greenish. The other has a slightly retracted spine and is cupreous. The
maculation is fairly wide in both specimens.

C. togata probably evolved within its present range. It has no close rela-

294 Tue University ScrENcE BULLETIN

tives, so it is difficult to say whether it was first a coastal or an inland species.
It eventually became widespread, evidently, in both coastal and inland
habitats in the southern United States. Differentiation into races probably
began in the form of a cline along the Rio Grande valley. The drier climates
of late Tertiary or interglacial Pleistocene times no doubt exterminated many
of the intermediate populations of the cline, resulting in evolution into the
modern coastal and inland subspecies. Dispersal of the inland form to the
northeast probably occurred during the Pleistocene. It could not have reached
the Lincoln, Nebraska, area until after the Nebraskan glaciation.

The pattern of maculation and color variation seem clearly to be a result
of selection to match the substrate color. In the drier Southwest, where saline
habitats are more often covered by a white crust of salt, the width of the
maculation is greatest. Populations with color values around 4 (cupreous)
are found in the area with red soil (Figs. 159 and 108). Dark soils are
prevalent in most other areas, and the beetles there are also dark. In some
southwestern habitats, however, the soil is light gray, nearly white. Here, the
elytra of C. togata are completely white or nearly so. The presence of the
basal dot seems to be evolving in southwestern populations; it has the effect
of increasing the amount of white on the elytra. Increased width of body
setae also has the effect of making the beetle appear more white, a trend that
has been noted in southwestern populations.

C. willistoni

This is a western species, most common in the Great Basin and Mojave
Desert, and reaching its eastern limit in Kansas and Oklahoma (Fig. 160).
Its distribution by county or state is the following:

KANSAS: Stafford; OKLAHOMA: Alfalfa, Beckham, Woods; TEXAS: Andrews; NEW
MEXICO: Roosevelt, Torrance, Valencia; WYOMING: Albany, Carbon, Uinta; UTAH:
Beaver, Box Elder, Cache, Davis, Emery-Rand County line, Iron, Juab, Millard, Salt Lake,
Tooele, Utah, Washington; ARIZONA; NEVADA: Churchill, Lyon, Nye, Storey, Washoe:
CALIFORNIA: Alameda, Inyo, Kern, Lassen, Los Angeles, Orange, Plumas, San Bernardino;
OREGON: Harney, Lake. Town that could not be located: Farr Post (—=Farrwest, Weber
County ?), Utah. Doubtful localities: Kellwood, Manitoba; Colorado Springs, El Paso County,
Colorado. The Los Angeles County (Seal Beach), Alameda Co. and Orange Co. (Irvine),
California, localities (listed above) are somewhat questionable.

The following forms have been described within this species:

Cicindela willistoni Leconte, 1879. Type locality: Lake Como, Wyoming Territory.

Cicindela echo Casey, 1897. Type locality: Great Salt Lake, Utah.

Cicindela pseudosenilis W. Horn, 1900. Type locality: Owens Lake, Inyo County, California.

Cicindela echo amedeensis (emendation of C. e. amadeensis Casey, 1909). Type locality: Ame-
dee, California. This form is clearly named for the type locality, but in the original descrip-
tion, the name is apparently misspelled.

Cicindela spaldingt Casey, 1924. Type locality: Callao, Utah.

Cicindela willistoni amargosae Dahl, 1939. Type locality: 4 miles north of Furnace Creek, Inyo
County, California. This form is now recognized as a separate species.

BIoNoMIcs AND ZOOGEOGRAPHY OF TIGER BEETLES 295

Fic. 160. Known distribution of C. willistont. Shaded square=C. w. willistoni, shaded
circle=C. w. echo, shaded triangle=C. w. pseudosenilis, open triangle=C. w. praedicta, open
square=C. w. estancia, open circle=C. w. hirtifrons, half shaded circle=population of uncertain
status, star=state record.

Cicindela willistoni praedicta Rumpp, 1956. Type locality: 3.5 miles south of Shoshone, Inyo

County, California.

Cicindela willistont estancia Rumpp, 1961. Type locality: 7 miles east of Willard, Torrance

County, New Mexico.

In recent years, five subspecies have been recognized: C. w. willistoni, the
reddish brown form with marginally expanded markings; C. w. echo, the
brownish form with usually “average” markings; C. w. pseudosenilis, the
blue or blue-green form, sometimes with expanded markings; C. w. praedicta,
the small blue or blue-green form with reduced markings or none at all; and
C. w. estancia, the feddish brown form with markings so expanded, the
elytra are nearly completely white.

In the study of geographic variation, the following characters were meas-
ured: (1) length of left elytron; (2) width of left elytron; (3) width of
labrum; (4) length of labrum, including tooth; (5) shape of base of middle
band in arbitrary units (Fig. 161); (6) smallest width of transverse portion
of middle band (Fig. 162); (7) number of setae on vertex and frons, except

296 Tue University SciENCE BULLETIN

7
=
=“
a?

I62

Fics. 161-162. Illustrations of certain characters measured on elytra of C. willistont; Fic.
161, arbitrary units for shape of base of middle band; Fic. 162, smallest width of transverse
portion of middle band.

the several supraorbital sensory setae near the medial margins of the eyes;
(8) color, using color wheel.

Specimens from seven localities were measured. The localities and sample
sizes are given in Table 44. Sample sizes are adequate except for females
from Oregon (specimens from Oregon were lumped into one sample) and
Kern County, California. Unfortunately, these seven localities do not en-
compass all the recognized forms mentioned above; no specimens of C. w.
estancia were available.

The means for the characters and localities are given in Tables 45 and 46.
Analyses of variance of the individual characters showed that there are
significant differences (p < 0.01) among the means of all characters in males
and all characters in females except number 1, which is significant at the 5°%
level. Pooled within-locality variance-covariance matrices are given in Tables
47 and 48. The sums of the among-locality variance components are 218.88
for males and 429.67 for females. The first four components for males and
the first three for females are highly significant (p < 0.01), and the fourth
component for females is significant at the 59% level. The first three functions
account for 96.66°/, of the variance (among localities relative to that within)

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 297

Taste 44. Localities and sample sizes of the specimens of C. willistoni measured.

N

Locality 3 2
IPRRRCAINGAS s Stattorda@os milslamaiy ine Oee Wehr SOM 23 eco sac 2 eee aso ee on neers vecccecieeee 12 12
DEE Vavi© MIN Gon Garbonmco= Comon Wake je ee a ee 12 12
3) UPAR; Salt Lake Co., vicmity of Salt Lake City and Saltair 12. 12
4. OREGON, Lake Co., Rest Lake, Summer Lake; Harney Co., Alvord Hot Springs 8 3
5a GALTEORNIAG Inyo) (Conn @lancho: (Owens Ie2)) Sees see ce eeeceeee pea ae 12 12
Cte OAR ORNDARs Kern: Gores altclall eyo e te ee ee ee ene eee 12 +}
J. (GINUTZORINIAS lintie) Co, S55) Sore Gs Wik Coto nolo eee ee ae ae, 12 12

Tape 45. Means of seven localities (see Table 44) and seven characters for males
of C. willistont. Values for characters 1-4 are in mm.

Character
Locality 1 2 3 4 5 6 i
1 7.61 73) 1.82 0.93 3.98 0.59 31.0
2 7.66 2.39 1.82 0.94 6.09 1.38 22D
3 Tes | 2:31 1.74 0.81 4.18 0.80 Df
4 7.96 2°39 1.87 0.88 Bai) > 0.85 5.6
5 7.69 2.30 1.82 0.88 3.50 0.86 6.2
6 7.63 2.33 1.81 0.88 4.79 iS) 4.8
i 7.96 2.03 1.63 0.79 11237 0.15 ee.

Taste 46. Means of seven localities (see Table 44) and seven characters for
females of C. willistoni. Values for characters 1-4 are in mm.

Character
Locality 1 2 3 4 5 6 7
l 7.84 2.59 1.84 0.96 4.01 0.56 38.3
2 7.84 2.64 1.85 1.00 5.99 1.28 28.6
3 7.64 253 1.81 0.86 3.54 0.79 6.8
4 7.58 2.46 1.80 0.89 3.63 0.74 6.3
5 7.80 2.50 1.85 0.91 2.88 0.77 5.8
6 7.82 2.56 1.87 0.92 4.98 1.23 5.3
7 Le 2.26 1.70 0.85 1.00 0.00 5.8

in males and 98.80°% in females. The distributions of the means in the first
three discriminant functions are shown in Figures 163 and 164. Localities
3-6 form a loose cluster (especially in males), while the others are widely
scattered.

The results of the simultaneous testing procedure show that all combina-
tions of pairs of means are significantly different at the 59% level except 4 vs.
5 for males and 3 vs. 4 for females.

298 Tue UNiversiry ScrENCE BULLETIN

Taste 47. Pooled within-locality variance-covariance matrix for males of C.
willistoni. Variances are along the diagonal, and covariances compose the rest of
the matrix. Values have been multiplied by 10-4; thus “1082”—0.1082.

Character

1 2 3 4 5 6 7

| 1082 286 246 148 540 224 1582

2 106 72 43 205 80 391

ol is) 69 40 113 63 427
SB 4 31 90 36 103
& 5 2521 580 —474
6 341 —508

7 184355

Taste 48. Pooled within-locality variance-covariance matrix for females of C.
willistoni. Variances are along the diagonal, and covariances compose the rest of
the matrix. Values have been multiplied by 10-4; thus “1748”—0.1748.

Character

1 2 3 4 5 6 7

l 1748 569 394 268 208 189-8 2328

2 235 136 93 45 58 523
sere: 105 67 64 4] —462
a 4 50 39 28 —361
= 5 1060 219 1186
6 183 —681

7 193125

Sets of vectors (Figs. 165 and 166), plotted for the first two functions,
show geographic trends of the seven characters when compared with Figures
163 and 164. Specimens with high values in Ke and low values in Ki
(localities 1 and 2) have a long, narrow labrum, while specimens from other
localities have wider labra. Specimens low in Ki and Ke (locality 7) have
long elytra and narrow markings, while specimens high in Ke (localities 3-6)
have wide elytra and broad markings.

The results of the color analysis are shown in Figure 167. Samples from
Oregon and Utah consist mostly of dark purplish to dark red-green (appears
brown-green) individuals. The southwesternmost sample in California is
similar but contains many green individuals. The other two California
samples contain mostly blue to blue-green individuals. The samples from
Wyoming and Kansas contain many individuals with brighter colors, rang-
ing from purplish to reddish to red-green (and green in Wyoming).

These analyses confirm the existence of four of the above five recognized
subspecies (the fifth, C. w. estancia, is clearly distinct because of its extremely
wide markings). Localities 2 (C. w. willistont) and 7 (C. w. praedicta) are

BroNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 299

lo4

Fics. 163-164. Distributions of the means of the seven localities for males (Fig. 163) and
females (Fig. 164) of C. willistoni in the first three discriminant functions (Ki, Ks, Ks), drawn
as three-dimensional models, with numbered balls representing the means and vertical supports
arising from the Ki, Ke surface. Different sized balls indicate different distances from the viewer.
The models are viewed from opposite directions, relative to the clusters.

widely separate from all other in the statistical analysis. Locality 5 (C. w.
pseudosenilis ) is quite similar to localities 3, 4, and 6 (C. w. echo) except in
color. Specimens from Kern County, California (locality 6, Saltdale, also
Mojave) differ from other populations of C. w. echo in having the middle
band nearly always very wide in the transverse portion, sometimes broadly
confluent with the humeral lunule. In other populations of C. w. echo, only
occasional individuals exhibit this tendency. Also many (about 1794) of the
Kern County specimens are dark green or blue-green. On the basis of these
characters and their geographical isolation, they could probably be separated
as a subspecies, although this will not be done at present. The Kansas sample

Fics. 165-166. Vectors for the seven characters of males (Fig. 165) and females (Fig. 166)
of C. willistoni for the first two discriminant functions. Each vector shows the change in the
discriminant function that the corresponding character would produce if it varied independently.
Units are same as in Figs. 163-164.

Fic. 167. Result of color analysis of C. willistoni, using the color wheel. Different degrees
of shading indicate different percentages of the sample. The positions of the three major colors
are shown at top center. Numbers show the sample size for each locality.

BroNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 301

Fic. 168. Cicindela willistoni hirtifrons, new subspecies; dorsal aspect of male (setae
omitted).

was widely separated from the others by the statistical analysis; it represents
what seems to be a clearly defined new subspecies, described below:

Cicindela willistoni hirtifrons, new subspecies (Fig. 168)

Head: Labrum with single median tooth, length usually more than half
the width; antennal scape with 10-25 stout erect setae; clypeus and genae
glabrous; frons and vertex with 16-51 fine erect setae medially, some long,
some short (not including several pairs of sensory setae near medial margins
of eyes); Thorax: pronotum glabrous medially, with long, erect to partly
decumbent setae laterally; proepisternum, proepimeron, procoxae, mesepi-
meron, mesocoxae, metaepisternum, and lateral parts of metaepisternum and
metacoxae with dense, long to medium erect setae; mesepisternum with a

302 Tue UNiversiry ScrENCE BULLETIN

few erect setae; mesosternum and medial portions of metasternum and
metacoxae with sparse short setae; prosternum glabrous; lateral margins of
pronotum subparallel, diverging slightly anteriorly; Abdomen: venter with
dense to sparse decumbent to erect setae; Elytra: male, gradually widened
to one-half to two-thirds their length, then gradually narrowed to apical fifth,
then abruptly narrow to rounded apex; female, margins much expanded at
middle from basal fourth to apical third, then rounded to apex; posterior
margin microserrulate; spine apical to slightly retracted; markings similar to
those of typical C. w. echo, but middle band often narrower; humeral lunule
and middle band usually connected by narrow marginal expansion of middle
band; apical lunule and middle band not connected; surface shiny or greasy-
appearing; Color: dorsum and front of head bronze or cupreous, with areas
of green and blue; genae blue and green; lateral portions of thorax cupreous,
green, and blue; venter green to purplish blue; pronotum bronze or cupreous
with depressions green and blue; elytra between markings cupreous to
purple-cupreous to bronze to greenish bronze, sometimes quite green when
viewed from an oblique angle.

Type locality: Big Salt Marsh, 11 mi. N.E. of Hudson, Stafford Co.,
Kansas. Holotype male, allotype female, and 15 paratopotypes (11 males,
four females), 7 April 1965 (Harold L. Willis) in the Snow Entomological
Museum, University of Kansas. Ninety-eight paratopotypes, 9 April 1964,
7 April 1965, 21 June 1965 (Harold L. Willis): ten in the U.S. National
Museum; ten in the American Museum of Natural History; 15 in the collec-
tion of N. L. Rumpp; five in the collection of G. C. Gaumer; four in the
collection of J. Stamatov, Armonk, New York; three each in the collections
of R. Freitag, R. C. Graves, R. L. Huber, and J. K. Lawton; two in the collec-
tion of J. F. Payne, and the rest in the author’s collection. Eight paratopo-
types, 23 May 1965 (Paul E. Slabaugh), in the collection of P. E. Slabaugh.

Fourteen paratypes: OKLAHOMA, 2.5 mi. S.W. of Plainview, Woods
Co., 3 May 1964, 3 June 1963, 8 June 1965, five specimens in the author’s
collection; 3 mi. E. of Cherokee, Alfalfa Co., 11 April 1931, 4 June 1963,
7 June 1931, 11 June 1931, 15 June 1935, eight specimens, five in the University
of Oklahoma, one in the U.S. National Museum, one in the Museum of
Comparative Zoology, Harvard University, and one in the author’s collec-
tion; NEW MEXICO, 4 mi. E., 1 S. of Arch, Roosevelt Co., 9 June 1965,
one specimen in the author’s collection.

Distribution: Central Kansas, western Oklahoma, west-central Texas,
east-central New Mexico, and possibly west-central New Mexico (Fig. 160).

Diagnosis: Differs from C. w. willistoni and C. w. estancia in narrower
markings, with the humeral lunule and middle band separate or narrowly
connected, not broadly confluent; from all other subspecies by the large
number of medial setae on the frons and vertex (more than 15, rather than

BIoNOMICS AND ZOOGEOGRAPHY OF TIGER BEETLES 303

10 or fewer), the relatively longer labrum (width/length ratio usually less
than 2.0, rather than more than 2.0), and the generally reddish brown dorsal
color, rather than (usually) dark brown, greenish, or blue. This subspecies
shares the very setose head, longer labrum, and reddish dorsal color with
C. w. willistoni and C. w. estancia.

Four out of 24 or 16.79% of specimens of C. w. hirtifrons have a labral
width/length ratio of 2.0 or greater. Out of 99 specimens representing C. w.
echo, C. w. pseudosenilis, and C. w. praedicta, 17 or 17.2°/%, have a labral
width/length ratio of less than 2.0; however, these values range from 8.4°%,
for C. w. echo to 29.1% for C. w. praedicta.

The one specimen from Roosevelt County, New Mexico, has quite wide
markings, the apical lunule and middle bands nearly being connected at the
margin. The Andrews County, Texas, record is based on a second instar
larva. About half the specimens from Oklahoma have slightly wider mark-
ings than most Kansas specimens. The record from Beckham County, Okla-
homa, is based on Ortenburger and Bird (1933); no specimens have been
examined from there. Only about 3°/ of the Kansas specimens have so much
green on the dorsum that they appear green-brown; however, it is a brighter
green than occurs in western subspecies. About 49% of the Kansas specimens
are a dark brown and might be confused with C. w. echo, but the number
of head setae easily separates them. All the Oklahoma specimens and the
New Mexico specimen are cupreous-brown, with no indication of green.
Occasional individuals of C. w. echo are reddish brown or reddish green;
however, as mentioned, the number of setae on the frons and vertex com-
pletely separates the two forms (if the head setae have been rubbed off, one
can find what their approximate number was by counting the punctures
from which they arose). The exact status of the population at Grants, Valen-
cia County, New Mexico, is not known. No specimens were available for
study; however, N. L. Rumpp (7 litt.) said that they are similar to the
Kansas specimens.

I do not know the subspecies of the population in Arizona (Fig. 160) be-
cause I have seen no specimens from there. Two specimens reputedly from
Orange County, California, that are dark blackish and have fairly wide
markings are being called C. w. echo for the present. One specimen labelled
Alameda County and two from Los Angeles County, California, are typical
C. w. pseudosenilis.

Wickham (1904a, b) thought that C. w. echo arose within the Great
Basin, that C. w. pseudosenilis has been isolated at Owens Lake, California,
since at least early Pleistocene, and that C. w. willistoni was separated from
the other forms of the species (known to him at that time) by the rising
mountains of late Tertiary. Rumpp (1961) postulated that C. qwillistoni arose
from an ancestor that lived in northern North America in the warm Creta-

5

304 Tue University ScrENCE BULLETIN

ceous. As climates cooled during the Tertiary, the ancestral species moved
south and became widespread from Colorado to the Pacific coast, gradually
assuming the character of C. willistoni. He stated that after its formation,
C. willistoni evolved only slightly, coming through the Miocene “in its
present form.” The rising mountains of the Cascadian Revolution isolated
populations in the Great Basin, Wyoming, and New Mexico. The various
subspecies began evolving during the Pliocene or earlier.

Wickham’s and Rumpp’s theories on the evolution of C. willistoni seem
quite plausible, although it is hard to imagine that little or no evolution has
occurred since the Miocene. The selective forces that acted to produce the
different geographic races (matching the color of the substrate seems to be
an important one, climatic conditions are another) would seem to be still
operating. Wickham and Rumpp did not know of the occurrence of C.
willistoni in the central United States, and neither mentioned that popula-
tions from east of the Rocky Mountains have quite setose heads, while those
in the West are sparsely setose (Wickham noted a difference, but only in
passing). The closest relatives of C. willistoni (it has no extremely close
relatives) are C. senilis and the C. tranquebarica complex, all of which have
very setose heads. Thus, I conclude that the primitive condition in C. willis-
toni is a densely setose head, and that the western forms are losing this char-
acter. The markings of the ancestral species were probably much like those
of C. w. echo or C. w. hirtifrons, from which expansions or reductions
evolved. The primitive color was most likely brownish; populations of C. w.
pseudosenilis still contain occasional brown or half brown individuals, hint-
ing of its origin from a C. w. echo-like ancestor. During the late Tertiary,
C. w. echo must have been distributed nearly throughout the Great Basin;
in fact, it transgressed into eastern Utah and southwestern and south-central
Wyoming, leaving relict populations (Fig. 160). During dry interglacials,
many populations were no doubt exterminated, leaving large gaps in the
range of C. w. echo, particularly in eastern Nevada. C. w. hirtifrons and
C. w. estancia probably evolved from a common ancestor (the same one that
left populations in Wyoming which became C. w. willistoni) that moved
south into New Mexico in the Tertiary and then dispersed northeastward
through Texas and Oklahoma into Kansas. A population was “trapped” in
the Estancia Valley of central New Mexico and became the very widely
maculate C. w. estancia, while C. w. hirtifrons retained a “normal” macula-
tion. The soil in the habitat of the former is tan and sandy, and is usually
covered by a white alkaline crust. In the range of C. w. hirtifrons, soils are
usually red clay or sand, and the climate is moister (especially in the eastern
part). Thus, these forms have apparently evolved toward a better match of
their substrate.

BronoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 305

GENERAL PATTERNS

The seven species discussed exhibit certain general patterns within the
central United States. The most striking are the increase in red color (in all
but C. fulgida) in the region of red soils, and the increase of white macula-
tion in the drier western regions, both of which have been discussed repeat-
edly above. Another is the distinctness of the populations in the Lincoln,
Nebraska, vicinity. Four of the species have more or less distinct forms
occurring there: the small, always reddish form of C. circumpicta, the black
form of C. fulgida, the narrowly maculate C. nevadica lincolniana, and a
form of C. togata in which the edges of the white maculation are brownish
and indistinct (occasional specimens from northern Kansas show this also).
As mentioned earlier, I think this means that these populations survived the
last one or two Pleistocene glaciations at this locality, while other populations
of their species were driven south or exterminated.

FAUNAL RELATIONSHIPS

The 13 species in this study are divisible into two groups, reflecting their
principal distribution and phylogeny: 1) the northern species; these are the
more primitive species that are adapted to cooler climates (most active in the
spring and fall in the central United States), and that have primarily northern
distributions; 2) the southern species; these include the more advanced
species that are adapted to warm climates (active in the summer in the cen-
tral United States), and that have primarily southern distributions. The
northern species are C. duodecimguttata, C. fulgida, C. hirticollis, C. repan-
da, C. tranquebarica, and C. willistoni. The southern species are C. circum-
picta, C. cuprascens, C. macra, C. nevadica, C. punctulata, C. schauppi, and
C. togata. The probable evolutionary relationships of these species were
discussed in the section on phylogeny.

SUMMARY

1. Aspects of the bionomics and zoogeography of 13 species of Cicindela
(Coleoptera: Cicindelidae) living in saline habitats of the central United
States (southern Nebraska, western Missouri, Kansas, and Oklahoma) were
studied.

2. A review was made of the published works concerning the bionomics
of Cicindela. By watching beetles in the field and rearing them in the Jab-
oratory, new knowledge was obtained on such subjects as oocyte develop-
ment, gross embryology, length of stadia, pupal development, post-emergence
changes of the adult, feeding of larvae and adults, mating behavior, oviposi-
tion, and ecological relationships. The larvae of four species were described

5

306 Tue University ScrENCE BULLETIN

for the first time (C. fulgida, C. nevadica, C. togata, and C. willistoni) and
one was redescribed (C. circumpicta ). The species show a tendency to avoid
competition by inhabiting different microhabitats and by being active at
different times of the year; however, there is much overlap. Adaptations for
living in saline habitats were noted; most are possessed by species not in-
habiting such areas, and many of the species in this study are more common
in nonsaline habitats.

3. The zoogeography of seven species was studied for their entire ranges
(C. circumpicta, C. cuprascens, C. fulgida, C. macra, C. nevadica, C. togata,
and C. willistoni). Geographic variation of morphological characters was
studied using generalized discriminant functions. The results were used to
help confirm or reject the existence of subspecies. The subspecies C. circum-
picta salinae and C. togata fascinans were rejected. One new subspecies,
C. willistont hirtifrons, was described.

4. Using the patterns of geographic variation and evidence from past
geological history, hypothetical schemes of evolution and dispersal were
proposed. An important selecting pressure acting on all species but C. fulgida
seems to be increasing the resemblance of the dorsum of the adult to the
color of the substrate. In regions having red soil, populations of beetles show
a pronounced tendency toward reddish colors. In the drier western parts of
their ranges, many species have an increased amount of white on the body,
particularly the white elytral markings. This seems to result from the fact
that in these areas, saline habitats are more often covered by a crust of white
crystalline salts; in moister areas, the salts are more often dissolved and the
color of the soil is apparent.

5. It is suggested that the Lincoln, Nebraska, vicinity was a refuge for
at least five species during the late Pleistocene glaciations.

LIT eRATURE SC 2D

ALLEE, W. C., A. E. Emerson, O. Park, T. Park, ann K. P. Scumipr. 1949. Principles of ani-
mal ecology. W. B. Saunders Co., Philadelphia. 837 p.

AxeLrop, D. I. 1950. Studies in late Tertiary paleobotany. Publ. Carnegie Inst. No. 590:1-323.

Baatman, R. J. 1965. Vegetation of the Salt Plains National Wildlife Refuge, Jet, Oklahoma.
Ph.D. Thesis. Univ. Oklahoma. 138 p. Univ. Microfilms. Ann Arbor, Michigan.
(Dissertation Abstr. 26:1302).

BaLpuF, W. V. 1925. The feeding of a common tiger beetle (Coleop., Cicindelidae). Ent. News

36:275-276.

. 1935. Bionomics of entomophagous Coleoptera. New York. 220 p.

BarKA, T., AND P. J. ANDERSON. 1963. Histochemistry, theory, practice, and bibliography. Harper
and Row, Inc., New York. 660 p.

Batra, S. W. T. 1965. Organisms associated with Lasioglossum zephyrum (Hymenoptera:
Halictidae). J. Kans. Ent. Soc. 38:367-389.

Buair, K. G. 1920. Cicindela germanica L. and its larva. Ent. Monthly Mag. 56:210-211.

BuaIsDELL, F. E. 1912. Hibernation of Cicindela senilis (Coleop.). Ent. News 23:156-159.

Biancnarp, F. N. 1921. The tiger beetles (Cicindelidae) of Cheboygan and Emmet Counties,
Michigan. Pap. Michigan Acad. Sci., Arts and Letters 1:396-417.

Buickie, R. L. 1958. Notes on Aegialomyia psammophila (O. S.) (Tabanidae: Diptera).
Florida Ent. 41:129-131.

BIoNoMics AND ZOOGEOGRAPHY OF TIGER BEETLES 307

Bouwman, B. E. 1909. Uber die Lebensweise von Methocha ichneumonides Latr. Tijdschr. voor
Ent. 52:284-294.

Branpt, E. K. 1888. Observations on insect anatomy. [In Russian] Horae Soc. Ent. Rossicae
22 :133-137.

Braun, E. L. 1947. Development of the deciduous forests of eastern North America. Ecol.

Monogr. 17:211-219.

. 1955. The phytogeography of unglaciated eastern United States and its interpretation.

Botan. Rev. 21:297-375.

BRIDWELL, J. C. 1912. Untitled note. Im forty-sixth meeting of the society, December 3, 1911.
Proc. Pacific Coast Ent. Soc. 1:46-48.

BroMteEy, S. W. 1914. Asilids and their prey. Psyche 21:192-198.

Bryson, H. R. 1939. The identification of soil insects by their burrow characteristics. Trans.
Kans. Acad. Sci. 42:245-253.

Burpvick, D. J., anp M. S. WassauER. 1959. Biology of Methocha californica Westwood (Hy-
menoptera: Tiphiidae). Wasmann Jour. Biol. 17:75-88.

Caper, E. E. 1922a. New Cicindelas of the fulgida group. (Coleop.). Canadian Ent. 54:62.

.1922b. Change of names in Cicitndela. Canadian Ent. 54:191.

CarpPENTER, F. M. 1953. The geological history and evolution of insects. Am. Sci. 41:256-270.

Casey, T. L. 1897. Coleopterological notices. VII. Ann. New York Acad. Sci. 9:285-684.

. 1909. Studies in the Caraboidea and Lamellicornia. Canadian Ent. 41:253-284.

. 1913. Studies in the Cicindelidae and Carabidae of America. Mem. Coleop. 4:1-192.

. 1914. Studies in Omus and Cicindela. Mem. Coleop. 5:1-24.

. 1916. Further studies in the Cicindelidae. Mem. Coleop. 7:1-34.

. 1924. Additions to the known Coleoptera of North America. Mem. Coleop. 11:1-347.

Cazier, M. A. 1939. Two new western tiger beetles, with notes (Coleoptera-Cicindelidae). Bull.

Brooklyn Ent. Soc. 34:24-28.

. 1948. The origin, distribution, and classification of the tiger beetles of Lower California.

Am. Mus. Novitates. No. 1382:1-28.

. 1954. A review of the Mexican tiger beetles of the genus Cicindela (Coleoptera, Cicin-

delidae). Bull. Am. Mus. Nat. Hist. 103:227-310.

Cuampion, H. G. 1915. Addendum to observations on the life history of Methocha ichneu-

monides Latr. Ent. Monthly Mag. Ser. 3, 1:40-42.

, AND R. J. CHampion. 1914. Observations on the life-history of Methocha ichneumontdes,

Latr. Ent. Monthly Mag. Ser. 2, 25:266-270.

Cuaney, R. W. 1947. Tertiary centers and migration routes. Ecol. Monogr. 17:139-148.

CuHapMan, R. N., C. E. MickeL, J. R. Parker, G. E. MILuter, anD E. G. Kettey. 1926. Studies
in the ecology of sand dune insects. Ecology 7:416-427.

CuHapmMan, V. J. 1960. Salt marshes and salt deserts of the world. Interscience, New York.
392 p.

Cuauporr, M. pe. 1843. Mémoires sur quelques genres et espéces de la famille des Carabiques.
(from Bull. Soc. Imperiale Nat. Moscou, 1837, 1838, 1842), Moscow.

CHRISTIANSEN, K. 1958. Geographic variation and the subspecies concept in the collembolan
Entomobryoides guthriet. Systematic Zool. 7:8-15.

CockERELL, T. D. A. 1920. Eocene insects from the Rocky Mountains. Proc. U.S. Nat. Mus.,
Washington 57:233-260, pls. 32-36.

Crippte, N. 1907. Habits of some Manitoba “tiger beetles” (Cicindela). Canadian Ent. 39:

105-114.

. 1910. Habits of some Manitoba tiger beetles (Cicindelidae) No. 2. Canadian Ent.

42:9-15.

. 1919. Popular and practical entomology. Fragments in the life-habits of Manitoba

insects. Canadian Ent. 51:97-101.

Crowson, ee 1946. Distribution and phylogeny of Cicindelidae (Col.). Ent. Monthly Mag.
82:278.

Dani, R. G. 1939. A new California tiger beetle (Coleoptera-Cicindelidae). Bull. Brooklyn
Ent. Soc. 34:221-222.

DaruincTon, P. J., Jr. 1957. Zoogeography: the geographical distribution of animals. John

Wiley & Sons, Inc., New York. 675 p.

. 1965. Biogeography of the southern end of the world. Harvard Univ. Press, Cam-

bridge, Massachusetts. 236 p.

Davis, L. V. 1962. Insects of the herbaceous strata of salt marshes in the Beaufort, North
Carolina, area. Ph.D. Thesis. Duke Univ. 259 p. Univ. Microfilms. Ann Arbor,
Michigan. (Dissertation Abstr. 24:2191-2192.)

Davis, W. T. 1910. Miscellaneous notes on collecting in Georgia. J. New York Ent. Soc. 18:
82-85.

308 Tue UNiversiry ScrENcE BULLETIN

. 1921. Cicindela tranquebarica and its habits. Bull. Brooklyn Ent. Soc. 16:111.

DeBacu, P. 1966. The competitive displacement and coexistence principles. Ann Rev. Ent.
11:183-212.

Deevey, E. S., Jr. 1949. Biogeography of the Pleistocene. Part I: Europe and North America.
Bull. Geol. Soc. Am. 60:1315-1416.

Diercxx, F. 1899. Les glandes pygidiennes des Staphylinides et des Cicindélides. Zool. An-

zeiger 22:311-315.

.1901. Les glandes pygidiennes des Coléoptéres, seconde memoir. Carabides (bom-

bardiers, etc.), Paussides, Cicindélides, Stephylinides. La Cellule 18:255-310, pls. 1-3.

Ditton, L. S$. 1956. Wisconsin climate and life zones in North America. Science 123:167-176.

Dorr, E. 1960. Climatic changes of the past and present. Am. Sci. 48:341-364.

Dow, R. P. 1916. Plaster-casting insect burrows. Psyche 23:69-74, pl. X.

Dungar, C. O. 1960. Historical geology. Second ed. John Wiley & Sons, Inc., New York.
500 p.

Durrant, S. D. 1955. In defense of the subspecies. Systematic Zool. 4:186-190.

EcxuorF, D. E. 1939. Cicindelidae of Iowa (Coleoptera). Iowa St. Coll. J. Sci. 13:201-230.

Epwarps, J. G. 1954. A new approach to infraspecific categories. Systematic Zool. 3:1-20.

Empen, F. I. van. 1935. Die Larven der Cicindelinae I. Einleitendes und alocosternale Phyle.

Tijdschr. voor Ent. 78:134-183.

. 1943. Larvae of British beetles. IV. Various small families. Ent. Monthly Mag. 79-209-

223, 259-270.

Enock, F. 1903. The life history of Cicindela campestris. Proc. Ent. Soc. London 1903:xv-xix.

Evans, G. O., J. G. SHEALS, AND D. MacrarLaNne. 1961. The terrestrial Acari of the British
Isles. Vol. 1. Introduction and biology. London. 219 p.

Evans, M. E. G. 1965. The feeding method of Cicindela hybrida L. (Coleoptera: Cicindelidae).
Proc. Roy. Ent. Soc. London, A. 40:61-66.

Fackier, H. L. 1918. The tiger beetles of Kansas (family Carabidae; subfamily Cicindelinae)
order Coleoptera. M.S. Thesis. Univ. Kansas. 51 p., 5 pls.

FeNpeR, K. 1945. Insect dispersal by the U.S. mails. Bull. Brooklyn Ent. Soc. 40:66.

Fircu, A. 1856. Third report on insects, supplement. Trans. New York St. Agric. Soc. 16:315-
490.

Fircu, H. S. 1963. Observations on the Mississippi Kite in southwestern Kansas. Univ. Kansas
Publ. Mus. Nat. Hist. 12:503-519.

Fox, H. 1910. Observations on Cicindelidae in northern Cape May County, N.J., during the
summers of 1908-09. Ent. News 21:75-82.

Freirac, R. 1965. A revision of the North American species of the Cicindela martima group
with a study of hybridization between Cicindela duodecimguttata and oregona. Quaes-
tiones Ent. 1:87-170.

Frick, K. E. 1957. Biology and control of tiger beetles in alkali bee nesting sites. J. Econ. Ent.
50:503-504.

Friepricns, H. F. 1931. Beitrage zur Morphologie und Physiologie der Sehorgane der Cicin-
delinen (Col.). Zeitschr. Morph. Okol. Tiere 21:1-173.

Frye, J. C., anp A. B. Leonarp. 1952. Pleistocene geology of Kansas. Bull. 99, St. Geol. Surv.
Kansas. 230 p.

Geicer, R. 1965. The climate near the ground. (Transl. from 4th German ed.) Harvard Univ.
Press, Cambridge, Massachusetts. 611 p.

GittHaM, N. W. 1956. Geographic variation and the subspecies concept in butterflies. Sys-
tematic Zool. 5:110-120.

Gityarov, M. S., anp I. Ku. SHarova. 1954. Larvae of tiger beetles (Cicindelidae). [In Rus-
sian] Zoologicheskii Zhurnal 33:598-615.

GosLinE, W. A. 1954. Further thoughts on subspecies and trinomials. Systematic Zool. 3:92-94.

Goutp, C. N., pir. 1930. Oil and gas in Oklahoma. Bull. 40, Oklahoma Geol. Surv. 3:1-663.

Granpi, G. 1951. Introduzione allo studio dell’ entomologia. Vol. I]. Endopterigoti. Bologna.
13325 p:

Graves, R. C. 1962. Predation on Cicindela by a dragonfly. Canadian Ent. 94:1231.

. 1963. The Cicindelidae of Michigan. Am. Midl. Nat. 69:492-507.

GressitT, J. L. 1958. Zoogeography of insects. Ann. Rev. Ent. 3:207-230.

Go.tpsmitH, W. M. 1916. Field notes on the distribution and life habits of the tiger beetles
(Cicindelidae) of Indiana. Proc. Indiana Acad. Sci. 26:447-455.

HacMeier, E. M. 1958. Inapplicability of the subspecies concept to North American marten.
Systematic Zool. 7:1-7.

Hamitton, C. C. 1925. Studies on the morphology, taxonomy, and ecology of the larvae of
holarctic tiger-beetles (family Cicindelidae). Proc. U.S. Nat. Mus. 65(17):1-87.

Hamiuton, J. 1885. Hibernation of Coleoptera. Canadian Ent. 17:35-38.

Bionomics AND ZOOGEOGRAPHY OF TIGER BEETLES 309

Hertey, H. M. 1937. Ecological studies on the Canadian River flood-plain in Cleveland County,
Oklahoma. Ecol. Monogr. 7:345-402.

Hirscuter, J. 1932. Sur le développment de la symétrie bilatérale des ovocytes chez Cictndela
hybrida L. Arch. Zool. Exper. Gén. 74:541-547.

Hoop, L. E. 1903. Notes on Cicindela Hentzu. Ent. News 14:113-116.

Horn, G. H. 1876. Notes on some Coleopterous remains from the bone cave at Port Kennedy,

Pennsylvania. Trans. Am. Ent. Soc. 5:241-245.
. 1878. Descriptions of the larvae of the North American genera of Cicindelidae, also of
Dicaelus, with a note on Rhynchophorus. Trans. Am. Ent. Soc. 7:28-40.

Horn, W. 1897. Drei neue Cicindelen und tiber Neolaphyra Bedel. Ent. Nachr. 23:17-20.

. 1900. Uber einige U.S.A.-Cicindelen. Ent. Nachr. 26:116-119.

——. 1906. Ueber das Vorkommen von Tetracha carolina L. in preussischen Bernstein und
die Phylogenie der Cicindela-Arten. Deut. Ent. Zeitschr. 50:329-336.

———.. 1908, 1910, 1915. Coleoptera. Adephaga. Family Carabidae, subfamily Cicindelinae.
Genera insectorum dirges par P. Wytsman. Louis Desmet-Verteneuil, Brussells. Fasc.
82A (1908), 82B (1910), 82C (1915):1-486, 23 pls.

———. 1926. Carabidae: Cicindelinae. In Junk, W., Coleopterorum catalogus. Berlin 1:1-345.

———. 1927. Ueber “Monstrositaten” und verwandte Vorgange bei Cicindelinen, Teil I. Ent.
Mitteilungen 16:471-477, pl. 8.

———. 1930. Notes on the races of Omus californicus and a list of the Cicindelidae of America

north of Mexico (Coleoptera). Trans. Am. Ent. Soc. 56:73-86.

. 1938. 2000 Zeichnungen von Cicindelinae. Ent. Beih., Berlin-Dahlem 5:1-71, 90 pls.

Howpen, H. F. 1963. Speculations on some beetles, barriers, and climates during the Pleisto-
cene and pre-Pleistocene periods in some non-glaciated portions of North America.
Systematic Zool. 12:178-201.

Hussett, T. H. 1954. The naming of geographically variant populations. Systematic Zool.
3:113-121.

Hure, L. H. 1915. The bionomics of the tiger beetle (Cicindela campestris). Proc. Roy. Phys.
Soc., Edinburgh 20:1-11.

Inertnc, H. von. 1926. Zur Verbreitungsgeschichte der Cicindeliden (Col.). Ent. Mitteilungen
15:156-161.

Incram, W. M. 1934. Field notes on five species of the genus Cicindela of the family Cicindel1-
dae from Balboa Bay, Orange County, California. J. Ent. & Zool. 26:51-52.

Iwata, K. 1936. Biology of two Japanese species of Methocha with the description of a new
species. Kontyu 10:57-89.

Jackson, H. H. T., anp H. E. Warrer. 1933. Notes on the occurrence of mammals in the
regions adjacent to the salt plains of northwestern Oklahoma. Publ. Univ. Oklahoma
Biol. Surv. 5:65-72.

Joticorur, P. 1959. Multivariate geographical variation in the wolf Canis lupus L. Evolution
13 :283-299.

Kapp, R. O. 1963. Pollen analytical investigations on Pleistocene deposits on the southern High
Plains. Ph.D. Thesis. Univ. Michigan. 256 p. Univ. Microfilms. Ann Arbor, Michi-
gan. (Dissertation Abstr. 24:1359.)

Ketty, K. L., ano D. B. Jupp. 1955. The ISCC-NBS method of designating colors and a dic-
tionary of color names. Nat. Bur. Standards Circ. 553. U.S. Gov. Printing Off., Wash-
tington. 158 p.

Kenpatt, H. M., R. M. GLENDENNING, aND C. H. MacFappen. 1958. Introduction to geog-
raphy. Harcourt, Brace & Co., New York. 684 p.

Kine, P. B. 1958. Evolution of modern surface features of western North America. p. 3-60.
In Hubbs, C. L. [ed.], Zoogeography, Publ. no. 51, Am. Assoc. Ady. Sci., Washing-
ton, D.C.

Koize, H. 1935. Biogeographische Bemerkungen zu K. Mandl’s Abhandlung tiber Cicindela
lunulata Fabr. und ihre Rassen auf Grund meiner Theorie der morphologisch-progres-
siven Tierverbeitung. Arb. Morph. Tax. Ent. 2:114-121.

LaFEertTé-SENECTERE, F. pE. 1841. Description de dix carabiques nouveaux du Texas et d’une
espéce nouvelle de buprestide de France. Rev. Zool., Paris 4:37-50.

Leconte, J. L. 1852. Remarks on some coleopterous insects collected by $. W. Woodhouse,

M.D., in Missouri Territory and New Mexico. Proc. Acad. Nat. Sci. Philadelphia 6:65-68.

. 1860. Revision of the Cicindelae of the United States. Trans. Am. Philos. Soc., new

ser. 11:27-64, pl. 1.

.1875a. Address of Dr. John L. Leconte, the retiring president of the Association. Proc.

Am. Assoc. Ady. Sci. 24:1-18.

.1875b. Address of Dr. John L. Leconte, the retiring president of the Association. Am.

Nat. 9:481-498.

310 Tue University ScrENcE BULLETIN

. 1875c. Notes on Cicindelidae of the United States. Trans. Am. Ent. Soc. 5:157-162.
. 1879. The Coleoptera of the alpine Rocky Mountain regions—Part II. Bull. U.S. Geol.
& Geog. Surv. 5:499-520.
Lene, C. W. 1902. Revision of the Cicindelidae of boreal America. Trans. Am. Ent. Soc.
28:93-186.
. 1912. The geographic distribution of Cicindelidae in eastern North America. J. New
York Ent. Soc. 20:1-17.
LENGERKEN, H. von. 1916. Zur Biologie von Cicindela maritima Latr. und Cicindela hybrida
L. Deut. Ent. Zeitschr. 60:565-575, pl. V.
. Die Salzkafer der Nord- und Ostseekiiste mit Beriicksichtigung der angrenzen Meere
sowie des Mittelmeeres, des Schwarzen und des Kaspischen Meeres, Zeitschr. Wiss. Zool.
135:1-162.
Lesne, P. 1897. Sur le terrier de la larve de Cicindela hybrida L. (Col.). Bull. Soc. Ent.
France, new ser. 2:273-274.
.1921. Le Cicindela silvicola Latr. sur les plateaux du Jura. Comptes Rendus Congr.
Soc. Savantes Paris Depts., Sect. Sci. 1921:102-109.
Linstey, E. G. 1939. The origin and distribution of the Cerambycidae of North America, with
special reference to the fauna of the Pacific slope. Proc. Sixth Pacific Sci. Congr.
4:269-282.
. 1958. Geographical origins and phylogenetic affinities of the cerambycid beetle fauna
of western North America. p. 299-320. In Hubbs, C. L. [ed.], Zoogeography, Publ.
no. 51, Am. Assoc. Adv. Sci., Washington, D.C.
MacGinirie, H. D. 1958. Climate since the late Cretaceous. p. 61-79. In Hubbs, C. L. [ed.],
Zoogeography, Publ. no. 51, Am. Assoc. Adv. Sci., Washington, D.C.

Macnamara, C. 1922. Tiger beetle larvae. Canadian Ent. 54:241-246.

Manpi, K. 1954. Aedeagus-Studien an Cicindeliden-Gattung (Col.). Ent. Arb. Mus. Gg. Frey
5:1-19, 4 pls.
Mayr, E. 1954. Notes on nomenclature and classification. Systematic Zool. 3:86-89.
. 1963. Animal species and evolution. Harvard Univ. Press, Cambridge, Massachusetts.
797 p.
Mitter, W. J. 1952. An introduction to historical geology, with special reference to North
America. Sixth ed. D. Van Nostrand Co., Inc., Princeton, New Jersey. 555 p.
MiskIMEN, G. W. 1961. Zoogeography of the coleopterous family Chauliognathidae. Systematic
Zool. 10:140-153.

MircHe.1, J. D. 1902. Observations on the habits of two Cicindelidae. Proc. Ent. Soc. Wash-
ington 5:108-110.

Montcomery, B. E., anp R. W. Montcomery. 1931 (1930). Records of Indiana Coleoptera.
I. Cicindelidae. Proc. Indiana Acad. Sci. 40:357-359.

Moore, R. 1906. Notes on the habits of Cicindela. Ent. News 17:338-343.

Moore, R. C., anp W. P. Haynes. 1917. Oil and gas resources of Kansas. Bull. no. 3, St. Geol.
Surv. Kansas 391 p.

Munrog, E. 1963. Perspectives in biogeography. Canadian Ent. 95:299-308.

OrrTEeNBuRGER, A. I., AND R. D. Biro. 1933. The ecology of the western Oklahoma salt plains.
Publ. Univ. Oklahoma Biol. Surv. 5:48-64, 81-87.

Owen, D. F. 1963. Variation in North America screech owls and the subspecies concept.
Systematic Zool. 12:8-14.

PacpEeN, H. T. 1925. Observations on the habits and parthenogenesis of Methocha ichneumo-
nides Latr. Trans. Ent. Soc. London Ser. 4, 58:591-597.

Papp, H. 1952. Morphologische and phylogenetische Untersuchungen an Cicindela-Arten.
Unter besonderer Berticksichtigung der Ableitung der nearktischen Formen. Osterr.
Zool. Zeitschr. 3:494-533.

Parkes, K. C. 1955. Sympatry, allopatry, and the subspecies in birds. Systematic Zool. 4:35-40.

Payne, J. F. 1964. A study of the surface temperature preference in relation to the habitat of
Cicindela repanda and Cicindela rufiventris. M.S. Thesis. Memphis St. Univ. 33 p.

Pearse, A. G. E. 1960. Histochemistry, theoretical and applied. Second ed. Little, Brown &
Co., Boston. 998 p.

Pearse, A. S., H. J. HumM, anp G. W. WuHarton. 1942. Ecology of sand beaches at Beaufort,
North Carolina. Ecol. Monogr. 12:135-190.

Perers, J. A., W. H. Burt, C. G. Srptey, C. M. Bocrert, T. H. Huspeiy, anp W. J. CLENCH.
1954. Symposium: subspecies and clines. Systematic Zool. 3:97-125.

PimentEL, R. A. 1959. Mendelian infraspecific divergence levels and their analysis. Systematic
Zool. 8:139-159.

PonseLtLe, A. 1900. Contribution a |’ étude des moeurs des Cicindéles. Feuille des Nat. 31:67-

68.

BroNomics AND ZOOGEOGRAPHY OF TIGER BEETLES el

Pratt, R. Y. 1939. The mandibles of Omus dejeani Riche as secondary sexual organs. (Coleop-
tera, Cicindelidae.) Pan-Pacific Ent. 15:95-96.

Rapp, W. F. 1946. Zoogeographic distribution of Cicindelidae (Col.). Ent. Monthly Mag.
82:224-225.
Rav, P. 1938. Additional observations on the sleep of insects. Ann. Ent. Soc. Am. 31:540-556.
REDFIELD, J. S. 1927. Mineral resources in Oklahoma. Bull. 42, Oklahoma Geol. Sury. 130 p.
Retnecx, G. 1923. Bemerkungen tiber die Lebensweise von Cicindela sylvicola Latr. Deut.
Ent. Zeitschr. 67:277-280.

Remmert, H. 1960. Uber tagesperiodische Anderungen des Licht- und Temperaturpraferen-
dums bei Insekten (Untersuchungen an Cicindela campestris und Gryllus domesticus).
Biol. Zentr. 79:577-584.

Renscu, B. 1957. Aktivitatsphasen von Cicindela-Arten in klimatisch stark unterschiedenen
Gebieten. Zool. Anz. 158:33-38.

Ruen, J. A. G. 1958. The origin and affinities of the Dermaptera and Orthoptera of western
North America. p. 253-298. In Hubbs, C. L. [ed.], Zoogeography, Publ. no. 51, Am.
Assoc. Ady. Sci., Washington, D.C.

Ricuarps, L. A., Ep. 1954. Diagnosis and improvement of saline and alkali soils. Handbook
60, U.S. Dept. Agric., Washington. 160 p.

Ripcway, R. 1912. Color standards and color nomenclature. Washington, D.C. 43 p., 53 pls.

Rivauier, E. 1950. Démembrement du genre Cicindela Linné. Rev. franc. d’Ent. 17:217-244.

. 1954. Démembrement du genre Cicindela Linné. II]. Faune americaine. Rev. franc.
d’Ent. 21:249-268.

. 1957. Démembrement du genre Cicindela Linné. III. Faune africo-malgache. Rev.
franc. d’Ent. 24:312-342.

. 1961. Démembrement du genre Cicindela L. (suite). IV. Faune indomalaise. Rev.
franc. d’Ent. 28:121-149.

. 1963. Démembrement du genre Cicindela L. (fin). V. Faune australienne. (Et liste
récapitulative des genres et sous-genres proposés pour la faune mondiale). Rev. franc,
d’Ent. 30:30-48.

Roginson, H. C. 1903. Report on the tiger beetles (Cicindelidae). Fascieoli Malayensis, Zool.,

pt. I, London. p. 179-183.

Ross, H. H. 1953. On the origin and composition of the nearctic insect fauna. Evolution
7:145-158.

. 1958. Affinities and origins of the northern and montane insects of western North
America. p. 231-252. In Hubbs, C. L. [ed.], Zoogeography, Publ. no. 51, Am. Assoc.
Ady. Sci., Washington, D.C.

Rumpp, N. L. 1956. Tiger beetles of the genus Cicindela in southwestern Nevada and Death
Valley, California, and description of two new subspecies (Coleoptera-Cicindelidae).
Bull. So. California Acad. Sci. 55:131-144.

. 1957. Notes on the Cicindela praetextata-californica tiger beetle complex. Description
of a new subspecies from Death Valley, California (Coleoptera-Cicindelidae). Bull. So.
California Acad. Sci. 58:144-154.

. 1961. Three new tiger beetles of the genus Cicindela from southwestern United States
(Cicindelidae-Coleoptera). Bull. So. California Acad. Sci. 60:165-187.

Say, T. 1823. Descriptions of coleopterous insects collected in the expedition to the Rocky

Mountains. J. Acad. Nat. Sci. Philadelphia 3:139-216.
SCHAFENER, J. H. 1898. Notes on the salt marsh plants of northern Kansas. Bot. Gaz. 25:255-

Scuaupp, F. G. 1879a. List of the described coleopterous larvae of the United States with some

remarks on their classification. Bull. Brooklyn Ent. Soc. 2:1-3, 13-14, 21-22, 29-30.

.1879b. Larvae of Cicindelidae. Bull. Brooklyn Ent. Soc. 2:23-24.

. 1883-1884. Synoptic tables of Coleoptera. Bull. Brooklyn Ent. Soc. 6:73-108, 5 pls.

ScHILDER, F. A. 1953a. Nomenklatorische Notizen zu Cicindela (Coleoptera). Beitr. Ent.

3):3112-3119:

.1953b. Studien zur Evolution von Cicindela. Wiss. Zeitschr. Martin-Luther Univ.,

Halle-Wittenberg. Math.-Naturwiss. Reihe 3:539-571.

Seat, H. L. 1964. Multivariate statistical analysis for biologists. John Wiley & Sons, Inc., New
York. 207 p.

SHELFORD, R. 1902. Observations on some mimetic insects and spiders from Borneo and Singa-
pore. Proc. Zool. Soc. London 1902:230-284, pls. 19-23.

SHELForD, V. E. 1907. Preliminary notes on the distribution of the tiger beetles (Cicindela)

and its relation to plant succession. Biol. Bull. 14:9-14.

. 1908. Life-histories and larval habits of the tiger beetles (Cicindelidae). J. Linn. Soc.

London 30:157-184, pls. 23-26.

312 Tue University ScrENcE BULLETIN

—. 1911. Physiological animal geography. J. Morph. 22:551-618.

—. 1913a. The life-history of a bee-fly (Spogostylum anale Say) parasite of a tiger bectle
(Cicindela scutellaris Say var. Lecontei Hald.). Ann. Ent. Soc. Am. 6:213-225.
.1913b. The reactions of certain animals to gradients of evaporating power of air. A
study in experimental ecology. Biol. Bull. 25:79-120.

.1913c. Noteworthy variations in the elytral tracheation of Cicindela (Coleop.). Ent.
News 24:124-125.

—. 1913d. Animal communities in temperate America as illustrated in the Chicago region.
Bull. Geog. Soc. Chicago, no. 5, Univ. Chicago Press. 362 p. [second ed., 1937].

—. 1915. Abnormalities and regeneration in Cicindelidae. Ann. Ent. Soc. Am. 8:291-294.

. 1917. Color and color pattern mechanism of tiger beetles. Illinois Biol. Monogr. 3:1-134.

SHERMAN, F. 1908. Notes on tiger beetle elevations. Ent. News 19:360-362.

Simpson, G. G. 1965. The geography of evolution. Chilton, Philadelphia. 263 p.

Sma.tey, A. E. 1960. Energy flow of a salt marsh grasshopper population. Ecology 41:672-677.

SmirH, H. M., anp F. N. Wuire. 1965. A case for the trinomen. Systematic Zool. 5:183-190.

Snow, F. H. 1877. Hunting Amblychila. Am. Nat. 11:731-735.

Sokar, R. R., anv R. Rinker. 1963. Geographic variation of alate Pemphigus popult-transversus
in eastern North America. Univ. Kansas Sci. Bull. 44:467-507.

SPANGLER, P. J. 1955. Habitat notes and description of the larva of Cictndela circumpicta john-
sont Fitch. Coleop. Bull. 9:81-84.

Strains, H. J. 1956. The raccoon in Kansas, natural history, management, and economic im-
portance. Univ. Kansas Mus. Nat. Hist. & St. Biol. Surv., Misc. Publ. no. 10. 76 p.

STEPHENS, J. J., III. 1959. Stratigraphy and paleontology of a late Pleistocene basin. Harper
County, Oklahoma. Ph.D. Thesis. Univ. Michigan. 73 p. Univ. Microfilms. Ann
Arbor, Michigan. (Dissertation Abstr. 20:1745-1746.)

STEVENIN, A. 1948. Relaciones antagoénicas entre un Asilidae y un Cicindelidae. Rev. Soc. Ent.
Argentina. 14:165-166.

SwrecimskI, J. 1957 (1956). The role of sight and memory in food capture by predatory beetles
of the species Cicindela hybrida L. (Coleoptera, Cicindelidae). Polskie Pismo Ent. 26:
205-232.

Tanner, V. M. 1927. A preliminary study of the genitalia of female Coleoptera. Trans. Ent.
Soc. Am. 53:5-50, pls. II-XV.

Tear, J. M. 1962. Energy flow in the salt marsh ecosystem of Georgia. Ecology 43:614-624.

TuHorNnsBury, W. D. 1965. Regional geomorphology of the United States. John Wiley & Sons,
Inc., New York. 609 p.

Townsenp, C. H. T. 1884. The proportion of the sexes in Cictndela vulgaris Say, and other

notes on the species. Canadian Ent. 16:227-231.
. 1886. Protective coloration in the genus Cicindela. Canadian Ent. 18:46-49.
Unear, I. A. 1964. A phytosociological analysis of the Big Salt Marsh, Stafford County, Kansas.
Trans. Kansas Acad. Sci. 67:50-64.
. 1965. An ecological study of the Big Salt Marsh, Stafford County, Kansas. Univ. Kansas
Sci. Bull. 46:1-99.
Van Dyke, E. C. 1929. The influence which geographical distribution has had in the production
of the insect fauna of North America. Trans. IV Intern. Congr. Ent. 2:555-566.
. 1939. The origin and distribution of the coleopterous insect fauna of North America.
Proc. Sixth Pacific Sci. Congr. 4:255-268.
Vaurie, P. 1950. Notes on the habitats of some North American tiger beetles. J. New York
Ent. Soc. 58:143-153.
. 1951. Five new subspecies of tiger beetles of the genus Cicindela and two corrections
(Coleoptera, Cicindelidae). Am. Mus. Novitates. No. 1479:1-12.

Watts, J. B. 1913. Robber-fly and tiger beetle. Canadian Ent. 45:135.

. 1961. The Cicindelidae of Canada. Univ. Toronto Press. 74 p.

WickHaM, H. F. 1904a. The influence of the mutations of the Pleistocene lakes upon the

present distribution of Cicindela. Am. Nat. 38:643-654.
. 1904b. Insect distribution in the Great Basin considered in the light of its geological
history. Ann. Rept., Ent. Soc. Ontario 35:42-46.

WiccLeswortH, V. B. 1929. Observations on the “Furau”’ (Cicindelidae) of northern Nigeria.
Bull. Ent. Res. 20:403-406.

Witte, A., anp C. D. MicHENER. 1962. Inactividad estacional de Megacephala sobrina Dejean
(Coleoptera, Cicindelidae). Rev. Biol. Trop. 10:161-165.

Wiutams, F. X. 1916. Notes on the life-history of Methocha stygia Say. Psyche 23:121-125.

. 1919. Descriptions of new species and life history studies. Part II, Bull. Exp. Sta.

Hawaiian Sugar Planters Assoc., Ent. Ser. 14:19-180.

Us

Bronomics AND ZOOGEOGRAPHY OF TIGER BEETLES 31

.1928. Pterombus, a wasp-enemy of the larva of tiger-beetles. Im Studies in tropical
wasps—their hosts and associates (with descriptions of new species). Bull. Exp. Sta.
Hawaiian Sugar Planters Assoc., Ent. Ser. No. 19:144-151.

, AND H. B. HuncErForD. 1914. Notes on Coleoptera from western Kansas. Ent. News

25:1-9.

Witson, E. O., anp W. L. Brown, Jr. 1953. The subspecies concept and its taxonomic applica-
tion. Systematic Zool. 2:97-111.

Woop, G. R. 1965. A trifurcate tiger beetle antenna (Coleoptera: Cicindelidae). J. Kans. Ent.
Soc. 38:392-394.

Wooprurr, R. E., ano R. C. Graves. 1963. Cicindela olivacea Chaudoir, an endemic Cuban
ne beetle, established in the Florida Keys (Coleoptera: Cicindelidae). Coleop. Bull.
17:79-83.

Wricut, H. E., Jr., anD D. G. Frey, ED. 1965. The Quaternary of the United States. Princeton
Univ. Press. 922 p.

Z1KAN, J. J. 1929. Zur Biologie der Cicindeliden Brasiliens. Zool. Anz. 82:269-414.

i

n
eo
oe

*

xe
nan aie

du OF : a sony Soil

Pete) b bey oe
ie

Wa - oa
er 4 ye 4

=) Ava Say @ :
) bees “th Sie on

hog - prise rau

-
io
2

-
ul bal
<

,

i

Be ws 3

THE UNIVERSITY OF KANSAS

SCIENCE BULLETIN

INTERACTION OF THE BULL’S-HORN ACACIA
(ACACIA CORNIGERA L.) WITH AN ANT
INHABITANT (PSEUDOMYRMEX
FERRUGINEA F. SMITH) IN
EASTERN MEXICO

By

Daniel H. Janzen

VoL. XLVII Paces 315-558 Ocroser 11, 1967 No. 6

ANNOUNCEMENT

The University of Kansas Science Bulletin (continuation of the Kansas Unt-
versity Quarterly) is issued in part at irregular intervals. Each volume contains
300 to 700 pages of reading matter, with necessary illustrations. Exchanges with
other institutions and learned societies everywhere are solicited. All exchanges

should be addressed to
LIBRARY OF THE UNIVERSITY OF KANsAs,
LAwRENcE, Kansas 66044

PUBLICATION DATES

The actual date of publication (z.e., mailing date) of many of the volumes of
the University of Kansas Science Bulletin differs so markedly from the dates on
the covers of the publication or on the covers of the separata that it seems wise to
offer a corrected list showing the mailing date. The editor has been unable to
verify mailing dates earlier than 1932. Separata were issued at the same time as
the whole volume. Beginning with Volume XLVI, publication was by separate
numbers and the date on each number is the actual publication date.

Vol. XX—October 1, 1932. Vol. XXXIV, Pt. I—Oct. 1, 1951.
Vol. XXI—November 27, 1934. Pt. II—Feb. 15, 1952.
Vol. XXII— November 15, 1935. Vol. XXXV,Pt. I—July 1, 1952.
Vol. XXIII—August 15, 1936. Pt. II—Sept. 10, 1953.
Vol. XXIV—February 16, 1938. Pt. III—Nov. 20, 1953.
Vol. XXV—July 10, 1939. Vol. XXXVI,Pt. I—June 1, 1954.
Vol. XXVI—November 27, 1940. Pt. Il—July 15, 1954.
Vol. XXVII,Pt. I—Dec. 30, 1941. Vol. XXXVII, Pt. I—Oct. 15, 1955.
Vol. XXVIII, Pt. I—May 15, 1942. Pt. II—June 29, 1956.

Pt. II—Nov. 12, 1942. Vol. XXXVIII, Pt. I—Dec. 20, 1956.
Vol. XXIX, Pt. I—July 15, 1943. Pt. II—March 2, 1958.

Pt. Il—Oct. 15, 1943. Vol. XXXIX—Nov. 18, 1958.
Vol. XXX, Pt. I—June 12, 1944. Vol. XL—April 20, 1960.

Pt. II—June 15, 1945. Vol. XLI—Dec. 23, 1960.
Vol. XXXI, Pt. I—May 1, 1946. Vol. XLII—Dec. 29, 1961.

Pt. II—Nov. 1, 1947. Vol. XLII—Supplement to, June 28, 1962.
Vol. XXXII—Nov. 25, 1948. Vol. XLIlI—Aug. 20, 1962.
Vol. XXXIII,Pt. I—April 20, 1949. Vol. XLIV—Sept. 1, 1963.

Pt. II—March 20, 1950. Vol. XLV—June 7, 1965.

Vol. XLVI— March 3, 1967

BGttORs «sists eh Pe ens R. C. Jackson

Editorial Board ........ GerorGE Byers, Chairman
KENNETH ARMITAGE
CHARLES MICHENER
Pau. Kiros
RICHARD JOHNSTON
DELBERT SHANKEL

TABLE OF CONTENTS

PAGE

MING ROU CUAL@IN eee ene aloe eee tecact ee ree ee Ae ie ek 316
ANG KEN @IWaTS BIS GIB MIB INGS) 2 ooeesoe ees eae ceo ccecd cect ave nec deceactcerecticcee pees : eRe eee 320
TE ESTRU MIMS (ORION No I ae re ee 322
BIONOMIGS @F ACAGIA CORNIGERA _......22.cc<.-0c0c-eesceeecc- Ssh ae ho Beco SR RA ee 323
SYCSSIOITIN ANB ONS Oa ee Se a ee ee ad 323
AGG CON ILO CAC Car OR Ie RI Ie ae Sr eae eae tae eee at Se ee SEE a LY)
Acacia sphaerocephala _............---....-------- Brae he Med gS tae fevte AO es. 324
TUETIET OD EYEE TRPOTIGTS Ve scp ee als ty ele NCE se oe ne, Re ee 324

IO SSIS HEU OV GN sd OR a Bae a See yee ae eee nee oF en ee oe Ae Ee re eee 324
GROSS) Ny HORA EN et COUN OY CON a re ee eae ee 331
DN EGCa eames 11 0 Ure oe ee ae tae Ne PE bs eh eed Ses Sa na gah ated ete ceca 331

TE agp ONO ce ey Bee NED is eters cadeee edad PRL AD Bien ei OR Ec esd ete SOR 341
ibelirctey [betas a2 tes Rees pens ee eee ee ener er eerie ee 344
LRRGMIT Ne FAVS UT a ee eta on ee eee 346
SMES, cc beste tee ee AEE ge NE ee ee res OP ee 346
Mevelopmentsot sepetativne sheatunes: 5. o-ccece cee secee sect cece ce ccs cccce case cesecne ete eee seeeecer ice eeeeseneess 348
REPRODW GWE MBIOQBOGN. 22-so cess es seco pees ee ge ee natn 2s es ee et 350
Poll gaat co rime eae es ea a A lan Nis) oe a eg 350
Seeayal peeve See Ee a Re a Ce ee ee 351
ITN LSU TA VRC LATA WU ASUCCG I He ee on es IE ea ee eee eee eee 352
REQUIREMENTS IN THE PHYSICAL ENVIRONMENT ..._......-----.-----------... 2-2 ----- 353
IMIGRGURR Sr Sl SN SP Pea Pe Cas fe Bele oe eet eens ae ere eee eee tar ee ee BBD od 2 na: . oo
lhemnagae Leiter ne meee ee sateen nt eek aE Ul SEN Phe ue oS ee eA eee ee Se oe 355
Gye SI yc Se ei ee Pi 35)

SS ULI I cle ee 2 aw we eg eee ee ean eee ee Se ee 355
TENURE) ye cates sae he A sme ete a ge DO ee CR Ae ne eee 356
POR WIEAMIO NI MGIROW aii AURA IME TEAR Sy cae ce ac ade ee ccc resect enee 357
A esy cla Canin) Geel TN fg eee ee eee hel a ee soe eet) Rs B>i/,
INGE AIRES Ei DIRECTION ERCY 01 Meee RR RRA Oe ee eer er Cee en he 359

IN Veotstcal ltr aam het ClO TS hese es sees a toe ease rd i 2 Ee ad es are Ss a De A 360

YD) eer ea TN Need CUO Hes Pes sc ene sR Sesh ee ov enh oe een DE oe kare 363
Seasaiall maby win clan cope eee seoxe orece te eit 6 oe ne ences ie nce oe cer eee eee 365

ESOS GANS PEGI Clan Saree ec teen ede ee oes ome Ae ae pe Ee aoe SSeS A Pe ee ay ae 367

Ant tolerant insects-contrasted -with ether insects -........-.---<--2.c20c2-ecceeseeeoeees eee 368
INaturalland eman-maadexcistiurbamece sites, =e eso e scene cree cesses eee ce nce 371
Othendsareswo fest eps tarc yi ne ay tose se he see cee eee ee ee 372
Iinsectsmncportedainmtheitenatune: 0. .ege Ee ee 373
Sanna el kira Fayed ed evcayS “ce a eee 373
Maina sie sMClet co licitea tas teste ee eee ees = ee acy ee Oe ee De eee 374
Bgevellecte "celhareve’, ate greet ESN ee are re ee Ae er ae A Rae Serene ee 374
VERRPEG) | | BS es ene PARE) on Se ar eee ae SR ie SENT 4 I RR Sl 8 chee ce 376

(Oat Men eRe ee pte and a eee ee Ree et Oe A ee oe 377
CONTRAST OF 4. CORNIGERA, A. SPHAEROCEPHALA, AND 4A. CHIAPENSIS ........ 379
APPENDIX I. DETAILS ‘OF DEFOLIATOR ACTIVITIES __.........---..-.cc22cccceccceeecceeeeeeees 384
BIONOMIGSSOF -RSEWDOMVARMEX JRERIRIUGGING A. oo. o5 ice cece cree ese seee ee 389
IN IGAW TSE ERIVASIES eo AUINIID) a NAIES alted GNIERS peek ee ce eect oe ON ae ae eee econ ced ck Sueas ecedece 389
Wollectan oyaco Louies yee ee ee he ES ea eee egs ee 389
Gicanspecting wine mcOlonies ses. = ae BARR cape Re ere ee Pee ea 390

Recording vantsactimity outside sof ithe tinea 22 sees asec cscs see esc ee eee cee ene 390

Marksin paw On kensye se eee ae enn ee consscseahvoccedstea Meno. cee kee ee 390
Mating <Aiohits: = tet ener tee ee eee ene niceties sen Se oe ee a 390
Maintaining laboratory colonies Sete eS nd ee ee ee ree ee 390
SIGMA CS op Neen ar tae 0 ak ee 391
POTSWRRMB WG LONG se: ccs cas aca ia ee seo cele gs ea 394
MORPHOEOGYOEF NIDIVIE WANES 5 soe2 = 2 acces eee 2 oe a ee ese 394
Morpholosy:orswonken m4 sco crated ee va ees ee 395

1 C0) 00 Veen ee a ee eee CS ee Pee Pcenceoabaocce 395
ISS pV EX 0) (0 uh yee tee ee Ee te NR OP AAR Sar CREME SCPE See es gn cece 395
Variatiom im colonies cee ee ee a 396
Bemiporanyy vaniatiomyimsforin =o. t oo eecee en ee eee eee 396
Morpholosy (of alates 0. a ee ee ee 396
Bremmallie 2.2 eee ee Oe tel eee nae ea nt CO RR See es 396

Malle se 8 ec gee Oe ea A ee a i a 398
COLONY MORPHOLOGY AND DEVE TEO PD MIEIN Geena eee 398
Structureor colonies minesin ell eyslio Oks eee eee eee 398
location xofdthe? queen. =-! =e ee ee 398
Placementiofsbrood) ..ces- et So ee ee ee eS ee 400

IDG oyenaKohol Gre INE VOTE MOMS spk cencteed ee ecrsecee earn ase cnenereea-cenneneneeneeeerenocsecesensoce ee 402
Distributionvor sthevalates see eee oe te a ei AAO ae te 406
ColoniestoccupyineamorentivanmOnens! OC teense eee ean ne 407
Presence..of ‘ac. queen’ fs 24.25. 38 oh ee ee ee ee 407
Placement: of sthexbrood!. 4... o.cecoes se eee 408
Distributions ofthe’ worker tonce meee eee eee 408
DistributionsoL othervanimalseinsanGdsonmShoo tomes eee eee ee 409
AT ECtsHoristressnONECO] OM ya OT pio) Oc yee eee eee ae ae 410
MORPHOLOGY AND DEVELOPMENT OF THE POPULATION OF COLONIES _.... 411
ENVIR © INIMIS Nie ANIC REFS © DIT NATE NE Sf ese 412
TIOS t REQULFEMENtS coe ec OS oon nee oe ree ne ae ee 413
INOIs tune ee ate X Ben 0 PEt ee, eon Oe ie Ocho ees eee wigs ils a 413
Dem perature. -<ic-sccc see enc Beh eee eee ae oe a 415
Sunn ss eh I a oe en 416
Waundlg 25 oe ces oe ans ae a Or Aes oe 416
BEHAVIOR: :OF INDIVIDUALS (22.2.5. ee ee 416
wilbar valiant! pupa -.cisoeeti cel Sets Seo acess chs bec de cases ee 416

e Worker: behav oii 2 cocoa cee ae eee ee re A tee 417
Worker activities am: the: thom) ocececcccccc creas cece este ere ae 417
Worker activities outstc en thre thirties sees eee ee ke eee a nee nsec eee eee nee 423
[Routine svactivities sone unGistiir beds topes sess eae 423

A, Beltian body jhatwest! 2.c3.05.03.0 3 424

B Nectar: icollection! ..2..-4 2. hc hee en 425
COWorkersiremowin ecatmaly ec acaGiciaitl S511 es eeeereeene esas eee pnd ee 426

DACleaning shoo testmracess b ym wOrnkensmeses sarees smear eran Eeekes AS

BP Patrolling shootasunacess bya WOLKers aden cemetery ner ADT

F Brood and! fadulltistranspont: este ccr ete: a ie aera 429

2: Trails” between! ‘shoots: 42.0 tse est Se A A 429
AVIRelation Of Col Omyasize wmto tical aera eee es eee ee 430

B: Distance-\betweeniithe! shoots) e222 -= he ee. ee ee 431

@iRood Vavailabilityeandetratletonmiatio reese eee eee eee 431

D) Relation tor timeotyday sand sweathier tomy tial sjeeesse eee eee nn nen 43]

E Relation of worker activity to trail formation —._.
33 ID iGO OEP Toy. .e Bunmawayes Vio e SAS eo ee eee eae eee eS ere oe oe

A Relation of size to division of labor —...... ee
Biel atom oiema cmt Om iiS1@ Tl Oi cl yee ny ee ea ce ones
(C) Rellaiivarm Git aligiterd oars fio) GliniiGitopnl (one WAVY ee
AMREAGIO MM tOMIMUnIISIV.eM OD} ECLSM DY i aWOLKCIS) tiescee senescent een
INS TENS exh OY AVGSO\! Seen eke Yk ye ree gee nee eee ey Pes Senter pr Ee ke Sey a eres een onae eatae

B Reaction to inanimate objects —.........

GpReactonmtominsect=sizeclumalmnttuiall Suge e=ss eee eee eee ae ee a enone oe EE
DwReactiomstomlances auniialls ice see sey oe coe oe a ceearcere snes sae eet nce os Seah
1s, \Worlker ceimerse to) jolene ORIG 2 ceeseeee ececeeccees eee Seer are eere eres ee ere memes
[BYE ONE Gk HeyOROVATNLSANWS SUMAN seca greens eee eee cee ne ee Eee Ponce Saree meee
asic emthem thon bELOremCO Plato mses seen serene sete eee eee Beh EN)
Outside the thorn before copulation —............ Br eA ee Tree See ere ee
(ruts el etl ert co rp treat ce ea Co fC a 1 ea ae
Inside thesthonnatter copulatvomy eee eet ee ee a eee. eee
[alny SO pa Sti ei] OM Van] CCM mares ree to see ne wea pe De eo eae Se eee eee rae
PS ee cag Ngan © egg 01g ene OS ee etd ae ee RON ae ef aE
Ohutigidle wrt Gime tavern loweriorre (oyorulleywioyny <a cc eece
INTE GT oy OY SVEN COS gm ate ae eRe eet
INO RaTATE TA Vas A GIOR'S: (OB WiO RIE GS yas ae ova sp, ope es ssrecese ee see eee nee
TROUT OV GUAYS? x seurgiee be nat Es ous ee Sa cel co Ne eat ee ANON att oe ore Ne PPE we OE Ce ob ad gle
BV feline Fit cree ere ee ea Mik alae ce em, it ae NT ee SS i een eee
Broductser kno nian thes GUC iy a tetateeetern ctlctan Sse WOR Ea a are oe se cee eee
ITU Gia ee ae wl ts yell» Dita g CR AT AL a etirae Siirota eed intr ai te ee
BTC] Syn os ween er ree Get ree 4 ato AU hoot ah a ate OS tere ee eS eee ee
PNTAUITG 190 Gigs ORE ALORS te att ees eee ae et ee EN Ee ek ee
ST CMe Ae ee cre ie Rab ee S/o even Cals EE EYES TORT Ry OPEL PMNS ART ee
BO vystilia Gale ee ee re eed eA toe ee eek ere ee Sle ie nay och ee BE
OM TERSACAGIACAINTES SIN aI: s Sa WD VeeANRInyA Vy aaee eee ee
TBS GIS ST @ IN ete ee eee cet crane ee Ne oe ee bi EN Br he ee eee,

FACTORS INFLUENCING THE PARAMETERS OF ACACIA CORNIGERA
12 @ PUTS ASATRL ONS race seers esc ak sn eee eRe: oe A Via erento eee ee tet diem e
MIAN ATES AUN) Ne Iii © D) Sie eae eect: any CEs ese Be ee ee ee ee
S CUT yarn eo eros acre, A MB ek Dede I Ea alee ae Fa i ee Oe
SiS gamere sae ee el 29 A cone eee EB Bt BR oe eh Re A Eee
PEGI tA ti OT) meee ee cene eek eee were eee Be gt nN in eg A ou eT, cs Set d ng e
“ECS amOXENTER LTRS re ee eee oo ey Ce econ wee wee Oe A NRT ery Rea aU SUE OVS. oon Og OG bee eae gk
Ieee a ee ea a, oe eee Se eRe or Oe OMB S RN SF ART 8 WOE ee EA er oct
WG Cha tio tps ty DES ate: = sees oe ee eee ee ee ee at SS hoe eee a ee
Choicevots plotsi.-secea ere eee eee a eI ae ee a Oe, Se
MDT a Cinl eM ts wares Careers eee ee PR ens pee ee i a) ee pete NE CL a A Se oe I ee

Ne tilayysl ero alta tilt © Koma Ca TCT 61 OT eee eee oe ern
Giga poi gt rin Sipe ee ee ee Le ee 2G ee eee Oe
Recolonizationyofstreatedsplotsys= eee a Ae ers Og ne LT ee
Removalmounverctatloneadjacentmtom 1pm Go i771 Cr (meen ean meee en
Clearin saofeeallley es etatio 1 get asec eee te eee cena pet ce Se ead ree ven ee
Enclosures ime banbedmewiter sss et ete oe ee en sete ne ee tr ee
PATA Glin Ger UIT CS hppa ee ieee ee eae Seen cet arin a So rela ee ee eR Pee ade oe EE

PAGE
431
432
432
qv?
433
433
433
435

438
439

Multiple: treatment: <.22=. 3) eee Ae a a ee ee 471
Retording data: 2-6-5 aoe eo ee eee oe ee 471
Shoot identifications 22.415. tn ee ee ee 471
Fede > sevic sce gh sn EE a ea SY eet ee 471
Height of the surrounding vecetatiom sce eee 471
Condition*of the? shoot #523. 4.352 on. es 3a ee ee 472
Causal'vagentiof condition! <2 en ee eee 472
Presence) of foreign objects: 22.ccs2cti ance eee eee ee 472
Presencetol Po ferrite oie eas os wc casacsescessesee eee ee 472

Tid eile) Ot eicose eee P See se ete a ehe 472
Statistical treatment, of data 223222. oe as,
PLOW DESCRIPTIONS: 02) sc ane ey oo Bee 480
Bi ee ns oh ks ke Real BM 8 ot uns 8 bas 481
RAS ISSUING) Ol Pee tee esse ste a oS eg ae ae de Pe rey ele ee ee 482
CO Se ere ee ree ee eee, Oe UE oP rare | MEER Vaan SO APP Maes Rew te pera eras en Mae dee ne oe 485
RE GS tS es ay ot ee a eet ole Nine p Seee rAd) Seen, Sak eee See 485
GPE SV ose ae ee Sn ee 2 487
Te a ee a we can Anteeeda deo ieek cee ca seers eee 489
TeV oe teed ced P Bes a8 oe le 6 I Oe Oe | SO aie: aici Lis 7 ke 490
ARID Re See beet no ect ee Ea oe Ge sy oe eh ce 491
DATASRROMSUBPILOUS) S22 ast re 8 ees ee eee ee ee 492
Eleiobteimeneme mt ese cn ee cee es eee 493
STpmifiGam Gey rie Sete Ae DY oe Ps NO ces te ee er ae 493
Subplot length increment records for suckers and stumps —.........-------------------------- 494
Subplot height increment records for existing shoots ......-.....--------------------0e-0---e-e--eee- A995
Height or length increments of occupied shoots contrasted to unoccupied shoots _.. 502

1 Length increment ofisuckets: 2.0.0 ee ee 504

Ze ileic itm NGLeMents moter x1STN CIES 11 OLS eee eee 507
DisGuissiGm ie. nce ese A I ae ee ee 507
Gondition:of; the ..shoot) 22.2 eck ee ee eee eee 515
Sipimi fiance occ ie Sec occa n cae ae ck a eed a ge 5)
Subplot records of condition for suckers from stumps ..............----------------------------- 515
Subplot mecords of condition foneexistiney shoots eae =e 519
Contrasts of the condition of occupied shoots with unoccupied shoots —.....----- 522

ig Suckeis:: © sesh 55 Se. crete ee See es aoe ee 22
Zexisting Shoots: 22. accede eee 523
Discussion: tJ Bee ee ee ee 525
Mortality: of An cornigenal cts 5 ek ee 528
Leal (production! by A cornigera (ee ne, eer eee 530
Biomass productionnofed. vcorni pena. see eee eee ee 532
Presence: Of winesi mee ee A he 3393)
Presence ‘of basal circles: 223 2st oe Se ee ee ee 536
ADDITIONAL, OBSERVATIONS iio ee ee ee 540
Development-o£ colomiessumisityy sence ects se og ee 540
Aurxaliary-shootieftects).ces2t:c. 5. R ete teeta ee eee ae 543
Reproductive biology: s.ccccc fe 5-2 sacks seen che ee ee Bart
Naturally “unoceupied shoots: 22:2 se:2: ee So ee 547
ECONOMICS? «,c2iczte et ie ee eS Ue en ee 548
DISCUSSION?” | ccccsec heen ec ee ee ne eo ren 550

THE UNIVERSITY OF KANSAS

SCIENCE BULLETIN

Voi. XLVII Paces 315-558 Ocroser 11, 1967 No. 6

Interaction of the Bull’s-Horn Acacia (Acacia cornigera L.)
with an Ant Inhabitant (Psewdomyrmex ferruginea F. Smith)
in Eastern Mexico’

Danie H. JANZEN

Department of Entomology, The University of Kansas

ABSTRACT

The interaction between a swollen-thorn acacia, Acacia cornigera, and an ant
inhabitant, Pseudomyrmex ferruginea, is described from the lowland coastal plain
in eastern Mexico. A detailed study of the bionomics of the acacia and the ant is
presented to aid in the interpretation of experiments with the interaction. The
experimentation lasted a year (Sept. 1963 to Aug. 1964), and was conducted in
the second growth vegetation in pastures, roadsides, and fallow corn fields in the
area between Temascal, state of Oaxaca, and La Granja, state of Veracruz. Addi-
tional observations of the acacia and the ant in other parts of their range north of
Coatzacoalcos, Veracruz, showed that the populations in the Temascal-La Granja
area are representative of those in eastern Mexico.

The bull’s-horn acacia, A. cornigera, is a representative swollen-thorn acacia
with well-developed foliar nectaries, enlarged stipular thorns, and small nutritive
organs (Beltian bodies) borne at the tip of each leaf segment. The colony of P.
ferruginea living in the enlarged stipules obtains sugars from the foliar nectaries,
and oils and proteins by eating the Beltian bodies.

A. cornigera is a woody plant with an extremely high growth rate. It is intol-
erant of shading and is a member of the canopy, or is emergent, during the first
6-12 years of regeneration. While its original habitat was natural disturbance
sites (river banks, arroyos), it has invaded with great success the man-made dis-
turbance sites that are not burned on an annual basis; nearly all of its regeneration
in such sites is as suckers from old root stocks and requires that the ant colony
move from the cut acacia into the new suckers. When the acacia is not occupied

1 From a thesis submitted in partial fulfillment of the requirements for the Ph.D. degree in
Entomology, University of California, Berkeley. Contribution No. 1346 from the Department of
Entomology, The University of Kansas, Lawrence. Supported by National Science Foundation
Grant GB-1428 (Dr. R. F. Smith) and Tropical Biogeography Research Grants 66 and 81. In-
vestigations in Costa Rica during the summer of 1963 were made while attending the National
Science Foundation Advanced Science Seminar in Tropical Biology at the Universidad de Costa
Rica.

316 THe Universiry ScIENCE BULLETIN

for natural causes, or has its ants removed experimentally, it is subject to severe
damage by phytophagous insects; these are insects which normally feed on other
species of plants.

Morphologically, P. ferruginea is a representative pseudomyrmecine ant; be-
haviorally it has a number of outstanding characteristics associated with its inter-
action with swollen-thorn acacias. The workers patrol and clean the surfaces of
the acacia, and bite and sting animals of all sizes that contact the plant. The
workers maul any other species of plant that contact the acacia and in many cases,
any that grow under the acacia. The colony attains a very large size and up to
25 per cent of the workers may be active on the surface of the acacia both day and
night. The larger the colony becomes, the smaller is the damage sustained by the
plant from defoliators. The colony enhances its own probability of survival by pro-
tecting the acacia on which it is completely dependent for food and domatia.

In the course of the study, 50 control or treatment subplots were established.
The ants were removed from the acacias by spraying with parathicn, clipping the
thorns, or cutting and removing the occupied shoot. Measurements throughout
the year of the height increment, condition, freedom from vines, leaf and thorn
production, and biomass production, showed that all of these parameters are
greatly reduced if the ants are removed from the acacia. In the experimental plots,
this was almost entirely due to phytophagous insect damage, and subsequent shad-
ing of the stunted shoots. Based on these data, and observations cf naturally
unoccupied acacias, it is concluded that a shoct of A. cornigera must be oc-
cupied by a colony of P. ferruginea for a substantial part cf its life to preduce
seeds and become a part of the reproductive population. The ant is dependent
upon the acacia for survival and the acacia is dependent upon the ant fer normal
population development; the interaction between the two can therefore be properly
termed one of obligatory mutualism.

INTRODUCTION

Within the mimosaceous plant genus Acacia, at least nine neotropical
species have an obligate or consistent relationship with ants. These are often
called “ant-acacias.” In all the known obligate relationships, the ants belong
to the genus Pseudomyrmex (in most previous literature cited as Pseudo-
myrma), and at least five species are involved. They are often called “acacia-
ants” and live in the swollen stipular thorns which are characteristic of ant-
acacias. They feed almost entirely on the nectar and nutritive bodies (Beltian
bodies) produced by the acacia. Most acacia-ants have been recorded only
from living ant-acacias and there is little doubt that they are dependent on
the acacia for existence. However, the question of whether or not the ant-
acacia is benefited by the presence of the acacia-ant has been outstanding in
the literature dealing with ants and plants since Belt’s (1874) often-quoted
description of the relationship of Pseadomyrmex belti Emery with Acacia
costaricensis Schenck in Nicaragua (Wheeler, 1942). The study described in
the present paper is an attempt to resolve this question. The species pair
Pseudomyrmex ferruginea F, Smith and Acacia cornigera L. was chosen for
intensive study. In previous literature, P. ferruginea is discussed as Pseudo-

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 317

myrmex belti fulvescens Emery or Pseudomyrmex fulvescens Emery (Jan-
zen, 1967a).

The aggressive behavior of many species of acacia-ants toward humans
coming in contact with the acacia has led to numerous published opinions
that the ants “protect” the plant from phytophagous animals (Belt, 1874;
Delpino, 1886-1889; C. Darwin, 1877; F. Darwin, 1877; Schimper, 1888;
Wasmann, 1915, 1916; Schwarz, 1917; Safford, 1922; Standley, 1928; Alfaro,
1935; and Brown, 1960). However, others have considered that the thorns
themselves are sufficient protection against browsing mammals, and that the
acacia-ants have little impact on the action of phytophagous arthropods
(Rete, 1904; Ule- 1905, 1906; Wheeler, 1913, 1942; and Skwarra, 1930,
1934a, 1934b). With the exception of Brown’s (1960) question-posing paper,
this argument has not been treated in detail since Wheeler’s (1942) and
Uphoff’s (1942) literature review. The portions of the populations of A.
cornigera and P. ferruginea treated in this paper have received little specific
attention in these discussions (except Skwarra’s papers). It was apparently
believed that all swollen-thorn acacias had approximately the same type of
interaction with the species of Psewdomyrmex obligatorily associated with
Acacia. As a consequence, little effort was made to understand each ant-
acacia interaction as a separate system. A demonstrated or postulated point
about the interaction of one species of acacia with one species of ant was
usually regarded as applicable to all. Associated with this, there was little
hesitation to propose hypotheses which were intended to encompass all
swollen-thorn acacias, and at times, all plants with obligate ant associates.
The data presently available indicate that the various interaction systems
between ants and acacias cannot be discussed in general terms until more
information has been gathered. The same must be said for other genera of
plants and the ants associated with them.

The data in the literature dealing with swollen-thorn acacias that have
been summarized by Wheeler (1942:94-116) do not merit further detailed
discussion until the acacias and their ants have been further studied. How-
ever, there are a number of papers in which the authors arrived at a definite
conclusion about the presence of a symbiotic relationship between the swol-
len-thorn acacias and their ant inhabitants.

Of those who supported the idea of a symbiotic relationship, Belt (1874)
is the most often referred to. His feelings can be summed as “These ants
form a most efficient standing army for the plant, which prevents not only
the mammalia from browsing on the leaves, but delivers it from the attacks
of a much more dangerous enemy—the leaf-cutting ants.” He then restates
this as “I think that these facts show that the ants are really kept by the
acacia as a standing army, to protect its leaves from the attacks of herbivorous
mammals and insects.”

318 Tue University ScrENCE BULLETIN

Brown (1960) feels that “On the face of it, Belt’s opinion has long seemed
to me more attractive than that of the exploitationists.” The “exploita-
tionists” are those such as Wheeler who felt that the ants were merely ex-
ploiting the acacia and that the acacia was not affected by their presence.
Brown continues on to discuss the evidence available that supports the idea
that the ants may protect the acacias from browsing mammals. He con-
cluded with, “The claims of the protectionist school for the adaptive nature
of some kinds of extrafloral nectaries and ant domiciles in attracting insect
‘guards’ against phytophagous insects also seem reasonable, but proper study
of the problem has scarcely begun.”

While never having seen the ants or the acacias in their natural habitat,
PF, Darwin (1877) has rather enlarged on Belt’s discussion in saying, “The
ants form a standing army for the tree, and not only prevent cattle etc.
browsing on it, but also protect it from the ravages of the leaf-cutting ants.
So serious is the latter danger, that the tree is actually unable to exist without
its guard of colonists.”

After observing the stands of Acacia sphaerocephala in the Tampico,
Tamaulipas, Mexico area, Schwarz (1917) had the following comments.
“|. . the writer can fully corroborate the original observations of Mr. Belt
to the effect that the ants effectually defend the acacia trees against the
attacks of man, cattle and insects. No leaf-eating caterpillers, no aphids, nor
coccids are seen on the trees; no leaf-cutting ants ever defoliate the same.”

After examining the ants and acacias in the field in Costa Rica, Alfaro
(1935) came to the conclusion that the aggressive nature and well developed
sting of the ants constituted the best defense of the plant against the small
enemies that could attack it. He appears to agree with Belt’s conclusion that
the ants serve to keep leaf-cutter ants from defoliating the shoot.

Wasmann expressed the view both in 1915 and 1916 that we may still
speak of a true symbiosis between the ant-acacias and the acacia-ants, rather
than merely a case of parasitism by the ant. The reiteration of this view
(1916) was in response to Wheeler’s (1913) paper. Wasmann’s feeling that
a true symbiosis exists is accompanied by an exposition of a possible means of
evolving such a symbiotic system.

While the above authors who believed that a symbiotic relationship
existed were not referring explicitly to A. cornigera and P. ferruginea in
the geographic area where the present study was conducted, this study has
shown that their beliefs were in general correct in respect to this pair of
species. The degree to which they are correct in respect to other species pairs
must remain undecided since there is not sufficient data in the literature to
discuss the question. The author is at present gathering data on these other
species pairs.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 319

Of those who supported the idea that the ants bear only an exploitationist
relationship to the acacia, there were only two, Wheeler and Skwarra. In
1913, Wheeler (borrowing von Ihering’s (1907) conclusion about plants in
the genus Cecropia and their ants) stated, “I believe, therefore, that we may
adopt von Ihering’s point of view, and say that Acacia cornigera, hindsi,
and sphaerocephala have no more need of their ants than dogs have of their
fleas. If this is true, the relation between the ants and plants is not one of
symbiosis, but one of parasitism.” By 1942 Wheeler had not changed his
mind. He stated that “. . . the 54 different organisms . . . listed above as
associated with the bullhorn Acacias .. . are sufficient . . . to show, first, that
these plants have plenty of natural enemies and are in this respect like other
nonmyrmecophytic trees and shrubs in the tropics, and second, that the
obligate ant tennants, though more virulent than those of Cordia alliodora
and the Cecropias, are nearly or quite as tolerant of alien ants and other
insects on the same plants. I therefore reiterate my statement of 1913 that
the relations existing between the Acacias and the obligate Pseudomyrmas
are not properly those of symbiosis, in which the plants have adapted them-
selves to the ants, but those of host and parasite, in which the adaptations are
solely on the part of the ants.”

After field observations in the state of Veracruz, Mexico, Skwarra (1934a)
concluded that the acacia-ants are not capable of protecting the leaves, flowers
and fruits of the shoots from destructive insects and that the ants are
therefore not useful to the plant. Her observations were based entirely on
A. cornigera and A. sphaerocephala and their various ant inhabitants. Her
conclusions were primarily based on finding bird nests and several species
of insects on occupied shoots. Skwarra’s findings are not included in
Wheeler’s 1942 paper, but had they been, he undoubtedly would have used

them to support his own opinion.

While a definitive statement cannot be made about the correctness of
Wheeler’s opinions in respect to other ant-acacias, the present study has
shown that his opinion was incorrect in respect to A. cornigera and P.
ferruginea in the study area. Skwarra’s statements of facts about acacias
and their ant inhabitants appear quite correct but her conclusions based
on them have been refuted by the evidence gained in this study. The major
problem of both workers was in failing to observe what happens to the
acacia when it is not occupied by a colony of P. ferruginea; they were only
concerned with what happens to the acacia when it is occupied.

Both A. cornigera and P. ferruginea range from northeastern Mexico
to Guanacaste Prov., Costa Rica. The present study is concerned with the
populations in the coastal lowland and bordering foothills of the area from
Coatzacoalcos, in the state of Veracruz, Mexico, northwest to the northern
range limits around Tampico and Cd. Mante, in the state of Tamaulipas

320 Tue UNniversiry ScIENCE BULLETIN

(henceforth designated “study area”). The other acacias with swollen thorns
and the Pseudomyrmex that occasionally are found in them in the study
areas are discussed for comparative purposes (Acacia chiapensis Safford and
Acacia sphaerocephala Schl. and Cham.; Pseudomyrmex gracilis mexicana
Roger and Pseudomyrmex nigrocincta Emery).

The experimental part of this study was primarily the comparison of
plots containing acacias from which the ants had been removed (by use of
insecticides or physical means), with plots containing acacias with their
normal ant colonies. Height increment and condition of the shoots were
the principal acacia characteristics recorded and compared. These plots were
established in vegetation of various ages, and in types of plant communities
that are representative of the communities occupied by the majority of the
population of A. cornigera. The experimental portions of the study were
conducted from September 1963 through August 1964, in the lowland coastal
plain immediately east of Temascal, in the state of Oaxaca, and in the foot-
hills around Temascal.

It is difficult to evaluate the present experimentation without understand-
ing the bionomics of A. cornigera and P. ferruginea in more detail than can
be gained from the brief discussions in the literature (Belt, 1874; Safford,
1914, 1922; Skwarra, 1930, 1934a, 1934b; Alfaro, 1935; Wheeler, 1913, 1942;
Wasmann, 1915). Therefore, the first two sections of this paper are con-
cerned with the bionomics of the acacia and the ant. The depth to which
the bionomics of P. ferraginea is described is justified from its ecological
pertinence and because the bionomics of a pseudomyrmecine ant have not
been previously treated in detail. Gathering the bionomic background in-
formation began in July 1962, at Campo Cotaxtla (Cotaxtla Experiment
Station), state of Veracruz, and continued through October 1964, in Mexico,
and at Berkeley, California. Most of the information was obtained in the
Temascal area between September 1963 and August 1964. The experimental
examination of the ant-acacia interaction is discussed in the third section of
this paper.

Other studies and reports in the literature show clearly that the inter-
action described in this paper is not representative of the relationship be-
tween acacias and other ant genera. Furthermore, there is no substantiated
evidence that these interactions between other genera of plants and Pseudo-
myrmex, or other genera of ants, are mutualistic. However, there is super-
ficial evidence that there are mutualistic interrelationships between ants and
plants, and this is a field worthy of considerable experimental study.

ACKNOWLEDGMENTS

I wish to express my deepest appreciation to Dr. Ray F. Smith of the Depart-
ment of Entomology and Parasitology, University of California, Berkeley, for

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 321

patient guidance in this study and for having obtained the funds and materials
for its support. I have greatly appreciated the consultation of Dr. H. V. Daly of
the same Department in organizing and treating the data presented in this paper.

Drs. R. F. Smith, H. V. Daly, H. G. Baker, and C. B. Huffaker of the Uni-
versity of California at Berkeley, W. L. Brown, Jr. (Cornell University), C.
Rettenmeyer (Kansas State University), E. O. Wilson (Harvard University), and
V. Rudd (U.S. National Herbarium) have contributed by their criticisms of the
manuscript.

An ecological study of this nature is especially dependent upon those who
identify the organisms involved. For interested and prompt assistance in deter-
minations, I would like to thank the following ecg Drs. E. O. Wilson (Har-
vard University), N. Weber (Swarthmore College), I. H. H. Yarrow ( British
Museum), and C. Rettenmeyer (Kansas State University), Formicidae; C. D.
Michener (University of Kansas), A. Wille (Universidad de Costa Rica), H. V.
Daly (University of California at Berkeley), and R. M. Bohart (University of
California at Davis), other aculeate Hymenoptera; G. I. Stage (University of
California at Berkeley), C. F. W. Muesebeck (U.S. National Museum) and B. D.
Burks (U.S. National Museum), parasitic Hymenoptera; G. Steyskal (U.S. Na-
tional Museum), C. W. Sabrosky (U.S. National Museum) and G. E. Shewell
(Entomology Research Institute, Canada Department of Agriculture), Diptera;
J. M. Kingsolver (U.S. National Museum) and C. D. Johnson (University of
California at Berkeley), Bruchidae; H. F. Howden (Entomology Research Insti-
tute, Canada Department of Agriculture), O. L. Cartwright (U.S. National
Museum), M. W. Sanderson (Illinois Natural History Survey), and P. Vaurie
(American Museum of Natural History), Scarabaeidae; G. B. Vogt (U.S. National
Museum) and R. F. Smith (University of California at Berkeley), Chrysomelidae
and Buprestidae; R. D. Warner (U.S. National Museum), Curculicnidae; J. A.
Chemsak (University of California at Berkeley), Cerambycidae; T. J. Spilman
(U.S. National Museum) and J. Lawrence (Harvard University), Cucujidae; W.
Connell (U.S. National Museum), Nitidulidae; V. Whitehead (University of
California at Berkeley), Coccinellidae; the late H. J. Grant (Academy of Natural
Sciences of Philadelphia) and D. Rentz (California Academy of Science), Or-
thoptera; J. T. Medler (University cf Wisconsin), H. L. McKenzie (University of
California at Davis), T. E. Moore (University of Michigan), D. D. Jensen (Un1-
versity of California at Berkeley), J. P. Kramer (U.S. National Museum), J. W.
Beardsley (Hawaii), H. Schroder (Senckenbergische Naturforschende Gesell-
schaft, Germany) and D. Hille Ris Lambers (Bennekom, Netherlands), Homop-
tera; C. J. Drake (U.S. National Museum), J. L. Herring (U.S. National Mu-
seum), and R. L. Usinger (University of California at Berkeley), Hemiptera;
S. F. Bailey (University of California at Davis), Thysanoptera; J. A. Powell
(University of California at Berkeley), D. Davis (U.S. National Museum), and
R. W. Hodges (U.S. National Museum), microlepidoptera; C. D. MacNeill
(California Academy of Sciences), W. D. Field (U.S. National Museum), E. L.
Todd (U.S. National Museum), J. G. Franclemont (Cornell University) and F.
H. Rindge (American Museum of Natural History), macrolepidoptera; P. W.
Wygodzinsky (American Museum of Natural History), Lepismatidae; W. J.
Gertsch (American Museum of Natural History), Araneae; R. E. Beer (Univer-
sity of Kansas), Acarina; S. B. Benson (University of California at Berkeley),
mammals; and R. W. Dickerman (Cornell University), birds.

The Mexican government was especially helpful. Through the courtesy of
Dr. Alejandro Ortega, Instituto Nacional de Investigaciones Agricolas, S. A. G.,

oS)
ii)
is)

Tue University ScrENCE BULLETIN

and Dr. Francisco Cardenas, Director of Campo Cotaxtla, living facilities were
made available for the author and his family during the summer of 1962, and
equipment and greenhouse space were offered during the entire study. Mr.
Guillermo Hernandez made living facilities available in the buildings of the
Comicion Federal de Electricidad at Temascal during the period September
1963 through August 1964; this study would have been much more difficult with-
out these quarters.

Mr. Eusavio Farfan, Mr. Juan Torrealva and Mr. Segundo Verde of the
Temascal-La Grandja area have generously allowed these experiments and obser-
vations to be conducted on their land, and have frequently modified their man-
agement plans in deference to the study. Leocardio Torrealva helped with the
held work necessary in preparation of the experimental plots and gathering data;
without his assistance many phases of this study would not have been possible
within the time available.

The greenhouse crew at Oxford Tract, University of California at Berkeley,
has been especially considerate in caring for potted seedlings during the study.
Dr. W. W. Allen contributed a sprayer and insecticides. Feeding experiments and
nectar analyses were conducted in Dr. H. T. Gordon’s laboratory in the Depart-
ment of Entomology and Parasitology, University of California at Berkeley.
Chromosome and pollen counts were made in Dr. H. G. and I. Baker’s laboratory
in the Department of Botany, University of California at Berkeley; anatomical
studies of Beltian bodies were done by Mr. F. Rickson of that department. Mr.
S. Snedaker of the Department of Botany, University of Florida performed soil
analyses. Special efforts were made by Dr. C. D. Michener (University of Kan-
sas), Dr. J. M. Savage (University of Southern California) and Mr. Snedaker to
collect Pseudomyrmex and swollen-thorn acacias in areas inaccessible to the
author.

To my wife goes more than the usual acknowledgments. She assisted in both
the laboratory and field, and her patience in caring for her family under un-
familiar and often unpleasant circumstances was a tremendous help.

TERMINOLOGY

Several terms used in this paper must be defined at the outset so as to
avoid confusion in their meaning. “Shoot” refers to all of the plant which
is above ground, originating from one point on the ground, and does not
denote age or degree of branching. “Sucker” refers to the shoot that has
recently regenerated from a cut or burned stump. “Seedling” refers to the
shoot that has grown from a seed without being cut or burned since the
seed germinated. Therefore, a shoot may be either a sucker or a seedling but
it includes all of the branches originating from a single point on the ground.
Statements about the bionomics of the shoot do not necessarily apply to the
root system as well. The various branches are referred to as lateral branches,
vertical branches, central axes, or short axillary branches. The “shoot tip” is
the terminal 5-15 cm of one of these branches; “axillary tufts” are the clusters
of leaves often produced in the axils of swollen thorns after the branch
bearing them is over a month old. “Type A” thorns are swollen thorns in

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 323

the form of a flat V. “Type B” thorns are swollen thorns that are highly
twisted; they are usually borne on short axillary branches.

When a shoot is referred to as being “occupied,” this means that the
outer surfaces of the shoot have more than one or two workers of Pseudo-
myrmex ferruginea on them. Occupied shoots almost invariably have brood
in the thorns. An “unoccupied” shoot is one that has essentially no workers
on the outer surface of the shoot. There may be workers and/or brood (or
founding queens) inside the thorns but the workers may not be active on
the surface of the shoot owing to cold weather or small colony size. Both of
these conditions are of a temporary nature. A “founding queen” is a queen
that is rearing her first brood in a thorn. A “colony queen” is the single
queen in a maturing colony. The entire colony may occupy a single shoot,
or may be divided among several shoots which can only be reached by trails
across the ground or foreign vegetation. The “queen-shoot” is the shoot that
contains the colony queen, and the other shoots that contain a part of the
colony are referred to as “auxiliary-unit.” This type of colony structure is
common to many tropical species of Pseudomyrmex. A colony may move
into a new shoot as a mature colony, or it may develop “in situ” from a
founding queen in a thorn on the shoot. The “basal circle” is the area cleared
of living foliage under the shoot by an active colony of P. ferruginea.

Throughout this paper, the names of collection sites consist of only the
precise site. The state and country can be obtained by referring to Figure 6.
The plot designations refer to the plots described in the third section of the
paper and illustrated in Figure 7. A “plot” is the area that includes the con-
trol “subplot” and the experimental “subplot.” The vegetation within a
plot and immediately surrounding it is relatively uniform in appearance and
disturbance regime. The plots are designated by letters; a letter followed by
an “A” indicates the plot was abandoned during the study. All subplots
numbered “1” (e.g., C-1) are experimental subplots. The last subplot (e.g.,
C-2, C-3) is always the control subplot.

BIONOMICS OF ACACIA CORNIGERA

Systematics. The three plants discussed below are the only native swol-
len-thorn acacias found within the study area. The swollen-thorn acacias
in the study area can be recognized by the following diagnoses (adapted
from Rudd, 1964), and Figures 1 to 5.

Acacia cornigera (L.) Willd.

Inflorescence two or more times as long as broad; pinnules with costa and secondary vena-
tion clearly evident in dried specimens; almost never more than 24 pinnae per leaf; petiolar
nectaries elongate trough shaped, from one to eight on mature leaves; yellow to red Beltian
bodies on tips of 50-100 percent of leaf segments; short axillary branches bearing slightly to
highly recurved thorns that may encircle the branch shaft bearing them; cylindrical legume

324 Tue Universiry ScrENcE BULLETIN

never dehiscent along a pre-formed suture, sometimes splitting along unpredictable lines follow-
ing drying, but frequently with thick hard walls that never split.

I recognize the following as synonyms; Acacia spadicigera Schl. & Cham., Acacia cubensis
Schenck, Acacia nicoyenss Schenck, Acacia hernandezu Saft., Acacia furcella Saft., Tauroceras
cormgerum (Schl. & Cham.) Brit. and Rose, Tauroceras cornigerum (L.) Brit. & Rose.

Acacia sphaerocephala Schl. & Cham.

Inflorescence globular to less than one-half again as long as broad; pinnules with costa clearly
evident but secondary venation not in dried specimens; almo:t never more than 24 pinnae per
leaf; petiolar nectaries elongate trough shaped, usually only one per mature leaf; yellow to orange
Beltian bodies on tips of 50-100 percent of leaf segment; no thorns highly recurved, some
with slight twisting along longitudinal axis and sha!lowly curved ups; cylindrical legume never
dehiscent along a pre-formed suture, usually sphiting along unpredictable lines following drying,
almost never with walls so hard that they cannot be broken by gentle pressure.

I recognize the following as synonyms: Acacia veracruzensis Schenck and Acacta doll-
chocephala Saft.

Acacia chiapensis Saff.

Inflorescence a distinct sphere; pinnules with costa clearly evident but secondary venation
not in dried specimens; usually over 40 pinnae per leaf and over 80 pinnules per pinna; petiolar
nectaries round, eroded cratesiform, one at the base of 30-100 per cent of the pinnal pairs and
one to six on the petiole; yellow Beltian bodies usually only on basal two to twelve pinnules of
each pinna; no thorns highly recurved or twisted, some on lower trunk of large trees bent back
out of plane of thorn “V”; flat legume partly to completely dehiscent along two sutures.

Synonym: Acacia globulifera Soff.

In respect to thorn morphology, it is the large volume and the easily re-
moved pith which is of importance to the ant. I propose that the acacias
with enlarged thorns that are commonly used as nest sites by ants be called
“swollen-thorn acacias” and restrict the name “bull’s-horn acacia” and _ its
forms to Acacia cornigera. The following species are definitely swollen-thorn
acacias: Acacia cornigera, sphaerocephala, chiapensis, hindsiu Benth., collinsi
Saff., and melanoceras Beur. This terminology provides a common name at
the sub-generic level for this group with strongly variable thorn morphology.
A. cornigera and A. sphaerocephala have the same common name in the
Spanish language where they occur sympatrically. They are both called
cornizuelo, cuernita, cuernos de toro, and cuernitos, with the first name being
nearly universal. Where sympatric in the study area, A. chiapensis, macra-
cantha Humb. & Bonpl., and farnesiana (L.) Willd. are all called gaisache
in Spanish. The latter two species are not swollen-thorn acacias though they
have stipular spines up to 8 cm in length.

Disrripution. A. cornigera is a widely distributed plant. At present, it
has been collected from northeastern Mexico south to the Isthmus of
Tehuantepec across the Isthmus and thence south to the province of
Guanacaste in Costa Rica. It also has a patchy distribution on the Yucatan
Peninsula and eastern Guatemala. A. sphaerocephala is more restricted; it
has been found in the coastal lowlands along the northern edge of the range
of A. cornigera and around the city of Veracruz, Mexico (Fig. 6). Within
the study area, A. chiapensis is very local in distribution, having been
collected only in eastern Oaxaca and southeastern Veracruz (Fig. 6).

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 325

i.

Fic. 1. a. A flowering branch of Acacia cornigera with flowers unopened, opened, and shed-
ding dead florets. The shoot was three years old, 3 m tall, and growing in brushy pasture
7 km east of Temascal. Photo 17 Jul. 1964. b. Foliar nectaries of Acacia cornigera on a tuft of
axillary leaves from the axil of a type A thorn that has lost its leaf. Same sleet as that in Fig.

la. Photo 17 Jul. 1964.

326 Tue Universiry ScrENcE BULLETIN

Fic. 2. Swollen stipular thorns taken from Acacia cornigera. The upper three twisted thorns
are classified as type B, the middle two as type AB, and the lower five as type A. All the thorns
are from the shoot in figure la. This range of thorn variation is characteristic of thorns from
occupied shoots over two years old. Photo 17 Jul. 1964.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 327

Fic. 3. a. Two mature shoots of Acacia chiapensis. The narrow shoot is representative of
that portion of the population that resembles Acacia cornigera, and it is occupied by Pseudo-
myrmex ferruginea. The spreading shoot is representative of that portion of the population that
resembles Acacia macracantha; it is not occupied. Both shoots are about 6 m tall, and 5-6 years
old. Shoots growing on the roadside 0.5 km west of Temascal. Photo 3 Jul. 1964. b. Flower-
ing branch, legumes, leaf, and type A thorn from Acacta chiapensis. Collected from the right-

hand shoot in Figure 3a on 12 Jun. 1964.

328 Tue University ScrENcE BULLETIN

Fic. 4. Swollen stipular thorns from Acacia chiapensis. The upper two thorns should be
classified as type B since they came from short axillary branches on the main trunk and are
slightly recurved. The lower four thorns are type A from long lateral branches. All of these
thorns came from a shoot 11 m tall that was representative of that portion of the population
that resembles Acacia macracantha (see Fig. 3a). Shoot growing on the roadside 0.5 km west
of Temasceal.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 329

Fic. 5. a. A lateral branch with flowering branches of Acacia sphaerocephala. Taken from
a 2 m tall roadside shoot growing at the northern margin of its geographic distribution (4 km
south of the point where highway 85 crosses the Tropic of Cancer, Tamaulipas, Mexico). Shoot
leafless except for small leaves on the flowering branches. Photo 4 Mar. 1964. b. Mature shoot
of Acacia sphaerocephala. Growing 25 m back from the high tide level on the beach 9 km
north of Anton Lazardo, Veracruz, Mexico. This shoot was occupied by Crematogaster sp.
and is representative in life form of those shoots growing on the sand dunes 20-3000 m of the
ocean. At inland sites, Acacia sphaerocephala is very similar in life form to Acacia cornigera.

Photo 22 Dec. 1963.

1. Hwy. 85 at the Tropic of
Cancer, Tamaulipas.
2. 20 mi. N. W. Ciudad El Maiz,
Hwy. 80, San Luis Potosi.
- Ciudad El Maiz, S. L. P.
- 11 mi S. Ciudad Mante, Hwy. 85,
Tamps.
. 13.6 mi. S, Ciudad Mante,
Hwy. 85, Tamps.

3
4,
5
| 6. 11 mi. N. E. Ciudad Valles,
BZ Hwy. 110, S. L. P.
——s 7. Tampico, Tamps.
8

- 27.5 mi. S. Panuco, Hwy. 122,

ZEEE
Bz a eee

2. EEE Huy. 85, S. lL. P.

Tee : 10, Tuxpam, Ver.

CHWs AAA ll. Tecolutla, Ver.

ince ZZZZgal 12. 8 mi. S. W. Martinez de la
; (2 Torre, Hwy. 125, Ver.
13. 6.5 mi. S. E. Jalapa, Hwy. 140,
EE Ver.; 20 mi. N. W. Conejos, Hwy.
140, Ver.

14. 28.5 mi. W. Conejos, on highway
to Huatusco, Ver.

15. 22.8 mi. W. Conejos, on highway
to Huatusco, Ver.

16. 10 mi. S. E. Conejos, Hwy. 140,

Ver.

Ndeeee

17. Veracruz, Ver.
18. Anton Lazardo, Ver.

19. Alvarado, Ver.

20. San Andres Tuxtla Mts., Ver.

21. Acayucan, Ver.

22. Coatzacoalcos, Ver.

23. San Ignacio de la Llave, Ver.
24. Campo Cotaxtla, Ver.

25. Cordoba, Ver.

26. Acatlan, Oaxaca,

27. Temascal, Oax,

28. Valle Nacional, Oax.

29. Ciudad Aleman, Ver.

30. La Granja, Ver.

31. Villa Azueta, Ver.

32. 10 mi. E. Villa Azueta, Ver.

Fic. 6, The distribution of Acacta cornigera, Acacia sphaerocephala, and Acacia chiapensts
within the study area. The range of Pseudomyrmex ferruginea is congruent with the range
of these three acacias except for a narrow strip along the dunes to the south of the city of
Veracruz. Solid lines indicate roads along which Acacia cornigera has a population density
of at least one plant per ten linear miles. Lines of “X” indicate roads along which there is at
least one Acacia sphaerocephala per ten linear miles. Collection sites are indicated by solid
squares for Acacta chiapensis and solid circles for P. ferruginea. Fine stippling represents the
postulated continuous distribution of 4. cornigera occupied by P. ferruginea. Diagonal striping
represents the postulated continuous distribution of 4. sphaerocephala occupied by P. ferruginea.
Only landmarks and key localities are listed.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 33]

Observations were made in many parts of the study area. Experimenta-
tion was done at Campo Cotaxtla, and the Temascal-La Granja area (Fig.
7). Unless otherwise indicated, the discussions in this paper are believed to
be representative of the A. cornigera and P. ferruginea populations in the
study area. P. ferruginea was found to be relatively uniform in behavior,
morphology and colony structure throughout the study area. A. cornigera
showed some morphological and behavioral variation that was apparently
associated with different habitats and distributional discontinuities. Visits
were made to the following areas within the study area; Coatzacoalcos
(Mar.); San Andres Tuxtla and Tlacotalpan (Dec., Mar.) ; Tuxtepec (Oct.) ;
Veracruz, Cordoba, and Jalapa (Sept., Dec., Jan., Mar., Apr.); Tamazun-
chale and Tampico (Jan.); and Cuidad Victoria (Mar.).

Gross Morrnotocy. In the following discussion of the morphology of
Acacia cornigera, only those aspects which are of direct importance to the
understanding of the ant-acacia interaction have been elaborated.

Mature shoot. Figures 8-15 show representative stages in the development
of a sucker shoot of A. cornigera. At the time of flowering, the shoot may be
nine months to 15 years old. The shrub or small tree has a life form which
is in great part dependent upon the physiognomy of the surrounding plant
community, the age of the shoot, the phytophagous insects present during
the rainy season, and the effectiveness of the ant colony in removing damag-
ing organisms. It usually has one main trunk and this is often developed
from an old root stock left after cutting or burning. In open pastures and
on riverbanks, it is often of spreading habit and less than 5 m tall (Fig. 13).
Where competing with surrounding vegetation for light, it is usually emer-
gent with a long thin trunk (Fig. 14-15). In second growth vegetation, oc-
cupied shoots of A. cornigera continue to stay at the general canopy level
or above until the competing species heights have reached 12-20 m. At this
time, A. cornigera has a D.B.H. of 14-18 cm. When heavily shaded, the
shoots rarely grow over 2 m in height and are then very slender (Fig. 21a).
When the ruteline scarab Pelidnota punctulata Bates and the larvae of the
noctuid moth Coxina hadenoides Guen. are common, their persistent re-
moval of the shoot tips of certain occupied shoots during the rainy season
produces flat-topped, thick canopied shoots. In most plant communities,
phytophagous insects are sufficiently abundant during the rainy season so
that they severely stunt the growth of unoccupied A. cornigera by destroying
mature foliage and shoot tips.

The growth pattern is affected by the same factors that affect the life-
form of A. cornigera. During the first three months to a year, shoots from
old root stocks characteristically lack long lateral branches; vertical growth
of the main axis is emphasized. The extent of later growth of lateral branches
is associated with the amount of light received by the part of the crown

332 Tue University Science BULLETIN

hn \ Presa
u ie (2) =

Temascal
pe Seno
H\ the
[ teet am abl
7)
[| & 7 pe
°
o |EL ve, 2
© |house % m
b
5 < N 2 moa Paha s
: ot
qj.
is \4
2
a os
09 rN
Y
= cue
we)
co gi AS

I LE
A H

Sr. Torrealva's
house

to La Granja

Fic. 7. Diagrammatic representation of the experimental plots and subplots established in
this study between Temascal and La Granja. Area (1) lies to the north of Temascal and below
the earth-fill dam, the Presa Miguel Aleman. Area (2) is on Seftor Farfan’s land around and be-
hind El Mocho’s house. Area (3) lies across the road from Seftor Torrealva’s house. The plots
are not drawn to scale; for dimensions see the plot descriptions.

ws)
Ww

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT _ 3:

Fic. 8. A pair of 21 day old sucker shoots from a 14 mm diameter burnt stump of Acacia
cornigera in a pasture burned 32 days previously on 1 Dec. 1963. Each thorn contains a
founding queen. The uppermost shoot tip is eaten off by a larva of Coxina hadenoides while the
lower one is undamaged. The shoot is 15 cm tall and growing 7 km east of Temascal.

oS)
LON)
are

Tue Universiry ScieENcE BULLETIN

Fic. 9. Unoccupied 5 month old sucker shoot of Acacta cor nigera growing in an area that
was burned in May 1963, 7 km east of Temascal. Each thorn contains a founding queen, and
five founding queens of Pseudomyrmex ferruginea can be seen on the outside of the shoot.
Four hours after sunrise, this shoot had the exceptional number of 31 founding queens on it
outside of the thorns. A larva of Halisodota sp., probably H. pura, was observed to eat out
the damaged green thorn on the short lateral branch. The upper terminal shoot tip was eaten
by an unidentified insect; its mandibular scars are visible on the right hand half of the terminal
thorn. The founding queen near the apex of the fourth swollen thorn from the base is attempt-
ing to pull the occupant founding queen out by her antenna. Photo 10 Oct. 1963.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 335

Fic. 10. An occupied 5 month old shoot of Acacia cornigera growing in vegetation that was
cut but not burned in late May 1963 (plot P). It was 97 cm tall. This is an auxiliary-shoot of
the queen-shoot 70 cm to the left that was cut at the same time. Of the 37 type A thorns on
this shoot, 29 contained brood and workers of Pseudomyrmex ferruginea, and there were ap-
proximately 400 workers on the shoot. The surrounding vegetation is almost entirely sucker
shoots from old root stocks of Croton miradorensis, Tournefortia hirsutissima, Eupatorium
odoratum and Leguminosae, and annual Solanum torvum and Labiatae. Photo Oct. 1963.

336 Tue University ScteNcE BULLETIN

Fic. 11. An occupied 15 month old shoot of Acacia cornigera (near plot P). It was 164
cm tall. The surrounding vegetation was 95 cm tall but was cleared to expose the shoot. The
shorter shoot to the right is a four month old auxiliary-shoot with 161 workers, while the
queen-shoot had 1,686 workers on and in it. Photo 26 Mar. 1964.

Fic. 12. An occupied 17 month old sucker shoot of Acacia cornigera in subplot E-3 before
treatment. It was 4.5 m tall. The surrounding vegetation was about 3 m tall but was cleared
to fully expose the shoot; it was composed of Bixa orellana, Croton glabellus, Jatropha urens,
Cassia bicapsularis, many species of woody vines, and other less common shrubs. The shoot on
the right is of the same age and history as the central one. Shoots of this size usually contain
a queen-unit with about 10,000 workers by the end of their second growing season. Photo
late Sept. 1963.

338 Tue University ScrENcE BULLETIN

~~} :
tac
o
of oe
*~ oe Sere,
0 Phi
ih ?.

mae GE

Fic. 13. An occupied 4 year old shoot of Acacia cornigera. It was 520 cm tall. This shoot
had apparently produced all of its growth in a heavily browsed and grazed pasture. A three
year old shoot of this size contained a colony of Pseudomyrmex ferruginea with 12,269 workers.
Directly in the center of this shoot is a recently vacated nest of Ptangus sulpheratus, the Derby
Flycatcher. Photo early Oct. 1963 in a pasture 3 km west of Las Tinajas, Veracruz, Mexico.

INTERACTION OF THE BULL’s-HoRN ACACIA WITH AN ANT INHABITANT 339

Fic. 14. An occupied 4 year old emergent shoot of Acacia cornigera in subplot V-2. It was
7 m tall and a queen-shoot with 9,257 workers. This shoot produced its growth in vegetation
that was ungrazed after the first year after it was cut. The surrounding vegetation is mostly
Compositae, Guazuma ulmifolia, Helicteres guazumifolia, Tabebuia pentaphylla, Cassia bicap-
sularis, Malvaceae, and other shrubs and trees. Photo late Sept. 1963.

Fic. 15. An unoccupied 7 year old shoot of Acacia cornigera removed from 8 year old re-
generation east of subplot H-1. It was 11 m tall. This shoot’s canopy was about 50 cm below
the general canopy of the vegetation from which it was removed. It had 39 mature leaves
and no intact shoot tips. This shoot would have probably died within the next year although
its stump produced quite vigorous sucker growth. The straight and slender trunk with naturally
pruned branches is characteristic of shoots that have developed in ungrazed and unburned
regeneration. Photo early Jul. 1964.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 341

producing the branches. The axis of an elongating branch is bright green
down to the fifth to tenth swollen thorn. A dark green photosynthesising
layer is present, and usually externally visible, beneath the extremely thin
epidermis over most of the shoot. It is probable that photosynthesis under
the epidermis constitutes a significant portion of that of the shoot as a whole,
especially during the dry season when the shoot may bear as few as 5 per cent
of the number of leaves present during the rainy season. A. cornigera is
quick to respond to light; this is shown by the many trees whose shoot tips
have bent toward a hole in the canopy and then grown up through it. The
wood of trees over 3 months old is tough and springy. During the first two
years of growth, leaves are usually produced nearly to ground level; other
species of trees and tall shrubs of the same height in the same plant com-
munity usually have leaves only on the uppermost part of the shoot. In
dense vegetation the production of leaves low on the shoot by A. cornigera
is most strongly associated with the pruning activities of P. ferruginea work-
ers; this pruning often allows sufficient light penetration for the leaves of
A. cornigera to be functional below the canopy of the surrounding vegeta-
tion. Dead branches, and thorns over three months old, are usually shed.

The short flower-bearing branches (Fig. la) are produced during the
middle of the dry season but contribute little to the form of the shoot. These
branches have small leaves with fewer pinnae, do not develop swollen
thorns, and die after the seed crop (Fig. 16) has fallen during the following
dry season.

The roots have numerous nodules; these are first obvious when the
seedling is about three weeks old and has three to five leaves. A three year
old root system may have a tap root over 3 m long. In wet areas, lateral
roots are well developed and vertical shoots sometimes develop from them.
Root stock cleaned of all its shoots can be transplanted and produces healthy
shoots. Root grafting has not been ascertained.

Leaf. The dark green, bipinnate leaves (Fig. 17) show high variation in
numbers of pinnae and pinnules per leaf on the same shoot. A representa-
tive leaf has 10 pinnae and 300 pinnules. A leaf of this size (18 cm long) is
usually subtended by a large swollen stipular thorn and is on a long lateral
branch or the main axis (Fig. 9). Small leaves are produced in tufts in
the axils of swollen thorns, or on flowering branches, especially during the
dry season.

As a result of the annual leaf drop, rapid vertical growth, and lack of
retention of large branches, there are rarely over 600 leaves on a shoot. Dur-
ing the first year of growth from a stump, a shoot occupied by a large colony
of P. ferruginea produces about 300 leaves. This number is most commonly
reduced by two types of defoliation. The larvae of the syssphingid Adeloce-
phala mexicana Bsdvy. may eat nearly all of the mature leaves from a shoot

Tue University ScreNcE BULLETIN

342

yi

l.

after the rains had started, but before this shoot had responded by pro-

A section of the canopy of a mature shoot of Acacia cornigera with a normal heay
This shoot was 4 m tall and growing in a bushy pasture 1 km north of Temasca

16.

seed crop.

Fic.

Photo 26 May 1964,
ducing new shoot tips.

Fic. 17. a. Lateral branch from an occupied shoot of Acacia cornigera. While the shoot
tip is still intact, this branch has nearly ceased lateral extension. Note that the leaves project
well past the ends of the thorns. b. The same lateral branch as in Fig. 17a, but 5 minutes
later, after a tame Brocket Deer (Mazama americana temama; iemazate) had browsed the hand-
held branch (worker ants had been removed from the outside of the branch). Photo 20 Jul.
1964, of a branch from a roadside shoot growing 18 km east of Temascal.

in the 1-2 m height range, but is deterred through attack by P. ferruginea,
trom destroying the shoot tips. In cases where all of the shoot tips have been
removed from an occupied shoot, the adults of the rutelline scarab Pelidnota
punctulata or the larvae of the noctuid moth Coxina hadenoides are usually
responsible. There is a subsequent severe reduction in leaf production until
new shoot tips are initiated.

In the Temascal area, the pattern of leaf drop of A. cornigera differs in
some respects from that of other deciduous plants. Nearly all leaves pro-
duced during the rainy season of the previous year are gradually dropped
during the warm and dry season (Mar.-May). Leaf drop begins with the
oldest leaves and gradually moves upward. Therefore, the growth since the
end of the last dry season can be identified as that above the first swollen
thorn with its subtended leaf still present. The shoot is rarely leafless, and
the length of time without leaves varies greatly among individuals. Though
little lateral and vertical growth occurs during the dry season, a few axillary
leaf tufts are continually produced on shoots over a year old and flowering
branches with small leaves are produced in the middle of the dry season.

The time of production of the new crop of leaves varies with the age of
the shoot and the individual plant. Shoots less than a year old have only

344 Tue Universiry SciENCE BULLETIN

partial leaf drop during the dry season and shoots less than six months old
usually have no loss of leaves. The new period of rapid vertical growth starts
during the last one to two months of the dry season in plants under one year
of age. In these shoots the new branches are often produced from a thorn
axil nearly at the top of the shoot and growing upward, immediately produce
a height increment as well. On the other hand, shoots over a year old begin
vertical growth near the time of the first rains. This growth is in the form
of vertical branches from thorn axils well below the tips of the highest of the
previous year’s branches, but still at or above the level of the surrounding
canopy of other species of plants; there must be several weeks to a month of
growth before a height increment can be recorded.

The shoot tip of A. cornigera differs in several important ways from that
of acacias not inhabited by ants. It is glabrous and slightly sticky; the work-
ers of P. ferruginea lick off this sticky material and thus aid in the separation
of the pinnules. While the two terminal unexpanded leaves are clasped
tightly over the shoot apex as is the usual case in Central American acacias,
the terminal portion of the shoot tip is extremely fragile and can be broken
off by bending or snapping with the finger; the shoot tips of Acacia macra-
cantha, Acacia farnesiana and the pale-leafed form of Acacia chiapensis are
very tough and fibrous. The shoot tip of A. cornigera is bland to pleasant
tasting to humans in contrast to that of acacia spp. not associated with ants.
In the latter, the shoot tips are very bitter tasting.

Beltian bodies. The oval to tear-drop shaped Beltian bodies are the
modified ends of the leaf segments (Fig. 18) that are borne one to the tip of
each pinnule, pinna, and leaf rachis. This omnipresence generally applies
only to shoots over 1 m tall growing in full sunlight. The Beltian bodies are
usually removed within five days of their appearance by P. ferruginea work-
ers on occupied shoots, or after some indeterminate period by P. ferruginea
founding queens, P. gracilis mexicana or Solenopsis geminata Fabricius
(rarely) on unoccupied shoots. They are cut up by the Psewdomyrmex
species and fed to the larvae. Due to the continual harvesting activity on
occupied shoots, Beltian bodies are usually found only on new shoot tips.

During the first two months of the rainy season, groups of shoots oc-
cupied by one P. ferruginea colony sometimes produce more Beltian bodies
(new foliage) than the colony will harvest. In this case, many of the leaves
retain their pinnule-end Beltian bodies because the P. ferruginea colony does
not allow other possible harvesters on the shoot. Beltian bodies on the ends
of pinnae and the leaf rachis are always harvested if the shoot is occupied.
Depending on the weather, those that are not harvested eventually dry and
drop off, or rot.

No direct use to the plant of Beltian bodies is postulated in the literature.
They were regarded by F. Darwin (1877) and Schimper (1888) as homolo-

WI

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 34

i
ee

Fic. 18. Two shoot tips of Acacia cornigera from plot N. The left-hand shoot tp is un-
damaged and was taken from an unoccupied shoot; the unharvested Beltian bodies on the ends
of the raches and pinnae are mature (ripe) while the Beltian bodies on the pinnules are not yet
expanded. The right-hand shoot tip was eaten off by Coxina hadenoides; the shoot from which
it was taken was occupied by Crematogaster sp., one of which is walking between the shoot

ups. Photo June 1964.

gous to serration-glands on the leaf margins of other plants. Preliminary
examination indicates clearly that they are modified ends of the leaf parts,
and that the rachis extends out into them. Darwin considered that they were
full of oils and proteins. Preliminary experiments to determine the food
value of Beltian bodies indicate that they are similar to yeast in quantity and
quality of nutrients. This degree of nutritive value of foliar tissue is unusual
(H. T. Gordon, personal communication). About 3000 Beltian bodies ripen
on a2 m tall occupied tree per day during the first three months of the rainy
season; this means about 450 mg of solid food per day is available to the

346 Tue University ScIENCE BULLETIN

ant colony. With the exception of several species that require considerable
discussion, Beltian bodies are not found on acacia species that lack a rela-
tionship with ants in the genus Pseawdomyrmex.

Foliar nectaries. The foliar nectaries are nectar-producing glands on the
dorsal side of the petiole and/or rachis of all leaves of mature shoots (Fig.
lb). They are elevated trough-shaped and 05-6 mm in length. Large
nectaries produce a sphere of clear, sticky nectar about 2 mm in diameter
during a 24 hour period; the nectar flow is heaviest about one hour before
sunrise. The workers of P. ferruginea collect the nectar as it appears; if the
shoot is unoccupied, many species of Hymenoptera visit the nectaries. A
shoot 2 m tall with 200 leaves produces about | cc of nectar during a 24 hour
period. The nectar consists of about equal parts by weight of sucrose and
fructose at an approximate concentration of 40 mg/cc HzO (determined in
the laboratory of Dr. H. T. Gordon, University of California, Berkeley).
While the foliar nectaries are mentioned casually in almost all accounts of
the swollenthorn acacias, no detailed accounts of their morphology or
physiology are published.

Stipules. Over 99 percent of the bilaterally symetrical stipules of A.
cornigera can be placed in two size classes: minute and swollen. Minute
stipules are 3-8 mm in length, and hard and needle-tipped when dry. They
accompany about 50 percent of the leaves produced by a shoot during its
lifetime and are for the most part confined to the leaves in axillary leaf tufts
and on flowering branches. The ants do not enter them. Mature swollen
thorns (Fig. 2 and 19) are 2-25 cm in length, hard-walled, and hollow with
sharp tips when dry. The “V” shaped type A thorns are part of the large
leaves found on lateral branches and the main vertical axis of the shoot. The
highly convoluted type B thorns are found on short branches originating in
the axils of type A thorns; though fewer in number, the type B thorns have
1-2 cc of space in each of them and thus constitute the major source of space
that the colony has for rearing brood (Fig. 20). The volume of space
available to the colony in a 3 m tall shoot is about 400 cc. Completely dry
thorns that have not been cleaned out by P. ferruginea are about one-fourth
empty, but are not available to the ants because they cannot cut into the
dry thorn.

The walls of swollen thorns have a number of characteristics which are
of importance to the ants living within them. By two weeks of age, a
swollen thorn is usually completely dead and dry except for a narrow median
ventral strip of living vascular tissue connecting the leaf petiole with the
branch. The hard and polished outer surface does not readily absorb water
but the inner walls are quite absorbent. The walls are 0.7-4 mm thick and
in general the type B thorns have the thickest walls. Type B thorns often
have a raised ridge of hard tissue running along an inside ventral surface;

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 347

ee emer ad at

5 6

78 9 401-22
* . * Sean * *

Fic. 19. a. A 75 cm long section taken from a point one-half way up a 4 m tall occupied
shoot of Acacia cornigera that was cut a week before the photograph. The type B thorns form
tight clusters surrounding each short axillary branch that bears them, while the type A thorns
are borne on the elongate lateral branches. This shoot was about 24 months old and growing
in dense, ungrazed regeneration along the Rio Tonto, 8 km east of Temascal. b. A 45 cm
long section taken from the terminal 150 cm of the main axis of the shoot in Fig. 19a. This
portion of the shoot characteristically bends over at the end of the rainy season and the next
year’s main vertically lengthening axis develops from a thorn axil near the base of this portion
of the shoot.

this may serve as a reinforcement. They are extremely tough and hard to
split by hand; they do not break across the grain but split along the longitudi-
nal axis. Type A thorns are more brittle and break across the grain. Both
types are most easily opened by cutting through the living tissue at the thorn
base.

The walls of the thorns have a noticeable effect on the microclimate in-
side the thorn. Temperature recordings made with thermistor probes
(Yellow Springs Instrument Co. #402 probe) show that 1) fully insolated
dark thorns can be as much as 2.5° C. hotter inside than light colored ones,
2) at high air temperatures outside when the air temperature is rising or
falling, the temperature inside the thorn rarely lags more than an hour be-
hind that outside the thorn, and 3) at low temperatures, the temperature

348 THe University ScIENCE BULLETIN

SUM OF THE LENGTHS OF THE
TWO HALVES OF THE THORN (mm)

0.0 1.0 2.0
TOTAL THORN INTERNAL VOLUME (cc)

Fic. 20. a. Graph of the relation between thorn volume (abscissa) and the sum of the
lengths of the two sides of the thorn (ordinate). Type A thorns are represented by dots, type
AB thorns by circles, and type B thorns by x’s. This graph contains all of the swollen thorns
that had been hollowed out by Psewdomyrmex ferruginea on the queen-shoot in Fig. 11. The
one small type AB thorn and the two small type B thorns had only one-half of the thorn
swollen. b. Graph of the thorns on the auxiliary-shoot in Fig. 11.

inside the thorn is rarely more than one-half degree C. higher than the air
temperature even when the thorn is fully insolated. While the absorbent
inner thorn wall probably soaks up any excess fluids from the ants in the
thorn, the outer thorn wall probably serves as an effective barrier against
water loss during the dry season, and prevents the thorn contents from being
drowned during the rainy season. The entrance hole of the thorn is usually
plugged by an ant’s head, making the thorn an almost watertight container.

The rate of swollen thorn production varies greatly in respect to the
time of year and the size of the shoot. Since only about one-half of the
leaves produced have swollen thorns, it is normally only when new branches
are produced that the number of swollen thorns on a shoot increases. A
shoot occupied by P. ferruginea produces about 200 swollen thorns during
its second year of growth (changing from 2 to 3.5 m in height). These
thorns would weigh at least 600 g. In five years of growth a shoot produces
on the order of 4 kg of thorns.

Development of vegetative features. In repeatedly disturbed areas such
as pastures, crop fields, and roadsides, the majority of Acacia cornigera shoots
grow from root stocks older than their respective shoots. In natural dis-
turbance sites, the shoot is more often of the same age as the root stock. It
appears that most root stocks of A. cornigera die after succession has pro-
ceeded without interruption (no grazing, burning or cutting) for 10 to 20
years. Therefore the shoots are usually of the same age as their roots follow-
ing the first clearing of an old forest. In this case, the regeneration of A.
cornigera is by seed. The shoots under experimental observation in this study
were nearly all younger than their roots because the study sites had been

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 349

repeatedly cleared during the past 30 years. A. cornigera shoots derived
directly from seed are very difficult to find except during the first two
months of the rainy season. Once a seedling growing in full sunlight has
reached a height of 75 to 100 cm, it is difficult to distinguish it from a new
sucker growing from a 2-3 year old root system unless the upper part of
the root stock is exposed and examined. Large older roots produce new
suckers that are very mature in appearance.

Very young seedlings do not initially possess the diagnostic characteristics
of a swollen-thorn acacia: swollen-thorns, Beltian bodies, and foliar nec-
taries. The number of pinnae and pinnules per leaf increases gradually
until the 10th to 20th leaf. One or two Beltian bodies appear on the basal
pinnules of the basal pinnae of the 7th to 16th leaf. Each successive leaf
bears a few more Beltian bodies, both further out on the leaf and further
out on the pinna. The first type A swollen thorn is produced at the 9th
to 12th node. The first axillary branch with a type B thorn does not appear
until the shoot has 75 to 100 nodes. The first traces of a foliar nectary do
not appear until the 5th to 8th leaf and the first large functional nectary is
usually on the petiole of the leaf associated with the first swollen thorn.
Multiple nectaries on the leaves appear at about the same time as the first
type B thorns. On well-watered soil in full sunlight, it takes a seedling 30
to 45 days to produce its first swollen thorn. It should be noted that while
Beltian bodies and foliar nectaries are not present on the newly germinated
seedling, they are present by the time the first swollen thorn is produced
and therefore a new founding queen in the first swollen thorn has an im-
mediate source of food.

Shading suppresses the development of Beltian bodies, nectaries, and
swollen thorns. A heavily shaded seedling, by the time it produces its 40th
leaf, is still producing thorns, Beltian bodies, and nectaries with the same
form and size as those in the 10th to 20th nodes of fully insolated shoots
(Fig. 21). Under the climatic regime at Temascal, a fully insolated seedling
on well-watered black soil will grow at least 200 cm in a year. However,
since seedlings are often not occupied by an effectively protective colony of
P. ferruginea during their first year of growth, they receive moderate to
severe damage to the shoot tips by phytophagous insects and thus rarely
grow more than 100 cm in the first year.

Suckers from cut or burned stumps have much more the appearance of
mature shoots than do seedlings. The leaves on a new sucker from a large
stump often have a Beltian body on the end of every leaf segment, large
type A thorns and a few type B thorns, and well developed foliar nectaries.
The immediate production of these properties by the acacia are of obvious
importance to a large ant colony moving from a shoot that has been cut
into the new sucker growing from the shoot’s stump.

350 THe Universiry ScreENcE BULLETIN

Fic. 21. a. An occupied shoot of Acacia cornigera that had grown for 7 months from a
cut stump in the dense shade in plot R. Photo 17 Mar. 1964. b. An occupied shoot of Acacia
cornigera that had grown for 7 months from a cut stump in the open brushy pasture adjacent
to plot F. Photo 17 Mar. 1964. Both shoots were growing on the same kind of soil and are
representative of the differences found between shaded and insolated shoots. The shoot in Fig.
21a had about 150 workers in the ant colony while that in Fig. 21b had about 2,000 workers;
both were colonies that had invaded from other shoots that had been cut. Both photographs to
same scale.

The growth rate of occupied sucker shoots is noticeably higher than that
of occupied seedlings. It is commonplace for a sucker from a large stump to
grow 400 cm upward in a year. Sucker regeneration takes place at any time
during the year, but as the sucker ages it becomes more responsive to seasonal
climatic changes (leaf drop during the dry season, not initiating growth until
the first rains). In evaluating height increments for shoots of various ages,
the time of year during which the growth took place must be considered.

Repropuctive Brotocy. Pollination. Acacia cornigera is pollinated by a
number of species of bees (e.g.. Bombus medius, Ceratina extimia, Mega-

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 351]

chilidae, Halictidae). These bees are occasionally chased by Pseudomyrmex
ferruginea, but the workers are not effective in keeping bees off the flowers.
While the workers of P. ferruginea are commonly seen digging into the
pollen covered surface of the inflorescence, it is doubtful that they effect polli-
nation by carrying pollen from one flower to another since they are constantly
cleaning themselves. The major flowering period varies widely from one
part of the study area to another, but is usually within the period from the
middle of the dry season to the first month of the rainy season. The buds of
the flower bearing branches are first evident about three months before
flowering begins. Small green legumes are present shortly after flowering,
but these require 10 to 12 months to mature. This means that the acacia
must live this long after flowering to produce viable seed.

Seed production. A. cornigera produces comparatively few seeds for a
plant of disturbed sites. In its lifetime, a shoot produces about 60,000 seeds.
Better than 99 percent of these seeds are destroyed by the larvae of bruchid
beetles. In the Temascal area, the common species were Acanthoscelides
oblongoguttatus Fahr., Mimosestes sp., and Stator subaeneus Schaefler. The
adults of these beetles may be found on mature pods at any time during the
year, and the author has yet to find a mature seed pod of A. cornigera that
was over two months old and not attacked by one of these species of bruchids.
Some natural enemies of the bruchids have been discovered, but an investiga-
tion of their biology has not been made. When a seed crop was picked be-
fore bruchid exit holes were evident in them, as many as 75 percent of the
seeds were undamaged. However, when the first generation of beetles
emerges, the beetles oviposit in those legumes which are not infested and
complete the seed destruction. The legumes may remain on the tree for two
to five months after maturation.

That some of the seeds escape destruction is due to one of the two
methods of seed dispersal used by A. cornigera: birds and water. Birds
readily split the pods to eat the sweet yellow pulp around the seeds and in
doing so ingest some seeds. In the Temascal area, the commonest species
are the Black-headed Saltator (Saltator atriceps), Grayish Saltator (Saltator
coerulescens ), and the Plain-tailed Brown Jay (Psilorhinus mexicanus ). The
seeds that are distributed by defecation normally fall in relatively good sites
for germination since these four birds are common in new second growth
vegetation. Secondly, when seeds are removed early from the site of bruchid
infestation, some escape damage. They are eaten shortly after the pods
ripen, as food for these birds is normally scarce during the dry season. Seeds
taken from the intestines of these birds germinated normally.

A second group of birds open the pods but do not eat the seeds (Black-
throated Oriole, Icterus gularis, Derby Flycatcher, Pitangus sulphuratus,
and the Melodious Blackbird, Dives dives). They eat the yellow pulp around

352 Tue University ScIENCE BULLETIN

the large seeds, but let the seeds fall to the ground where they sometimes
germinate directly under the parent tree. By opening the pods they likewise
remove some seeds from the site of bruchid infestation. The workers of
P. ferruginea are often consumed in large numbers by the Black-throated
Oriole but rarely found in the other species mentioned.

Throughout much of the study area, there are two general forms of
legumes; one has a very thin wall that splits by twisting when dry, and the
other has a thick wall that does not split. There are intermediates. The birds
appear to find the thin walled form much easier to open; many more of these
are opened than the thick walled form. Associated with this, it is the thin
walled form which has spread away from the natural disturbance sites along
watercourses in areas newly opened to agriculture. If not opened by birds,
the thin walled form eventually twists open and drops its seeds; by this
time the bruchids have destroyed better than 99 percent of the seed. The
thick walled form appears very well suited to dispersal by water. In old
established pastures, the thick walled form is generally the dominant form.

Invasion of new sites. Land recently exposed to plant colonization (land-
slides, sand bars in rivers, lake beaches, river banks) and newly cleared
forest (over 10 to 20 years old) is very slowly colonized by A. cornigera
in the area between Temascal and La Granja. Despite the nearby presence
of seed bearing shoots, the invasion rate rarely exceeds ten established plants
per acre per year and is usually much lower. This appears to be most directly
associated with the high mortality rate of unoccupied seedlings. However,
most of the man-made disturbance sites are repeatedly cleared of their above-
ground vegetation by fire or cutting. Roots and seeds of A. cornigera are
often not destroyed by this clearing procedure. As the seed invasion con-
tinues, and the shoots already present drop seed around their bases, the
density of root systems increases to levels as high as 2,000 per acre (estimate
from swamp pasture east of Tuxtepec, Oaxaca, Mexico) and 800 per acre
(counts from dry pasture in plots N and O). Brushy pastures, with their
occasional burning and/or clearing (2-5 year intervals), and vegetation
cover opened by browsing, have the optimal conditions for producing large
populations of A. cornigera.

Not all root stocks bear maturing shoots. During any one regeneration
cycle, 10-75 percent of the root systems have a vigorously rising shoot oc-
cupied by P. ferruginea. The remainder have short shoots that are usually
less than 75 cm tall with 4-20 swollen thorns. Of these thorns, 10-100 percent
have founding queens of P. ferruginea in them. As a colony develops in situ,
or the shoot becomes an auxiliary shoot, a rapidly rising sucker develops if
not heavily shaded.

Under well developed canopies that are young enough still to have mature
A. cornigera in them, there is a new crop of A. cornigera seedlings during

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 353

the first two months of the rainy season. During the following three years,
this cohort suffers nearly 100 percent mortality; individuals rarely grow over
1 m tall and are poorly developed (Fig. 21a), even when occupied by an
established colony of P. ferruginea. On these shaded shoots, there is a very
low rate of production of new colonies in situ. While ample founding
queens are present, some factor in the shaded site often prevents successful
colony growth. This is very likely associated with the low rate of nectar and
Beltian body production.

REQUIREMENTS IN THE PuysicaL ENviroNMENT. Following Holdridge’s
classification (1964), Acacia cornigera is found in the study area from
Tropical Tropical Arid Forest to Low Subtropical Dry Forest to Low Sub-
tropical Wet Forest to Tropical Tropical Moist Forest. Following Leopold
(1962), it is found in Pine-Oak Forest, Mesquite-Grassland, Rain Forest,
Tropical Evergreen Forest, and Savanna. Despite this occurrence in a wide
range of formation types and vegetation life zones, there are several common
denominators to its habitats: 1) they are disturbed sites where sunlight reaches
very close to ground level, 2) since such sites are frequently cleared of all
vegetation above ground, the regenerating occupied suckers of A. cornigera
are able to initiate growth as canopy members or emergents, 3) being fully
exposed to sunlight, the canopies of A. cornigera experience the highest air
temperatures that can be achieved by the prevailing weather, 4+) the other
species present as canopy members during the early stages of succession have
rapid growth as well, 5) there are high populations of phytophagous insects
present and these sites are frequently subjected to cattle grazing and
browsing, and 6) the moisture content of the soil fluctuates greatly, depend-
ing on the season.

Within a wide range of temperature, rainfall, and soil type regimes, the
presence and density of A. cornigera is primarily a function of the disturb-
ance history of the plant community. In the literature, it is characterized as
a plant of the land-ward side of ocean dunes to upland foothill oak forest
(1,200 m) associated with pasture and frequently cleared areas (Skwarra,
1934a, 1934b; in the state of Veracruz); as a plant of the tropical littoral to
400 m elevation in clearings, open country, and roadsides (Wheeler, 1913;
in Guatemala and Costa Rica); and as a plant of ravines, riverbanks,
savannas, and other naturally disturbed areas (Belt, 1874; near Matagalpa,
Nicaragua). It appears that Wheeler’s discussion of A. cornigera in Costa
Rica applies to Acacia collinsti Saft. since A. cornigera (—=Acacta nicoyensis
Schenck) has not been found at Alajuela or other sites over 200 m elevation
during intensive searches for this plant in Costa Rica. In all cases A.
cornigera is regarded as a plant of lowland, tropical, disturbed plant com-
munities.

354 Tue University ScIENCE BULLETIN

At the margins of its range, the general density of shoots is often very low,
there being only an occasional streamside, roadside or pasture plant. Yet
within 10-15 km or less of this margin, there are often very dense popula-
tions in swamps, pastures or roadside ditches. If a site at the margin of the
range is suitable for the growth of one shoot, the population often builds
up to densities equal to those found in the most central parts of the range.

Moisture. Within the study area A. cornigera is more restricted in respect
to aridity than is P. ferruginea. This is evident from Figure 6 where it can
be seen that P. ferruginea crosses the interface between the ranges of A.
cornigera and A. sphaerocephala at 11 mi. N.E. Ciudad Valles, 27.5 mi. S
Panuco, 22.8 mi. W. Conejos, and between Veracruz and Anton Lazardo
without interruption. At each of these sites, the climate becomes drier as
one moves into the range of A. sphaerocephala. At the range limits of the
distribution of A. cornigera to the north, northwest, and northeast, and at
the break at 28.8 mi. W. Conejos and 10 mi. S.E. Conejos, its disappearance
appears to be in agreement with regions of about 1,000 mm annual rainfall
and a dry season of about 6 months duration. Except at 20 mi. N.W. Ciudad
El Maiz, it is replaced by A. sphaerocephala at these points. At this place,
the last plants of A. cornigera are found in barren rocky pastures with
Opuntia spp. and Acacia farnesiana. These sites appear to have been covered
with oak forests at one time. A. sphaerocephala may be prevented from
reaching this area by a band of east facing slopes of the Sierra Madre Oriental
in the area of Ciudad El Maiz with hiehe: rainfall (approximately 2,000 mm
annual). There is a strong possibility that the inability of A. cornigera to
extend its range into drier sites is associated with its inability to hold its
leaves and produce new axillary leaf tufts when the dry season is too long.
If the shoot is completely leafless for more than a month, the ant colony dies
or has its numbers greatly reduced by starvation. If this happens, the acacia
enters the rainy season in an essentially unoccupied condition with conse-
quent poor growth. A. sphaerocephala can hold some leaves and produce
new leaves at least a month longer than can A. cornigera.

In the area to the south of the city of Veracruz along the beach above
Anton Lazardo, A. cornigera is replaced by A. sphaerocephala on the dunes
within 500-3000 m from the high tide line. At Tuxpam, Tecolutla, and
Coatzocoalcos, A. cornigera has been found as an occasional plant within
50 m of high tide, being in some cases the first woody plant. Veracruz has a
longer dry season than either of the three later sites, and it appears that the
relatively well drained new dunes are too dry for A. cornigera. The A.
sphaerocephala on these sites forms a low dense mat which is not occupied
by P. ferruginea.

There are also inland sites within the range of A. cornigera that are ap-
parently too dry for it. In the dry hilly oak forest around Villa Azueta, the

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 355

acacia is virtually absent even on the roadsides; only 25 km away in the
wet river bottoms it is very common. Along the road southeast of Conejos
the first appearance of A. cornigera is in roadside swamps as the altitude
decreases moving down the arid shelf from the west. In the Temascal area,
the oak-grass covered knolls often have no A. cornigera on them. This may
in part be due to the rapid rainfall runoff through laterite soils although
the high frequency of fires is probably of greater importance.

Temperature. At 6 mi. S.W. Tamazunchale, 8 mi. S.W. Martinez de la
Torre, 28.5 mi. W. Conejos, and Valle Nacional, the failure of A. cornigera
to extend its range appears to be associated with lower temperatures. At
each of these sites, the number of days with the maximum temperature be-
low 24° C. is about 90-110 per year. During weather of this frigidity, the
workers of P. ferruginea are generally not active outside of the thorns. One
result of this is that the colony is reduced in size, apparently through starva-
tion. During the cold period at Temascal in December through February,
almost no vertical growth occurred in experimental plots. Even those plants
with intact shoot tips had almost no growth. Shoot tips are rapidly removed
by phytophagous insects during cool weather owing to the lack of patrolling
workers of P. ferruginea; these shoot tips are only very slowly replaced by
the acacia during cool weather. A. cornigera is apparently cold-hardy as is
shown by experience with cultivated and escaped shoots in Florida. They
are not damaged by occasional frosts and survive the coldest winters. Fire
rather than cold seems to be the factor which prevents naturalization (C. F.
Dowling, personal communication).

Soil type. High densities of Acacia cornigera have been found on beach
sand, red and yellow laterites, black soil over limestone, black swamp soils
laden with high organic matter, and on many other unidentified soil types.
Within the study area and its range, no soil types have been found on which
A. cornigera is consistently lacking. The low densities commonly recorded
on red-yellow laterite hills are largely due to the disturbance history (regular
burning) rather than unsuitability of the soil. Once these areas are farmed
sporadically and/or grazed, A. cornigera sometimes becomes common on
well watered laterites.

Sunlight. As is the case with many plants of early succession, A. cornigera
requires direct or intense indirect sunlight for normal development of
swollen thorns, Beltian bodies, flowers, and height. Most of the occupied
shoots in the study area are canopy members or emergents, or growing in
open communities lacking surrounding vegetation. If the acacias were able
to grow well in shade, the interrelationship with the ant would lose much of
its significance inasmuch as the single most important role of the colony of
P. ferruginea is to prevent damage to the shoot tips, which, in their growth,
keep the shoot canopy in the sunlight.

356 Tue University SciENcE BULLETIN

The major portion of A. cornigera shoots growing in heavy shade can be
placed in one of two size classes. Most frequently, they are 5-100 cm tall
with few mature leaves, few swollen thorns, and no intact shoot tip. The
thorns are often partially rotted and many leaves have only minute stipules.
While founding queens are sometimes found in the thorns, more often they
ere empty and only very rarely are such shoots occupied. In the other size
class are placed the rare slender vertical shoots of 100-250 cm height. They
have small swollen thorns and internodes one and one-half to three times
the normal length (Fig. 21a). The number of Beltian bodies per leaf is
reduced and they are pale in color. The foliar nectaries are small with a
reduced nectar flow. These slender shaded shoots are usually occupied by
I’. ferruginea, associated with small gaps in the canopy and developed from
old root stock of senescent emergents. Most shoots in both size classes do
not reach the canopy or flower; they usually disappear within three years.
Jf they do reach the canopy, they develop normally.

The height of the canopy has some effect on the survival of shaded shoots.
Where the canopy is very low (e.g., 120 cm following six months of regenera-
tion after cutting or burning), shoots in the smaller size class occasionally
have a burst of growth during the dry season which carries them into or
above the canopy. This is apparently due to the nearly full sunlight that they
receive due to the leafless nature of the general canopy. This circumstance
is often associated with the invasion of the shoot by a large colony of P.
ferruginea. Shoots in the taller size class are usually old enough so that they
respond to the dry season by reduced vertical growth rates and partial leaf
drop.

Fire. The shoot of A. cornigera is easily killed by fire. Light ground-
level fires are common during the dry season throughout the study area. The
heat is often of sufficient intensity to kill the branches and leaves of shoots
3-6 m tall. Shorter shoots are often consumed entirely. If the fire only
scorches the bark at ground level, the shoot is killed but the ant colony sur-
vives to occupy the new sucker shoots (Fig. 22, 23). The pruning activities
of P. ferruginea in the immediate area of the shoot during the rainy season
(Fig. 35) lower the quantity of dry litter present around the base of the
shoot and thus are partly responsible when the fire is not hot enough to kill
the ant colony (and even in some cases, to kill the shoot).

The root stock does not appear to be damaged by fires. However, in
areas where fires occur every year (oak-grasslands, canefields, some pastures
and milpas), A. cornigera is usually not present 2-4 years after the first
burning. This is because the regular fires destroy all of the mature colonies
of P. ferruginea and the new sucker shoots are unoccupied. This system
is described in greater detail in Janzen 1967b.

~I

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 35

Fic. 22 (left). An occupied sucker shoot from the base of a fire-killed shoot. The parent
shoot was 4 m tall and the sucker shoot 122 cm tall. Only the terminal 25 cm of the shoot
is shown. There are 12 workers of Pseudomyrmex ferruginea in the photograph. The entire
colony will eventually move from the dead shoot into the new shoot as it grows. Photo 11

Jul. 1964.

Fic. 23 (right). Unoccupied sucker from the base of a fire-killed shoot. The parent
shoot was 4 m tall and the biggest sucker 15 cm tall. The upper 13 cm of the shoot is
shown in the photograph. This shoot was about 15 m from the shoot in Fig. 22. Sawdust from
wood-boring beetles in the dead shoot can be seen on the leaves, and the branches of the
living shoot have been damaged by Coxina hadenoids and other insects. Photo 11 Jul. 1964.

PopuLaTion GrowTH Parameters. Height increment. The height of a
shoot of Acacia cornigera must be considered in relation to the surrounding
vegetation and the ant colony occupying the shoot. In the course of this
study, comparisons of height increment between occupied and unoccupied
shoots have provided reliable indications of the efficiency of the ant colony in
keeping phytophagous insects off the shoots. However, changes in mean
and individual heights must be evaluated in respect to the type of shoots
(seedlings or suckers), age of shoots, time of year, height and density of
surrounding vegetation, and percentage of the shoots occupied by Pseudo-
myrmex ferruginea.

Even when occupied by a large, established colony of P. ferruginea, a
seedling in its first and second year of undisturbed growth usually does not
produce a shoot of more than one quarter the weight and size of the sucker
that can be produced in one year by a four year old root system with a 4 cm

355 Tue University ScrENCE BULLETIN

diameter stump. Seedlings and suckers do not become leafless during the
dry season of their first year of growth. Intact shoot tips cease vertical
growth only during the cool dry period. However, the seedling with its
relatively shallow root system (less than 1 m) is often not capable of replac-
ing a damaged shoot tip during the dry season. The dry season only slows
the replacement of shoot tips on suckers from large stumps, and thus height
increment continues.

Even though occupied by a large colony of P. ferruginea, shoots over a
year old usually cease vertical growth sometime between November and
February. The cessation is associated with a failure to replace shoot tips eaten
by defoliating insects rather than to dormancy of intact shoot tips. This
period is both cool and dry, and most new growth is in the form of axillary
tufts and leaves. The older the shoot, the longer after the first heavy rains
before a positive height increment occurs. Shoots in the 1-2 year age class
are variable; many show large increases in height within a month before or
after the first rains. Shoots over two years old often do not initiate vertical
growth until weeks or even a month after the first rains. In addition, the
main vertical branch usually starts a meter or more below the top of the
shoot. It is therefore often as long as a month after the rains begin before
any substantial change in height occurs. This type of growth pattern is
repeated each year while the shoot maintains its position in the rising gen-
eral canopy. It is clear that comparisons of height increments at different
times in the year must be identified with the time of year involved.

In almost all plant communities where A. cornigera has an even or in-
creasing population density, occupied shoots are canopy members or emer-
gents (Fig. 10-14, 24). In secondary succession under three years of age, in
which the vegetative cover is for the most part fromed by regeneration from
cut stumps, the tallest plants are A. cornigera and the plant population with
the greatest mean height is that of A. cornigera occupied by P. ferruginea.
Apparently associated with the need to maintain such a position in the
general vegetation canopy, partially shaded shoots grow very strongly ver-
tically while fully insolated shoots develop in a lateral direction as well.

A. cornigera has one of the largest height increment rates of the woody
plants in early secondary succession in the study area. On 10 July 1964, the
height and condition of 20 regenerating sucker shoots with 1-2 year old
colonies of P. ferruginea were recorded. These were emergents from a
canopy of herbs and shrubs 60-100 cm high that were occasionally browsed
by cattle. The acacias were chosen to be between 100 and 200 cm tall with
no evidence of present or previous insect damage to the uppermost six nodes.
For 18 days these plants were examined every third day and each plant with
a damaged shoot tip was rejected from the sample. On 28 July, nine shoots
remained. These shoots had an original mean height of 148 cm (s.d.=26

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 359

ee

Fic. 24. Emergent occupied shoots of Acacia cornigera in subplot O-2. These year old
sucker shoots were 150-300 cm in height and grew from stumps cut in June 1963. At least
the upper 100 cm of each shoot seen in the picture had grown since the first heavy rain on
20 May 1964. There are many unoccupied shoots of A. cornigera below the general vegetation
with its canopy at about 100 cm height. Photo | Aug. 1964.

cm) and final mean height of 193 cm s.d.=38 cm). The mean of the height
increments was 45 cm (s.d.=11 cm). This represents a mean height in-
crement of 2.50 cm per 24 hours by shoots which received on damage to the
shoot tip for 18 days well into the rainy season. At the Temascal weather
station, the mean maximum temperature for this period (10-28 July) was
30.4° C., and the mean minimum was 22.3° C. Precipitation was 284.8 mm.
It is believed that this rate of growth during the rainy season is representative
of undamaged shoot tips of lengthening branches on seedlings over one
year old and sucker regeneration of all ages. It is representative of the rates
of height increment in the absence of shoot tip damage. Once the acacia
shoot is part of, or emergent to, a gradually rising general canopy over 3-4
m in height, the representative height increment of occupied shoots is 1-2
m per year (this figure includes shoots receiving various amounts of insect
damage).

Age-class representation. The proportions of different aged trees in
natural vs. man-made disturbance sites are often quite different. In natural

360 Tue Universiry SciENCE BULLETIN

disturbance sites, it is common to find acacias of all ages and the only large
size class is that of new seedlings near the bases of seed bearing trees during
the rainy season. This type of age-class distribution is generally associated
with the more irregular disturbance history of natural sites. In man-made
disturbance sites that lack heavy cattle browsing, it is common for all oc-
cupied shoots to constitute one size class and the unoccupied another. All
the shoots are nearly the same age due to the total destruction of the vegeta-
tion often wrought by fire or cutting. The occupied shoots are canopy mem-
bers or emergents, and the unoccupied shoots are under the general canopy
of the plant community. This difference between the two groups is ac-
centuated as the unoccupied shoots are stunted and eventually removed by
biotic mortality agents. With increasing grazing pressure and the sub-
sequent opening of the plant community, the distinctiveness of these two
size classes is reduced because unoccupied shoots in full sunlight sometimes
live long enough to become occupied and then grow to maturity.

Mortality factors. Mortality of the above-ground shoot, and mortality of
the root system are two quite different parameters. In general, death of the
root system requires several successive destructions of the shoot, or else 6-18
months of repeated destruction of the new shoot tips. The amount of stress
tolerated by a root system appears to be directly related to the size of the
root system, and the size of the shoot that was destroyed. When uninter-
rupted shoot development is allowed for periods of two or more years be-
tween shoot destructions, an individual root system lives at least 30 years
and probably much longer. If suckers from cut stumps are heavily shaded,
the root system finally dies. However, when plant communities that are
10-15 years old are cleared, old and apparently dead A. cornigera stumps
sometimes produce small shoots which grow into completely developed
mature plants if they become occupied by P. ferruginea. In natural dis-
turbance sites such as river banks and arroyos, entire root systems are often
washed out. These plants occasionally root where they are deposited by
the receding waters.

In the Temascal area, biotic mortality agents of roots are almost never
observed. In one case, a localized aggregation of pocket gophers (Hetero-
geomys hispidus) killed about one-half of the A. cornigera in two half-acre
areas by uprooting them and eating the roots (73 plants). This damage oc-
curred during the dry season and stopped when the rainy season began.

In the Temascal area, there are various agents that destroy the entire
shoot of A. cornigera but leave the root system undamaged. When clearing
roadsides or around houses, the shoots were often cut along with the other
vegetation. While A. cornigera is occasionally left standing out of fear for
the ants, quite often it is singled out and cut because the people do not care
for the ants. Newly cleared cornfields, pastures and roadsides are often

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 361

burned if cut in the dry season. Since the entire shoot is lying at ground
level, the ant colonies are destroyed by these fires. Many acres of unused
land are burned by fires that escape from set fires. Since the shoots in these
areas are upright, the intensity of the fire regulates whether the trunk is
just scorched (killing the shoot but not the ants) or the entire shoot is con-
sumed. Many sites go for 2-3 years without accidental fires and then may be
burned each of 1-3 successive years. This random destruction of A. cornigera
has much less total impact on the acacia and ant population than does the
deliberate annual burning of some lands for agricultural purposes (e.g.,
sugar cane fields). In this case, the roots of A. cornigera are systematically
deprived of their shoots and eventually disappear from the site.

The most important biotic remover of entire shoots in the Temascal area
is the cricetid rodent Sigmodon hispidus (hispid cotton rat). It eats A.
cornigera most frequently during the last two months of the dry season; at
this time A. cornigera is one of the few shrub-sized plants that has green
foliage and tender stems. It cuts unoccupied shaded shoots, under partially
to entirely closed canopies 50 to 150 cm high. These shoots are usually 10 to
100 cm tall. The bark and supple branches are eaten. Usually the only
evidence of the rat’s feeding is a 2-10 cm tall stump, a pile of wilted leaves,
loose thorns, wood shreds, and light brown oval fecal pellets. S. Aispidus
also climbs to the top of emergent unoccupied shoots to cut off shoot tips.
It is extremely common in 50-200 cm high regeneration in pastures and
fallow cornfields. During the last month of the dry season, it may remove
as much as 75 percent of the unoccupied shoots below these low canopies.
S. hispidus is attacked by P. ferruginea when it attempts to cut an occupied
shoot and notches are occasionally found in the trunks of occupied shoots.

There are only a few insects which kill entire woody shoots, and only
one of these is numerically important. The larva of the buprestid beetle
Chrysobothris sp. near C. multistigmosa Manh. occasionally kills 2-3 m
shoots by girdling the trunk from the inside. This beetle is very common
and is very often responsible for the death of unoccupied stumps during the
first six months of sucker regeneration. The female beetles are usually
deterred from ovipositing in the newly cut stump by the ants. When pre-
paring oviposition sites, the rare cerambycid beetle Onicideres poecila Bates
cuts off unoccupied shoots of A. cornigera that are 60-120 cm tall (Fig. 25).
The cut is made 20-40 cm above the ground and the female oviposits in the
cut shoot. A closely related cerambycid, Lochmaeocles cornuticeps is com-
mon in the Campo Cotaxtla area. Adults of this species completely girdle
the trunks of unoccupied shoots up to 4 m tall and oviposit in the dead shoot.

There is very high mortality of very young seedlings. While apparently
very resistant to fungus attack, seedlings are readily eaten by insects. Moth
larvae (Noctuidae, Arctiidae) and orthoptera do the most damage. When

362 Tue University ScrENcE BULLETIN

Fic. 25. An adult female of the cerambycid Onicideres poecila immediately after cutting
an unoccupied shoot of Acacia cornigera in subplot N-2. The shoot was 12 mm in diameter.
She oviposited in the portion of the shoot that fell to the ground. The cut shoot had an
auxiliary-unit of Pseudomyrmex ferruginea living in the thorns but there were no workers active
outside of the thorns owing to cold weather. Photo I1 Jan. 1964.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 363

the leaves are removed from very young seedlings, the entire plant dies. In
addition to insects, black rats (Rattus rattus) and lizards (Iguanidae) on
occasion eat very young seedlings of A. cornigera.

Damaging factors. With the exception of the activities of man and dam-
age by Sigmodon hispidus, biotic mortality agents are rarely observed in
sign or action. However, damage to A. cornigera is very commonly en-
countered. Most of the insects responsible for this damage in the Temascal-
La Granja area are listed in Table 1. These 56 species are not of equal
importance and abundance throughout the year, and their density varies
from site to site. Some of these variables are expressed in Table 1. A select
group of these species are also discussed further in Appendix 1.

Tasre 1. Insects that feed on A. cornigera in the Temascal-La Granja area. L=
larva, A—adult, R—rare, O=occasional, V—very common, C=capable of feed-
ing on occupied shoots, S=stopped from feeding by P. ferruginea, N=nocturnal
feeder, D=diurnal feeder, Sh—=most commonly eating shoot tips and very young
leaves, M=most commonly eating mature leaves, and B=feeding on or in
branches and/or trunk. They are divided by the season because their appearance
on the acacia is closely correlated with the three seasons. It should be noted that
the largest number of species are present during the time when the most active
growth occurs in occupied shoots. Those species of particular significance to the
ant-acacia interaction are discussed further in Appendix I.

COOL SEASON

(December - February)

Lepidoptera
Goria ademordes-miNOCUING AG. meat eee ee reece I, OF CyN-ish
WCCASION AINE AT GLC Actes ee eh ee et DPA ONS SINe Sh
WG) lejonveloyoreira SU eas nS a eee eee ee ee L, O, S, ?, thorns
Orthoptera
Montezumina oblongoculata: Tettigoniidae —_.................--.-----.----------- A, O, S, N, Sh
Microcentrum rhombifolium: Tettigoniidae -...............-.-..---.--------------- ISO} SS ING Sin
(CQUMGERDRE TAILS. (HOB UCR HNEO SAVING ELE ee cee ree ee ee mee eS IX, (), S35 ING Sin
Coleoptera
Chrysobothris sp. near C. multistigmosa: Buprestidae —_............-....- A, O, S, D, MB
Ibe Why (G5 es 18
OPES jHORGHOS (CRIN OM OCES cece cece ces beer sec eee AG) RSS; D538
iby, RCS a 8
Goscinopicjammucidass Ghiysomielid acy seen A, O, S, D, ShM
Lechriopini, near Lechriops sp.: Curculionidae (Zygopinae) -........ INS O), (GID), Sin
PONG ers ithorns
Acanthoscelides oblongoguttatus: Bruchidae ...........--.------------------------ ibe Wie (Ga Is SeeGls
IML DOS2GTES. joys IVAUSUGES: ope en ee ee ee eee lbp Woz, 25, See
ISEGLO TASH D CT CUES STUNG IN lac eee cee oes Ibn I, (Gy Os Seals

364 Tue Universiry ScreENcE BULLETIN

WARM DRY SEASON

(March - May)

Lepidoptera
VAT ISLOLEIG GOT GILG a GELEC DIG dee ee Ip We Sy IN, Sin
Goxinaihadenoidess Noctuidae ee eee IBS We (Gy IN, Sin
IMGGro] ep 1clo pte reali etl emer t1n c15015 a ee 1G; (5S; 5 daourtas
Coleoptera
Chrysobothris sp., near C. multistigmosa: Buprestidae —_...........- A, O, S, D, MB
Ibs Wi iCs 5 1B
Goscinopiera smucidass Chiysomelid ac essen A, R, S, D, ShM
Lechriopini, near Lechriops sp.: Curculhonidae (Zygopinae) —._.. AS ONG] Dash
IL, ©; GC; Ps thorns

The three Bruchidae listed above

Heteroptera
IMOZENANOTHETILO SAA COTE! aC pee ene ee ASO; 'S; Ds Sh
RAINY SEASON
(June - November)
Lepidoptera
Coxina hadenoides: Noctuidae ................------.....-- tS Fics ner Ibs Wo (Gy ING Sin
SiS p) A272 GTO XICATZ 2a SSV SS) L111) 211 Cl cle ee ee I, OF CoN; M
ArISIOLEIANGCOTGLIIN GM GElCGITI CAC ee eee TERS aeNeES
INOTGp CaS Dec Nie eal 0 [0151 eee ese L, V, S, N, ShM
Leaf rollers: Gelechiidae, Walshiidae, Geometridae —.......---- IP, Ws Ss Fs Sail
Hiahisidotaisps probably lian pura eATClii Ga cee ee L, O, S, N, ShM
ROSEN C1 TT Cram NOLOGO DIC Cac menses ene EXORGINGM
ESTES TCR CMTTLC LLC 272. au N TAG ELI C1 Cee ee a L, R, S, N, ShM
Ue O Pst s UTC Gm G COTM CELI Cel eee eee Is 1S Sy INE, Sin
BARCIA ata este G COMMCLIT ac eee ne eee eee ne 15 Wa (GS ING Sin
Semiothisa adjacens, S. puntolineata: Gane nN: ok es ee ESOS, Nash
Microlepidopteralinyitine stho nisi see eee ener ee ga ee L, O, S, ?, thorns
Teel AOL bia MENTAL Si) Sc mlay CAC INI CAC esse seem ae oe L, R, C, D, ShM
Orthoptera
Montezumina oblongoculata, Microcentrum rhombifolium,
Chloroscirtussp .mOtners-ealett SOD11d ac pene ee L, V, S, N, ShM
(adults appear in late August)
HA CHIC GAS ae, cee Bee ed aA BRD os ESL SS Nes Sek oc Ane a L, O, S, D, ShM
Blattidacieet ee eee Be ys ea 5 fo ee oe Ae, Se A, R, S, N, Beltian
bodies
Phasmatidae (adults appear, smite eA oust) ee eens sense eee L, V, S, N, ShM
Coleoptera
Pelidnota punctulata: Scarabaeidae (Rutelidae) ~.....................--.------ ADV; Gane sit
Diplotaxis denigrata: Scarabaeidae (Melolonthinae) —.................-...... IX; Wo Se INI, Sin
Anomoea sp., probably A. rufifrons: (Chrysomelidae) ~................-... AN Vis son
Chrysobothris sp., near C. multistigmosa (Buprestidae) --................ A, V, S, D, MB
EVs G; rb

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 365

The three Bruchidae listed above
Goscinoptera mucida: Clhysomelidac <2 none ne nce JX WA
Saxinus sp., probably S. bastlarts: (Chrysomelidae) —.............-...... IN We Si 1D Sia
Griburis decoratus, Pachybrachis sp., probably P. femoratus,
Cryptocephalus trixonatus, Urodera cructfera:

(@hiy some lic ac) ements ae eee ee a8 Spe A, V, S, D, Sh
NGRMDS (RGT ALLIS? VENUS MENG ELS oe cece cae eo A, V, C, D, M
Pelidnota strigosa; Scarabaeidae (Rutelinae) ___.-..-2...-.-.---22------2-----. A, R, S, N, Sh
Phyllophaga vexata, P. anodentata, P. dasypoda, Diplotaxts

microtishia: Scarabaeidae (Melolonthinae) —....................----..-.- A, R, S, N, Sh
Euphoria leucographa: Scarabaeidae (Cetonniinae) —..........-..-... ASONSs Ds Sh
Cathartus annectens, Anchorius sp. near A. lineatus, Hapalips

SDennleatay en crais sm Cy p tor hia oC a eases LA, V, C, ?, legumes
Lechriopini near Lechriops sp.: Curculionidae (Zygopinae) ............ A, O, C, D, Sh

ONG. 2) thorns

Heteroptera

WORE GELONIEDLO SG COLIC ACH eee IL/ANG Wo Ss 1D Sn
Homoptera

FLO TRIOUISCOMGCOULULGL EM ASS) Ga Capers eee ee ae AG IVES ED INEEB

Oncomero pig aclarion-ma)| aSs1G ac upee a ee eee LA, V, S, DN, B

Dysmicoccus sp. in D. neobrevipes complex: Coccidae ~...................-- Ibye\. (OF 5 IDIN |, 18}

COCCI AC eee ee ee en EE. yO WET AL Net ed eee LA, R, S, DN, M

PA AISI CUIGER TYG = MeN) IN Gace eee eee ie ese ILS, G5 IDING Sn

Heteropsylla sp.: Psyllidae fa rte Nokon BIN 1th Dea Oe RO, ILANS RS Sy DINE, Sin

Wanboniaworoztimebos Nem bracic ae ye ee eee VAG RSS; DNB

DIGCTOPLOGLGES|) 30 GIGA C1 Cal Ce area ete ees ee Ay ONSs DINGS
Hymenoptera

FACE ONIN TIN CMOGLO STIL72 O S125 00 OGITIIC1 Gl Ae eee tees eee A, O, S, DN, ShM

SOMARODES GETAIAIOS AROSE aca ecto eco ee A, R, S, D, Beltian

bodies

Seasonal abundance. The most obvious division is between those insect
species that are present during the rainy season and those that are present

during the cool and dry season (Table 1).

During the cool season (Dec. through Feb.) the density of active insects
is much lower than during the rainy season. This is especially true of
nocturnal insects. Of those that feed on A. cornigera during the cool season,
there are only the larvae of Coxina hadenoides, adults of tettigoniid grass-
hoppers (Montezumina oblongoculata, Microcentrum rhombifolium, Chloro-
scirtus sp., and others) Onicideres poecila (adults cutting shoots for oviposi-
tion), Coscinoptera mucida, and Chrysobothris sp., near C. multistigmosa
(females ovipositing in damaged unoccupied shoots). However, the com-
bined activity of C. Aadenoides on occupied and unoccupied shoots, and the
tettigoniid grasshoppers on unoccupied shoots, is sufficient to accomplish
nearly 100 percent destruction of the shoot tips of A. cornigera that are

366 Tue University SciENCE BULLETIN

emergent. Apparently due to the cool weather and subsequent dry weather,
many of the damaged shoots, especially those over one year old, do not
initiate new vertically lengthening branches until shortly before or after
the first heavy rains in May.

From the return of warm weather in March until the first heavy rain
on the 20th of May, the density (species and numbers) of active insects in
the plant community decreases rapidly. By the first part of May, insect
densities are less than one percent of those observed during the peak of the
rainy season. Of those that feed on the shoot tips of A. cornigera, the larvae
of Coxina hadenoides, the adults of Mozena tomentosa, the larva of
Aristotelia corallina, and Conscinoptera mucida are sufficiently abundant to
completely halt the height increment of only a small part of the unoccupied
shoots in the 60-150 cm size class. Only C. hadenoides feeds externally on
occupied shoots but its damage is considerable; it is doubtful that there was
a single occupied shoot tip in the control subplots that grew continuously
from 1 Mar. to 20 May without being fed on at least once by a larva of C.
hadenoides. The larvae of A. corallina are very effective in removing the
shoot tips of unoccupied A. cornigera in the 5-50 cm height range; on many
sites where the surrounding vegetation is very low, and therefore the small
acacias are fully exposed, 80-100 percent of the shoots have either the webbing
or a larva of this moth on them during April and early May. The normal
height increment rate of A. cornigera with intact shoot tips is undoubtedly
slowed during the dry season. However, in those small local areas where
neither C. hadenoides nor A. corallina are abundant, those shoots of A.
cornigera with vertically lengthening shoot tips grow 15-60 cm in height
during the dry season.

Most of the shoot tips produced by A. cornigera during the dry season
are destined to become axillary tufts of leaves and flowering branches. These
constitute the major part of the diet of C. hadenoides, M. tomentosa, and C.
mucida. The removal of one of these shoot tips does not have the same effect
on the plant as that of removal of a main shoot tip on a lengthening branch
or central axis during the rainy season. It should be noted that the general
vegetation surrounding the shoot is not increasing in height during the dry
season and therefore the lack of height increment in the part of A. cornigera
does not place the plant in immediate danger of being shaded by the sur-
rounding plants. The surrounding vegetation is leafless and this allows sun-
light to reach the lower leaves on the A. cornigera shoot. These leaves are
shaded during the rainy season.

The remaining 47 insects listed in Table 1 are present during most of
the rainy season. Coxina hadenoides, and Mozena tomentosa are much more
abundant at this time. CArysobothris sp., near C. multistigmosa adults are

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 367

more common during the rainy season than the dry season but the larvae
cause considerable mortality to unoccupied stumps throughout the year.
Large populations of grasshoppers do not develop until the middle of
August but these are also present through the cool season when most of the
other rainy season insects have been gone for several months. The zygopine
weevil and the microlepidoptera larvae in green thorns are present through-
out the year but the weevil is present only as a larval stage in dry thorns
during the dry season.

Host specificity. The insects listen in Table 1 can be divided into a group
that may be found feeding on either unoccupied or occupied shoots, and a
group that is virtually always found feeding on unoccupied shoots. The
former group is composed of the larvae of Syssphinx mexicana, Rosema den-
tifera, Coxina hadenoides,:Euacidalia sp., and green-thorn microlepidoptera,
and the adults of Pelidnota punctulata, Agrilus bicarinatus, and the zygopine
weevel. While it is not know with certainty that Syssphinx mexicana, Ro-
sema dentifera, P. punctulata, and Euacidalia sp., are completely host specific
on A. cornigera, they were not found on any other species of plant in the
Temascal-La Granja area despite extensive searching. However, Syssphinx
mexicana, Rosema dentifera, Agrilus bicarinatus and Pelidnota punctulata
are recorded from areas (literature records and museum specimens as local-
ity sources) where A. cornigera does not occur. Coxina hadenoides and the
zygopine weevil are commonly found on Acacia chiapensis. Coxina hade-
noides is occasionally found on Acacia macracantha but definitely prefers
A. cornigera and A. chiapensis as a host plant in choice experiments.

The bruchids and cryptophagids in the legumes can not properly be
placed in either group since they live for the most part outside of the sphere
of activity of Pseudomyrmex ferruginea. The same can be said for nectar
and pollen collecting Hymenoptera. They visit the plants frequently, when
other sources are scarce, but are all species which are commonly seen visiting
other species of flowers in the area.

The group normally found feeding on unoccupied shoots is much larger
than the previously mentioned group. The 41 species in this group feed,
with but two exceptions, on other species of Mimosaceae in the area as well
as unoccupied A. cornigera. Aristotelia corallina has been found only on A.
cornigera. As a population, Mozena tomentosa is believed to be host specific
on A. cornigera despite the single female found feeding on Mimosa albida.
Associated with this host specificity, M. tomentosa appears to be the most
adept, of all the species in Table 1, at finding small unoccupied shoots of A.
cornigera that are heavily obscured by the surrounding vegetation. When
the ants are removed from a shoot, it is in great part the neighboring Mimo-
saceae that contribute phytophagous insects. With the exception of Acro-
myrmex octospinosus, the two Thecla species, Pelidnota strigosa, Euphoria

368 Tue University SciENCE BULLETIN

leucographa, the scale, and Dysmicoccus sp. in the neobrevipes complex, the
remaining 33 species are found feeding on Acacia chiapensis or Acacia mac-
racantha. The Chrysomelidae, Lepidoptera, Orthoptera, Diplotaxis deni-
grata, and the two leafhoppers are commonly found feeding on A. chiapensis
along with some other insects which are not found on A. cornigera. The
Chrysomelidae, most of the Lepidoptera, Orthoptera, and D. denigrata are
also found feeding on other species of Mimosaceae (Calliandra houstoniana,
Acacia farnesiana, Mimosa albida, Mimosa sp.).

With the exception of the scales, mealy-bugs, and microlepidoptera, all of
the insects feeding on unoccupied shoots are highly mobile in the stage found
on: A. cornigera. Associated with this, they are often known to move from
plant to plant until they find one on which they are not molested by P. fer-
ruginae. Under natural conditions, this search for a suitable host plant is
sufficiently successful to insure that shoots of A. cornigera from which the
ants have been removed begin to show phytophageous insect damage to the
shoot tips and mature foliage in 1-2 days. In the treatment subplots in the
experimental part of this study, the large groups of unoccupied shoots act
as a trap for these insects. Their density per unit volume of plant com-
munity is considerably higher in some treatment subplots than in the neigh-
boring plant community. However, the damage to unoccupied shoots
scattered among occupied shoots (mixed subplots) is so severe that no
differences are noted between the grouped and scattered unoccupied shoots.

Ant-tolerant insects contrasted with other insects. The nature and
periodicity of the damage done are quite different between two groups of
insects: those found normally on occupied and unoccupied shoots, and those
found normally only on unoccupied shoots. Of those found normally on
occupied shoots, Rosema dentifera, Agrilus bicarinatus, Euacidalia sp., the
zygopine weevil, and the microlepidoptera larvae in the green thorns have
no noticeable effect on the growth rate or condition of occupied A. cornigera.
The effect of the bruchids on the seed crop is very great but it appears to be
beyond the influence of P. ferruginae. A. cornigera normally exists in large
numbers in spite of the high rate of seed destruction.

The other insects found on occupied shoots have a strong impact on indi-
vidual shoots, but their cumulative impact on the entire population is low
compared to that of the insects on unoccuped shoots. The larvae from a
clutch of five eggs of Syssphinx mexicana usually strip an occupied shoot 'of
its mature leaves. However, at any one site, only 0.5-3 percent of the shoots
are defoliated by S. mexicana during the rainy season. Further, a single
occupied shoot usually has enough mature leaves to rear the group of larvae
to maturity after which they pupate without feeding on other shoots. A
shoot defoliated by S. mexicana generally produces a new crop of leaves
within two weeks and the probability of a single shoot being defoliated twice

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 369

during the growing season is very small. Since the shoot tips are not eaten,
the removal of the mature leaf crop does not noticeably affect the emergence
of the shoot.

Pelidnota punctulata has more of an impact than S. mexicana on A. cor-
nigera form and growth rate because it eats the shoot tips. The removal of
all of the shoot tips by an adult P. punctulata (Fig. 26) is very detrimental
to the height increment rate of the shoot. After the beetle moves to another
plant, 1-2 weeks are often required for a new vertically lengthening shoot tip
to reach the level of the older damaged shoot tip. The density of P. punctulata
usually does not exceed six beetles per one hundred shoots and usually the
density is 1-3 per one hundred shoots. In the experimental plots, at least 50
percent of the shoots did not have feeding damage of P. punctulata during
the period 20 May through 8 Aug.

While S. mexicana and P. punctulata feed on the acacia only during
rainy season, Coxina hadenoides is present in the larval stage throughout the
year and has a heavy impact on the height increment rate at all times.’ It is
a very rare occurrence that any one shoot of A. cornigera over 30 cm tall
passes through an entire year of growth without having the main shoot tip
eaten off by C. hadenoides at least once (Fig. 18). However, C. hadenoides
is sensitive to the density of worker ants patrolling outside of the thorns; the
frequency of feeding damage of this larva is inversely proportional to the
density of workers outside of the thorns and the constancy of their patrolling
(as affected by the size of the colony and the time of year and day). There-
fore, the larger the ant colony, the longer there is uninterrupted vertical
growth. It is largely due to C. Aadenoides that weakly occupied shoots have
height increment rates that are often less than one-tenth that demonstrated
by heavily occupied shoots.

The damage inflicted by the three insect species listed above is normally
received by occupied shoots and the population does very well in spite of it.
At any given site during the rainy season where 90-100 percent of the readily
visible shoots are occupied by mature colonies (the usual case), from 5-35
percent of the main shoot tips are destroyed by P. punctulata or C. hadenoides
at any given time. However, the degree of damage stays relatively constant
and the damage is rotated among the shoots. It can be said with certainty
that during any part of the rainy season, the total number of vertically
lengthening shoot tips present on a cohort of occupied shoots is 2-20 times the
amount eaten during the same period by P. punctulata and C. hadenoides.

The damage inflicted by the individuals of insect species that feed on
unoccupied shoots is less impressive than that described for the occupied
shoots. However, the cumulative effect of the omnipresent diurnal and
nocturnal defoliators of unoccupied shoots is much more severe than the
effect of the relatively sporadic damage by the insects that feed on occupied

ao ett =

Fic. 26. The terminal end of the main axis of an occupied shoot of Acacia cornigera. The
shoot tip of the main axis and the terminal axillary branch bearing the two type B thorns were
eaten by an adult Pelidnota punctulata. 'The mature leaves were removed by a larva of Syssphinx
mexicana. The short branch on the right that is growing out through the cluster of type B
thorns is a flowering branch. Photo in subplot M-2 in June 1964.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 371

shoots. Insect damage to unoccupied shoots is normally severe enough to kill
the shoot either directly or indirectly after six months to two years. During
the rainy season the diurnally active Chrysomelidae and Mozena tomentosa,
together with the nocturnally active Orthoptera, Lepidoptera larvae, and
Scarabaeidae generally remove 90 percent or more of the new axillary and
terminal shoot tips as fast as they are produced. This is the case with both
isolated and clustered (experimental plots) unoccupied shoots. Unoccupied
shoots receive the usual damage from S. mexicana, P. punctulata, and C.
hadenoides, but the density of these insects per unoccupied shoot is generally
much lower than per occupied shoot. Unoccupied shoots rarely have the
well developed succulent shoot tips that P. punctulata and C. hadenoides
prefer as foods; they have already been eaten off. Unoccupied shoots often
have so few mature leaves that the larvae of S. mexicana defoliate their shoot
and wander off in search of more food.

In addition to the obvious damage to the foliage, it appears that the rate of
new shoot tip production in unoccupied shoots drops off sharply 6-8 weeks
after the rainy season begins. This was impossible to demonstrate con-
clusively because the removal rate by phytophagous insects is usually so high
that it is not possible to determine how many new shoots have been pro-
duced over a given period of time.

One response to severe insect damage is flowering. Shoots in the 100-200 cm
size range that are unoccupied from March through August produce many
more flowering branches than do those that are heavily occupied. By July,
most of the unoccupied shoots in this size range have some opening flowers
on them and almost none of the heavily occupied shoots in this size range
have flowers. Excessive flower production in shoots of this size and smaller,
appears to be a reaction to the heavy and continuous damage to the un-
occupied shoots. The majority of these unoccupied flowering shoots usually
die before the end of the next dry season if they remain unoccupied, and
therefore do not contribute to the reproduction of the population since it
takes nearly a year for the seeds to mature.

Natural and man-made disturbance sites. Every species of insect in Ap-
pendix I was encountered at least once while feeding on unoccupied shoots
of A. cornigera in both natural and man-made disturbance sites. In most
arroyo and river-bank sites, the shoots are widely dispersed (up to 1000 m
between shoots) in contrast to those in fields, pastures and roadsides. There
is no evidence that wide spacing reduces the incidence of damage by de-
foliators of either the group of insects found on both occupied and unoc-
cupied shoots or the group found only on unoccupied shoots. However, at
lower insect densities this is probably not the case.

The shoots of A. cornigera in natural disturbance sites are usually among
vegetation that is very irregular in height, species composition, and density.

Us
“NI
bho

Tue University ScIENCE BULLETIN

Associated with this, many of the shoots are growing in partially shaded
sites, rather than being either clearly emergent or submergent as is the usual
case in the relatively uniform man-made disturbance sites. There appears to
be a reduction in the frequency of Chrysomelidae, Buprestidae, Scarabaeidae,
Lepidoptera, and Homoptera in these partially shaded sites as contrasted
with sunny sites. A reduction in damage to foliage is, however, not notice-
able. Variation in damage is difficult to evaluate because it is often so severe
that maximal damage is present.

Certain defoliators of unoccupied shoots are correlated with different
plant communities. The melolonthine scarabs which are such severe de-
foliators of unoccupied shoots at night during the first two months of the
rainy season (especially Diplotaxis denigrata), are much more abundant in
heavily grazed grass pastures. Orthoptera are much more abundant in the
brushy pastures and ungrazed sites than they are in the open heavily grazed
pastures. Chrysomelids are equally abundant in most sites but there are local
variations in the species composition of those feeding on unoccupied A.
cornigera. Lepidoptera and Homoptera are most easily observed feeding
in the open heavily grazed sites but they are definitely present in the denser
vegetation as well. Mozena tomentosa is undoubtedly the most omnipresent
feeder on unoccupied shoots and its characteristic feeding damage (wilted
shoots tips) is found in all types of sites at all times of the year. Coxina
hadenoides is the most omnipresent feeder on both occupied and un-
occupied shoots.

Heavily shaded unoccupied shoots are less severely attacked than are ex-
posed ones. This is especially true of those growing under the dense vine
tangle in ungrazed regeneration for the first 1-3 years after a fire or cutting.
However, the rates of growth, and rates of recovery from damage, are so
low that the undamaged shoot tips rarely become emergent. If and when
they do, they are then rapidly eaten. The emergent shoot protruding above
the general canopy is more readily found by phytophagous insects.

Other parts of the study area. With the exception of the Cotaxtla and
Temascal-La Granja areas, populations of A. cornigera have not been
examined intensively for defoliators in the study area. S. mexicana, C. hade-
noides, P. punctulata and M. tomentosa were collected throughout the range
of A. cornigera in the study area with sufficient frequency to indicate that
they were probably present throughout. Judging from the damage found on
unoccupied shoots in other areas the feeding pressure of phytophagous in-
sects is as great as in the Temascal area, even if the same species are not
involved.

In the Cotaxtla area, with a few exceptions, the species listed in Appendix
I have been recorded from either direct observation or the finding of damage.
The Cotaxtla area has more acreage of pasture and the pastures are more

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 373

heavily grazed than in the Temascal-La Granja area, and therefore the
population structure of A. cornigera differs somewhat. However, the in-
cidence of insect damage to unoccupied shoots is, if anything, higher than in
the Temascal area. This appears to be associated with the very exposed
nature of many of the shoots in the open pastures.

Insects reported in the literature. Wheeler (1942) lists the insects reported
in the literature as feeding on various swollen-thorn acacias, or being as-
sociated with them in some manner. No insects are recorded specifically for
Acacia cornigera but two are recorded from Misanhtla, Veracruz from
Acacia sp. It is likely that the acacia referred to is A. cornigera. Bruchus sp.
is recorded from the legumes and Adelocephala xanthocroia is recorded as
entering the exit hole of the Bruchus sp. to feed on the pulp in the legume.
The bruchid is probably Acanthoscelides oblongoguttatus or Mimosestes sp.
It is likely that the feeding habits of Adelocephala xanthocroia are incor-
rectly recorded since the exit hole of the bruchid is three or less millimeters
in diameter and A. xanthocroza larvae quickly reach a diameter of 5-10 mm.

It is interesting that in Wheeler’s list of 54 species of arthropods there are
only 13 which could possibly act as defoliators. Of these 13, only five are
likely to feed on swollen-thorn acacia foliage or fruit. This list refers to all
the species of swollen-thorn acacias known to Wheeler and it is very small for
a group of common plants that range from north-western and north-eastern
Mexico to Panama and Colombia. Studies in progress show that such a list
should contain several hundred species of insects.

Sampling problems. The insects on the shoots in this study were found
by visual searching; shoots were not beaten over a sheet or surveyed in
other ways because the shoots were for the most part in experimental plots
and could not be disturbed. The undersides of the leaves were examined
but not as thoroughly as the upper sides. When searching at night with a
flashlight, it is especially difficult to locate insects on the trunk and interior
part of the canopy. It is necessary to examine a shoot from above and/or on
all sides. Shoots over 250 cm tall are not easily examined, and the list of
insects in Table 1 undoubtedly omits some of the species that feed on taller
shoots, since no observations were made on insects feeding on shoots over
> m tall. However, feeding damage characteristic of the species on shorter
trees is definitely present on these taller shoots.

It is probable that more than 80 percent of the species of insects that
regularily fed on occupied and unoccupied A. cornigera in the Temascal-La
Granja area are included in Table 1. There are many other species of in-
sects, representing at least 23 families, that are seen on the foligae of un-
occupied shoots; these are not listed as there is no evidence that they feed
on A. cornigera.

374 Tue UNiversity SCIENCE BULLETIN

Mammals as defoliators. In addition to the insects listed in Table 1, there
are some mammals in the Temascal area that eat the foliage of A. cornigera
under certain circumstances. In no case examined was the interaction be-
tween ant, acacia and mammal found to be as clear as that between ant, acacia
and insects. The brocket deer (Temazate: Mazama americana temama),
burro, horse, and cow were examined in this context. Of prime importance
was the determination if these animals would eat A. cornigera in the absence
of P. ferruginea. None of these animals have the direct impact on A. cornt-
gera that phytophagous insects do, but cattle may drastically alter the structure
of the vegetation and thereby have a strong indirect effect on A. cornigera.

Brocket deer. In past years, brocket deer have been common in the
Temascal area, and their tracks were occasionally seen during this study.
There is no evidence that they entered the experimental plots. Since they
could have been an important herbivore in past years, the following observa-
tions were made to determine the reaction of a brocket deer to A. cornigera
and P. ferruginea. A tame female (three years old) was used; she had never
been allowed to feed outside of the owner’s yard and he was convinced that
she had no previous experience with A. cornigera.

She readily ate shoot tips and mature foliage from A. cornigera that did
not have ants outside of the thorns (Fig. 17). These lateral branches (40-70
cm long) were cut from occupied shoots, and the ants shaken and picked
off. The branches had fully developed type A thorns, intact mature leaves,
and undamaged shoot tips. The deet nipped off the shoot tips and mature
leaves with its molars or incisors. Large mature leaves often have raches or
petioles that are too tough to bite through with the incisors, so the deer
shifted her head and bit them off with her molars. The leaves project 7-15
cm past the tips of the thorns and her thin face fit easily between the widely
spaced halves of the swollen thorns. She rarely shifted her feeding position
as a consequence of being stuck by a thorn. Occasionally, a dry and hard
swollen thorn was bitten off with the molars and chewed up with the leaf.
On one occasion, she ate 40 inflorescences that had been hand-picked from
A. cornigera.

When feeding on unoccupied A. cornigera and on other plants, she fed
rapidly. On one occasion, she ate 50 mature leaves and five shoot tips from
five branches in 12 minutes. As soon as she finished one branch, another was
offered. She would eat 15-50 shoot tips (5-10 cm long) as fast as they were
offered. The deer would not feed indefinitely on A. cornigera alone, but
when shoots of this were alternated with other acceptable plants, she would
eat until satiated.

Of the foods offered her, she had definite preferences. A. cornigera,
Malvaceae, Convolvulaceae, Leguminosae (Mucuna pruriens and other
vines), grass inflorescences, and sedge inflorescences were readily accepted.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 375

When first feeding after a night without food, she would eat foliage of
Acacia macracantha, Acacia chiapensis and Acacia farnesiana; later she re-
fused these after having eaten a small amount of other foliage. Even after
two days of starving, she would not eat foliage (shoot tips or mature leaves
of Solanum torvum, Bixa orellana, Croton glabellus, Eupatorium odoratum,
Calliandra houstoniana, and Bauhinia ungulata. These six plants are all
shrubs that are commonly found in heavily grazed cattle pastures.

It can be said with some assurance that the foliage of A. cornigera is not
distasteful to the brocket deer, and that the type A thorns on lateral branches
do not deter it from feeding. Following a number of days of feeding the
deer unoccupied foliage, she was offered a branch about 1 m long that was
swarming with excited workers of P. ferruginea. This was in the early
morning after the deer had passed the night without food. She was normally
not allowed to feed at night. She began to nip the leaves from the branch
immediately. Several ants ran onto her nose and facial region and began to
sting her. She stopped feeding, turned her head away, and by the use of her
very long and flexible tongue, and sharply pointed forefoot, removed the
ants. She then returned to the branch and began to feed again. Again the
ants ran on her and again she stopped feeding to remove them. This was
repeated 2-3 times before she finally turned and walked away from the
branch, leaving about one-half of the leaves still intact. About 15 minutes
later a new branch was offered to her with ants on it, and the same pattern
was repeated.

This experiment was repeated four times that morning at 15 minute
intervals, and each time her reaction was the same. On two following oc-
casions, 3 and 4 days later, the same experiment was tried again. In each
case the same thing occurred. The deer always reacted to the ants by turn-
ing casually or suddenly away from the branch and removing the stinging
ants with her tongue or forefoot. She removed all ants before she resumed
feeding; she was very sensitive to the presence of an ant biting her, stinging
her, or just running on her hair.

There were two immediate effects of her reaction to P. ferruginea. Some
leaves were often left on the branch, and she required 3-10 times as long to
remove the leaves that she did eat. However, the greatest apparent signific-
ance of this form of feeding deterrence is in neither of these direct effects.
In natural circumstances, where there are a large number of species of
plants to choose from, any factor that causes the deer to turn away from the
plant is of significance to the plant; it is unlikely that she would return to
the same plant immediately. A second point of consideration in the case
of A. cornigera is that since the deer can only reach vegetation which is less
than about 1.5 m tall, any shoot taller than this would have the most im-
portant part of the foliage (shoot tips) above the deer’s foraging range.

376 THe Universiry ScrENCE BULLETIN

Burro. A six year old female burro refused to eat loose leaves, or leaves
off of branches, of unoccupied A. cornigera. These were offered many times
between September 1963 and July 1964. She rejected the foliage either on
seeing or smelling that it was A. cornigera. She did not contact the thorns
before turning away. Since she sniffed at the offered foliage, it is likely that
she recognized it by smell. When leaves of A. cornigera were cut and
offered mixed with the leaves of grasses, mints (Labiatae) or guacimo
(Guazuma ulmifolia), the acacia leaves were usually spit or sorted out of the
mixture, and the other leaves eaten. Occasionally, the acacia leaves were
chewed and swallowed with the other plant matter. This burro had previous
experience with occupied A. cornigera since there were several thousand
shoots of A. cornigera in its pasture. Since she did not eat the foliage con-
sistently under any circumstances, it is impossible to determine if her refusal
of A. cornigera is purely a function of chemical properties of the acacia or
is in some way connected with the presence of P. ferruginea.

Behind the laboratory at Temascal, another female burro grazed around
20 potted unoccupied seedlings of A. cornigera for about one month. She
never fed on the plants. However, her weanling burro was sampling plants
behind the laboratory and ate four of the seedlings in entirety. Each shoot
had 2-5 fully developed swollen thorns on it that contained founding queens.
The leaves, stems and thorns were chewed thoroughly and he made no
attempt to spit them out. Shortly after he was observed to try, and then re-
ject, foliage of Callinadra houstoniana and Bauhinia ungulata. This indicates
that the foliage of A. cornigera is in itself not distasteful to young burros,
despite the fact that adult burros long exposed to ant-inhabited acacias were
very unwilling to eat it. In view of the reaction of the brocket deer to P.
ferruginea, it can be surmised that the avoidance of the foliage of A.
cornigera by burros is learned and enables them to avoid contact with P.
ferruginea.

A simple test of this was conducted. The air blown through a thoroughly
washed aspirator into the nose of a six year old female burro caused no reac-
tion. However, when the aspirator was filled with 100 workers of P. ferru-
ginea, the burro rapidly jerked her head away from the stream of air. Ants
were not being blown onto the burro. This experiment was repeated several
times a day for three days; after this time, she also jerked her head away from
the stream of air from a clean aspirator. It is of course possible that the
alarm odor of the ants (discussed in a later section) was merely an un-
pleasant odor to the burro. It is also possible that the alarm odor served as
an indicator of the presence of P. ferruginea; when a large mammal comes
near to, or brushes against, a heavily occupied shoot of A. cornigera, the
alarm odor can be distinctly smelled at a distance of a meter or more and
might serve to identify the acacia at night.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 377

Cattle. During the cool season, cattle will, on occasion, browse on A.
cornigera. In January, members of Farfan’s and Torrealva’s herds of cattle
were observed seeking and eating foliage from standing shoots of A. cornt-
gera in the 15-4 m height range. Since the weather was cool (13-23° C.),
there were almost no workers of P. ferruginea outside of the thorns. The
cattle were eating mature leaves from lateral branches 1-2 m above the
ground (Fig. 27).

The availability of alternative foods at this time of year is very difficult
to evaluate. Some grasses, Convolvulaceae, Leguminosae (vines) and other
readily accepted plants are still green and are eaten by cattle. Certainly the
volume of food available is less at this time than during the rainy season, but
it is also more than during the latter part of the dry season; during this warm
dry period the cattle do not normally browse on A. cornigera. During the
following weeks, until the warm season began, these cattle were frequently
observed feeding on standing A. cornigera. A single cow would remove

Fic. 27. A seven year old cow browsing mature foliage on lateral branches of a tall un-
occupied shoot of Acacia cornigera during the cool season (January 1964) in one of Juan Tor-
realva’s pastures. A colony of Pseudomyrmex ferruginea was living in the thorns of this shoot
but it was too cold for the workers to be active outside of the thorns. This same cow was
observed to browse mature foliage from heavily occupied tall shoots that had been cut with a
machete in July 1964.

378 Tue University ScrIENCE BULLETIN

0.25-25 percent of the mature leaves from a shoot that were less than 2 m
above the ground. During the rainy season, these cattle do not normally
browse standing shoots of A. cornigera. However, they were occasionally
observed to eat a shoot tip or mouthful of mature foliage from both occupied
and unoccupied shoots.

During the cool season, these cattle also followed people cutting brush.
At this time there was a large amount of green foliage above the browse line,
and by following the workers, the cattle could get this food. The foliage from
occupied canopies of A. cornigera was sorted out by the cattle and eaten from
the piled mixture of Croton glabellus, Bauhinia ungulata, Mimosa albida,
Helicteres guazumifolia, Coccoloba sp., and other 2-4 m tall shrubs. It should
be noted that at this time of year, occupied shoots of A. cornigera have re-
duced ant populations on the surface of the shoot.

During the rainy season, the five oldest cattle in Torrealva’s herd came
to the canopy of any cut tree and began to browse. This was especially
noticeable in the case of large trees. The cattle were apparently attracted by
the sound of cutting and the movement of the other cattle toward the fallen
tree. The leaves of some of these trees are not normally eaten when the tree
is standing even if the leaves are within reach. In this case ( Ateleia ptero-
carpa, Tabebuia pentaphylla, Bauhinia ungulata), the cattle stop feeding
after a couple of mouthfulls. On the other hand, Guazwma ulmifolia and
Cordia alliodora are eaten readily, whether the tree is standing or cut. A.
cornigera is a special case; at this time of year (Jul.) it is almost never
browsed when standing, but when cut it is readily browsed. The cut A.
cornigera shoots had mature foliage growing at ground level and the cattle
had grazed on many other species of plants within a few inches of this foliage.
When the acacia was cut, it was not stripped of all its foliage by the cattle, but
10-75 percent of the mature leaves and shoot tips were often eaten. These
shoots normally had a thousand or more ants on their surface at the time the
cattle were feeding, but relatively few of these ants found their way onto the
cattle due to the disorganization of the colony on a newly cut acacia. The cat-
tle did not show an obvious response to the few ants that found them. It ap-
peared that the ant was not able to sting through the skin either on the bare
nose or the hairy face to a sufficient depth for the cow to feel it. The cow’s
nose was surprisingly tough or insensitive; needle-sharp spines sticking into it
often did not cause her to turn away.

The thorns of A. cornigera do not noticeably affect the cattle. They do
not recoil when stuck in the nose or cheek. A cow’s long tongue reaches
around and behind the leaf, and plucks it from the branch by pulling it
toward her. Occasionally the leaf petiole is cut with the incisors. Entire new
green branches up to 40 cm long are sometimes eaten. When a cow has her

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 3/79

head well among the foliage and finds herself being stuck when withdraw-
ing, she closes her eyes until her head is free.

When the cattle were feeding on standing A. cornigera during the cool
season, it was impossible to determine their reaction to P. ferruginea since
the shoots had very reduced ant populations outside of the thorns. During
the warm part of the year, both before and after the cool season, the cattle
were only very rarely observed to browse standing shoots of A. cornigera.
In the case of both occupied and unoccupied shoots, they usually turned
away from the shoot after a mouthful. However, on one occasion in late
April, an adult cow was observed to browse the mature leaves from un-
occupied shoots in treatment subplot N-1 for nearly 15 minutes. The cow
moved from shoot to shoot and did not eat more than five leaves from any
one shoot. On several occasions in the general area of plots N and O, cattle
in Farfan’s herd were observed to start to take a mouthful of foliage from
heavily occupied low shoots, but then turn their heads away without taking
any further leaves. There were large numbers of excited workers on the
surface of these shoots.

When the branches or leaves of A. cornigera were offered to any of these
cattle, they were usually rejected even when offered as alternative members
of series of highly acceptabl plants (Guazwma ulmifolia, Cordia alliodora,
grasses, Labiatae, Convolvulaceae, and Fabaceae (vines)). When small
branches of A. cornigera foliage (both de-spined and with thorns) were
offered as a mixture with grass or other acceptable plants, the A. cormigera
was usually sorted by taste and/or by vision from the bunch. The acacia was
not accepted, or if accepted it was often spit out. However, two old cows
were found that would eat equal mixtures (by count) of leaves picked from
A. cornigera and G. ulmifolia.

In summary, it can be said that at least under certain circumstances, a
brocket deer, a young burro, and numerous cattle ate the foliage of A.
cornigera. At least to these animals, it is not as distasteful as other plants
that are commonly left unbrowsed in these animals environment. This sug-
gests strongly that the repeated avoidance of A. cornigera by browsing
mammals may be due to some other factor than the taste of the plant itself.
Part of this avoidance factor maybe due to the aggressive behavior of P.
ferruginea, but considerable experimentation is necessary to settle the matter.

Contrast oF A. CORNIGERA, A. SPHAEROCEPHALA, AND A. CHIAPENSIS.
While Acacia cornigera is separable from Acacia sphaerocephala on the basis
of inconspicuous but relatively consistent details of external morphology,
Pseudomyrmex ferruginea treats them as if they were the same species of
plant. The thin-walled legume characteristically found on A. sphaerocephala
is indistinguishable from the thin-walled type of legume found on A. corni-
gera in the Temascal area. The short inflorescence of A. sphaerocephala

380 Tue Universiry ScIENCE BULLETIN

(Fig. 5a) is very similar to that of A. cornigera (Fig. la). It has the same
floret length and density, position on the shoot, and peduncle with a large
diameter. Though usually lacking nectaries on the leaf rachis, occasional
plants of A. sphaerocephala have been found with 1-3 nectaries distal to the
petiolar nectary. The petiolar nectaries of both species are indistinguishable
in gross morphology. The Beltian bodies of A. spaerocephala differ from
those of A. cornigera only in that those of the former are slightly broader and
have less red epidermal pigment. With the exception of the bicolored thorns
and completely untwisted type B thorns found in many parts of the range of
A. sphaerocephala, subpopulations and individuals of A. cornigera have
been found that display almost every variation in thorn form that is known
in A. sphaerocephala. While A. cornigera does not develop into low dense
mats of vegetation like those of A. sphaerocephala on sand dunes (Fig. 5b)
even when growing on sand dunes, there are many areas where A. sphaero-
cephala has an upright life form.

The principle ecological difference between these two species is that within
the study area, A. cornigera is a plant of wetter areas and A. sphaerocephala
is a plant of drier areas. Within each of their respective ranges, A. cornigera
and A. sphaerocephala have expanded into man-made disturbance sites in the
same manner and here they play ecologically equivalent roles in the plant
community. That they are two distinct species, and not merely two forms
associated with two different rainfall regimes, is shown by the following
facts. In areas where the species are sympatric, they flower at different times
of the year, and the plants are morphologically distinct from each other.
Apparent hybrids are rarely encountered. There is no doubt that the plants
determined to be A. cornigera and A. sphaerocephala in the study area are
properly associated with their type specimens.

A. cornigera has a much greater range than A. sphaerocephala but both
species are about equivalent in respect to the amount of morphological varia-
tion per unit area of range. While these two species seem very closely re-
lated, sufficient evidence is not yet available to postulate the actual course of
events that has led to their distinctness.

A. cornigera and A. sphaerocephala are much more similar to each other
than either is to Acacia chiapensts. P. ferruginea does not interact with A.
chiapensis in the same manner as it does with A. cornigera and A. sphaero-
cephala. At any particular site, 4. chiapensis is extremely variable in respect
to morphology and life form. The individual plants can be arranged from
those that are phenotypically similar to the sword-thorn acacia, Acacia
macracantha, to those that look like hybrids between A. chiapensis and A.
cornigera (Fig. 3a).

The similarity of the hybrids to 4. macracantha is based on the 1) pale
green leaves; 2) straight, thin, and slightly flattened thorns which are dark

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 381

brown, swollen, and have a tough, central parenchyma; 3) very low ratios
of swollen thorns to small stipular thorns; 4) spreading life form with the
shoot canopy as wide as the shoot is tall; 5) extremely bitter tasting foliage
that has little insect damage to the shoot tips; 6) very low number and small
size of Beltian bodies produced per leaf (A. macracantha lacks Beltian
bodies); and 7) fact that the shoot is invariably not occupied by a self-
sufhcient unit of P. ferruginea but always has founding queens of P. fer-
ruginea and many thorns previously opened by founding queens (A. mac-
racantha thorns are only very rarely entered by P. ferruginea queens).

The similarity of the hybrid plants, at the other end of the range of varia-
tion, to A. cornigera is based on the 1) dark green leaves; 2) thick and round
dark brown thorns that occasionally are curved and have soft central paren-
chyma; 3) very high ratio of swollen thorns to small stipular thorns; 4)
slender life form with most thick branches being vertical; 5) bland tasting
foliage that often exhibits repeated damage to shoot tips by insects of the
same species that feed on unoccupied A. cornigera; 6) medium number and
normal size of Beltian bodies per leaf; and 7) fact that the shoot is almost
invariably occupied by a unit of P. ferruginea whose size is proportional to
the number of Beltian bodies produced by the shoot.

While it is plausible and likely, sufficient evidence is not available to
support the hypothesis that the plants which are phenotypically similar to A.
cornigera are in fact exhibiting the expression of genes or chromosomes
gained through occasional crossing between the two species. The same
genera of bees pollinate both species. In the Temascal area, most shoots of
A. chiapensis flower in January through March. However, occasional shoots
of both species are found in flower in all months of the year.

Before the existence of man-made disturbance sites in the Temascal area,
A. cornigera and A. chiapensis grew as scattered and mixed populations on
river banks and in arroyos, based on observation of the few remaining un-
disturbed sites. When man entered the area, A. cornigera spread rapidly into
the fields, pastures and roadsides. A. chiapensis has stayed in the natural
disturbance sites and is now relatively much rarer than A. cornigera. A
major reason for this is that while dissemination of seeds of A. cornigera by
birds is very common, there is no evidence of overland seed dispersion of
the seeds of A. chiapensis by birds. In the latter species the seeds are im-
bedded in a dry and to me, tasteless pulp. The legumes of A. chiapensis
often drop from the shoot before they completely dehisce and they will float
for at least an hour; they are most likely normally water dispersed.

There is a second reason why A. cornigera has spread much more rapidly
than has A. chiapensis. Like A. cornigera, A. chiapensis needs full sunlight
for normal growth. Man-made disturbance sites often grow a dense shading
canopy very rapidly. When occupied by a colony of P. ferruginea, A. cornt-

oS)
CO
bo

Tue University ScrENCE BULLETIN

gera is quite capable of growing fast enough to keep above the rapidly
growing surrounding vegetation. As one moves across the range of varia-
tion of A. chiapensis, from being phenotypically like 4. macracantha to being
phenotypically like A. cornigera, the ability to produce a large annual vertical
height increment increases; the susceptibility to insect destruction of the
shoot tips also increases. However, these rapidly growing plants are not able
to sustain an ant colony large enough to insure freedom from insect damage.
Therefore, most of the A. chiapensis shoots do not grow fast enough to re-
main above the surrounding canopy. In natural disturbance sites, the canopy
is generally more interrupted and the plants are often able to lean out toward
the water. In these areas the slowing of growth by insect damage is not so
serious to the plant due to the lack of shading plants.

Within the study area, A. cornigera has a much greater geographic and
ecologic range than does A. chiapensis. A. chiapensis appears to be much
more closely related to the Guatemalan Acacia donnelliana Sat. and the
Costa Rican and Panamanian Acacia melanoceras Beurling than it is to
either A. cornigera or A. sphaerocephala.

In respect to the members of the genus Acacia as a whole, the swollen-
thorn acacias possess a distinct set of characteristics. If those swollen-thorn
acacias which are consistently found to be occupied by some species of
Pseudomyrmex are separated as a group from those which are inconsistently
found to be occupied by some species of Pseudomyrmex, then Acacia cor-
nigera can be regarded as representative of the former group and Acacia
chiapensis as representative of the latter group. This is not to say that A.
cornigera is equal in all quantitative respects to a plant such as A. sphaeroce-
phala but rather that it is very similar to it in qualitative characteristics.

Foliar nectaries are commonplace throughout the genus Acacia in
Mexico and Central America. However, they do not reach the degree of
morphological proliferation in other species that is attained by those of the
swollen-thorn acacias. A. cornigera and A. sphaerocephala undoubtedly
produce more nectar per leaf than do any other acacias in the study area.

The leaves and their products are essential to growth and survival of the
ant colony. While A. cornigera leaves do not generally remain on the shoot
for more than a year, it is notable that during the dry season the loss of
leaves is gradual. New leaves are continually produced in axillary tufts so
that living shoots are rarely leafless. The leafless condition is common during
the peak of the dry season in other acacias in the study area and especially
those that are sympatric with A. sphaerocephala.

Within the study area, Beltian bodies are peculiar to the swollen-thorn
acacias. A shoot of A. cornigera that is growing in full sunlight, occupied
by P. ferruginea and over two years old, produces a Beltian body on the tip
of every leaf segment. Within its life span, A. cornigera undoubtedly pro-

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 383

duces more weight of Beltian bodies than either A. sphaeocephala or A.
chiapensis. In contrast to the usual case with leaf parts, Beltian bodies have a
very high nutritive value; yet there is no evidence that their production by
the plant is of direct value to the plant itself. They possess a vascular trace
and appear to be a modification of the tip of the leaf segment rather than an
outgrowth of the leaf margin. The rate of leaf production during the dry
season is less than 20 percent of that during the rainy season. But since small
numbers of new leaves are produced as leaf tufts in thorn axils, small num-
bers of Beltian bodies are produced on many shoots during the dry season.

Within the study area, there are five species of acacia that regularly bear
stipular spines which are over 3 cm long. Of these, A. cornigera, A. sphaero-
cephala, and A. chiapensis are regarded as swollen-thorn acacias; Acacia
macracantha and Acacia farnesiana are not placed in this group because they
do not have swollen thorns, are not inhabited by ants, and do not produce

Beltian bodies.

Of these five species, A. cornigera produces the largest volume of space
in its thorns (although A. sphaerocephala may rarely do just as well). As
much as one-half of this volume may be contained in the type B thorns
which are highly twisted back on themselves and are not located in the
area of most new leaf production. Both type A and type B thorns are much
shorter than their leaves. Both cattle and the Brocket Deer are not deterred
from eating mature leaves by the presence of the type A thorns on lateral
branches (Fig. 17 and 27). The long tongue of the cow and the long narrow
face of the deer are both well suited to reaching between the short, widely
spread thorns to remove leaves. The thorns contain considerably more
lignified tissue than is necessary to make them as strong as those of the same
length that are produced by A. macracantha. However, this lignification is
necessary if the thorns are to contain a large internal lacuna and still be as
strong as the solid thorns of A. macracantha which are much thinner. While
new swollen thorns are generally not produced by A. cornigera or A.
sphaerocephala during the dry season, they are produced with great regularity
during the rainy season. A. chiapensis is less consistent in its rate of produc-
tion of swollen thorns.

A. cornigera appears to have the highest growth rate of any species of
acacia in the study area. This applies to both lengthening of specific branches
and to shoot height increment. Such a trait is important to a plant which is
dependent on full sunlight for normal maturation and yet must attain this
maturation while growing among vegetation which is constantly rising in
height and producing heavy shade. Possession of a very tender shoot tip that
lacks fibrous tissue, and external epidermal modifications for protection, is
apparently associated with the very rapid rate of branch elongation.

To the author’s taste, the shoot tips of 4. cornigera and A. sphaerocephala

384 THe UNiversity ScIENCE BULLETIN

have a pleasant bland flavor. The shoot tips of the pale-leafed form of A.
chiapensis are very bitter to the author’s taste. The shoot tips of A. macra-
cantha, A. farnesiana and seven unidentified species of Acacia in the study
area are likewise very bitter tasting. Apparently associated with the physical
tenderness and relative lack of distinctive flavoring compounds of the shoot
tips of A. cornigera, they are accepted as food by a number of insects that
normally feed on other species of plants. It is also eaten by a cricetid rodent,
the Brocket Deer, and livestock. Loss of foliage to these phytophagous
animals takes place only under special circumstances, if a large colony of
P. ferruginea lives in the shoot. It is almost guaranteed to take place if the
shoot is unoccupied.

A seedling of A. cornigera or A. sphaerocephala requires at least three
full growing seasons before it will flower in the following dry season. The
first early flower crop is produced in the fourth dry season. A sucker of A.
cornigera from a large stump may flower when only nine months old but
it usually does not flower until the end of its second dry season. The seed
does not mature until 10 to 12 months after flowering. A. chiapensis seed-
lings and regenerating suckers require one to two years longer to flower for
the first time and the seeds mature at the time of the next year’s flowering.
In natural disturbance sites, reproduction of A. cornigera, A. sphaerocephala,
and A. chiapensis occurs often from direct growth from seeds as it does
from suckers from cut or burned stumps. Yet in man-made disturbance sites
that are repeatedly cleared, these three swollen-thorn acacias are heavily de-
pendent upon regeneration from root stock to maintain their existing
densities.

Throughout the study there has been no evidence that indicates that any
of the traits described above are directly dependent upon the presence of a
colony of P. ferruginea for their expression.

APPENDIX I. DETAILS. OF .DEFOLIATOR, AGTIVIiES

Mozena tomentosa Ruckes.

Mozena tomentosa is a 2cm long brown coreid bug. The adults are present all year and
both adults and nymphs are very common during the rainy season. They feed both day and
night and suck the sap of the terminal and penultimate leaf rachis and shoot tips of Acacia
cornigera. This causes permanent wilting of the portion of the leaf distal to the puncture and
cessation of growth of the punctured shoot tip. A cluster of 3-15 nymphs or 1-4 adults may
feed on one shoot tip or axillary tuft of new leaves. With the exception of one female on the
shoot tip of Mimosa albida, M. tomentosa was not found feeding on other species of plants.
The adults are very adept at locating young A. cornigera shoots that are submerged in dense
second growth, and are responsible for the wilted shoot tips that are very commonly found on
young shoots that are not occupied by a worker force of P. ferruginea. In an experimental plot
where the acacia sucker shoots were nearly dead following nine months of defoliation by all
types of phytophagous insects (H-1), M. tomentosa were often the only insects that were still
attracted to the tough and calloused branches that were still living. It found the new axillary
buds almost as soon as they appeared above the epidermis. During the dry season, M. tomentosa
adults, larvae of the noctuid Coxina hadenoides, and larvae of the gelichiid Aristotelia corallina
were the three major active defoliators of A. cornigera.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 355

Workers of P. ferruginea readily attacked adults and nymphs of M. tomentosa. If the workers
invaded a shoot on which a bug was feeding, it sometimes required repeated attacks by single
workers to induce the bug to withdraw its mouthparts and leave. Occasionally ants wiped their
mouthparts on the leaf surface following these attacks. When a bug lit on a shoot tip that was
being patrolled, it was immediately attacked and departed. On 12 July 1964, an adult female
of M. tomentosa was placed on the shoot by the author. She was bitten 18 times in 62 seconds
as she fell from branch to branch and finally flew off the occupied shoot.

Umbonia orozimbo Fairm.

Umbonia orozimbo is a 6 mm long green and red membracid. The adults were present on
A. cornigera all year and the nymphs were found from Aug.-Jan. (Fig. 28). Adults were also
found feeding on the stems of Acacia chiapensis and Acacta macracantha. While not abundant
in the adult stage, on two occasions a female was found with her brood of nymphs feeding
on the main vertical branch 20 cm below a wilted shoot tip. U. orozimbo was found both in
experimental plots and on shoots that had naturally lost their colony. When a stem covered
with nymphs was bent over to contact a queen-shoot occupied by P. ferruginea, 82 nymphs and
the female were forced to jump off in 2 minutes and 42 seconds due to biting and stinging
attacks by the worker ants.

Pelidnota punctulata Bates

Pelidnota punctulata is a 25 mm long rutelline scarab with yellow elytra and a green to
brown pronotum. The adults were common from June through September; they fed on the
shoot tips of A. cornigera only at night. Neither the bettle or its feeding damage was found on
other species of acacia. During the day, the beetles rested in the most heavily shaded part of
the shoot and this was usually toward the center of a mass of type B thorns. P. punctulata ate
the shoot tips of emergent shoots of all sizes. The beetles were not found in shaded sites.
Commonly, two or three beetles were found feeding on the same shoot. The beetle remained
on a single shoot for one or two nights, and during this period usually ate all the shoot tips.
Single beetles required about 30 minutes to eat a shoot tip. They then crawled down the
branch and out onto another branch to another shoot tip. P. punctulata ate 5-60 percent of the
shoot tips that were destroyed by insects during the first part of the rainy season in the control
subplots (Jun. and Jul., 1964). Once the beetles had removed the shoot tips from all of the
plants in a small area (e.g., one-fourth acre), they left and a new crop of shoot tips developed.

P. punctulata ignored the attack of P. ferruginea. Very commonly there were up to 100
workers attacking a single beetle while it fed. The only time the beetle appeared to notice was
when stung under the labrum. Then the beetle brushed off the ant with a foreleg. The beetles
were rarely found on unoccupied shoots because unoccupied shoots usually lacked undamaged
and succulent shoot tips, and the adults did not remain long on such shoots.

Diplotaxis denigrata Bates.

Diplotaxis denigrata is an 8 mm long dark brown melolonthine scarab. From late May
through the first week in August the adult beetles were commonly found feeding at night on
Acacia macracantha, Acacia chiapensis and on unoccupied shoots of A. cornigera. They ate the
shoot tips and new leaves, and flew to the plants shortly after sunset. While up to 1,000 in-
dividuals were counted on one shoot of A. macracantha, the number on A. cornigera ranged
from 1-50. When there were over 10 bettles on a shoot, they usually destroyed every new shoot
tip in one evening of feeding. In the area where the beetles were very common (plot M and L)
they were rarely found on shoots lacking undamaged shoot tips, but if a dying plant had as
few as one shoot tip remaining, it often had D. denigrata adults feeding on it.

Adults alighting on shoots occupied by P. ferruginea were attacked immediately and they
dropped off or flew away. On one occasion, a series of heavily occupied shoots of A. cornigera
had shoot tips eaten by D. denigrata because the evening was too cool for the ants to be active
outside of the thorns without substantial disturbance of the colony. The beetles did not provide
this disturbance. Feeding D. denigrata did not shake the foliage to any noticeable degree.

One specimen of Diplotaxis simplex Bates was also taken while feeding on A. cornigera
along with a large group of D. denigrata.

Anomoea sp.

Anomoea sp. probably A. rufifrons is a 1 cm long clytrine chrysomelid with yellow males,
and black females with a red thorax and a broad red elytral band. The adults were present from

386 THe UNIversity ScIENCE BULLETIN

late May through late August and were most abundant in June. They fed during the day on
undamaged shoot tips and rested at night on the ends of lateral branches. A. ?rufifrons was a
voracious feeder and concentrated on the tips of the shoot tips. One male was observed to eat
the terminal 1 cm of a shoot tip in 16 minutes. Adults of both sexes were found feeding on
A. chiapensis as well as on A. cornigera. Normally, A. ?rufifrons was found more commonly on
A. chiapensis but it became very common on the A. cornigera in the experimental plots that
lacked P. ferruginea. This beetle was the most desrtuctive of the species of clytrine chrysomelids
that fed on unoccupied A. cornigera. Its importance was accentuated because it was active during
the major part of the growing sesaon (rainy season) and was very adept at finding unoccupied
shoots that still had undamaged shoot tips.

The adult beetle was very sensitive to the approach of worker ants and usually flew from
the shoot before the approaching ant could contact it. If a worker did catch the beetle, the
beetle usually fell to the ground and eventually dislodged the ant. When the beetle flew from
the shoot, it did not return to the same plant but flew on until it found another. A female was

observed to land on 11 different occupied shoots in 3 minutes and 14 seconds and was chased
from each one by P. ferruginea.

Chrysobothris sp.

Chrysobothris sp. near C. multistigmosa is a 25 mm long, brown buprestid beetle. The
adults were common throughout the year on A. cornigera and A. chiapensts. They flew readily
between the shoots and rested on unoccupied shoots at night. They were occasionally seen eating
the midribs of leaves and pinnae. The females oviposited in freshly cut stumps and shoots, and
in dying shoots that were still standing. The larvae were occasionally found to kill shoots with
a 1.5-3 cm basal diameter by internal girdling. They were a very common cause of mortality to
unoccupied living stumps of both 4. cornigera and A. chiapensts.

The adults walked with a quick, jerky pace that attracted the patrolling and cleaning workers
on occupied shoots. Usually, a single ant attack was sufficient to cause the beetle to fly from the
shoot; it then flew directly to another shoot. The larvae were unaffected by the presence of the
ants since they are under the bark.

Acanthoscelides oblongoguttatus Fahr.

Acanthoscelides oblongoguttatus is a 6 mm long, mottled brown bruchid beetle. Associated
with the fact that there was a small percentage of shoots with seed pods present throughout the
year, adult beetles were likewise present throughout the year. They were also occasionally seen on
the foliage and at the foliar nectaries of A. cornigera. This bruchid was reared only from A.
cormgera and A. sphaerocephala in this study. The larvae of the few females that found the
seed-bearing shoots at the time when the pods were still green but the seeds were blackening,
achieved 5-75 percent destruction of the seed crop. Those females that arrived later and the
ones of the first generation offspring completed the infestation of the remaining seeds. The
mature pods usually remained on the shoots long enough for at least two generations of
beetles to pass through them (2-3 months). The only seeds that survived were those that were
carried from the shoot by birds or fell to the ground after birds opened the pods, and were not
infested before leaving the shoot.

Adults of A. oblongoguttatus were most active on the pods in the late afternoon. The
worker ants that were patrolling the pods attacked them whenever they were encountered but
the beetles just flew to another pod. Since the worker ants did not bore into the pods and the
beetle larvae were inside of the seeds, the ants had little or no effect on the larval bruchid in-
festation. Once the adults had emerged through their exit hole, the ants did not enter the pods
but other insects did.

Coxina hadenoides Guen.

Coxina hadenoides is a very common noctuid moth with a 15 mm wingspread. The fore-
wings are finely mottled brown, white and black, and the hindwings cream with slight dark
markings. The mature larvae are about 3 cm Jong and 4-5 mm in diameter. They may be
yellow, gray, or brown and are characterized by two large black spots on the ventral side of
the abdomen. When resting during the daytime, a larva holds itself appressed to a branch or
else straight out from the branch at an acute angle while holding on with just the prolegs and
looks like a dead twig. The first instar and older larvae were found at night feeding on the shoot
tips of A. cornigera, A. chiapensis and rarely on A. macracantha. Mature foliage was not eaten
except when the larvae were starved. During the day, the larvae rested upon unoccupied shoots.
If the shoot was occupied, they usually walked off onto other plants. The cocoon was spun in

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 387

the leaf litter on the ground. The moths and larvae were present in the field throughout the
year.

C. hadenoides did more damage to A. cornigera in the Temascal area than did any other
species of insect. The larvae wandered from shoot to shoot, and a third instar larva could eat
3 to 6 shoot tips in one night. The larvae fed unhindered on unoccupied shoots. The degree
of damage inflicted by C. hadenoides to occupied shoots was a function of the density of patrolling
and cleaning workers at night. If the attack of the workers was sufficiently strong as a larva
started onto a shoot, it turned around and left. This required more than an occasional worker.
Each worker that bit the larva was grabbed with the mandibles and forelegs and thrown from
the shoot by the larva. An attack frequency of about two workers per second for more than 2-3
seconds caused a larva to leave. If the workers came in waves, a larva could remove up to
four per second.

If not deterred when first climbing onto the shoot, the larvae walked to the top of the shoot.
If undamaged shoot tips were found, the larvae began to feed almost immediately. The
workers cleaning and patrolling in the area of the shoot tip usually attacked immediately and
were thrown from the shoot. If the frequency of attack was high enough (about 2 workers per
second) a larva would run down onto a thorn or leaf and take up a resting or defensive posi-
tion. In the latter case, the larva backed out onto the thorn, with its head toward the branch
and held about 5 mm above the surface. The attacking workers usually contacted the larva
just anterior to the first pair of prolegs and were easily grasped and removed.

On 3 July 1964, a last instar larva that was placed on an occupied shoot at 9:00 a.m., was
backed into a defense position on a type A thorn on a lateral shoot. In 32 minutes, it threw 75
workers off of the shoot. If the evening became cooler and worker activity outside of the thorns
decreased, the resting larvae moved back to the shoot tip to feed. As many as seven larvae were
found on the single shoots that normally had a large worker force outside of the thorns but for
some reason did not on that particular night.

First and second instar larvae were not strong enough to throw ants off of them but they
ran (looped) almost as fast as the workers. When chased off the shoot, the larvae often dropped
a few centimeters on a silk thread and later climbed back up to the leaf. The only shoots that
went through an entire growing season without being damaged at least occasionally by C. hade-
noides were regenerating suckers from large stumps which were occupied by a very large queen-
unit which had moved in from the canopy of the cut shoot.

On one occasion, a worker was observed to bite a larva of C. hadenoides behind the head in
such a manner that the larva could not grab the worker. However, each time the worker brought
her gaster around to sting, the larva bit the gaster and the worker retracted. After about 15
minutes of this struggle, the author picked up the larva . The larva regurgitated a clear fluid
onto the author’s fingers. Some of this contacted the worker and she was completely paralyzed
in less than 5 seconds.

Syssphinx mexicana Bsdv.

Syssphinx mexicana is a yellow syssphingid moth with 5 cm wingspread. The larva is green
with silver and red markings; when fully mature it is three inches long and 1.5 cm in diameter.
Both adults and larvae were present from late May through September. The eggs were glued
in rows of 2-5 eggs on a leaf midrib, thorn or branch. Egg parasites were found in one egg
clutch. The eggs required 6 days to hatch and the larvae matured in about 20 days. Starved
larvae pupated when about one half normal size and produced apparently normal but smaller
adults. The larvae burrowed into the surface litter to pupate. Out of 51 last instar larvae col-
lected in the last two weeks of July, 28 were parasitized by 1-4 large tachinid larvae. Five
larvae consumed essentially the entire crop of mature leaves of a 200 cm tall shoot (approximately
200 leaves) during their maturation.

When ovipositing, the female was attacked by any workers of P. ferruginea that were present.
She hung on the branch or thorn and curled her abdomen up to lay the eggs on the upperside of
her support. The abdomen and legs are covered with a dense yellow pile. When the workers
bit her, they only succeededhin pulling out a tuft of this pile. Workers found the moth’s abdomen
difficult to walk on and fell off frequently. It took the female about five seconds to lay five
eggs. On a shoot with many active workers, about five workers attacked her while she was
Ovipositing on a lateral branch. More than six larvae or eggs were not found on a single oc-
cupied shoot. The adult female usually flew 3 to 15 m between ovipositings. She did not
oviposit on plants that where below the level of the canopy. The eggs were bitten but un-
damaged by the workers but they did not persist in this attack. When eggs were found on the
shoot that were present for a day or so, there were no workers chewing on them. However,
once the larva had hatched and the smooth hard chorion was broken, the workers gradually
chewed away the remaining chorion.

388 Tue University SciENCE BULLETIN

It is not known to what extent the newly emerged larvae were attacked by workers. By
the time the larvae had reached a length of 3 cm they were not attacked by the workers patrolling
the cleaning around them. Larvae were only found on the mature leaves and confined their feed-
ing to them. They fed at night and usually rested on the underside of the leaves during the
day. When a shoot was stripped of its leaves, the larvae occasionally tried to feed on the shoot
tips. When they crawled up onto the shoot tips, they were attacked by any workers present
and usually driven off and back down the stem. By the time the shoot had lost its leaves, the
larvae were usually large enough to pupate and they left the shoot. After the shoot was
defoilated and the larvae had left, it immediately put out a new set of leaves from the thorn
axils.

If a larva was taken from a shoot occupied by one colony and placed on a shoot occupied by
another colony, it was usually attacked. It reacted to this attack by smearing regurgitate on
the biting and stinging worker. Workers often ran down the body of a larva on their own shoot
as if the larva was part of the plant. Even when the larva was eating foliage, workers were
observed to run across the chewing mandibles and to completely ignore the damaged area.

It appeared that the attack of the workers affected the number of eggs laid by the female.
She normally oviposited her small egg clutches on occupied shoots since these were by far the
most common ones. However, it was common to find clutches of 10-45 eggs laid on 3-4
different branches of an unoccupied shoot in an experimental plot (E-1). After these larvae
hatched, they very quickly removed the leaves (already in reduced number) and began to
wander in the plot. Some found other unoccupied shoots and completed their development.

Clutches of four or five empty egg shells were often found on shoots that had one or two
maturing larvae. Predators were almost never observed on occupied shoots except for Polistes
which did not forage intensively on occupied shoots, apparently due to the constant attack of
the wasps by worker ants. A certain percentage of the first instar larvae may have been removed
by the workers before the larvae became accepted by the colony.

The females did not appear to fly up to the canopies to oviposit on shoots that were over
3 m tall. About 2 out of each 100 shoots had larvae of S. mexicana on them. Though loss of
mature leaves was very high, it appeared to have little effect on the plant since the shoot tips
were not destroyed, the larvae left and did not continue to eat the new foliage, and the shoot
produced a new set of leaves within two weeks.

Aristotelia corallina Wals.

Aristotelia corallina is a gelichiid moth with a 1 cm wingspread and pink, yellow and black
markings. The mature larva is 10-12 mm long with the anterior one-half of the body ringed
with black and white and the posterior one-half striped with black and white. Adults and larvae
were present throughout the year but were most abundant during the last three months of the
dry season. During these months, the larvae were very common; in plot J almost every shoot
had one to four larval webs spun among the leaves. The larval webs were spun in the tops of
shoots within the 10-50 cm size range, and during the night the larva left its web to feed on
the shoot tip. The larvae of A. corallina were in large part responsible for the lack of height
increment of shoots in this size range. During the rainy season the webbing became soaked
and this was associated with high mortality among the larvae. At this time, larvae were oc-
casionally found in the tops of taller shoots.

When a colony of P. ferruginea invaded a shoot with the larvae and webbing of 4. corallina
on it, within a matter of a day or less, all webbing and larvae were removed. On two occasions
the workers caught a larva and tore it into pieces, some of which were carried into the thorns.

Halisidota sp.

Halisidota sp. probably H. pura is an arctiid moth with a 2.5 cm wingspread. It has yellow
and brown forewings and cream colored hindwings. The larva is densely covered with long
yellow hairs. The larvae were common from September through November. They ate the new
green thorns on the shoot tips of unoccupied shoots (Fig. 29). While they did not appear to
seek the terminal point of the shoot tip, the last 1 cm was usually eaten along with the last
green thorn. During the fall growing season, the larvae were responsible for at least 10 percent
of the damaged shoot tips in plot L. They fed at night.

Larvae of this moth placed on occupied shoots were immediately attacked by the workers.
They bit at the hairs and legs. The larvae tried to escape by running down the leaf or branch
and the workers chased them until they ran off the end of a leaf or branch or were forced to
let go and fall. Many of the long hairs were pulled out during this harassment but the workers
could not get close enough to the body to bite the larva unless they came in from underneath

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 389

and bit a proleg. The larvae regurgitated on workers that were biting them. This regurgitate did
not kill the workers and often did not cause them to let go.

Rosema dentifera Draudt.

Rosema dentifera is a notodontid moth with a 2.5 cm wingspread. The forewings are bright
green and cream colored hindwings. The mature naked larva is about 3 cm long and green with
widely spaced thin black and white bars. The larvae were common from late July through
December and fed on the mature foliage of A. cornigera at night. They were found only on
shoots with a large worker force. The thin cocoon was spun in the axil of one of the lower
thorns, usually incorporating a section of a leaf into the cocoon wall.

The workers cleaned and patrolled around the resting and feeding larvae without disturbing
them. They likewise showed no reaction to the presence of the workers. However, if a larva
was removed from a shoot occupied by one colony and placed on a shoot occupied by another
colony, it was usually attacked immediately in the same manner as were most lepidopterous
larvae. The larvae took evasive action by running out onto the end of a thorn and if too many
ants followed, they jumped off. Lacking long hairs, they were very vulnerable to the workers.
R. dentifera was never found on a part of the shoot that was heavily patrolled such as a major
shoot tip. It may be that the ants did not tolerate their presence in this area just as they did not
tolterate Syssphinx mexicana larvae on the shoot tips.

Acromyrmex octospinosus Reich

Acromyrmex octospinosus is a 5-10 mm long red attine (leaf-cutter) ant that was common in
the area. On three occasions, foraging columns were observed to remove the shoot tips, green
thorns and most of the mature leaves from a young unoccupied sucker regenerating from cut
stumps in subplot A-1 and H-1l. Their damage was not found on unoccupied and occupied
shoots that were over 100 cm tall.

When a queen-shoot was cut and placed against the base of a stump that was being stripped
by A. octospinosus, the workers of P. ferruginea that ran out onto the shoots attacked the leatf-
cutter workers immediately. These later workers either ran or fell off the shoot and the trail
to the shoot was not maintained.

BIONOMICS OF PSEUDOMYRMEX FERRUGINEA F. SMITH

MATERIALS AND MetTHops

Collecting colonies. A unique feature of ants that have an obligate relationship with a plant
is that the entire colony can be collected as a unit, killed, and later examined without having
seriously damaged the distribution of brood within the colony. The ants occupying a shoot less
than 100 cm long can best be collected by cutting the shoot into sections and dropping the pieces
into a large cyanide jar. The workers of very young colonies often can be induced to enter the
thorns by tapping the shoot trunk two or three times. This reduces the loss of workers when
the shoot is collected. If large numbers of workers are active on the surface of such a small
shoot, they can be easily aspirated from the surface. Except for the few knocked off by the blow,
cutting a shoot does not cause the ants to leave and thus the shoot can be transported to a
more convenient site for working. The ideal time to collect a large colony is on a cool day
(less than 21° C.). At this time, a very low disturbance reaction is produced as the thorns are
stripped from the tree.

If it is necessary to collect a large unit on a hot day, the best method is to strip the thorns
off one at a time and drop them into a large jar in which a small strong cyanide jar has been
placed. As the thorns are removed, there often is a very strong swarming reaction on the shoot
surface. These ants are easily aspirated and the entire shoot content is obtained. When placed
in the cyanide jars, most workers die in the thorns and the brood is not thrown out. Thus, when
shoot sections or thorns are numbered, the morphology of the colony can be reconstructed.
Thorns with brood can be preserved with contents intact by dropping them directly in a liquid
preservative. Large colonies were also collected by spreading a sheet under the shoot and spraying
the latter with parathion. After several hours had elapsed, and most of the workers outside the
thorns had dropped from the shoot, the thorns were stripped off. However, this method is
awkward and time consuming.

When collecting samples of workers that are seen doing particular things on the shoot
surface, extreme care must be taken not to disturb the colony so that the collection site does
not become covered with disturbed workers. If the queen is desired, she is obtained most easily

390 Tue University ScireNcE BULLETIN

by choosing a large colony on a shoot at least 10 m from any other shoot. The four or five
largest clusters of type B thorns are stripped off and each thorn is cut open until one is found
packed with eggs. If the queen is not in the egg-filled thorn, she is usually in an adjacent one
having an enlarged entrance. If she is in none of these thorns, she has recently moved to an-
other thorn higher in the acacia, or the shoot contains an auxiliary-unit without a queen.
Founding queens are easily collected by opening the thorns of 20-50 cm tall shoots that have no
workers on them. Alates are most readily obtained throughout the year by opening type A
thorns on lateral branches of shoots 3 or more m tall.

Transporting live colonies. Units can be easily transported when in the shoot. A whole shoot
is cut and tied to the outside of the car; the wind prevents the ants from leaving the thorns.
Shoots with their roots can be dug up and transplanted. The leaves usually fall off but new
leaves are produced so rapidly that the colony is not lost through starvation. The occupants of
a thorn can be kept inside by plugging the entrance with a piece of wood (a broken off
thorn tip is convenient); the workers chew through tape.

Recording ant activity outside of the thorn. The number and distribution of worker ants
outside of the thorn is of great importance to the acacia. The most reliable fast counts are
obtained using a hand-counter and starting at the base or top of the shoot and moving up or
down. After this count on the main trunk, or concomitant with it, those on the lateral branches
are counted. When over 50 workers are present of the surface, it is reasonable to count by fives
or tens; this method gives about 90 percent accuracy. In making a census of this type, the
faster the counts are made the more accurate they are; workers move around on the tree but
tend to stay in one general area for 1 to 10 minutes. Only those visible from one side of the
shoot can be reasonably counted at one time but since there is much more lateral movement
than vertical, a count on one side only takes in 75 to 90 percent of those on the surface. The
wider and taller the shoot the less accurate the surface count is but since most experimental plots
contained roughly even-aged shoots, comparisons seem valid within the plot. There is no
strict relation between the numbers of workers on the shoot surface and the type of unit living
in it. In general, the larger the colony the more workers there are on the shoot surface under
given environmental conditions, but this is a highly variable parameter.

Marking workers. Individuals can be marked with a small dot of white paint (Testor’s
Dope) on the thorax or gaster. It is best to apply this with a single horsehair while the worker
pauses, since handling them introduces the possibility of injury to the ant.

Mating flights. Once the site of a mating flight of P. ferruginea is located it is very simple
to observe mating behavior in detail. Mating activity usually continues well into the early day-
light hours and can be closely observed by climbing the object around which the swarm is
taking place. The presence or absence of an early morning mating flight is easily recorded by
hanging a 20 x 20 cm piece of sheet metal covered with sticky tree-banding compound high up
on telephone poles or on the fronds of prominent outstanding palms in the area. Both males and
females adhere to the surface.

Maintaing laboratory colonies. The reactions of workers to brood and reproductives is best
determined by observing laboratory colonies in glass tubing. Large colonies (over 500 workers)
can be maintained in the laboratory by feeding them honey and Beltian bodies. The colonies
are started by stripping about 30 of the largest thorns off of a 180 cm queen-shoot. These thorns
are then chilled in the freezer compartment of a refrigerator until the workers are immobile.
The contents of six thorns are then dumped into a 14 by 500 mm glass tube with one end
plugged with cotton and a strip of paper toweling within the tube. Five of these tubes are laid
at right angles across six 4 by 400 mm glass tubes that are supported 10 inches above a 100 x 50
cm metal sheet. The margins of this sheet are ringed with tree-banding compound. Water
moats are unsuccessful since large numbers of workers drown themselves. Honey is dripped on
plastic strips laid parallel to the large diameter tubes. To provide Beltian bodies a shoot tip with
several mature subtending leaves is placed in a bottle of water of the metal sheet so as to con-
tact the tubes. Any swollen thorns on the shoot tip are clipped. To provide new shoot tips,
the 5-10 workers on the wilted shoot tip have to be sacrificed.

The chilled workers revive almost immediately and after a day pile the brood at one site
in each of the big tubes. After a period of several days to a week, the queen and her tube’s
contents are moved by the workers into one of the long 4 mm diameter tubes. Gradually the
remainder of the brood is moved into these narrow diameter tubes. For the first week, the
colony is given only honey and the queen becomes less physogastric and ceases egg production.
When a shoot tip is presented, Beltian body harvest occurs immediately. Within 6 to 12 days
the queen regains her physogastric condition and egg laying is resumed. These colonies appear
relatively normal and workers patrol the outside of the glass tubes in much the same manner
as they patrol a shoot. They collect the major part of the honey just before dawn even though
it is available for 24 hours. The queen with her eggs remains roughly segregated from the
other brood in one of the narrow tubes but larvae are present in other parts of the same tube. The

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 391

queen moves around much more than she would have been abel to in a thorn on the shoot.

Founding queens with their first brood and a few workers can be kept temporarily in pill-
boxes with cover-slip glass tops. A small vial of honey is placed inside. Groups of workers with
brood can also be kept in this manner.

Systematics. Substantial confusion has surrounded the taxonomy of the
reddish-brown pseudomyrmecine ant discussed at length in this paper. With
the exception of the type description (F. Smith, 1877), Pseudomyrmex
ferruginea has been primarily discussed in the literature as Pseadomyrmex
belti fulvescens Emery (1890) or Pseudomyrmex fulvescens Emery (1890).
Emery’s type specimen has clearly come from the population discussed in
the present paper; F. Smith’s type from Mexico (contained in the British
Museum) has kindly been compared with representative specimens from
eastern Mexico by Dr. I. H. H. Yarrow and there is little doubt that the

common reddish-brown obligate acacia-ant in eastern Mexico 1s P. ferruginea.

The rare, smaller and yellower obligate acacia-ant in eastern Mexico is
Pseudomyrmex nigrocincta Emery (1890); it becomes progressively more
abundant along the Pacific lowlands of Central America down to Costa Rica.
See Janzen (1967a) for a discussion of the taxonomic problems associated
with the obligate acacia-ants in Central America.

A few species of ants besides P. ferruginea (Fig. 40) may be found living
in dead or live swollen-thorn acacias within the study area. Since these ants
do not have the same influence on the ecology of the swollen-thorn acacias as
does P. ferruginea, these ants are mentioned below so that a shoot inhabitated
by them is not recorded as being occupied by P. ferruginea. The other two
species of Pseudomyrmex occasionally found living in live swollen-thorn
acacias, Pseudomyrmex nigrocincta and Pseudomyrmex gracilis mexicana
Roger are discussed in the section entitled OTHER ACACIA-ANTS IN
THE STUDY AREA. At least three other species of Pseudomyrmex may
be found on very rare occasions living in thorns opened by other ants; they
use these thorns in the same manner as hollow twigs. Ants in the genus
Crematogaster often live in the thorns of large dead or living unoccupied
swollen-thorn acacias. They enter through entrances made by founding
queens and workers from previous occupation, and though the holes made
by microlepidoptera and zygopine weevils when the adults leave the thorn.
These latter holes are usually on the sides of the thorn, or near the base. In
the population of A. sphaerocephala growing along the dunes south of
Veracruz, Crematogaster is the usual occupant of the plant. Crematogaster
is most easily distinguished from P. ferruginea by the former’s small size
(2-4 mm long), petiole attached to the dorsal side of the gaster, heart-
shaped gaster, and slow rate of movement. Camponotus planatus Roger is
occasionally found living in thorns on unoccupied, and more rarely oc-
cupied, shoots; it cuts its own entrance holes and/or uses those made by

392 Tue UNiversiry SciENCE BULLETIN

Fic. 28. A brood of nymphs of Umbonia orozimbo on an unoccupied 110 cm tall shoot of
Acacia cornigera in subplot N-l. The adult female is out of sight above the brood. When
this stem was placed on an occupied shoot, the workers of Pseudomyrmex ferruginea chased the
membracids from the shoot. Photo November 1963.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 2393

Fic. 29. Thorn and shoot tip damage by the larva of Halisodota probably H. pura. This
was the terminal node of an unoccupied Acacia cornigera in subplot J-1 in November 1963. The
shoot was 20 cm tall. Mandibular scars can be seen on both sides of the thorn. The feeding
damage of tettigoniid grasshoppers 1s very similar to this,

394 Tue Universiry SciENCE BULLETIN

P. ferruginea. These entrances are larger and rounder than those of P.
ferruginea. C. planatus is a short, broad ant (4-6 mm long) with a spherical
gaster and one petiolar segment (P. ferruginea has two petiolar segments).
It has a reddish-brown head and/or thorax, and a black gaster.

Distripution. The geographic distribution of P. ferruginea in the study
area is congruent with the combined ranges of A. cornigera, A. sphaeroce-
phala, and A. chiapensis (Fig. 6). The ecological distribution of P. fer-
ruginea is discussed in the later section entitled ENVIRONMENTAL
REQUIREMENTS. Aside from seedlings or suckers less than a year old,
single unoccupied shoots of A. cornigera and A. sphaerocephala are only
occasionally encountered. No local populations of these two swollen-thorn
acacias are known or reported without colonies of P. ferruginea, except for
a band of A. sphaerocephala about 30-3000 meters wide immediately ad-
jacent to the hightide line on the ocean beach from a point about 3 miles
south of the city of Veracruz, south to Anton Lazardo. Shoots of A.
chiapensts frequently lack a colonoy of P. ferruginea.

P. ferruginea was not collected thorughout the study area, at every site
where it was observed. However, enough samples were taken to assure that
its distribution shows continuity and discontinuity in the same manner as
that of the swollen-thorn acacias. The geographic range of P. ferruginea,
like that of A. cornigera, has apparently not been greatly extended by man’s
disturbance of the vegetation. However, prior to human destruction of most
natural plant communities in the study area, the distributional pattern of
P. ferruginea was the network of arroyos, rivers, and ocean beaches. Man’s
activities have filled the lacunae in this network with many acres of highly
suitable habitats for A. cornigera, A. sphaerocephala, and P. ferruginea. Now
these two acacias and their ant have a much more continuous distribution.

Morreuotocy oF Inpivipuats. Morphology of immature stages. When
foliar nectar and Beltian bodies are available, the large white cylindrical eggs
are laid continually by the single colony queen. They are only very slightly
sticky, and clusters of them are easily shaken apart.

Numerous pseudomyrmecine larvae are described by Wheeler and
Bailey (1920) and Wheeler and Wheeler (1956). Pseudomyrmex ferruginea
is treated in both papers under the name Pseudomyrmex belti subsp. ful-
vescens. As is the case with most pseudomyrmecine larvae, those of P. fer-
ruginea are elongate with well developed dorsal hooked hairs. These hairs
serve as anchors to hold the larva in place on the slanting inner walls of the
hollow thorns, and on other larvae. The larvae range from 0.7 mm (first
instar larvae) to 6.7 mm (larvae that will produce alates) in length. Well
fed larvae are nearly circular in cross section, but under starvation conditions
a deep longitudinal ventral depression develops; at this time the larva be-
comes shorter and much of the visible white fat body disappears. The inactive

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 395

prepupa is easily distinguishable from the larva by the former’s lack of
dark gut contents. The pupae are not enclosed in a cocoon. Development
from newly laid egg to freshly emerged worker takes about 35 days during
the dry season and rainy season. Since the larvae can live at least a month
without food, shortage of food undoubtedly lengthens this time of develop-
ment.

Morphology of worker. Form. P. ferruginea is a representative, medium
sized pseudomyrmecine. The worker has short legs and an elongate body
that is held low to the substrate. The large convex compound eye covers
about one-half of the side of the head. The ocelli are well developed. The
head is wider than the thorax and bears short mandibles with 3-4 teeth. The
two petiolar segments are sufficiently flexible to allow the sting to be
brought to bear on any point touching the worker. The small, tear-drop
shaped gaster is carried straight out to the rear.

Dimensions. While there are distinct behavioral roles within the colony,
the workers display monophasic allometry in respect to head width con-
trasted with thorax width (terminology of Wilson, 1953). From Figure 30
it can be seen that there may be considerable variation in thorax width for
a given head width but there is no indication in any of the colonies examined
of polyphasic allometry or dimorphism among the workers.

There are obvious size differences among the workers of colonies of all
ages. Workers range from 3.5-7.0 mm in length. The workers produced

1.000
0.983

0.966

HEAD WIDTH ACROSS EYES (mm)

0.950
2DSOh BISGs" .2O%) ~WOTa 9.555) 1.592, 000 608" .ol7

MAXIMUM THORAX WIDTH (mm)

Fic. 30. Head width across the eyes plotted against the maximum thorax width of 50
workers of Pseudomyrmex ferruginea from a one year old colony. There is no indication of
polymorphism or strong allometry in respect to these two measurements. The relationship is
likewise not linear. Sample from a 2,000 worker colony from Acacia cornigera 1 km west of
Temascal.

396 Tue Universiry SciENCE BULLETIN

by a colony living in a young shoot become progressively larger until the
colony is about two years old. This change may be in part due to the in-
creasing amount of food that is generally available to the colony as it grows.
It is also associated with the increasing efficiency of older workers to harvest
food. In figure 31, the head widths of workers (A: X=1.002 mm, s.d.=0.0026)
from a relatively young colony (345 workers) and the head widths of 345
workers (B: X=1.084 mm, s.d=0.035) from a three year old colony are plotted
(about 5,000 workers). Measurements of other colonies indicate that these
two histograms may be regarded as two successive stages in the development
of worker size in a colony, with (B) being close to the maximum. The work-
ers representative of the size and age class in the first histogram (A) have
probably all died in the colony from which the second sample was drawn
(B). However, this latter colony still is producing an occasional worker
in the size class represented by histogram (A) in Figure 31. The significance
of this to an occupied shoot is that not only do older colonies have more
aggressive workers, but they have larger workers as well.

Variation in color. Throughout the study area, the gaster of P. ferruginea
varies from reddish-yellow to black, but teneral workers require about three
weeks to be completely darkened. Other body parts range from yellow to
dark brown. In the Temascal area, the integument of the teneral worker
is pale yellow to beige at the places where it will be reddish-brown when
fully tanned. It is purple-gray where it will be black. The two most common
color phases are some shade of light reddish-brown with slight or no darken-
ing of the gaster, and some shade of reddish-brown with a very dark brown
gaster. These are found throughout the study area. All intergrades of these
two patterns may occur in one colony but usually the colony conforms to
one color phase or the other. Colonies with very dark workers are found
scattered throughout the study area (most commonly around Veracruz and
Tampico). In the Temascal area there were several colonies with almost
yellow workers. By far the commonest color phase is that with the dark
gaster. There is no evidence that these color patterns represent sibling species.

Temporary variation in form. When collecting nectar, a worker visits
many successive foliar nectaries. The gaster sometimes enlarges as much as
three times the usual volume and the sclerites become separated like those
of a physogastric queen. When a worker is confined without fluids for a
period of two or more days, the volume of the gaster drops to about one-half
that usually observed.

Morphology of alates. Female. The alate female does not differ from
the worker in form except in the longer head, two times larger ocelli,
greater body size, and thorax modified for flying. While larger than the
workers, alate females are much less variable in size. They range from 7.7-8.2

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 397

180

160

140

120

100

20

NUMBER OF WORKERS

ie)
0.93 0.97 1.01 1.05 1.09 1.13 1.17

HEAD WIDTH ACROSS EYES (mm)

Fic. 31. Frequency distribution histogram A is the head widths across the eyes of all the
members of a 345 worker colony of P. ferruginea. This colony was about one year old.
Histogram B is the head widths across the eyes of a sample of 345 workers from a colony
about three years old with about 5,000 workers. Both colonies were collected from <Acacta
cornigera about 7 km east of Temascal.

398 THe University SciENCE BULLETIN

mm in total length (sum of mandibles, head, thorax, petiole, postpetiole, and
gaster).

The color changes that take place as the teneral worker ages are the
same as those described for the worker. The alate females have the same
color phases that the workers do. Usually the color phase of the queen of a
colony matches that of the workers and alate females in the colony, but on
two occasions a queen with a pale gaster was found with workers with dark
gasters.

When a founding queen is gathering nectar from the foliar nectaries, her
gaster increases to about twice the usual volume. The gaster of a fully
physogastric queen may be as large as 7 mm long and 3-4 mm wide. In this
case, the sclerites are about 0.75 mm apart. The normal dimensions of a
founding queen’s gaster are about 25 mm long by 1.5 mm wide. Under
starvation conditions, the colony queen’s gaster returns to nearly normal size
within two weeks.

Male. The males are representative, medium sized pseudomyrmecines
(7.9-8.2 mm long, including mandibles). Like the alate females, they vary
little in size. The compound eye is very large and protruberant, covering
nearly one-half of the side of the head. The ocelli are very prominent (0.15
mm in diameter); this development is very likely associated with the mating
flight which begins before dawn.

Cotony MorpHo.ocy anp DeveLopMENT. Structure of colonies in single
shoots. A Pseudomyrmex ferruginea colony is initiated when a dealate queen
opens an acacia thorn for her own use, or occupies a thorn opened by an
earlier worker or founding queen. This occurs at all times of the year. From
this single thorn the colony grows to occupy, either by use for brood or by
patrolling, all of the thorns on the shoot. Under the proper circumstances,
the colony also spreads to other shoots of swollen-thorn acacias. Thus, the
mature colony lives in and on a series of dispersed containers (thorns) with
exposed routes connecting them (branches of the shoot or trails between
shoots). Since a colony confined to one shoot has a different structure from
one living in several shoots, the former condition is discussed separately from
the latter. The variation in colony structure is primarily a function of shoot
age and form, colony age and history, and time of year.

Location of the queen. The shoot in which the founding queen settles
may have from one to at least 2,400 thorns. In younger shoots (1-200 thorns)
she is alone or ‘shares the shoot with as many as 34 other founding queens
that have colonies in various stages of development. In a shoot that lacks an
active worker force, the founding queens may be in almost any thorn with
at least one side over 15 mm long. On 23 Jul. 1964 the 90 thorns on 29 shoots
in subplot J-1 contained 14 founding queens without any brood, 23 founding
queens with eggs only, 28 founding queens with eggs and larvae, 10 founding

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 399

queens with eggs, larvae and pupae, and one founding queen with eggs,
larvae, pupae and three workers. There were 2 thorns with more than one
founding queen and each of these thorns had two entrance holes. All thorns
were under 90 days old. Of the 16 remaining thorns, 5 had no entrance, 3
were damaged or rotten, 2 had dead queens, 3 were empty, 1 had a salticid
spider web, and 1 harbored a colony of a small yellow Solenopsis. The maxi-
mum age of these thorns is known because all thorns were clipped from
these shoots on 25 Apr. 1964. The shoots had a mean height of 28 cm (s.d.=9
cm) and were among herbaceous and woody sucker regeneration in a fallow

corn field.

In large unoccupied shoots having over 200 thorns, founding queens tend
to be in the type A thorns on lateral branches. Damaged or old rotting thorns
are usually only tenanted by queens that are hiding and usually when all
other thorns are tenanted. There is no indication that thorn size influences
the probability of the founding queen, becoming the queen whose colony
eventually controls the entire shoot, and destroys the other incipient colonies.

Isolated sucker shoots over a year old and seedlings over two years old
(often over 100 thorns) usually contain a queen-unit. The consequence is a
reduction, often to zero, in the number of founding queens present. In such
a shoot, the few founding queens present are found in rotting thorns, or
thorns very marginal to the general area of thorns, leaves and growing points.
The number of founding queens in other parts of a shoot reflect the size,
age and type of unit, i.e, whether a queen-unit or auxiliary-unit.

During the latter part of the dry season (Mar.-May) and during the cool
season (Dec.-Feb.) there is a reduction in the number of searching queens
in the field with a consequent reduction in the number of founding queens
in the thorns. Even so, on the average there is still more than one founding
queen per regenerating shoot. This reduced number could increase produc-
tion of incipient colonies since it could lower competition among founding
queens for thorns, Beltian bodies and nectar. On the other hand, there is
high mortality among the small colonies that have only eggs and larvae. If
this is due to some inherent physiological cause, such that it is necessary to
have many queens starting to get a few maturing colonies, then the periods
following the cold and dry weather might be expected to show a lowering
in production of mature colonies. In fact, the rate of colony production at
this time is not lowered.

Until there are 15-20 workers, the queen stays in the same thorn with all
of her brood and most of the workers. However, she may move with all
of her colony members to a new thorn. On 27 May 1964, 21 sucker shoots
with a mean height of 33 cm (s.d=23 cm) were dissected. The surrounding
vegetation was 100-150 cm tall and was completely burned on 1 Dec. 1963.
On 13 Jan. 1964, only 66 of the 142 thorns on these new shoots had founding

400 Tue University ScIENCE BULLETIN

queens in them. During September, there would have been greater than 95
percent occupation by founding queens. In May, the dissected shoots had
210 thorns with no entrance holes; 83 empty thorns with no founding queen;
23 thorns, each with a founding queen and eggs (138 eggs); 10 thorns, each
with a founding queen, eggs, and larvae; 3 thorns, each with a founding
queen, eggs, larvae, and pupae; 55 thorns, each with a founding queen, eggs,
larvae, pupae and workers( mean number of workers per thorn was 3.7,
s.d.=3.1); and 2 thorns occupied by one small expanding colony (12 workers).
The largest colony had 16 workers and had a maximum age of 163 days.

The founding queen can not stay in the same thorn throughout the life
of the colony. By the time a queen has produced 10-40 workers, her gastric
sclerites begin to separate. By the time she has 100-300 workers she is usually
completely physogastric. At some undetermined time she moves into one
of the largest thorns on the shoot, usually an undamaged, 1-6 months old
type B thorn. This thorn is roughly in the center of the present year’s thorn
and leaf mass. As the location of the thorn mass moves upward with the
growth of the tree, she changes thorns at least 1-3 times per year. She there-
fore remains roughly in the center of the thorn mass in one of the largest
thorns.

Placement of brood. In the early stages of colony development, most of
the brood is located in the same thorn as the queen. Founding queens are
usually accompanied by 4-15 eggs before the first larvae are present. These
eggs are usually scattered separately in the lowest part of the thorn. During
the first 2-9 months of shoot regeneration in an area completely cleared of
old colonies, queens with only eggs, or with eggs and a few larvae, are most
commonly encountered. As more members are added to the colony, it re-
mains in the original thorn until it is packed with brood and adults. Egg
production remains low at this time; this first thorn usually contains less
than 50 eggs.

As the worker force grows, control of new thorns is gained (e.g. 1 thorn,
11 workers; 2 thorns, 19 workers; 5 thorns, 55 workers; 13 thorns, 125
workers). As the number of workers in the colony increases, the percentage
of thorns used for brood decreases to a level of about 50 percent. This per-
centage is highest during periods of shoot growth activity which is highest
in the rainy season and lowest toward the end of the dry season. At the end
of the dry season, entire trees may be abandoned. This low use of thorn
space is in part associated with the unsuitable nature of some thorns for the
brood. It is also associated with a lack of food. About 50 percent of the
leaves produced are subtended by swollen thorns. If the Beltian bodies
produced by the four leaves associated with 2 swollen thorns are only sufh-
cient to support the larvae of one thorn until four more leaves are produced,
then only about 50 percent of the thorns available to the colony can contain

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 401

brood (based on the fact that only 50 percent of the leaves have swollen
thorns). It is probable that the number of Beltian bodies produced by two
leaves is not sufficient to nourish a thorn full of brood to the adult stage.
Even when there is a surplus of Beltian bodies the colony is not able to
grow fast enough to use all of the old thorns and the new thorns that are
produced with the new leaves. On this basis, a colony provided with the
products of a single shoot could never fill all the thorns of one year with
brood, especially since it has last year’s thorns to fill as well.

The number of larvae in a brood thorn is variable. The upper limit of
brood dispersion among the brood-thorns is probably a function of worker
efficiency. On any given shoot, the number and volume of brood in a thorn
is only roughly proportional to the size of the thorn. As the proportion of
later instar larvae in a thorn increases, the number of larvae in the thorn
decreases.

When the colony is first expanding, the worker force occupies those thorns
that it can and uses them for brood. These are often, but not necessarily,
adjacent to the thorn occupied by the queen. Relatively soon, when the
colony has 4-8 thorns and 40-100 workers, new green thorns are occupied
as they appear. As control of the entire shoot is achieved by destruction of
competing colonies, the brood-thorns are chosen so that the brood is con-
centrated roughly in the center of the mass of thorns and the degree of ex-
pansion from this center is a function of the amount of brood. Colonies five
or more years old often have the brood-thorn area fragmented into clusters
that are spread throughout the canopy and are located closer to the major
areas of new leaf production than to the center of the canopy.

The area of thorns with brood is effected by the dynamics of A. cornt-
gera. As the area of 1-2 year old thorns moves upward with growth of the
shoot, so does the area of brood thorns. This area roughly conforms in shape
to the shape of the canopy of the shoot. This is one way in which the colony
structure can be influenced by the surrounding vegetation. Slender shoots
have long, thin brood areas, while spreading shoots have laterally extended
brood areas.

Within the brood area there are occasional empty thorns. These often
are green, damaged (thorn cavity opened), or strongly distorted (only one
side of the thorn swollen). These thorns rarely lack entrance holes. The
number of empty thorns in the brood area increases and the brood area be-
comes smaller as 1) the colony becomes senescent in old shoots, 2) the
growth rate of the colony decreases due to lack of food during the cool and
dry season, or 3) the number of thorns produced increases without conco-
mitant colony size increase. The number of thorns available to the colony
increases during the first 3-4 years of colony growth and then decreases.

402 Tue UNiversiry ScIENCE BULLETIN

The number of immature individuals in a colony increases from year to
year. In a shoot old enough to react to the cool or dry season by reduction
in growth rate, the rate of increase in larvae is highest during the first six
months of the rainy season and slows to zero only with cessation of all new
leaf production during the cool or dry season. Except when starving, the
colony always has eggs present. The rate of egg production approaches zero
when the queen moves to a new thorn but is re-established immediately
following her settlement in a new thorn. With the exception of eggs, there
is no segregation of the immature forms between thorns. There is a sug-
gestion that the new first instar larvae (or eggs) are brought to the brood
thorn in even-aged groups but during their growth are reshuffled through
the thorns and unevenly fed. This results in the thorough mixture of instars
and sizes found in most brood thorns. Green to tan thorns that are less than
two weeks old are often found with all larval ages, sexes, and castes in them.

The eggs are partially segregated from other brood. Except when the
physogastric queen has recently moved into a new thorn, she is accompanied
by 400-800 eggs, 0-70 first instar larvae, and 0-10 later instar larvae. Pupae
are normally absent from the thorn with the queen in it. This quantity of
brood occupies about 1.2 cc or 60 percent of the volume of the thorn with
the queen in it. The eggs are not clustered but adhere very lightly to each
other and can be shaken apart. Other eggs in the colony are found in three
places. Small groups of 5-40 are occasionally found in brood thorns with
first and later instar larvae. These eggs apparently represent a new cohort
ready to begin growth in the thorn. Commonly the thorn next to that of the
queen is as full of eggs as is the one with the queen. These eggs are occasion-
ally accompanied by small numbers of larvae of all instars and pupae. Oc-
casionally a large thorn full of eggs is found 20-60 cm below that occupied by
the queen. This thorn is in various stages of being filled with the normal
large numbers of larvae and pupae. It has an enlarged entrance and is the
thorn previously occupied by the queen.

Alate brood appears during the second year of colony growth and is mixed
with worker brood. Brood of both sexes may be found in the same thorn.
The larvae of the first alate forms appear in the margins of the brood-thorn
area (type A thorns near the ends of lateral branches). During the third
year of colony growth, alates are found in progressively higher numbers and
closer to the central area of brood thorns. After the third year, alate last
instar larvae and pupae may be found in almost any brood-thorn except the
queen-thorns. Alates are often the last adults to be produced by a dying
colony or fraction of a colony. Contrary to the usual case with temperate
ants, alate brood is present all year.

Distribution of the worker force. Worker numbers increase throughout
the year, if Beltian bodies and nectar are available and diurnal temperatures

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 403

stay above the 24-27° C. level. Increase in worker numbers during the cool
and dry season is probably insignificant in comparison to that when large
numbers of Beltian bodies are available during the rainy season. The leaves
of the flowering branches that are produced during the later part of the dry
season bear enough Beltian bodies to cause a small increase in worker num-
bers. A colony occupying sucker regeneration less. than six. months old is
likely to have sufficient food for growth because shoots of this age from large
stumps continue to grow during the dry season. The size of a colony is more
strongly associated with the number of growing seasons passed than its age
in months.

In determining the number and distribution of workers in a colony, very
young colonies may present some difficulty. Usually a shoot with numerous
small colonies has one larger colony occupying several thorns and several
small colonies occupying only one thorn each. A representative shoot might
have 16 thorns: three empty, five with founding queens and only eggs, five
with founding queens with eggs, larvae and one to five workers, and three
that are packed with brood and workers. Two of the last three thorns would
lack a queen and one would have a partly physogastric queen. In this ex-
ample, the six colonies would have 1, 2, 2, 2, 5, and 31 workers, respectively.
Almost all of the workers found outside of the thorns would belong to the
last colony.

A shoot with a colony of 200-400 workers, or more, is cleaned of other
small colonies by the worker force; counting the workers is therefore no
problem. From these low levels, the colony may increase to as high as
12,269 workers in 930 days. This colony was an exceptionally rapidly grow-
ing colony in that its shoot appeared to have continued to grow throughout
each dry season and therefore the colony grew as well. A more representative
colony of this age would have 4000-8000 workers. The actual number of
workers produced by the large colony during the 930 days undoubtedly
was much higher. A correlation between shoot size or age, and number of
workers is only found where a colony occupies a single shoot. The oppor-
tunity for the colony to disperse and occupy shoots with different histories
destroys this correlation.

In using the age of a sucker shoot to determine the possible age of the
colony, care must be taken to use shoots that are growing in areas that were
burned as well as cut. The following examples of carefully aged colonies
will give some idea of the growth rate of colonies of P. ferruginea. (1) In
May 1963, a cornfield that had lain fallow for one year was burned. On 26
Jan. 1964, the 14 tallest shoots in a one acre section of this field had a mean
height of 90 cm (s.d=32 cm). The mean number of workers per shoot was
146 (s.d=117). However, 12 of the 14 shoots had more than one young colony
with workers in it. These 12 shoots had 101 founding queens and over

4()4 Tue University Science BULLETIN

one-half of the queens had workers in their thorns. This pasture was heavily
grazed until Sept. 1963. (2) In Dec. 1962, a brushy pasture along the south
margin of plot C was cleared and burned. In Oct. 1963 the ten tallest shoots
in a one acre section in the center of this pasture had a mean height of 111
cm (s.d.=14 cm). Each shoot contained a single large colony and 0-6 found-
ing queens. None of the founding queens had anything other than eggs
in their thorns. The mean number of workers in the ten maturing colonies
was 306 (s.d.=1.27). This pasture was lightly grazed during the entire period
of colony growth. (3) In Jan. 1962, a brushy pasture along the south side of
plots L and M was cleared and burned. In March 1964, nine shoots were
chosen so as to be so far from other shoots that they would lack auxiliary
units. They had a mean height of 294 cm (s.d.=37 cm) and each shoot con-
tained a single colony with no founding queens. Each colony queen was
fully physogastric. The mean number of workers per colony was 1149
(s.d.=297).

Workers are not found outside the thorns unless they are collecting
Beltian bodies or nectar, until the colony has 50-100 workers and patrolling
begins. Until control is gained over most of the thorns, so that the brood
can be distributed and need not be packed in the thorns, the number of
workers per thorn with brood remains high (20-40). By the time the colony
is 1-2 years old, the number of workers per brood thorn has dropped to 4-20.
Thorns without brood but within the patrolling range of the workers con-
tain 0-5 workers. Even under extreme disturbance conditions it is difficult
to induce more than 50 percent of the worker force to leave the thorns.

When undisturbed, workers are distributed somewhat evenly over the
external shoot surface with the highest densities on the growing shoot tips
and in the area of brood concentration. This density is known to rise and
fall with the weather, time of day, time of year, disturbance, colony size,
colony age, colony activity, the individual colony, and unknown factors. Less
than five percent of the worker force is off of the shoot and on the ground
or surrounding vegetation.

Some of the aspects of normal changes in the density of workers outside
of the thorns of undisturbed shoots are demonstrated by the following study.
Beginning on 1 Apr. and on 8 June, the numbers of workers active on the
surface of the same 17 shoots along the margin of plot N-2 were counted at
hourly or somewhat greater intervals over a 24 hour period (Fig. 32 and 33).
The mean height of the shoots on 1 Apr. was 158 cm (s.d=68 cm) and on
8 Jun. was 240 cm (s.d.=61 cm). Of the 17 shoots, 9 were queen-shoots and
8 were auxiliary-shoots. From both recordings, it can be seen that 1) there
is a sudden rise in activity around the time of sunrise and sunset, 2) the
absolute numbers of workers on the shoot is not a function of the temperature
at the moment of counting, and 3) the numbers of workers per shoot is much

INTERACTION OF THE BULL’s-HoRN ACACIA WITH AN ANT INHABITANT 405

5000

4000

a
WwW
wn
4
3000 a
| a
‘
‘
‘
2000

w
Ww
H
a
=z
=)
wn

NUMBERS OF WORKERS OUTSIDE OF THE THORNS OF 14 A. CORNIGERA (A)
TEMPERATURE (degrees C)

0800 1100 1400 1700 2000 2300 0200 0500 0800 1100
TIME (hours)

Fic. 32. Curve A represents the total number of workers of Pseudomyrmex ferruginea
active on the surface of 14 occupied shoots of Acacia cornigera from 8:30 a.m., 1 April to
8:10 am. 2 April 1964 (dry season). The mean height of the shoots was 158 em. A rather
sudden increase in activity near sunrise and sunset, and distincitly lower activity at night,
are evident from the curve. Curve B represents the air temperature 1 m above the ground
throughout the period of observation.

5000

a@— SUNSET

@— SUNRISE

4000 34

NUMBERS OF WORKERS OUTSIDE OF THE THORNS OF 14 A. CORNIGERA (A)

3000 30
=

oO

2000 A
a

o

+s

Lou)

ao

vw

1000 22 &
=)

B

-

x

ty

a

=

Ww

o 1g &

0800 1100 1400 1700 2000 2300 0200 0500 0800 1100

TIME (hours)

Fic. 33. Curve A is as Curve A in Figure 32 except that the counts were made on 8 June
to 9 June 1964 (rainy season). The same shoots were examined. On this date their mean
height was 240 cm. The same peaks at sunrise and sunset are evident but nocturnal activity
was not lowered as it was during the dry season. Curve B is the air temperature.

406 Tue University ScteENcE BULLETIN

higher during the rainy season than during the dry season. During the dry
season, the early morning peak in activity is correlated with a peak in the
flow of nectar from foliar nectaries. During the rainy season it is not clearly
associated with any aspect of the bionomics of the tree. The reason for the
increase in numbers at sunset is not clear on a direct cause-and-effect basis.
Indirectly, it may be associated with the fact that just after sunset a large
number of phytophagous insects become active. While in general the num-
bers of workers active on the surface is influenced by the temperature, within
the range of 18-43° C. there are factors both internal and external to the
colony which can cause substantial increases and decreases in the number
of workers active outside of the thorn. Below about 18° C. the workers have
trouble walking without falling off the shoot. Above 43° C. they seek the
coolest part of the shoot and do not behave normally. The total number of
workers in each shoot probably did not increase more than ten percent be-
tween | Apr. and 8 Jun. 1964. However, on 8 Jun., there were 7-27 undam-
aged tips on each shoot whereas on 1 Apr. there were 0-3.

Irrespective of other influences, it has been noted throughout this study
that the highest numbers of workers are present on the surface of the shoot
at the time when the shoot has the largest numbers of undamaged shoot tips.
While many of these workers are in the vicinity of the undamaged shoot tips,
the number of workers on other surfaces of the shoot also increases. This
system appears to be self perpetuating because the more workers on the
outside of the shoot, the higher the probability that a given shoot tip will
remain undamaged.

Distribution of the alates. The first alates appear in the colony 15-24
months after the first workers. The colony contains 1000-1500 workers at
this time but old senescent colonies with as few as 200-500 workers often have
alates. Since alates leave the shoot daily, they do not build up to large concen-
trations except when cold weather prevents mating flights. A colony may
contain all males, all females, or both. They do not segregate themselves nor
are they segregated by the workers. Their exit from the colony does not ap-
pear to be influenced by the workers. During the cool season and the dry
season, the numbers of new founding queens are lowered (i.e., fewer queens
in the mating flights) and brood is absent from many shoots that contained
only part of a colony (auxiliary units, which produce the majority of alates,
are reduced in number and size). Following a month of cold weather which
prevented mating flights, a colony with 12,269 workers had 3,421 alate females
and 1,311 males on 20 Jan. 1964. Assuming an accumulation in the colony
for the 30 days, this indicates a daily production rate of 0.0093 females and
0.0035 males per worker. This is probably an underestimate since there may
have been an occasional morning that was warm enough for alates to leave.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 407

Colonies occupying more than one shoot. When a colony in one shoot
expands to occupy a neighboring shoot, it establishes a unit of brood and
workers (auxiliary-unit) with much the same behavior and morphology of a
colony in a single shoot (queen-unit) that lost its queen 1-10 weeks before.
Instead of gradually dying as does a colony without its queen, the auxiliary-
unit is maintained for the life of the colony except under certain weather con-
ditions when the shoot looses nearly all of its leaves for two weeks or more.
A connection is not maintained continuously between the queen-shoot and
auxiliary-shoot.

The fluctuations from a status very similar to that of a queen-unit to that
of an almost dead queenless unit depend on frequency of re-establishment of
the trails between the queen-shoot and the auxiliary-shoot. This frequency is
a function of 1) the amount of food being produced by the queen-shoot and
auxiliary-shoots, 2) the size of the queen-unit, and 3) the distance between
the two shoots. The number of auxiliary-units maintained by a queen-unit
is likewise a function of these three variables. When the branches of two
shoots are in contact, the two are occupied by the same colony if it is large
enough to occupy them. If one of the two shoots contains the queen, the dif-
ference in colony structure between the two shoots is a function of colony
size; if a very large colony, it treats the two shoots as if they were just two
large branches from the same stump.

The maximum number of auxiliary-shoots is maintained during the rainy
season. For example, on Jul. 31, 1964, the 108 emergent and occupied shoots
in plot N-2 were occupied by 14 colonies. In the subplot, the number of
auxiliary-shoots used by each colony was 27, 14, 14, 10, 9, 5,5, 4, 2, 2, 2, 1, 1,
and 1. This record does not include auxiliary-shoots outside of the subplot.
The auxiliary-shoots of the colony with 27 were all in branch to branch con-
tact. This colony was highly dispersed, and worker numbers were low on
each shoot. The colony with 10 auxiliary-shoots was the largest in the subplot
(queen-shoot had at least 10,000 workers) but its auxiliary-shoots were 1-6 m
from the queen-shoot. In this case, worker numbers were very high on each
auxiliary-shoot. During the dry season, the 14 colonies only had 63 auxiliary-
shoots in contrast to the 98 during the rainy season.

Presence of a queen. A colony does not have more than one queen; the
auxiliary shoots do not have queens. Founding queens are more commonly
found in auxiliary-shoots than in queen-shoots. When present in a queen-
shoot, they are in the thorns most marginal to the shoot canopy. In the auxil-
iary-shoot, they are occasionally found in thorns more centrally located.
Founding-queens on occupied shoots are removed when found by the work-
ers of the established colony. As a result, the number of founding-queens
present inversely reflects the size of the worker force in relation to the area of
the shoot that has to be patrolled. It is one of these founding-queens that estab-

408 Tue Universiry SciENcE BULLETIN

lishes a new colony should the auxiliary-shoot be abandoned during the dry
season by the established colony.

Abandonment of auxiliary-shoots occurs during the dry or cool season
when the production of Beltian bodies falls very low. It also occurs when the
queen-unit is destroyed. Abandoned auxiliary-shoots may be completely
empty and in this case they should perhaps be termed “vacated” since the
entire worker force and brood has moved to another shoot. More commonly,
the trail between the queen-shoot and auxiliary-shoot is not re-established.
This leads to gradual die-out of the auxiliary-unit (2-10 weeks); in this case,
the auxiliary-shoot usually contains a few thorns with moldy brood and an
occasional worker. The brood consists of alate pupae, and may be scattered
through the thorns. If an abandoned auxiliary-shoot is invaded by another
colony, workers coming from the original colony at a later time are killed by
the workers of the new resident auxiliary-unit. In the laboratory, a group of
several hundred workers and brood from a large colony, were isolated with-
out their queen for several weeks. They killed 14 different founding-queens
offered to them, but accepted a fully physogastric queen from another colony.

Placement of the brood. The fraction of the total inside volume of the
thorns on the shoot used for brood in auxiliary-shoots is a function of the
same three factors that influence the numbers of auxiliary-shoots maintained
by the colony. In general, there is less brood in each auxiliary-unit than in the
queen-unit, but the total in all of the auxiliary-units of a single colony may be
larger than that in the queen-unit. The area of the shoot canopy that contains
brood-thorns is usually interjected with empty but apparently acceptable
thorns. There are higher numbers of alate larvae and pupae, and a lower
number of first instar larvae in the brood-thorns of auxiliary-shoots than in
queen-shoots. There is less volume of larvae per unit volume of thorn used
for brood than in queen-shoots. Auxiliary-units show the same reactions to
stress by shading and cutting of the shoot as do queen-units. Shaded auxil-
iary-units may subsist on products from an isolated queen-shoot or auxiliary-
shoot.

Distribution of the worker force. The number of workers in an auxiliary-
unit is often much lower than that in its queen-unit, but the sum of the work-
ers in all auxiliary-units maintained by the colony may equal or surpass that
of the queen-unit. On 29 Jul. 1964, in subplot N-2, a colony with 9 auxiliary-
shoots (heights: 332, 226, 220, 180, 138, 126, 90, and 85 cm) and a 300 cm tall
queen-shoot had 19,452 workers. The queen-unit contained 7,640 of these
workers, and the auxiliary-units contained, by order of shoot height, tallest to
shortest, 3,210, 2,000, 1,576, 2,624, 1,262, 207, 412, and 321 workers, respec-
tively. Alates were present in all shoots. The four tallest auxiliary-shoots and
the queen-shoot had branches in contact at the time (29 Jul.), but during the
previous dry season, all of the shoots occupied by this colony were connected
only by trails at ground level. It is impossible to determine the age of this

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 409

colony because even though the subplot was burned in 1961, the colony was
on the margin of the burn and moved into the plot following a more recent
cutting of the neighboring pasture.

The number of workers in an auxiliary-unit is a function of the same
three variables that determine how closely it matches the queen-unit in mor-
phology. Since the connection between the queen-unit and auxiliary-unit is
not continuous, or rhythmic, either within the day or for longer periods, the
number of workers in the auxiliary-unit is likely to fluctuate much more than
in the queen-unit. The highest numbers are found in auxiliary-shoots that
have direct branch connections with the queen-shoot. The number of work-
ers per auxiliary-shoot brood-thorn does not vary from that of a queen-shoot,
as long as the trail between the shoots is re-established often enough to keep
the auxiliary-shoot worker force from being severely depleted by mortality
agents. Associated with the reduced quantity of workers in the auxiliary-
unit, the number of workers active outside of the thorns is often lower than
that of the queen-shoot under similar environmental circumstances; the work-
ers are distributed on the surface of the auxiliary-shoot in the same patterns as
on the queen-shoot.

Distribution of other animals in and on shoots. In queen-shoots and aux-
iliary-shoots, thorns tenanted by animals besides P. ferruginea are usually
found in the same positions as are those occupied by founding-queens, i.¢., in
thorns in areas infrequently patrolled. Prolepismina sp. (La Granja, Co-
taxtla), the thrips Diceratothrips horribilis Priesner (Cotaxtla), and salticid
spiders (all sites) are commonly found in dry empty thorns with entrance
holes. A small black bee (Hylaeus sp.) nests in empty dry thorns on unoccu-
pied shoots in the Temascal area. Microlepidoptera larvae are occasionally
found in the green or green-yellow thorns of queen-shoots. These larvae are
usually destroyed before maturity. In auxiliary-shoots they more frequently
complete development. The unoccupied shoots contain most such larvae.

Colonies of other species of ants are not found in queen-shoots having
enough workers to patrol the shoot. However, auxiliary-units of P. gracilis
mexicana (1-5 thorns) occasionally are found in the same shoot with a young
colony or weak auxiliary-unit of P. ferruginea. As the P. ferruginea unit
gains in worker strength, the P. gracilis mexicana unit is removed by the P.
ferruginea workers. Other species of ants are removed in the same manner.

Occasional colonies of other Pseudomyrmex, Cryptocerus, Camponotus,
Solenopsis, and more frequently Crematogaster, are found in shoots nearly
or completely lacking a worker force of P. ferruginea. These other ants usu-
ally live in the dried thorns. They use the entrance holes of P. ferruginea
but may modify their shape. Crematogaster commonly uses the exit holes of
microlepidoptera. P. nigrocincta is treated by P. ferruginea as if it were an-
other P. ferruginea colony and therefore the two species do not co-inhabit the

410 Tue Universiry ScrENcE BULLETIN

same shoot except as founding queens and very young colonies. In respect
to colony morphology, the interaction of P. nigrocincta with A. cornigera is
indistinguishable from that of P. ferruginea in the study area. Solenopsis
geminata tends Dysmicoccus sp. (neobrevipes complex) under earthworks at
ground level or in the exposed canopy of A. cornigera but does not keep
brood in the thorns. It also harvests Beltian bodies and nectar.

A few insects live a “persecuted” existence on the shoot; they are con-
stantly being attacked. The zygopine weevil (Lechriopini near Lechriops)
appears to be a mimic in color and behavior of P. ferraginea. It occasionally
is encountered running on the surface of queen-shoots but is more common
on auxiliary-shoots and shoots lacking a worker force of P. ferruginea. Its
larvae live in the thorns and feed on resins that are produced in reaction to
feeding damage to the strip of live tissue in the base of the thorn. The otitid
fly (Euxesta sp.) and a salticid spider (Metaphidippus maxillosus) are mim-
ics of P. fewuginea both in behavior and appearance. They are commonly
found running among the ants on the surface of queen-shoots and auxiliary-
shoots. The nests of polybiine vespid wasps are occasionally found on occu-
pied shoots. The nests of Polybia occidentalis pygmaea (Fabricius) are
placed on lateral branches and in shadier part of the shoot canopy. They are
more frequently encountered on auxiliary-shoots than on queen-shoots. The
vespids are attacked by workers of P. ferruginea if they land on the branches,
but the wasps actively chase the ants away from the paper nest.

The placement of bird nests depends on the species of bird rather than the
type of unit present in the shoot. The Derby Flycatcher (Pitangus sulphera-
tus) nests in the upper central branch forks as does the Barred Ant-shrike
(Thamnophilus doliatus). The Inca Dove (Scardafella inca) nests on hori-
zontal exposed branch forks.

Affects of stress on colony morphology. Acacia cornigera is a plant of
heavily insolated habitats and shading effects the occupant colony of P. ferru-
ginea. Heavy shading during the principal growing period (late May to
Dec.) slows its growth and reduces the rate of production of Beltian bodies,
nectar, and swollen-thorns. The Beltian bodies are paler yellow than those
of insolated shoots and they may have less nutritional value. The nectar pro-
duced is less in volume and tastes less sweet. The thorns have a smaller vol-
ume, and type B thorns are rarely produced (Fig. 21a). Heavily shaded
shoots only rarely have a queen-unit large enough to fill more than a single
thorn with brood. These shoots are occasionally maintained as auxiliary-
shoots by large queen-units and the larvae may be in part maintained on
products from the queen-shoot.

Heavily shaded unoccupied shoots usually do not live more than 1-2
years; this reduces the amount of time available for a queen-unit to develop
when compared with insolated shoots. Some mortality factor peculiar to
shaded sites may be responsible for the lack of maturing colonies in shaded

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 41]

habitats. It is more probable that the apparent reduced quality of Beltian
bodies and nectar, the definite reduced quantity of Beltian bodies and nectar,
and small amount of space in the thorns is responsible for the usual lack of
colonies.

During the cool and dry season, shoots over one year old often cease most
new leaf production and drop most of their leaves; the consequent loss of
Beltian bodies and nectar to the unit produces clear starvation symptoms. As
adults emerge, new larvae are not present to fill their places in the brood
thorns. In auxiliary-units this is because young larvae are not moved from
the queen-unit to the auxiliary-unit since food is not present in the auxiliary-
unit to feed them. In queen-units, the reduction of the number of thorns
with brood is due to reduced egg production and slowed larval growth. In
both cases, the number of larvae per brood-thorn and number of brood-
thorns declines. If all food is absent for two to four weeks, the queen usually
loses her physogastric condition and egg production stops. If starvation per-
sists, the brood area becomes fragmented and takes on the appearance of a
dying queen-unit or auxiliary-unit. Similar effects can be produced by killing
the queen and most of the members of the unit with insecticide, or by remov-
ing the queen-thorn. Queen replacement is not suspected since in all 22 cases
where the queen-shoot was deliberately destroyed, the associated auxiliary-
units died. Supercedure is unlikely since the alates mate outside the colony,
do not return to their own colony except by chance, and once they have left
the colony are attacked by the workers if they return.

Reducing the number of thorns available to a large colony causes over-
crowding. When the brood-thorns are destroyed by clipping the thorns, the
small amount of brood salvaged by the remaining workers is packed into any
thorn available (thorns missed and new green thorns). When intact thorns
are not available in sufficient quantity, the brood is stored in the open thorn
bases. The old intact thorns lying on the ground are not used as brood-
thorns though founding queens and workers may hide in them. If care is
taken not to kill the queen, fragmentation of the area of brood-thorns can be
accomplished by cutting them partly open, and then leaving them hanging in
the canopy. The brood is then removed by the workers and packed in with
other brood in the remaining thorns.

MorPHoLocy AND DEVELOPMENT OF THE POPULATION OF COLONIES. Due to
the “auxiliary-shoot phenomenon,” the number of colonies per hectare is not
equal to the number of shoots with a worker force per hectare. There is a 1:1
ratio only if the shoots are far enough apart so that they develop a colony of
their own before becoming occupied by the ants from a neighboring colony.
In addition, a few naturally unoccupied shoots are present during the latter
part of the dry season. The number of root systems is often much greater
than the number of shoots occupied by a worker force in man-made disturb-

412 Tut University ScieNcE BULLETIN

ance sites. In any given regeneration cycle only a small part of the shoots are
occupied and grow to maturity. On 12 Oct. 1964, following 285 days of regen-
eration without grazing after a cutting, plots N and O (850 m?) had 163

emergent A. cornigera shoots occupied by P. ferruginea (X height=93 cm,
s.d=16 cm) and at least 392 shoots below the canopy occupied only by found-

ing queens (X height=23 cm, s.d=11 cm). It is very unlikely that more than
5 percent of the submergent shoots would have become emergent auxiliary-
shoots or queen-shoots in this particular successional cycle. In natural dis-
turbance sites, the shoots are often spaced so far apart that there are no auxil-
iary-shoots.

At high shoot density, the number of colonies is lowered by colony inter-
action. Maximum shoot density does not produce maximum colony density
after the first year of regeneration. Direct branch contacts and trails between
shoots occupied by two different colonies lead to aggressive merger of colonies
since the workers will not tolerate foreign workers. The maximum distance
that a colony will range to find auxiliary-shoots is not known, but if shoots
are spaced 4-6 m apart (about 440 per hectare) the colonies in the different
shoots usually do not interfere with each other. A colony will move as far as
15 m to a new shoot if the shoot it lives in is killed. If all the shoots in an area
are occupied by colonies of sub-equal size, they could be closer together since
there is evidence that equal-sized colonies can resist invasion by each other if
the shoot branches do not touch.

At moderate densities, occupied vigorous shoots appear to be distributed
with contagion since a queen-shoot is often associated with a cluster of rapidly
growing auxiliary-shoots. As expected, the entire colony is distributed inde-
pendently of other colonies in the area if the shoots are far enough apart so
that the colonies do not come in contact with each other. This is not a static
system. Ownership of auxiliary-shoots may change or a shoot may develop
a colony of its own from a founding queen if it is abandoned. If an area with
moderate to heavy densities of A. cornigera shoots is mapped and lines are
drawn connecting the queen-shoots with their own auxiliary-units, the map
of the field is covered with roughly evenly spaced stellate figures with variable
numbers and lengths of arms. The arms occasionally cross each other and
may not be straight. While the position of the queen-shoot for a colony (the
center of the stellate figure) usually stays in the same place from year to year,
the various colonies do not always occupy the same auxiliary-shoot each rainy
season.

ENVIRONMENTAL REQUIREMENTS. Using Holdridge’s (1964) classification,
Pseudomyrmex ferruginea is found from Tropical Tropical Thorn Wood-
land to Low Subtropical Thorn Woodland to Low Subtropical Wet Forest to
Tropical Tropical Moist Forest in the study area. Following Leopold (1959),
P. ferruginea is found in Pine-Oak Forest, Mesquite-Grassland, Tropical De-

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 413

ciduous Forest, Tropical Evergreen Forest, Rain Forest, and Savanna. In all
these habitats a swollen-thorn acacia is a common denominator.

It is extremely difficult to separate the environmental tolerances of P. fer-
ruginea from those of Acacia cornigera. It is done with most ease in those
areas where A. cornigera stops and Acacia sphaerocephala starts or continues
(13.6 mi. S. Ciudad Mante, 11 mi. N.E. Ciudad Valles, 27.5 mi. S. Panuco,
22.8 mi. W. Conejos, 10 mi. S.E. Conejos). At these sites P. ferruginea con-
tinues into the drier habitats occupied by A. sphaerocephala while A. corni-
gera stops. Observations during the cool and dry season in the Temascal area
give some clues as to the limiting factors in the physical environment at the
margins of the range of A. cornigera and P. ferruginea. These are discussed
below.

In the literature, P. ferruginea (as P. belti and P. fulvescens) is described
as occurring at all 4. cornigera sites in Costa Rica and northward, as well as
at the localities of some other species of swollen-thorn acacia (Wheeler, 1913,
1942: Skwarra, 1934a, 1934b). From these reports, the general concensus is
that it is an ant of lowland tropical disturbed plant communities. Museum
records do not contradict this.

Host requirements. In its occupation of more than one species of Acacia
(chiapensis, cornigera, sphaerocephala, hindsu, and collinsii) P. ferruginea
demonstrates that it has a wider climatic tolerance than any one of them.
These host-plant associations are based on Wheeler’s (1913, 1942), Skwarra’s
(1934a, 1934b), and Safford’s (1921) discussions, locality records from the
literature and museum specimens, and this study. The present study has
yielded no records of P. ferruginea living on or in plants other than swollen-
thorn acacias. There are two such records in the literature. Skwarra (1934a)
reports finding a small colony in a bromeliad. This was possibly a colony
moving from a destroyed acacia or a misidentification. Emery (1890) reports
that the type of Pseudomyrmex belti fulvescens was collected from the cauline
swellings of Cordia gerascanthus by Beccari. This report is regarded by
Wheeler (1913, 1942) as very exceptional and I am inclined to regard it as a
migrant colony or an error in the placement of the host plant label, following
the search of several hundred Cordia trees in Central America.

Moisture. At the margins of the distribution of P. ferruginea, where A.
cornigera or A. sphaerocephala appears to disappear in response to increasing
aridity (Tropic of Cancer, 20 mi. N.W. Ciudad El Maiz), it appears that
the dryness does not kill the ant colony directly but rather that the plant lacks
leaves for such a long period during the year that the ant colony dies from
lack of foliar nectar and/or Beltian bodies. Most of the shoots found at these
sites still had their leaves in March, but they were growing in exceptionally
wet locations: arroyos and road banks. They were all occupied by well de-
veloped colonies of P. ferruginea. Six, two year old shoots were found grow-

414 Tue Universiry SciENCE BULLETIN

ing on dry hillsides at 20 mi. N.W. Ciudad El Maiz; they had dropped their
leaves and in all cases, there were no living remains of the ant colonies that
had tenanted them. These shoots would therefore have to pass the next rainy
season without the presence of a mature colony of P. ferruginea. These ob-
servations agree closely with those at Temascal where auxiliary-shoots that
lose all their leaves are abandoned and at times, are not reinvaded in the
rainy season.

An analogous situation is found between Veracruz and Anton Lazardo
where due to some factor (probably salinity) there is a very reduced produc-
tion of Beltian bodies by A. sphaerocephala which in turn prevents colonies
of P. ferruginea from becoming established.

During periods of total leaf drop, it is more the lack of foliar nectar than
Beltian bodies which kills the colony. This is shown by colonies that survive
for several months in the dry season on shoots that produce no new leaves but
do not lose all of their old leaves. Queen-shoots cut and hung in the dry
laboratory have no live workers or brood after two months. Single workers
confined in clean glass tubing rarely survive over 24 hours; if the cotton plug
is soaked with water or a sugar solution, they may live 5-10 days. During the
dry period, workers are willing to drink water and a variety of sweet fluids
during the day. The pre-sunrise nectar flow is collected completely during the
dry season.

With the entrance plugged by the worker, the thorn is a relatively mois-
ture-tight container. It has a polished, hard, and relatively waterproof outer
surface. The nectar brought in from the early morning nectar flow increases
the total moisture content of the brood and workers inside the thorn. This
moisture level probably decreases only slowly thereafter during the day even
when hot and dry winds are blowing. It is possible that moisture conserva-
tion during the dry season may be partly responsible for the concentrating of
brood in certain thorns rather than dispersing the larvae through all the
available space on the shoot. It should be much easier to maintain relative
humidity levels inside the thorn above a certain point when there is a large
mass of brood rather than a little. It is possible that type B thorns are pre-
ferred as brood-thorns because their thicker walls make better insulators
against drying of the contents. When it rains, thorn walls do not become
soaked unless the thorns are over two years old. The refusal of the ants to
use these older thorns for brood is very likely associated with this. Untended
brood is often killed by fungus (following abandonment of an auxiliary-
shoot or insecticide treatment). It is possible that the moisture resistant
thorn wall aids the workers in keeping the inside of the thorn dry during the
rainy season and thus avoid problems with fungus.

Rain has a depressant effect on the activity of P. ferruginea. Workers
walk on branches covered with water with extreme difficulty and slowness.
Therefore, considerable disturbance is necessary to bring workers out of the

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 415

thorn when it is raining hard. Light drizzle only reduces the numbers out-
side the thorns. If free water is not standing on the branches or thorns, the
workers run easily on the damp surfaces. Since most entrances are on the
under side of the thorn, rain does not enter. Brood and workers in clipped
thorns are drowned by rain.

Temperature. Low temperatures have a depressant effect on the activity
of P. ferruginea. At 6 mi. S.W. Tamazunchale, 8 mi. S.W. Martinez de la
Torre, 28.5 mi. W. Conejos, and Valle Nacional, P. ferruginea and A.
cornigera reach their range limits simultaneously. There is heavy rainfall at
each of these sites, so it is likely that it is the increasing number of days with
the diurnal maximum temperature below a certain level (about 24° C.), that
is acting as the range limiting factor, rather than dryness. It is not likely that
the colony is killed directly on the coldest days of the year at these sites; fol-
lowing a night when the minimum was 11° C. (Tamazunchale) and 12° C.
(Temascal), no colony mortality was recorded. At these temperatures, the
workers are barely able to walk and often fall off the shoot if they leave the
thorn. They can sting in this condition. A colony can be chilled until the
workers are immobilized, and it recovers completely.

During the Dec.Jan. cold season at Temascal, there was essentially no
growth of A. cornigera shoots. The consequent reduction of Beltian body
production caused auxiliary-shoot abandonment and the starvation of larvae
in some queen-shoots. Much failure of Beltian body harvest was noted due to
worker inactivity. However, on warm days when the workers were active,
there was a harvest of those Beltian bodies present. In areas where the cool
season is much longer than at Temascal, severe loss of colonies could take
place through starvation.

The slow physiological growth rate of A. cornigera at low temperatures
is decreased by the increased insect damage that occurs when P. ferruginea is
not active outside of the thorn for long periods. This damage is mostly done
by nocturnal insects. Adult katydids (Tetigoniidae) and the larvae of the
noctuid moth Coxina hadenoides do the most damage during the cool season.
It is the shoot tips that are removed by these insects, and in doing so, they are
not only removing Beltian bodies, but they are also slowing their production.

During cool weather, alates hatch but do not leave the shoot, which re-
sults in an increase in the number of alates in the colony. The failure of mat-
ing flights to occur during the long cool season at the western margins of the
distribution of P. ferruginea may be in part associated with its disappearance
at these sites. During the cool season, the growth of the colony is definitely
slowed. Trails between shoots are not established during periods of cool
weather, and this is associated with abandonment of auxiliary-units. In addi-
tion, the slowness with which the colonies of P. ferruginea develop in ex-
treme shade may be associated with the fact that diurnal temperature maxima
average at least 7° C. cooler in deep shade than in full sunlight.

416 THe University SciENCE BULLETIN

The most noticeable effect of cool weather on P. ferruginea is the reluc-
tance of the workers to leave the thorns, and the reduced running and walk-
ing rates when they do leave. However, the activity of the ant outside of the
thorn is not always directly related to temperature. This is shown by the fact
that the sunrise activity peak is often at a temperature roughly equal to that
of the nighttime activity low (Fig. 32, 33). The sunrise activity peak is at a
temperature much lower than the noon-time temperature at which lower
numbers of workers are usually on the shoot. It takes a temperature of 5-7°
C. higher at noon to bring equal numbers of ants onto the shoot surface than
at 9:00 a.m. At high air temperatures (45-48° C.), workers and alates are
driven from the thorn to lie in the shaded parts of the shoot. It is not known
what the mortality is under these temperatures.

Sunlight. In general, P. ferruginea is an ant of heavily insolated low vege-
tation. Founding queens may avoid deeply shaded areas in their search for
young shoots to colonize. Direct insolation can raise the temperature inside
of the thorn 1-3° C. higher than the outside air temperature; this could affect
larval growth rates. Light affects worker activity in that the number of work-
ers outside of the thorns is different at the same temperature, depending on
whether it is daytime or nightime. This is the case both during the dry and
the rainy season. There may be almost no workers active on the shoot sur-
face at noon when the air temperature is 22° C.; there are often many work-
ers outside of the thorns at this temperature at midnight.

Wind. Even under disturbance conditions, the workers do not leave the
thorn entrance when the wind is blowing over 25-30 mph. At this speed,
they are usually blown from the shoot. Workers active outside of the thorns
show no reaction to gentle shaking of the shoot by wind, but they quickly
show a disturbance reaction if the shoot is further shaken by an animal.

BEHAvior OF INDIVIDUALS. Larva and pupa. The larvae are relatively inac-
tive, even when feeding. If touched by a passing worker or with a probe, a
larva begins to rotate the head in and out of the trophothylax (pouch formed
by protrusion of the ventral side of the first abdominal segment; Wheeler and
Bailey, 1920) while opening and closing its mandibles. Fragments of Beltian
bodies are pushed deep into the trophothylax by the worker. If left protrud-
ing, the fragment is pulled in by the larva. The head is tipped completely
into the trophothylax while feeding. When large food pieces that distend the
pouch are eaten, the head is more externally directed so that mandibular ac-
tion is visible. A roughly spherical piece of food as wide as the larva can be
accommodated. If the food protrudes from the trophothylax, a worker may
remove a piece or take the entire piece away, cut it up, and redistribute it. A
worker may force open the trophothylax and deposit regurgitated drops of
fluid there. Presumably this is honey in the laboratory and would be petiolar
nectar on the acacia. While feeding, the larva ejects a clear fluid into the

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 417

trophothylax with the food bolus. Yellow fragments of Beltian bodies are
easily visible passing through the esophagous. These pieces are about 0.1 mm
in diameter in a larva 1 mm wide.

The larvae can be fed ripe Beltian bodies by hand. They accept frag-
ments of Beltian bodies forced into the trophothylax with fine forceps; about
% to % of a Beltian body from the end of a pinnule is most frequently ac-
cepted. This size fragment is that which is fed to the larvae by the workers.
Beltian bodies at the age chosen by the worker for harvest are almost always
accepted by the larvae when fed by hand. Less ripe ones are variably rejected
when offered. If the unripe piece is placed with the rind against the mandi-
ble, it is rejected immediately or within a couple of hours. If the softer inner
parenchyma is turned towards the mandibles, it is usually accepted.

Very rarely, workers are observed to capture small moths, lepidoptera
larvae, and brood of P. gracilis mexicana. These are taken into brood thorns
and presumably fed to the larvae. Wheeler and Bailey (1920) report insect
fragments in the buccal pellets of P. ferruginea. Buccal pellets also contain
pollen grains, fungal hyphae and spores, dust, and green plant tissue. On one
occasion the workers in a starving laboratory colony gave a live small larva
of P. ferruginea to a large larva as food. About one-half of the small larva
was eaten. Larvae in the laboratory accepted pieces of muscle and fat body
from the bruchid Acanthoscelides oblongoguttatus that were fed by hand.
However, such insect food constitutes much less than 1 percent of the food
eaten by a colony of P. ferruginea. There is no evidence that a colony de-
prived of Beltian bodies can subsist on other food.

Larvae can live for long periods without food. In the laboratory, larvae
may live at least 49 days without food or worker care. These larvae are ema-
ciated and shrunken but perform feeding motions and will accept food. This
long survival ability is undoubtedly due to the large volume of fat body in
well-fed larvae. Shrunken larvae are often observed in abandoned auxiliary-
shoots and in queen-units that have been starved for a month or more. The
ability of larvae to go without food is of clear value to isolated colonies that
live in shoots that lose most of their leaves during the dry season.

Worker behavior. Worker activities in the thorn. The workers are rela-
tively independent in their activities. They remove their own pupal exuvia
but other workers occasionally help. Self-cleaning is often observed in the
laboratory and on the surface of the shoot. The solid gleanings apparently
go into the buccal pellet. Following accidental dunking in honey, the worker
cleans iself but other workers may aid in licking off the honey. In labora-
tory colonies, workers are observed to stand relatively motionless for periods
of several minutes. Callow workers hatched within the past 12 hours are
especially inactive when in the presence of mature workers and larvae. It is
not known how long individual workers remain within a given thorn, and
how long before the callow leaves the thorn for the first time. One worker

418 Tue Universiry SctENcE BULLETIN

is usually on guard at the thorn entrance. Her antennae are often visible
just inside the opening. Treating the colony as a unit, more time is spent in
the thorns by the workers than in any other part of the shoot. It is not known

how much of the worker’s time is used in construction and cleaning in the
thorn.

Little care is given to the eggs in laboratory colonies. When eggs are to
be transferred from the thorn with the queen to a brood-thorn, they are
carried one at a time between the mandibles. In the laboratory they are
cleaned and occasionally rearranged but are not given the frequent attention
that larvae are. On occasion, workers starved more than 15 days will break
eggs with their mandibles and drink the fluid that is squeezed out.

Larvae receive much more cleaning and rearranging attention than do
eggs, prepupae, or pupae. It is possible that workers receive some exudate
from them. They lick the general body surface but do not concentrate at any
particular site such as the “exudatoria.”

In the laboratory, the founding queen’s first 2-3 workers do little tending
of the brood. These workers show little interest in retrieving spilled brood,
whereas the founding queen will pull her spilled brood back into her domatia
both in the field and in the laboratory.

Workers do not segregate the larval instars in laboratory colonies and
there is no indication that they do so within the thorn. A single worker can
pick up and carry a larva of any size. The larva is held out in front by the
worker when carried and often lifted completely off the substrate. It is held
at either end. Pupae are carried in the same manner. When a thorn is jarred,
the brood is often packed by the workers into the thorn apex opposite the one
with the entrance hole. If the thorn is progressively cut into from either end,
the workers retreat, pulling the larvae after them into the undamaged por-
tion. If the thorn contents are shaken out and the broken thorn placed on the
pile of larvae, the workers retrieve the larvae. They then take them in
through the regular entrance and break in the thorn.

Workers from inside different thorns on the same shoot, or from two dif-
ferent colonies, will work together to repack all of their spilled and mixed
brood into one thorn. If a larva from one colony is placed on a flat surface of
another colony’s shoot, or in the trail between shoots, it may be ignored,
picked up and carried into a thorn, picked up and thrown away, or bitten and
the juices drunk. Which of these alternatives is followed depends on the
duties of the worker which encounters the larva and how disturbed the colony
is. When the larva is carried into a thorn, it is not bitten and stung as is a
microlepidopterous larva before being taken into a thorn. Workers may
fight over the larva in a tug-of-war fashion. Worker pupae of P. ferruginea
from Acacia sphaerocephala near Veracruz were given to a laboratory colony
of P. ferruginea from A. cornigera from Temascal. The pupae were accepted

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 419

as well as the workers which emerged. When one colony invades a queen-
shoot or auxiliary-shoot of another colony, it seems probable that some of the
brood and callow workers are accepted by the victor colony.

Dead larvae are removed from the thorn. This is especially noticeable fol-
lowing parathion treatment. Many larvae are killed by contact with con-
taminated workers and perhaps by being fed contaminated buccal pellets. In
such cases, ejected dead larvae often litter the foliage in the brood-thorn area.

The amount of brood in a thorn is apparently regulated by the workers.
In general, the larger the thorn volume the larger the number and volume of
brood that are placed in it. The actual numerical ratios vary with type of
unit, age and history of the colony, and size of unit. When a small colony
comes into possession of a large shoot and many thorns, it tends to use the
largest thorns yet it maintains an area of brood-thorns with few unoccupied
thorns in it. When there is a shortage of thorn space, such as when a new
colony moves into a new set of young sucker shoots, all the thorns are packed
with brood. The mechanism for the relatively even brood distribution within
normal shoots may be rather simple. A worker would remove larvae when
the volume of larvae/volume of thorn ratio passes some maximum related to
1) the volume, condition, and position of the thorn, 2) the condition of the
larvae in the thorn, and 3) the number of workers on the shoot. It would
presumably place the larvae in some other thorn where the volume of larvae/
volume of thorn ratio is below the maximum. The triggering density must be
rarely reached since larvae are only occasionally seen being carried between
thorns. It is presumed that some adjustment process is used to keep below
the triggering density, such as selective feeding of certain larvae. If all the
eggs or first instar larvae found in some thorns were to grow equally fast,
they would overfill the same thorn when they reach the third or fourth instar.
Pupae are normally not moved between the thorns.

Unless larval cannibalism is an effective means of colony survival when
normal food is inadequate, the availability of Beltian bodies and foliar nectar
must set the lower limit to larval feeding rates. While some food comes from
small soft bodied insects captured on the shoot, this comprises less than one
percent of the food intake of a colony. The upper limit of larval feeding rate
is probably set by larvel satiation. An excess of Beltian bodies is present on ac-
tively growing shoots having only very young colonies or small auxiliary-
units in them; such shoots have many unharvested Beltian bodies on the
leaves.

Normally, Beltian bodies are harvested as soon as ripe. When a trail be-
tween an auxiliary-shoot and a queen-shoot is re-established, the auxiliary-
shoot may be stripped of its Beltian bodies that have appeared since the last
trail was established. These are then carried to the queen-shoot. P. ferruginea
does not ‘ncorporate Beltian bodies into the buccal pellet. In the laboratory
a worker gives its buccal pellet or Beltian body fragment to a larva making

420 Tue Universiry ScieNcE BULLETIN

bobbing motions with its head. Workers sometimes force open the tropho-
thylax to remove a food fragment and give it to another larva.

Nearly all direct unaggressive interaction between workers involves food
exchange. In the laboratory, the founding queen’s first two or three workers
show little interest in each other but as they become older, worker to worker
interaction becomes involved. In older colonies, the most common worker
interaction observed is liquid food exchange. Workers frequently rush head-
on into each other and there appears to be antennal interplay. When one of
the pair has been drinking honey, it often spreads its mandibles and a drop
of fluid appears between them. The other worker drinks from this. Often
the pair arrange themselves on the walls of the glass tube so that one member
is upside-down to the other. This brings the hypopharngeal areas in contact
without elevating the thorax. A third worker may also drink from the re-
gurgitated drop. In the laboratory, buccal pellet interchange has not been
observed. The worker that has brought a Beltian body into the tube drops it
and it is picked up by another worker. Another worker also may take it
directly from the bearer. Beltian bodies are carried through the thorn en-
trance, not passed to a worker inside. When a larva is being carried into a
thorn through a break in the thorn side, it may be taken from the entering
worker by one inside. Workers do not commonly clean each other except
when covered with sweet substances.

The number of workers per thorn is somewhat consistent with respect to
thorn volume and amount of brood in it. The thorn with the queen in it
often contains over 50 workers. Brood thorns have 4-20 workers, and thorns
without brood have zero to four or five. Since the thorn with the queen is
not proportionately that much larger, the smaller number of workers found
in brood-thorns may be related to the number necessary for brood mainte-
nance rather than maximum tolerance of workers for each other.

Workers in the thorns are probably just as aggressive to foreign organisms
as they are outside. In the laboratory, many workers in the glass tube do not
hesitate to attack foreign workers or founding queens. The worker guarding
the thorn entrance must recognize in some way and accept the passage of all
the workers in the thorn; she backs out of the way of an entering worker.

There are distinctly more workers in the thorn with the queen than in
other brood thorns. This may be due to her presence or that of the eggs’. Yet
in the laboratory, the physogastric queen receives little extra attention. Her
gaster is occasionally licked and workers may try to remove partly extruded
eggs, though she may resist by turning and biting at the worker or walking
away from her. In the laboratory, the queen is fed by the workers. Though
physogastric, she is quite capable of walking without worker assistance on
thorn, paper, or brood substrates. In glass tubes she has more difficulty but is
still not assisted by the workers.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 421

In moving to a new thorn, the queen is aided by the workers; they pull on
her legs to help her through the thorn entrance. The workers choose the
thorn with the highest volume that it located between the thorn occupied by
the queen and the upper green thorns. They enlarge the entrance of this
thorn in the same manner as they do that of the thorn she is leaving. She is
accompanied, but not chased, by a group of workers to the new thorn. In
some cases she is moved into a green thorn, but one that is hard and drying.
She may move as far as 100 cm on a tall shoot.

In the laboratory, a young colony of workers without its queen will accept
an old physogastric queen with only slight aggression towards her; she may
be bitten briefly or stung momentarily. They will not accept founding
queens. Thus a colony that has lost its queen in nature normally dies since
a physogastric queen would not be available to it. A physogastric queen is
not accepted by a laboratory colony that has a queen. Founding queens are
always killed by foreign workers in the laboratory, in thorns, and on the
shoot.

In the laboratory, newly hatched males and alate females are tolerated
until they leave or die. The males often require worker assistance in re-
moval of the pupal exuvie clinging to the genitalia. Alate females require no
assistance in molting. In the laboratory, alate females have been observed
receiving regurgitated liquid from workers.

The thorns are cleaned out by the workers. Unopened thorns develop a
natural lacuna equal to 14-34 the volume of that finally excavated by the
workers. They start cutting the entrance while the thorn is still full of moist
parenchyma. At this time, the thorn is green but stiff and fully expanded.
They remove most of the soft tissue before it has dried and shrunk. More
than one worker works on the entrance during its period of construction.
One worker works at a time, until the workers are well into the thorn cavity.
Two different entrance holes, one below each thorn apex, are only very
rarely encountered. When this occurs, they are made by two different found-
ing queens.

The entrance hole is nearly always in the same general position on each
thorn. It is on the ventral side 3-20 mm proximal from the thorn apex. At
this site the thorn is 25 mm wide. This is as close to the thorn apex as the
worker can excavate and make an entrance that is 1.5-3 mm long by 1-3 mm
wide. There is no preformed entrance or thin area in the thorn wall. The
long axis of the hole is parallel to the thorn axis and the hole passes directly
through the thorn wall. Dry thorns without an entrance are occasionally
entered by enlarging an exit hole of a microlepidopteran, or a zygopine wee-
vil. This entrance may be low on the thorn wall or at its base. Completely
dry thorn walls are not penetrated by P. ferruginea. Entrance holes are occa-
sionally started below the pointed apex of green legumes. They are not

finished.

422 Tue Universiry ScreENcE BULLETIN

The solid material is removed in the form of pieces of 0.2-1 mm in diame-
ter. This tissue is removed by cutting and pulling at partly freed pieces. It
is seen sticking to dew-covered leaves in the early morning under the thorns
in which construction is going on. These pieces have been carried to the en-
trance and dropped out. Smaller pieces may be incorporated in the buccal
pellet. After working for periods of 2-15 minutes, a worker often goes to a
brood thorn and returns several minutes later. Since the moist thorn pulp is
sweet, the worker may obtain some sugars from its sap. During the cool sea-
son an unidentified bird has been observed to crush the green thorns with its
beak; presumably this is done for the sweet fluid.

The entire thorn is emptied by the ants once the work has started. The
thorn walls are scraped down to the hard lignified tissue that is 0.5-2 mm
thick in type A and 2-4 mm thick in type B thorns. The non-entrance thorn
apex is excavated to the point where the diameter of the space is 1-2 mm. The
excavation of the inside of the thorn hastens the drying of the thorn walls.

Work on the entrance of a green thorn is not continuous but it is usually
completed in 24 hours. Cool weather may keep workers inactive so long that
some thorns are not hollowed out because the thorn wall becomes too hard to
penetrate. While still visible from the outside, a worker cutting an entrance
is easily distracted by disturbance of the colony. When an old entrance is be-
ing enlarged in a dry thorn to allow passage of the queen, there are as many
as four or five workers ringing the hole and gnawing pieces off the lip. This
work is very slow. To enlarge a hole from 1.5-2.5 mm to 2.5-4.0 mm requires
36-72 hours.

The outer margin of the thorn entrance is considerably larger than is nec-
essary to admit a worker. Once it is cleaned of the original parenchyma, the
thorn wall about 1 mm below the entrance is thickened with a hard matrix
of masticated parenchyma fragments. This results in an inner entrance with
a diameter closely approximating the width of the worker head (1-15 mm).
The worker guarding the entrance remains at this point with her antennae
extended almost to the margin of the outer entrance. The original paren-
chyma is probably too soft to be of use in this critical area. Workers that are
forced to use experimentally clipped thorns may close off the entrances, which
are 3-6 mm wide, with masticated parenchyma. In this case, the central en-
trance hole is of the same diameter as that of the inner entrance of an un-
damaged thorn.

Foreign and miscellaneous waste objects are removed from the thorn. In
laboratory colonies, prepupal meconia, cast exuviae, dirt, and dead larvae and
workers are dropped from the tube entrances. Some of this detritus may be
incorporated into the buccal pellet. Grains of sand put into the thorn entrance
are ejected. Shoots on sand dunes with the thorn entrances open to the wind
occasionally have thorns packed full of sand, and entombed larvae and work-
ers; the blowing sand entered faster than the ants could remove it. Thermis-

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 423

tor probes are attacked in any thorn with workers in it. The fiberglass or
fingernail polish coating of the probes is chewed off if they are left in place
for several hours. Workers chew through tape placed over entrance holes, and
the worker in the entrance bites introduced thread or straw.

Frass or larvae of microlepidoptera are removed from the green thorns. If
a microlepidoptera larva pupates in the thorn with the consequent construc-
tion of a silk crosswall, the workers may not be able to enter the obstructed
area until the moth has left. The hard and smooth silk partition is sharply
angled to the thorn axis. Zygopine weevils mature in dry, unopened thorns
but the workers enter the weevil’s exit hole and clean out the larval frass. No
external larval parasites of P. ferruginea have been found. Except when the
brood is in a thorn with a microlepidopteran crosswall, no arthropods besides
P. ferruginea have been found in brood-thorns.

A few insects are found in empty thorns. A single larva of a coccinellid
beetle (Brachycantha sp.) was found in an empty thorn on a shoot occupied
by P. ferruginea. It had an empty thorn on a shoot occupied by P. ferruginea.
It had an abdominal convex plate which was used to plug the thorn behind it
when taking refuge in the apex of the non-entrance half of the thorn. It
pupated behind the plate. The silver-fish Prolepismina and the thrips Dice-
raththrips horridus are found in older empty thorns in the Campo Cotaxtla
area; Prolepismina runs rapidly and evades workers that are chasing it on the
outside of the shoot. Salticid resting webs are generally found only in the
empty thorns of auxiliary-shoots or shoots with no worker force. The spider
can cope with an occasional worker but does not resist persistent removal
attempts.

Worker activities outside the thorn. Behavior outside the thorn is sepa-
rated from that inside the thorn because worker activity is somewhat differ-
ent in the two areas. Activity off the shoot is an extension of that on the
shoot, but outside the thorn. Activities on auxiliary-shoots differ from those
on queen-shoots but these are quantitative differences rather than qualitative.
These differences will be discussed under each aspect of behavior outside the
thorn.

1. Routine activities on undisturbed shoots. Activity outside of the thorn
is functional, either in respect to the individual worker or the colony as a
whole. There are definite activity patterns and duties. These are strongly
affected by disturbance and therefore are discussed separately. In general,
the number of workers active on the surface of the shoot increases as disturb-
ance of the colony increases, as the weather changes from cool to hot, as col-
ony size increases, as shoot size increases, as colony age increases, and as the
number of undamaged shoot tips on the shoot increases. A queen-shoot has
larger numbers of workers on its surface than does an auxiliary-shoot of the
same size even when the total number of workers on and in each shoot is the
same. Not only do these factors influence the number of workers on the sur-

424 Tue Universiry ScieNcE BULLETIN

face, but to some extent they influence the pattern of activity on the surface.
The routine activities are harvesting Beltian bodies, collecting nectar, remoy-
ing damaged plant tissue, cleaning shoot surfaces, and patrolling. Less com-
monly observed are transport of brood and cutting of entrances into thorns.

A. Beltian body harvest. The Beltian bodies are deliberately harvested.
A worker seeking a Beltian body goes directly from a brood-thorn to a shoot
tip or new auxiliary tuft without making detours onto mature leaves or
thorns. If pinna-end or rachis-end Beltian bodies are not available, she goes
along and across the rows of pinnules tapping the Beltian bodies with her
antennae and licking some. The one selected is cut and torn off at the junc-
tion with the pinnule. When engaged in removing a Beltian body, a worker
is difficult to distract. They may be touched by a paintbrush and marked
without showing any reaction. This “preoccupation” is accented by the fact
that if the leaf is touched with the finger, patrolling workers often run to it
from several centimeters away but even if the harvesting worker itself is
touched, it often shows no reaction. The Beltian body is carried directly back
to the thorn from which the worker came. Workers do not exchange Beltian
bodies outside the thorn and they do not co-operate in their removal. A
Beltian body taken from a worker with forceps and left lying on a thorn axil
is usually picked up by another worker within a few minutes.

It appears that as food is needed by the larvae, a worker goes from source
to source until she finds a ripe Beltian body rather than only visiting one
growing point and perhaps returning without one. This behavior results in
even distribution of food among the brood thorns. Beltian bodies are not
commonly found in the thorns; they are not stored by the workers, but rather
left on the leaves when there is an excess. In laboratory colonies, the Beltian
body is cut up and fed to the larvae shortly after it is brought back. Unripe
ones are ignored by the workers when placed in the glass tube.

The movements of harvesting workers are rather stereotyped. A worker
returning to a thorn with a Beltian body runs with little deviation. Her
speed is about 5 cm per second (32° C.). She runs to a node within one to
three nodes from its thorn and starts to run out the thorn. If it is not the right
thorn, she turns around before moving more than 1-2 cm along the thorn.
After returning to the stem, she runs down to the next thorn. She repeats
this at successive thorns until she finds her own. She avoids contacts with
other workers and will even run through the middle of a frenzied group of
disturbed workers without stopping.

Beltian bodies are harvested during the entire 24 hour cycle. Except when
a new shoot covered with unharvested Beltian bodies is invaded by an active
colony, less than five per cent of the workers active on the surface of a shoot
are involved in harvesting. The rate of removal is correlated with the num-
ber of workers normally present on the surface of the shoot. When a trail is
established between two shoots, there is a sudden increase in the rate of Bel-

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 425

tian body removal from the auxiliary-shoot. These are then carried to the
queen-shoot. Like other activities outside of the thorn the Beltian body har-
vest rate decreases as temperature decreases. During cool wather the shoot
tip grows more slowly but it can be assumed that the Beltian bodies left on
the leaves (due to the workers not leaving the thorns) are ripe since they are
harvested when diurnal temperatures rise. The pale Beltian bodies from
heavily shaded shoots are harvested in the same manner as those on insolated
shoots.

The workers do not relieve the founding queen of the duty of harvesting
Beltian bodies until there are 2-3 in the colony. This seems to be because by
the time 2-3 workers are produced, the first worker is old enough to assume
this duty. Beltian body harvest and nectar collection are the first duties io
appear among new workers outside of the thorn.

B. Nectar collection. Nectar collection is much less stereotyped than is
Beltian body harvest. Any worker may stop at a nectary to drink of her own
needs. However, this is different from the systematic visiting which results
in a distended gaster. Nectar collecting is most easily observed during the
activity period at sunrise during the latter part of the dry season. The worker
collecting nectar goes from nectary to nectary. When the gaster is swollen to
2-3 times its normal volume, the worker returns to a thorn and empties its
crop.

During the rainy season, nectar flow is more continuous and the collection
of nectar is more evenly distributed through time. At this time, workers with
the gaster distended with nectar are less frequently encountered. Workers
that are collecting nectar show variable responses to disturbance. Some will
attack intrusive animals but are not as agile as patrolling workers. They can
sting with a distended gaster.

Most of the nectar is collected by the occupants of the particular shoot,
since trails between shoots are not established until 1-2 hours after the main
nectar-collecting period at sunrise. Workers with distended gsaters are not
seen on trails between shoots but some nectar may be carried between shoots.
If the connection between two shoots is by direct branch contact, then work-
ers with distended gasters are seen to pass from shoot to shoot.

Some other sources of liquid are available, but provide a very small part
of the total liquid intake of the colony. During hot afternoons in the dry
season, workers are sometimes seen collecting sap from cuts in the shoot.
Some water soluble nutrients may be obtained from the parenchyma in
green thorns. When Beltian bodies are cut up, juices are released which may
be drunk by the workers. When small soft-bodied insects are caught and
dismembered, workers cluster around the prey and drink the body fluids.
Liquids from fresh bird droppings are utilized. If papaya, pineapple, or
mango junces are dripped onto the shoot on hot days, workers that encounter
them often drink. They do not, however, swarm to these liquids. Workers

426 Tue University SciENcCE BULLETIN

do not appear to regularly drink from drops of pure water placed on the
shoot surface on hot afternoons during the dry season.

C. Workers removing damaged acacia tissue. Worker ants normally cut
away damaged parts of the acacia. Despite the presence of an active worker
force, phytophagous insect damage to A. cornigera is common. This is done
by insects that are unmolested by the ants (e.g., Syssphinx mexicana), at-
tacked but not removed (e.g., Pelidnota punctulata, Coxina hadenoides), and
undetected. The stumps of shoot tips, leaf raches, and green thorns are the
commonest evidence of damage. The damaged surface dries and/or darkens
almost immediately. Within an hour or less a worker begins to cut away this
callus or necrotic tissue. This activity is first seen in maturing colonies with
200-300 workers. These damaged surfaces may be “regarded” as foreign tissue
by the worker and this excavation may be an extension of the constant clean-
ing of the shoot surface by the workers.

A worker removing damaged tissue acts like one cutting an entrance in a
green thorn. She braces herself and partly cuts and partly tears out a chunk of
tissue from 0.2-1 mm in diameter; she then turns around and drops it away
from the shoot surface. A petiole stump 15 mm long may be reduced to a
nub 1-2 mm high after 48 hours of this activity (with short interruptions).
Only rarely is more than one worker active at the same site. If another
worker comes to the site while the first is away dropping a piece of tissue,
the first worker may wait until the second leaves before resuming work.
Leaves and shoot tips that are wilted are pruned off. This is one of the main
reasons why heavily occupied shoots usually are very clean.

The worker may obtain some materials of use from this activity. Aside
from gathering sap, she may also use the fungal hyphae and partly rotted
tissue of the shoot to incorporate into the buccal pellet. A worker often stops
after 15-30 pieces have been removed and returns to a thorn for 1-11 minutes.
She may be giving her buccal pellet to a larva.

Only a small number of workers on the surface are active in removing
damaged tissue at any one time. This number depends on the amount of
damaged tissue sites that are present. Shoots with a large worker force usu-
ally have lower amounts of insect damage and therefore there is less occasion
for such work. Auxiliary-shoots tend to have more unremoved dead tissue
than do queen-shoots. This is due to the lower number of workers per unit
surface area of the shoot. Like other activities outside of the thorn, the ac-
tivity of removing damaged tissue disappears at low temperatures. However,
it is occasionally observed on cool nights when there are very few other work-
ers on the surface of the shoots.

D. Cleaning shoot surfaces by workers. When the colony has not been
disturbed, the majority of workers can be separated into two roughly even-
sized groups on the basis of activity. One group spends most of its time
running over the shoot surface and does relatively little cleaning; these are

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 427

the patrolling workers which will be discussed in the next section. The other
group spends most of its time licking leaf surfaces, throwing off pieces of
detritus, and probing corners and crevices. These are the cleaning workers
and they tend to stay in one place (e.g., on a shoot tip or mature leaf) for
periods of 1-30 minutes. This results in bark, thorns and leaves being very
free of foreign matter. They look as if they had been freshly washed. Clean-
ing workers are first evident when the colony has 100-300 workers.

The workers of very young colonies of 2-10 workers make buccal pellets
long before systematic cleaning of the shoot surface begins. In view of this, it
is possible that some necessary part of the larval diet comes from material ob-
tained from the cleaning activity. In mature colonies, apparently all of the
workers make buccal pellets, but the cleaning workers very likely make many
more than the others. Buccal pellets from workers on shoots held in the labo-
ratory often include sand.

The cleaning activities of P. ferruginea aid in normal expansion of the new
leaves. All surfaces of the expanding new leaves are intensively licked. The
pinnae and pinnules are forced apart as soon as is possible without breaking
them. These surfaces are sticky until cleaned. On unoccupied shoots, the
pinnae and pinnules are often stuck together until fully expanded. It is
probable that in addition to the adherent pollen grains, fungal spores, and
dust, the ants collect this sticky substance. Those workers cleaning and pa-
trolling the growing points are the last to leave the surface of the shoot as
the air temperature falls. As many as 10 workers may be found cleaning the
terminal five cm of a shoot tip.

Mature leaves, thorns, and branches receive similar but less intensive at-
tention. When cleaning a mature leaf, the worker licks exposed surfaces and
probes into corners. She tends to stay on a single leaf for periods up to 30
minutes. Rarely are more than four cleaning workers observed on a single
mature leaf. Usually there are less. The buccal pellets of these workers are
full of dust, pollen grains, fungal spores, and unrecognizable fragments of
similar size. They remove spider webs made during the night, dead pinnules,
and leaf-roller webbing. Any object lying on the leaf surface is picked up
and carried, or dragged, to the edge of the leaf and thrown or dropped off.
Thorn and branch surfaces are treated in the same manner as mature leaves,
but have a lower density of cleaning workers.

When the colony is disturbed, it becomes impossible to distinguish clean-
ing workers from patrolling workers because both exhibit the same attack or
disturbance reaction.

E. Patrolling shoot surfaces by workers. On an undisturbed shoot’s sur-
face, a large part of the workers are patrolling and at best, only momentarily
engaged in the duties described above. Many are walking or running on the
branch, thorn, and leaf surfaces. They reverse direction frequently and spend
little time cleaning the surface of the shoot. Most of the time, they appear to

428 Tue Universiry SciENCE BULLETIN

be of no direct service to the colony. These are called patrolling workers be-
cause there is a pattern to their movement, they are not distributed independ-
ently of each other, and they are responsible for most of the attacks on intru-
sive animals. With respect to any particular worker, she must be followed by
the observer for a few seconds before a decision can be made as to whether
she is patrolling or cleaning. The distinction between the two activities is
largely a function of the amount of time spent running about and the amount
spent cleaning the shoot surface. As the young colony grows, it is the patrol-
ling and cleaning workers whose numbers on the shoot surface show the
greatest increase. As with the cleaning workers, density is primarily a func-
tion of temperature, the time of day and the size of the colony.

The area patrolled by a worker is highly variable in size and part of the
plant covered. Some representative areas are 10-30 cm of a main axis, a 40 cm
lateral branch, a cluster of type B thorns, or the terminal 5-10 cm of a shoot
tip. She may stay in her roughly defined area for 1-20 minutes on a warm day
and for longer periods at night at lower temperatures. She stops and reverses
direction frequently, and is motionless for only a few seconds. During a
period of 10 minutes she might make 30 complete or partial trips through an
area composed of four internodes and four type A thorns. She often runs out
one or both sides of the thorns encountered. If cleaning workers are on a
leaf, she often does not go out on it. If it is bare she may go out on it. When
she leaves a particular patrol area, she may enter a thorn or go to another
part of the shoot and patrol. If the patrolling worker encounters dirt or web-
bing, she acts like a cleaning worker and attempts to remove it. Patrolling
workers also make buccal pellets.

The area patrolled by one worker usually overlaps that of another both in
time and space. The highest density of patrolling workers on undisturbed
shoots are found in the area of brood-thorns, area of growing points, and
occasionally on the lower trunk.

Her perception of foreign objects appears to be primarily visual, but vibra-
tions in the substrate are definitely of some importance. Since patrolling
workers rush towards small insect-sized objects that suddenly appear in their
vicinity, when these objects are 1-3 cm away, they probably are reacting to a
visual stimulus. The faster an object is moving, the further away it is per-
ceived and the quicker the worker turns toward it. In her short dashes toward
objects (insects, sand particles) entering her field of perception she may run
up to 15 cm per second. Patrolling workers frequently rush up to other
workers but they stop suddenly at the point of contact. Patrolling and clean-
ing workers are usually the first to contact intrusive organisms.

If patrolling and cleaning workers are removed from some point on the
shoot without disturbing the shoot, there is a definite rate of replacement of
these workers by workers from other surfaces of the shoot. During the first
few minutes of removing ants, the rate of replacement is high because those

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 429

ants that have patrolling or cleaning duties in the area are quick to reach the
point of removel. Then as the additional workers must come from farther
and farther away, the rate slows to a relatively constant value. The size of
this value increases as one approaches the growing points and the area of
most of the brood-thorns.

The following is a representative example of worker replacement rate. On
7 Jul. 1964, four lateral branches (35 to 56 cm long) were chosen for observa-
tion on a 205 cm tall shoot containing a queen unit. Each branch had an in-
tact shoot tip. The temperature was 34° C. (9:00-11:00 a.m.). The tip of the
author’s finger was held against the tip of the shoot tip and each worker that
ran out on it was collected. The branch was not moved and the workers
on the tip found the finger only through their normal activity pattern. This
was continued for 20 minutes at each branch. The four branches yielded a
mean of 20 workers (s.d=4) per branch and was virtually free of workers
with the exception of one to five cleaning workers on mature leaves. During
the last 8-10 minutes of the collection, each ant removed came from the main
trunk and ran out to the tip of the branch. One-half hour later, the number
of workers on the branch was back to the previous level.

F. Brood and adult transport. Workers are occasionally observed carry-
ing brood members from one thorn to another, and along trails between
shoots. The larger the member, the less likely that it will be transferred. As
when carrying a Beltian body, the worker runs directly, though more slowly,
and is not distracted by other worker activity. It avoids other workers and
may run around the branch or thorn when approached. A larva or pupa is
held in the mandibles by one end. It projects forward and is held above the
substrate. If the larva cannot be pushed through the entrance on the first
attempt, the worker turns, backs into the entrance and pulls the larva
through. At times she backs in without trying to go in forward first.

Workers are occasionally transported from thorn to thorn. The carrier
holds the labial area of the one being carried and she leans back over the car-
rier’s back. Her ventral side is against the dorsal side of the carrier. The car-
ried worker remains inactive, with legs held against the thorax and body
curved into a half-circle. The carried worker immediately becames active if
the pair is picked up. The carrier always backs into the entrance with a
worker. The worker being carried is usually not callow in respect to colora-
tion. Workers have not been observed to pick up other workers outside of
the thorns.

2. Trails between shoots. Passing over ground, leaves, branches, stems and
vines, the trail between shoots is a bidirectional odor path. Reversal of vine
or branch segments of the trail cause no confusion among workers. Even at
ground level the trail tends to be confined to branches and vines. Presumably
this is due to their smooth uninterrupted surfaces which act as a guide in
addition to the odor of the trail. It appears that the establishment of a trail

430 THe University ScrENcE BULLETIN

requires that the first worker wander from one shoot to the next in such a
short time that his trail stays fresh enough to be followed back. Since the
distance can be crossed most rapidly along smooth and straight stems, it is
probable that they will frequently become incorporated into the surface of
the trail.

Workers may miss the trail. At branch or vine forks, a running worker
occasionally takes the wrong fork. If such lost workers are constantly laying
new trail substance, this would establish a short false trail. This is supported
by the fact that the same mistakes tend to be made at the same points by dif-
ferent workers. Sometimes a worker does not stop when she takes a wrong
turn. New trails to the shoot are thus established in this manner. Once lost
she begins to wander. If she encounters the trail in this wandering, she con-
tinues along it. When lost, she usually continues in the same direction she
was going previously. Workers rarely reverse direction on the trail more than
a few centimeters. They run rapidly and do not stop to investigate foreign
objects. On the trail, in contrast to their behavior on the shoot, they are often
timid and run away from introduced objects rather than attacking them.

A trail is started by a worker wandering approximately in the direction of
the other shoot. In the two cases observed, the worker started from the
queen-shoot. Such a worker may be following remnants of a previous odor
trail or the queen-unit may “remember” where the auxiliary-shoot is. Other
workers follow and cover many segments of the path of the first worker. The
new trail often follows many of the same twigs used on previous days’ trails.

The trail does not always pass from the queen-shoot to an auxiliary-shoot.
As many as 14 shoots may be connected by active trails at the same time. The
usual number is 2-4. The trail may be as long as 12 meters between any two
shoots, but is usually 0.5-2 meters in straight line distance. By going up and
down vertical stems and vines, the trail maye be 2-3 times this length.

Once the trail is established, it may be retained for a few minutes up to
10-12 hours. Termination mechanisms are not understood. Presence and use
of trails is correlated with colony age, colony size, distance between shoots,
and the availability of food. How consistently the trails between shoots are
re-established is primarily a function of these same variables. Very old colo-
nies with few workers tend only to have auxiliary-units that are connected
by direct branch to branch contacts.

A. Relation of colony size to trail. In areas of moderate to high shoot
density, colonies with over 1,000 workers almost always have one or more
auxiliary-shoots that must be reached over surfaces of objects that are not A.
cornigera. The number of auxiliary-shoots increases with the size of the
colony. This may be because more workers leave a large colony; therefore
the chances are higher that a given unoccupied shoot will be encountered by
members of a large colony. In addition, the larger a colony is, the less likely
it is that its shoot will produce enough food for it. Many colonies in the

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 431]

Temascal area could never maintain their size on the products of any one of
the shoots that they occupy. In general, the larger the colony the longer the
trails, but medium sized colonies (2,000 workers) may also have very long
trails. The larger the colony the more consistently the trails are re-estab-
lished between the auxiliary-shoots and the queen-shoot.

B. Distance between the shoots. The discovery of a shoot that will become
an auxiliary-shoot must be done by random wandering of a worker. The
further away the shoot, the lower is the probability that the worker will
wander to it and return to the queen-shoot. The maximum distance over
which this can be done is shortened by the fact that the longer the shoot re-
mains without an auxiliary-unit, the greater are its chances of developing its
own colony or being destroyed by biotic mortality agents.

C. Food availability and trail formation. During the cool and dry season,
when growth slows in most shoots, there is increased trail forming activity
in order to find shoots with Beltian bodies. This attempt is unsuccessful for
the most part. During this season, worker populations are reduced in auxil-
iary-shoots. The total colony size decreases and many auxiliary-shoots are
vacated or abandoned. When new growth occurs at the end of the dry season
and during the rainy season, it is often the case that a single shoot does not
produce enough food for the queen-unit or large auxiliary-unit that occupies
it. New trails are then re-established. The new distribution of a queen-unit’s
auxiliary-unit may not be the same as in the previous season, since the same
colony may not find the same shoots that it occupied previously.

D. Relation of time of day and weather to trails. Trails between shoots
are established and maintained during clear or slightly overcast sky condi-
tions. They are not consistently begun or ended at any particular air tempera-
ture. Usually, the air temperature is over 29° C. and ground temperatures
may be as high as 50° C. Trails may become established, persist, or be aban-
doned at any time between sunrise and sunset. It appears that trail activity
will persist at lower temperatures in the morning and late afternoon than
during the midday hours. At night, workers have not been observed to travel
between shoots that do not have contacting branches.

E. Relation of worker activity to trail formation. A sudden 3-20 fold in-
crease in the number of workers outside of the thorns on the shoot is often
associated with the establishment of a trail to another shoot. Many of these
workers are not those using the trail. Between the time that the trail is
established, and the time that it terminates, the numbers passing in each direc-
ion are at best only subequal. Rates may be as high as 6 workers per minute
going one way for a 30-60 minute period. Workers are often carrying Beltian
bodies (usually from the auxiliary-shoot to queen-shoot) or brood (usually
from the queen-shoot to auxiliary-shoot). For example, on 28 May 1964, a
trail was observed between a 140 cm queen-shoot and a 46 cm auxiliary-shoct
from 5:15-5:48 p.m. (31° C.). It was at ground level and 520 cm long. Three

432 THe University ScIENCE BULLETIN

meters of this length were on a vine 1 cm in diameter. During the period of
observation, 16 ants bearing Beltian bodies, and 106 ants carrying no load
passed from the auxiliary-shoot to the queen-shoot. During the same period,
39 ants carrying each a larva 1-3 mm long, and 59 ants carrying no load passed
from the queen-shoot to the auxiliary-shoot. One worker was carried each
way. The next day, the queen-shoot contained 2,421 workers and the auxil-
iary-shoot contained 206.

3. Division of labor among workers. Among colonies of about the same
size occupying shoots of about the same size, there is consistency in the num-
bers of workers on the surface of the shoot doing the same things. These
numbers are highly influenced by weather and disturbance. Workers in one
thorn can only communicate with those in another thorn by going to that
thorn, or by chemical communication (e.g., alarm odors). Therefore, except
for patrolling and cleaning workers, it is probable that the groups doing dif-
ferent things are relatively independent of each other. That is to say, the
numbers of workers gathering Beltian bodies is independent of the number
of cleaning workers (although of course the two numbers are correlated). It
is only a small part of the workers in a unit that are on the surface of the
shoot at any one time (less than 20 percent unless the colony is disturbed).
What each individual worker is doing is apparently not related to its size but
rather to its age and the amount of disturbance the colony has experienced.

A. Relation of size to division of labor. Measurements of entire colonies
show only weak monomorphic allometry. Collections of workers performing
various duties do not indicate that any size class is restricted to a particular
duty or vice versa. Until a colony has produced about 2,000 workers, the
mean worker size rises as the colony ages. Young colonies characteristically
have smaller workers cleaning and patrolling; however, these are also the
oldest workers in the colony. Workers as small as those of a colony with 50
workers, may be found performing any duty in a colony with 4,000 workers.
These small workers are much more rarely encountered in a colony of such
a large size because few in this size range are produced; most of those pro-
duced when colony was young have long since died.

B. Relation of age to division of labor. Since workers of all sizes may
perform a given duty, it is suspected that the age of the worker dictates the
duty she performs. This may explain why some small young colonies are
more aggressive than some larger young colonies; the former, though smaller,
have older workers. As a colony ages, the possession of a larger and larger
cleaning and patrolling force may be in great part due to the increase in the
number of older workers. Callow and young workers in the laboratory are
less active in feeding the brood, investigating new objects, attacking intro-
duced founding queens, and running out of the tube. Patrolling workers are
almost always the darkest (and thus the oldest) on the shoot. Darkening of
workers with age in laboratory colonies is related to increased aggression to

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 433

foreign objects both inside and outside of the glass tube. It is probable that
the successive appearance of workers that harvest Beltian bodies, collect nec-
tar, transport brood, open green thorns, clean, and patrol in the maturing
colony is associated with increase in age of certain colony numbers rather
than the concomitant increase in colony numbers.

C. Relation of disturbance to division of labor. When a shoot is shaken
many workers emerge from the thorns. If no further disturbance occurs,
they gradually return into the thorns. The number left on the surface after
the colony is calm approximates the number that were there before the dis-
turbance. In some cases the number of workers that remain is lower. It is
not known if the workers that came out of the thorns were performing a
duty within the thorns. All workers will sting if handled roughly and often
sting even with less provocation.

However, even with severe shaking of the shoot, it is difficult to induce
more than about 50 percent of the workers to leave the thorns. On this basis,
it could be said that up to 50 percent of the workers are committeed to per-
form the duty of defending the colony outside of the thorns while about 50
percent are committed to tending the brood. These workers in the thorns are
are very aggressive towards objects entering intact or broken thorns. When
an auxiliary-shoot or queen-shoot is being invaded by another colony, very
large numbers of workers are outside of the thorns but it is impossible to de-
termine which belong to which colony.

4. Reaction to intrusive objects by workers. A. Perception. Cleaning and
patrolling workers are usually the first to encounter intrusive objects. Their
numbers are augmented from inside the thorns to a degree related to the
severity of the disturbance and the amount of alarm substance released. In-
trusive objects are anything besides a swollen-tthorn acacia or a member of
the colony. Seedling swollen-thorn acacias are destroyed as intrusive until
they are old enough to have developed the characteristics of swollen-thorn
acacias. Objects are intrusive when they contact the shoot or are in the circu-
lar area around the base of the shoot where plants are attacked.

The alarm substance, a musky, rancid onion odor, is released from the
mandibular glands of an excited worker. When perceived by another worker,
it causes a doubling or more in running and turning rate, and produces a
reaction that can be described as “rushing frantically about over the surface
of leaves and branches.” The worker in the thorn entrance is in an ideal site
to pick up this air-borne substance and join the other workers and/or com-
municate with workers in the thorn. The odors released by a large colony
can be carried at least 10 meters downwind to cause a disturbance reaction in
other colonies. Since the colony disturbed by this odor also releases alarm
substance, the colonies in a stand of shoots can become excited in a few min-
utes. A large disturbed colony can be smelled by the author at least 3 meters
downwind. The alarm substance of one worker can be smelled by crushing

434 Tue University SCIENCE BULLETIN

the worker’s head between one’s fingers. Once a worker has discharged its
alarm substance following rough handling, it is relatively odorless when
crushed. It is not known if a worker releases the alarm substance when in
individual combat with another worker, or when mauling intrusive vegeta-
tion. The odor is most evident when a shoot is shaken or a large animal
moves within about 1 m of it.

The worker with its very large compound eyes can detect movement of
man-sized objects at least a meter away from the shoot. A queen-unit with
as few as 75 workers on the shoot surface may show a strong disturbance
reaction to the movement of an observer 1 m downwind. The larger the
colony, the quicker it reacts. A sheet of white paper waved nearby causes
the same reaction. A group of patrolling and cleaning workers on a growing
point can be excited by waving a finger 10 cm away from the workers. Work-
ers carrying brood may be caused to run behind the branch by similar move-
ments. No reaction is observed if the finger is not moved. If a twig or finger
is waved back and forth 1 cm from a patrollling worker, she often reaches
out for it. At night, small shadows produced by the light from a flashlight
and that are moving on leaf or branch surfaces may be attacked. Cleaning
and patrolling workers often walk past motionless insects but attack them
immediately when they move. Patrolling workers frequently chase other
workers passing through their patrol area; the faster the worker moves the
more likely that it will be chased. While contact with the worker stops the
attack, a visual stimulus appears to initiate it. The worker sees moving ob-
jects behind, in front, above, and to the sides. The reaction to movement is
even more noticeable when the reaction is one of avoidance. A fleeing worker
is extremely difficult to grasp with forceps due to her dodging ability.

The odor of a man or cow appears to excite a unit; slow (to avoid a
visual stimulus) downwind approach often results in a disturbance reaction,
though it is not as strong as when the shoot is shaken. The lepidopterous
larvae (Syssphinx mexicana, Rosema dentifera) which are not attacked by P.
ferruginea appear to have developed a colony odor. If one is moved to a
shoot belonging to a different colony it is attacked with the same intensity as
other insect species. Even the foliage on which the caterpillar is sitting may
be attacked if placed on a foreign shoot.

Shaking the shoot or cutting into the trunk with a saw produces a general
disturbance reaction. It is assumed that the vibration caused by chewing of
Sigmodon hispidus has the same effect. The workers probably sense the
vibration directly through the substrate. Workers seem adept at concentrat-
ing at the point of contact of the disturbing object with the shoot. Those
workers accidentally contacting the area in their disturbed activity tend to
stay there. While the worker runs, her antennae are sometimes held against
the substrate and sometimes held in front or elevated.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 435

B. Reaction to inanimate objects. In general, the workers remove all
inanimate objects from the shoot. Dust, spores, and pollen grains are incor-
porated into the buccal pellet. Small dirt lumps are broken apart and the
fragments dropped off the shoot. Stones too large for two or three workers to
remove are eventually ignored. Spider webbing and leaf-roller webbing are
pulled off though some types take several days to remove. Embioptera webs
are cut through and pulled off but the workers are slow in doing it. Loose
dead twigs and leaves are pulled until they fall from the shoot. Wire, string,
and paint are attacked by biting and stinging. Eventually they are ignored,
presumably after developing the colony odor. Banding compound (“Stikem,”
Michel and Pelton Co., Oakland, California) is at first attacked, but later is
approached cautiously. Bird nests are usually not attacked even in queen-
shoots. They eventually rot away or an occasional straw may be removed by
a worker. They remain long enough for a set of nestlings to be reared. Adult
and nesting birds are not attacked. In plot B, the fruits of Muntingia calabura
were often found impaled on the thorns of A. cornigera after falling from the
tree. The workers worked both night and day at chewing out pieces of the
fruit and dropping them off the shoot. However, some workers became as
distended with the sap from this fruit as they did with foliar nectar.

C. Reaction to insect-sized animals. Insect-sized animals are with few
exceptions, attacked by the workers of P. ferruginea. These animals usually
leave the shoot following the attack of a single worker. They rarely cause a
general disturbance reaction. Caterpillars may require a numbers of work-
ers to chase them before they are willing to let go of the leaf and fall. Those
insects which are ignored (Syssphinx mexicana, Rosema dentifera) cause no
change in individual worker behavior. Those which are attacked yet im-
pervious to biting and stinging (Pelidnota punctulata) cause a local disturb-
ance involving 2-100 workers and lasting as long as the insect is present. If
an insect runs past a patrolling worker too fast for her to follow, she increases
her speed and runs in short dashes through the area where the insect passed.
If a worker encounters the leg of a large insect (i.e., walking-stick; Phasma-
tidae), and the insect just lifts its leg out of the worker’s reach, the worker
runs in tight circles and up and down the surface. If an insect shakes the
leaf on which a worker is cleaning or patrolling, she begins to run rapidly
over the leaf surface and may go back to the branch and out on neighboring
leaves. Thirty seconds to a minute are required for her to calm down if no
further disturbance occurs. Sufficient alarm substance is not released by a
single excited worker for the author to smell it, but workers within 10-15 cm
of the excited worker often become excited too. Some insects (especially
chrysomelid beetles) often sense the presence of a walking or running worker,
and jump or fall from the leaf before they are contacted.

A given worker does not invariably attack the animal encountered, but if
it does, it grasps the animal with mandibles, prothoracic legs, and meta-

436 Tue Universiry SctENCE BULLETIN

thoracic legs, and stings by bringing the gaster under the thorax. An attempt
is made to cling to the substrate with the mesothoracic legs. Workers pulled
away from the leaf, continue to wave the mesothoracic legs in an attempt to
grasp a support. If the sting cannot penetrate the object from the normal
position, the gaster is moved around until a soft place is located. The animal
usually leaves before the sting penetrates. At times it carries off the worker
still hanging to an appendage, but often the ant releases its hold immediately
after attacking. Small soft bodied insects (Lepidoptera larvae, moths, brood
of Pseudomyrmex gracilis mexicana, aphids), are sometimes held and stung
until they are paralyzed. A single sting of a few seconds duration behind the
head is usually sufficient to paralyze a 1-cm-long geometrid larva in 20-30
seconds. A larva the size of a worker may be carried to a thorn and presum-
ably used as food for the larvae. However, competition often develops be-
tween the workers that attack the larva, and often they try to hold it against
the pull of the worker trying to carry it away. A third faction develops in the
form of workers that attempt to take the dead larva and drop it off the shoot
as if it were a bit of detritus. If the larva is larger than a worker, it takes
several workers to hold it down and requires a series of stings to paralyze it.
As more workers gather at the scene each tries to pull the larva in a different
direction. A few workers cut into the larva. The larva is eventually
stretched and torn into pieces which are then carried to a thorn (rarely) or
dropped off the leaf (usually). There appears to be no co-ordination of ac-
tivity on the part of the workers. Workers never cooperate in the transport of
a piece of larva. Each one treats the larva as if she were the only ant attacking
it.

Founding queens are often killed by the workers of the shoot on which
the queen attempts to establish a colony. When founding queens are being
removed from thorns, the worker grabs one of the queen’s antennae with
her mandibles, and pulls until the queen is partly out. Then she tries to sting
the queen. If the worker is successful in removing the founding queen, they
may become entangled and drop to the ground with the worker stinging the
queen. The queen also tries to sting the worker. The worker kills the queen
in most cases. On the shoot, the worker may drag her around by an antenna
or leg until she escapes or the worker is joined by others. Several workers
may attack the same founding queen.

The only obvious relation between the size of the insect and frequency of
worker attack is that big insects are easier to find, usually slower, and move
the foliage more. Whether or not the insect is phytophagous of habit seems
to have little bearing on the attack. Movement by the insect, and particularly
the jerkiness of this movement, is of great importance. The placement of the
insect on the shoot influences the probability that a worker will encounter it,
and the probability that the worker will be aggressive toward it. For exam-
ple, an insect on the uppermost terminal shoot tip of an occupied shoot, in the

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 437

rainy season, is normally attacked within a second or less. That not all in-
sects are attacked is due to the insects’ properties, the particular worker en-
countering the insect, and the numbers of workers active on the outside of the
shoot.

Those variables which influence the number of undisturbed workers on
the outside of the shoot also affect the efficiency of the unit in removal of
small animals from the shoot surface. Since phytophagous insect activity de-
creases greatly during cool or dry weather, decrease of worker activity at these
times may be less critical than it would be during the growing season. How-
ever, there is enough phytophagous insect activity during the cool and dry
season so that when combined with the physiological reduction in shoot
growth rates, the result is that intact shoot tips on lengthening branches are
almost nonexistent during this period of relative worker inactivity. During
the dry season, workers are less active on cool nights and the larvae of the
noctuid Coxina hadenoides do heavy damage. During the growing season,
many workers are active at night as well as day and they are effective in low-
ering the incidence of many nocturnal feeders.

The workers of P. ferruginea are effective in keeping other insects off the
shoots both during the day and night. This is shown clearly in Table 2. This
data was obtained by direct counts of the insects present on occupied and un-
occupied shoots between 13 Jun. and 29 Jul. 1964. All species of ants, bees,
wasps, and flying lepidoptera are excluded in the counts. The shoots were in
control and experimental subplots and at each recording, all shoots in a given
subplot were examined. Since some colonies do not have workers active out-
side of the thorns on cooler nights, these plants were recorded as unoccupied
even though they contained a colony. However, these plants constitute less
than one percent of the sample. Night-time recordings were made with a
flashlight. No recordings were made in the rain. No subplot was recorded
within five hours of the previous recording. Over 90 percent of the unoccu-
pied shoots were in treatment subplots and over 90 percent of the occupied
shoots were in control subplots.

The insects found on occupied shoots (240) have quite different character-
istics from those found on unoccupied shoots (3124). Of the 240 insects, 88
were larvae of Syssphinx mexicana, 63 were larvae of Coxina hadenoides, and
40 were adults of Pelidnota punctulata. S. mexicana is not attacked unless it
begins to feed on the uppermost undamaged shoot tips; all 40 of the larvae
were on mature foliage. C. hadinoides is attacked by P. ferruginea, but un-
less the frequency of attack is high, it is not removed. P. punctulata is at-
tacked persistently but ignores the workers. Most of the other insects present
on occupied shoots were being chased off at the time of recording or were on
shoots with only a few (5-25) workers active outside of the thorns. There
were 3 P. punctulata, 43 S. mexicana and 182 C. hadenoides among the 3124
insects on unoccupied shoots. A sample of worker effectiveness of this type

438 Tue Universiry ScteENcE BULLETIN

Tapie 2. Incidence of phytophagous insects at night and during the day on
shoots of Acacia cornigera occupied by Pseudomyrmex ferruginea, and on un-
occupied shoots, during the first part of the 1964 rainy season.

Day Occupied Unoccupied
INOts Of SHOOLS MEXA TUM EC ieee meee ee eee en ee 1,241 1,109
INO SO tans OO ES i WiLL LenS CLS arene ee ae ne me a 33 427
INGORO tae LSCCESE O Dim cans 110 CO LS eee eee eee ee ae 48 977
No. of insects known to feed on A. cornigera ........-...-------- - 45 894
Night

INo. wore shootsy examined Bs. 5e ee ee ee ee 847 793
Non obeshoots with insects) ee ee ee ee 109 466
INKS CHE THORS ONT) TEMS: GVO cece eee ennnocecemercereeeeceneteenaneetareecurue 192 2,147
No. of insects known to feed on A. cornigera ................--.-..- 187 2,04

is complicated by the fact that once an unoccupied shoot has had its new
foliage and shoot tips removed, and has ceased growth, many phytophagous
insects are no longer interested in the shoot as a source of food and therefore
do not remain on the shoot.

Since queen-shoots have more activity outside of the thorns than do auxil-
iary-shoots, the thoroughness of removal of animals is likewise greater on
queen-shoots. Patrolling and cleaning activity is heaviest in the central thorn
area and on new growing points. Animals are almost never seen in these
areas on a shoot with a worker force outside of the thorns. Less continuous
patrolling and cleaning activity is necessary to insure the removal of seden-
tary phytophagous insects (aphids, mealy-bugs).

D. Reaction to large animals. Shaking the shoot by a large animal ap-
pears to cause a disturbance reaction by the entire colony. In a small young
colony, it may cause all the workers outside of the thorns to hide in the foli-
age or enter the thorns. During heavy winds or cold weather, only a few
workers will leave the thorns if the shoot is shaken and these soon fall off or
return to the thorns. The usual reaction to shaking the shoot is an increase
in the rate of movement and in the number of workers on the shoot surface.
For example, on 11 May 1964, the main axis of a 95 cm tall shoot was tapped
sharply 20 times with a pencil at 9:30 a.m. (34° C.). Within 30 seconds, the
number of workers outside of the thorns rose from 52 to 130. Four and one-
half minutes later (no further disturbance) the number had fallen to 65
workers on the outside of the thorns. This shoot contained a queen-unit with
about 500 workers in it. The number of workers outside of the thorns of a
250 cm shoot under similar conditions rises from about 250 to 350 workers,
and falls to the previous value in about 10 minutes. A similar rise in activity
may be associated with the recent establishment of a trail to another shoot,
invasion by another colony, perception of the body odor from a large mam-

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 439

mal, or visual perception of such an animal near the shoot. This increase in
number of workers outside of the thorns should not be confused with normal
increases in activity such as those at sunrise and sunset. Following severe
disturbance, it may take an hour or more for the activity on the surface to
return to normal levels. It is very difficult to determine normal levels since
worker behavior is frequently modified by animals, wind, rain, changes in
temperature, and the activity of the colony itself. The rate of return to normal
levels of activity increases as the temperature decreases.

The reaction to a large animal is varied in the individual worker, but con-
sistent in colonies with more than 200-500 workers. When foliage is torn or
chewed from the shoot, it may take a few seconds to a minute or more for a
general disturbance reaction to take place. The patrolling and cleaning work-
ers on the growing points react immediately. There are those workers which
do not attack (Beltian body harvesters, nectar collectors, etc.) but the colony
always contains workers that will. Further, it appears that the more alarm
substance released, the more workers there are that will leave the thorns. The
ability to react to alarm substance apparently increases with increasing age of
the worker. The attacking worker runs out onto the contacting animal and
usually bits, stings, and tries to pull away after a second to a minute or more.
The barbed sting sheath often holds her in place. She also releases alarm sub-
stance. It is possible that this alarm substance serves as an aposematic stimu-
lus to browsing mammals (cattle, horses, goats, deer, cricetid rodents). It
would be especially helpful in recognizing Acacia cornigera at night.

E. Worker damage to plant parts. When diurnal temperatures are over
26-28° C., any plant that does not have the characteristics of a swollen-thorn
acacia, and contacts the shoot, may be mauled by a worker (Fig. 34, 36-38).
There is a circular area under the shoot in which other plants are also mauled
(Fig. 35); this produces a circle of bare ground varying in diameter and dis-
tinctness that is usually not greater than the diameter of the canopy. Shoots
with over 200-500 workers that have occupied the shoot for several months or
more are free of vines and do not have growing branches or shoots of other
plants within the canopy. There are some exceptions to the above statements.

The plant part is mauled and this results in a blackened necrotic area.
The workers persist until contact with the A. cornigera shoot is lost due to
withering of the plant part (Fig. 37 and 38). On the ground they persist
until all green or soft tissue is gone . There are often large nobs of callus tis-
sue on the ends of intrusive branches that have repeatedly tried to produce
leaves. On branches with leaves, the petiole is attacked and the leaves fall off.
In open grassland, there may be no more than a patch of bare dirt around the
base of the trunk (Fig. 35b). When the shoot grows in roadside mats of vines
there is a small hole in the mat where the trunk comes through, and no vines
use the shoot as a standard. In dense vegetation, this activity produces a
cylindrical space free of foreign vegetation around the shoot. In the produc-

440 Tue University ScteNcE BULLETIN

Fic. 34. Workers of Pseudomyrmex ferruginea attacking the tendrils of a vine of Ipomoea
sp. that was placed over an upper node of an occupied shoot of Acacia cornigera a few minutes
before. Within five minutes, the more slender tendrils had been pruned off. Three of the
workers in the area of the thorn axil are biting into the tendril. The number of workers in the
photograph is representative of those usually present around a foreign object of this size that
does not move. Photo 1 Aug. 1964 in second growth vegetation 7 km east of Temascal.

tion of the cylindrical empty space around the shoot, the growing points of
other plants are mauled when they contact the shoot. The workers may chew
them for distances up to 30-40 cm back from the point of contact. The sub-
sequent death or leaflessness of the branch over this distance leaves a gap
between other green plant matter and the shoot. If the base of the other plants
are within the basal circle, the workers travel up the stems to the growing

points and destroy them if they are equal to the A. cornigera shoot in height
or lower (Fig. 38).

If a vine or branch falls across the shoot, the workers may not be able to
remove it but they often kill such a stem by girdling it. The reaction to a
foreign plant may be immediate or it may take a few days of warm weather
for damage to appear. A tender vine tip (Convolvulaceae) may be killed in
a few minutes. The rapidity with which an intrusive plant part is removed is

Fic. 35. a. The basal circle under a one year old queen-shoot occupied by a large colony
of Pseudomyrmex ferruginea. The shoot was 165 cm tall. The basal circle is 40 cm in diameter
and is completely devoid of living leaves of other species of plants. The leafless woody stems
are those that sprouted from old root stock at the site before the colony of P. ferruginea had
grown to a size sufficient to keep them pruned. All annual herbs and vines that sprouted in
the basal circle were killed in the seedling stage from mauling by the workers. Photo 27 Jul.
1964, in second growth 7 km east of Temascal. b. The basal circle under a two year old
queen-shoot occupied by a large colony of Pseudomyrmex ferruginea. The Acacia cornigera
was 400 cm tall, and was a sucker from an old cut stump that can be seen between the
bases of the two main suckers. The basal circle is about 100 cm in diameter. Its freedom from
foreign living plants of any kind is due to the cumulative effect of two years of mauling by
the ants. Further, any small seeds that fall into the basal circle are carried outside of it and
dropped by the workers. The density of workers in the basal circle was about 1 per 200 cm’.
This site was burned on 19 May 1964; the basal circle was sufficiently free of dead plant matter
so that the fire did not approach close enough to scorch the bark and kill the shoot. Photo
27 Jul. 1964 in regeneration after a fire 5 km east of Temascal.

44) Tue Universtry SciENCE BULLETIN

_

Fic. 36. a. On the left, a normal shoot tip of Tournefortia hirsutissima. On the right, a
shoot tip of T. Aursutissima that began to grow into the canopy of an occupied shoot of
Acacia cornigera and was mauled by P. ferruginea. Both of these shoot tips are from the
same individual plant and are of subequal age. It is this type of mauling activity that keeps
the occupied shoot free from vines and that produces a cylindrical space around the shoot
that is relatively free from foreign vegetation. Photograph taken in late June, in plot P.
b. As in Fig 36a except that the plant is a woody vine-shrub.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 443

Fic. 37. a. As in Fig. 36a except that the plant is the shrub Brxa orellana. b. As in Fig.
36a except that the plant is the vine Brgnonia unguis-cati. The upper shoot tip grew into an
occupied shoot.

an increasing function of colony size and diurnal temperatures. The size of
the lateral and basal clear area is definitely associated with the individual
colony as well as its size. The workers in young colonies start mauling other
plants when there are about 200 workers in the colony. Workers mauling
foreign plants are easily distracted; they are often aggressive to the distractor.

Some sap or plant tissue may be collected during the mauling. It is not
likely that mauling of plant parts is a simple gathering of food because it is

444 Tue University ScreNcE BULLETIN

least noticeable during the dry season when produce from the shoot is low-
est in quantity, and it ceases at such a short distance from the shoot.

Workers on trails between shoots usually ignore the green plant matter
that they walk on. However, the vines that are used for trails sometimes
come in direct contact with the shoot and are mauled at this point. When
mauling is done along a trail it is done to lateral growing points of the vine.

With the exception of some terrestrial bromeliads and Agavaceae, the
growing points or leaves of all intrusive plants are successfully mauled. Cases
of little or no damage to intrusive foliage are associated with unoccupied
shoots, small auxiliary-units in large shoots, cool weather, or very recent
growth of the intrusive vegetation. Very pubescent growing points, corky
branches, and coarse-leaved grasses are most slowly damaged. Seedlings of
swollen-thorn acacias without swollen thorns, Beltian bodies, and leaves with
petiolar nectaries, are destroyed.

Behavior of reproductive female. Inside the thorn before copulation. In
laboratory colonies, alate females that are 12-36 hours old have been observed
receiving regurgitated honey from workers. They spend much time standing
motionless in the tube and do not react to the presence of brood in a recogniz-
able manner,

Outside the thorn before copulation. Alate females leave the thorn 1-3
days after emerging. Workers have not been observed to restrain alates from
leaving in laboratory colonies or in the field. When a thorn containing alates
is broken open, no effort is made by the workers to restrain the alates that
leave. In the laboratory, they leave the brood thorns singly 4-2 hours before
actual sunrise (sunrise as here used means actual appearance of the sun above
the horizon). They run about on the thorn for 1-10 minutes and then fly to
the screened window. On this window, males pay no attention to females.

In the field, alate females have been observed with a flashlight to leave the
shoot 80-42 minutes before sunrise. They fly rapidly out and up in a straight
line and in the dark it is not possible to follow this flight. Some ignore the
flashlight but others fly to it. There are no records of P. ferruginea being col-
lected at lights. They are seen running on branches and the main axis of the
shoot, generally in an upward direction. While most fly from the top of the
shoot, some leave from lower points. There are large numbers of workers on
the shoot at this time but those encountering the alates ignore them. Air
temperatures are usually 22-23° C. (range 21-25° C.). No exit has been ob-
served during rain or drizzle. When diurnal maxima are in the 18-24° C.
range, the alates do not leave the shoot.

After leaving the shoot, she flies upwind to a tall object in the area of the
shoot; mating swarms have been observed on telephone poles, palms, roof-
tops, and broad leafed tree tops. All such substrates stand well above the sur-
rounding vegetation. She lands on the substrates 5-15 meters above the
ground. She stands for a few seconds and then elevates the gaster and opens

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 445

cia

Fic. 38. On the left is a portion of the canopy of an occupied Acacia cornigera. On the
right is a leafless shoot tip of an unidentified shrub in the Leguminosae. The base of the shrub
was in the basal circle of the acacia and the acacia grew upward to encompass the shrub. The
leaves of the shrub were then killed by the workers of Pseudomyrmex ferruginea, and any new
buds were likewise killed. Photo November 1963 in plot I.

446 Tue University Scrence BULLETIN

the sting chamber. The sting is fully extended and pointed upward. Pre-
sumably she is releasing a sex attractant at this time. She holds this position
until one of the males running on the surface encounters her and mounts to
copulate. Once the genitalia are joined and the male stops moving, copula-
tion lasted exactly 10 seconds for five timed pairs. At the end of copulation,
the female turns and begins to bite and push the male off. At this time, the
pair usually falls free. They break apart while spinning earthwards or im-
mediately after landing. During high swarming activity, as many as 500
pairs in copula may drop from the swarm in 15 minutes. On Jul. 28 1964, it
was estimated that 3,000 newly impregnated queens were produced by the
three mating swarms in the trees over 2 acres around plots K, P, N, and O.
During the later part of the dry season, there were some mornings when not
over 10 copulations took place in a single swarm. A large spider, Selenops
galapagoensis, and the vespids (Polybia occidentalis pygmaea, and Stelopo-
lybia areata (Say)) which prey on the males have not been seen to capture
females.

Outside the thorn after copulation. Immediately upon breaking away
from the male and alighting on a surface, the alate queen stands still, elevates
her gaster, and pumps the terminal segments of the gaster. This action lasts
30-60 seconds. This presumably works the sperm packet into the spermatheca
and closes the genital chamber. After the gaster is lowered, she stands for 1-15
minutes on the substrate and then runs or flies away. Observation of this
activity is facilitated by spreading a sheet below the mating swarm. These
females show negative phototaxis to strong lights. They will sting at any
time.

The search for Acacia cornigera shoots begins immediately. If a shoot
is encountered at this time, the newly impregnated queen reacts to it in the
same manner as a female which has been searching for at least a day. A trac-
ing of the route of a searching alate female has the appearance of a long string
scattered on the floor. She flies short hops of 5-200 cm and runs over leaf and
stem surfaces. At a distance, the shoot lacks an odor or sight attraction to
these females. Frequently a searching queen flies or runs within 5-20 mm of a
leaf or trunk of A. cornigera and does not turn toward it. Later, upon touch-
ing the same or a different shoot, she stops and investigates. She often covers
the same general area or plant surface two or more times and tends to stay
within 1-3 meters of some arbitrary point for 10 or more minutes. Alate
queens do not search at night or when diurnal maximum air temperatures
are below 24° C.

Upon touching a small unoccupied shoot, the searching queen runs onto
it and then over most of the branch and trunk surfaces. She shows a definite
avoidance reaction to other queens and insects by backing rapidly away and/
or running around to the other side of the stem, and then either running or
pausing motionless. She may jump off. If she does not wander off, and is not

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 447

scared off by another queen or insect, she begins to pull off her wings with
hind and mid-legs within a few minutes. This may take up to 10 minutes.
When dealate, she wanders about over the surface of the shoot examining en-
trances. If she finds a thorn without a queen, she enters and does not leave ex-
cept for food. If all are occupied, and she finds a new green thorn, she often
begins to construct an entrance hole. If not distracted, she will continue until
out of sight. This work is continued at night and requires 10-25 hours.

If a thorn is not available, she may wander off, stay on the shoot until
eaten by a predator, or be scared off. She also may enter and occupy a thorn
temporarily vacated by another founding queen searching for food. The
starved brood occasionally found with a founding queen indicates that the
invading queen may not adopt the strange brood. If she can gain entrance, a
searching queen may fight with the resident founding queen. On two occa-
sions a thorn has been opened when a founding queen and a searching queen
were fighting inside. A dead queen is occasionally found in a thorn with a
new founding queen in the entrance. On one occasion, a founding queen was
observed to throw the fragments of a decomposed founding queen out of the
thorn entrance. When there are two entrance holes in the thorn, one queen
positions herself in each entrance. There is no evidence that either of these
queens survives to rear brood in such a thorn. Two queens will not tolerate
each other and one cannot guard both entrances. In the morning, in an area
where there are large mating swarms, as many as 36 dealate females (search-
ing queens) have been found on the surface of 25-45 cm tall A. cornigera
shoots growing on bare ground (Fig. 9).

When a searching queen encounters a shoot with a worker force, she
searches it the same manner as an unoccupied shoot. Normally, her contacts
with workers are so frequent that she quickly drops off. If she is caught and
held, she is killed by stinging. Usually the worker that catches her leg drops
off with her to the ground. Since auxiliary-shoots often have a lower worker
force than queen-shoots of the same size, she has more chance to find an un-
occupied thorn and enter it. She cannot return to her parent colony without
being killed by the workers.

Once having left the shoot for any of the above reasons, she wanders until
she finds another shoot or dies. Some queens undoubtedly examine large
numbers of shoots before they die. Searching queens are commonly found
wandering on the ground and on low stems. They may leave shoots of A.
cornigera and hide in hollow twigs during hot afternoons, and return to the
shoot at night. In glass tubes without food and water, dealate queens may
live at least 18 days. They can therefore search for this length of time, and
probably longer since water and nectar are available in the field. They are
most active on the shoots in the early morning and late afternoon. During
the mid-day hours they are often found motionless on shaded parts of the
shoot.

448 Tue University ScrENcE BULLETIN

In the field, there are many more searching queens than there are new
thorns on unoccupied shoots. On hot days in the rainy season, searching
queens are as dense as three per m* of three months old regenerating vegeta-
tion. The number of searching queens seems greatly reduced during the cool
and dry season, but founding queens are still relatively abundant in the
thorns. Eventually, the maximum number of colony queens is one per shoot.
Therefore, high mortality must occur among searching and founding queens.
During the dry season, searching queens constitute a major part of the diet
of the lizard Sceloporus variabilis. It is commonplace for the stomach and
intestine of a single lizard to contain the remains of 5-30 queens. Bird drop-
pings frequently contain queen head capsules. Spiders (e.g., Dipoena oco-
singo, Tmarus ineptus, Corythalia sp.) are commonly found on and off the
shoot with searching queens as prey.

Searching queens are not aggressive. They see well and are very alert to
movement. They notice hand movements at distances of 20-100 cm. Queens
sting when picked up and squeezed. In a tender area of skin, they can cause
the same amount of pain as a worker.

When a searching queen encounters a shoot of Acacia chiapensis, she re-
acts to it in the same manner as to a shoot of A. cornigera, except it takes her
much longer to cut the entrance hole. On those shoots that have hard thorn
walls and a tough inner parenchyma, she often only clears out enough of the
parenchyma to admit her body. She is not successful in rearing brood in
these cases. A three meter tall shoot of A. chiapensis may have several hun-
dred thorns with entrances made by founding queens of P. ferruginea; these
thorns are almost invariably empty. An established colony of P. ferruginea
in a shoot of A. cornigera is quite willing to use an adjacent shoot of A.
chiapensis as an auxiliary-shoot (Fig. 39).

Inside the thorn after copulation. The queen that enters an empty or new
green thorn takes up a position with her mandibles 2-4 mm inside the ex-
terior margin of the thorn entrance. Her antennae can often be seen immedi-
ately inside the entrance. She is usually found at this point in thorns torn
from the shoot, or cut open while still on the shoot. Thorns on small shoots
that have never been tenanted by a worker sometimes have an entrance con-
stricted with masticated parenchyma. This construction work must have
been done by a founding queen. She cleans a green thorn completely, and
removes the parenchyma in the same manner as the workers.

Her eggs, brood, and first workers are found scattered throughout the
thorn. By the time the young colony has grown to where the thorn is packed
with brood and workers, a worker replaces the queen at the entrance. In
laboratory colonies, even when accompanied by her first 2-3 workers, it is the
queen that faces the 4 mm diameter tube entrance and goes to investigate
Beltian bodies and other objects placed on the tube lip. In laboratory colonies,
the founding queen spends some time licking and rearranging the larvae.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 449

Fic. 39. On the left, a shoot of Acacia cornigera occupied by a large colony of Pseudo-
myrmex ferruginea. On the right, a shoot of Acacia chiapensis that is occupied as an auxiliary-
shoot of the colony that has its queen in the 4. cornigera. The colony contains about 10,000
workers. The A. cornigera shoot is just beginning its third rainy season as an occupied shoot.
The shoot of A. chiapensis appeared to be of the same age. The 4. cornigera is in flower and
has not yet begun strong vertical growth. Photo early June 1964 in an abandoned quarry 1 km
west of Temascal.

The founding queen only leaves the thorn to collect Beltian bodies and
petiolar nectar. At sunrise founding queens have been observed to collect
petiolar nectar until the gaster is greatly distended with nectar and return
directly to the thorn from which they came. These thorns contain brood and
0-3 workers. Entire and partially eaten Beltian bodies are commonly found
in the thorns with the brood of queens that do not yet have workers. In the
38 shoots in plot J that were dissected (thorns opened) on 25 Apr. 1964 (a.m.)
six of the 16 thorns with founding queens and brood (no workers) had a
total of 40 Beltian bodies in them. Founding queens that have eggs and no
larvae do not collect Beltian bodies. The founding queen makes repeated

450 Tue University SciENcE BULLETIN

trips in the early morning and stores the Beltian bodies which are then fed to
the larvae during the day. As she develops a physogastric condition she is
dependent upon her workers to 1) defend the entrance, 2) enlarge the en-
trance so that she may occasionally move to new thorns, 3) bring food for
her and the larvae, and 4) distribute her eggs and larvae to other thorns and
shoots.

In addition to competition between queens causing the loss of very young
colonies there is some other mortality to founding queens. Up to one queen
per thorn may establish themselves and raise some brood in the thorns of a
single shoot. Only one of these becomes the queen of the colony that even-
tually occupies the entire shoot. A thorn may lose its founding queen to a
searching queen many times. Occasionally, moldy or dry queens are found
in thorns and sometimes they are accompanied by starving or dead brood.
Crematogaster and Solenopsis workers will drag moribund queens from en-
trance holes. Predating raids of a very small dolichoderine ant drives all
queens from all thorns on the shoot and the brood is lost to the invaders.
The very small size of this ant (2 mm long) may be a factor in its ease of
entry into the thorns. Salticids cannot get past the queen’s head but a spider
entering temporarily vacated thorns could capture the returning queen. Dead
queens have been found in thorns in the webs of spiders. In the Temascal
area, when growing in the sun, shoots with more than five thorns, and with-
out a worker force have not been found that are completely free of founding
queens. During the period Jul.-May the percentage of founding queens per
thorn drops. There is only a slight rise in the number of other organisms
occupying the thorns during this decline in the numbers of thorns occupied by
founding queens.

Parathion treatments are effective in removing some founding queens (for
the most part when they are foraging for food), but re-colonization is so rapid
that it is not possible to determine what percentage. Clipped thorns are not
attractive to founding queens though queens are occasionally found in natu-
rally broken thorns. They guard the large hole as if it were a normal entrance
but are not successful in raising brood in these thorns. Founding queens in
auxiliary-shoots and in queen-shoots have only a very short history due to the
constant efforts of workers to remove them, unless the shoot is abandoned
during the cool and dry season.

Physogastric colony queen. Even though confined to a thorn when fully
physogastric, the colony queen is very agile. On rough surfaces such as paper
or the walls of the inside of the thorn, she walks easily, but on the smooth
glass tube of a laboratory colony slips occasionally. When newly placed in a
laboratory colony she ceases egg laying in 1-9 days, but then when Beltian
bodies are made available she begins to lay in 6-12 days. When the physo-
gastric condition is reduced to the state where the gastric sclerites are nearly
approximated, she can run freely on a vertical or horizontal sheet of paper.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 451

She holds her gaster in place and above the surface even when running hori-
zontally on a vertical surface. Eggs are extruded by rhythmic pumping of the
gaster. It is suspected that the queen ceases egg production 3-4 days before
she changes thorns and therefore has a gaster with a reduced diameter when
she passes through the thorn entrance. Both of the queens observed while
changing thorns were only partly physogastric. When a shoot is cut, the
queen is transferred along with the remainder of the colony to whatever liv-
ing shoots are available (on the cut stump or a neighboring shoot).

No mortality factors have been observed for physogastric queens except
fire and other forms of complete colony destruction. Many workers may
leave the thorn when the temperatures in the thorn are over 43.5° C.; it is
possible that a colony queen could be killed at these high temperatures.
When a shoot is cut and placed where there is no access to a new A. cornigera,
the colony is gradually reduced in size as the larvae die and workers wander
off and are lost. The last remnants of the colony are 1-2 thorns with 5-30
workers and the colony queen with a gaster that is only slightly enlarged.
However, on one enigmatic occasion, a colony that had been stored in a bare,
dry room for 32 days still had 38 workers and one fully physogastric queen.
When first cut, this colony had about 1,000 workers.

Behavior of male: Outside of the thorn before copulation. When leaving
the shoot, the males’ activity is like that of the females but the males fly al-
most immediately after leaving the thorn. Both sexes may be observed to
leave the shoot at the same time. The males appear to fly directly to one of
the high points where the females are presumed to be releasing a pheromone.
The males always come from the downwind side of the tree (flying upwind).
In the area where mating flights were observed, most of the vegetation was
low (1-2 m tall) with outstanding objects (telephone poles, palms, occasional
broad-leafed trees). The flight pattern of males at daybreak is the following.
The males fly to the top of the outstanding object. With a hovering flight
they fly around and above any surfaces. The largest concentration is in the
area from which the most copulating pairs are falling. An estimated 10,000
males have been seen in the canopy of a single palm. However, most flights
are smaller. Many males land on the surface and run up, down, and around
in small circles. They buzz their wings continually. If a female is not found,
the male continues to fly around the object but he constantly loses altitude.
When 1-4 meters from the ground, he flies straight out and away from the
site of the swarm in an upwind direction. In this flight he gradually gains
height and by the time he gets to the next outstanding object he is back up to
a height of 3-7 meters above the ground. The swarming activity is then re-
peated. One male usually participates in several different swarms in a single
morning. As it becomes lighter and warmer in the morning, the number of
males swarming gradually diminishes. The swarm lasts until about 30 min-
utes after the last copulation takes place. This is 15-100 minutes after sunrise.

452 Tue University SciENCcE BULLETIN

Whether the morning is clear or cloudy seems to have little effect on the
duration of the flight. That the males continue to fly about a site after the
females have gone is probably due to attraction to a residual of the sex at-
tractant. Males have only been observed to swarm on high objects on which
females were known to be present that morning.

The reduction of the number of males in the swarm is for the most part
due to individuals settling on the underside of leaf surfaces where they remain
immobile for the rest of the day. They are alert to shaking of the leaves and
if disturbed will fly to a new leaf. They are usually 2-5 meters above the
ground and often in the same tree in which the swarming took place. That
the resting males take part in the next morning’s mating flight is substanti-
ated by the fact that at any given swarm, males outnumber the females 100 to
1 or more. Sex ratios in the colonies are approximately even. For how many
mornings the male is active is not known.

When the male lands on a surface on which there is a female he usually
finds her within a few seconds. It has not been possible to determine if he
finds her by sight alone or by odor as well. While males chase each other,
copulation attempts between males have not been observed. Once one male
is in copula with the female, another may find them and climb on. Occa-
sionally, as many as five or ten males may do this. The only result is that the
entire mass usually falls from the surface and breaks apart. Only one male
copulates with each female.

At certain times of the year, there is heavy predation of the males. During
the dry season, the two vespids, Stelopolybia areata and Polybia occidentalis
pygmaea, are constant predators of male P. ferruginea from about 15 minutes
before sunrise until the end of the swarm. They are not yet present when the
greatest part o fthe mating takes place. On palms, a large flat spider (Selenops
galapagoensis) is an ever-present predator of the males that alight and is
always found feeding on a ball of mashed males. While it feeds, any male
that lands nearby is caught. It appears that the two species of vespids take
about 200 males per morning per site and the spider about 100 out of a
swarm of 500 to 2000 males). S. areata flies off easily with a male; they are
captured while they are on the leaf surface. The smaller vespid P. occiden-
talis pygmaea often falls or glides to ground level. It may alight and cut the
head and wings off the struggling male before flying off. Male head capsules
have been found in passerine bird droppings but not as frequently as are
workers and queens. They are also occasionally found in the stomachs of
skinks, Anolis and other lizards.

After copulation. On Jul. 28, 1964, 16 males were observed to fly back up
into the swarm after each broke free from a female. This activity was ob-
served on a number of occasions. This indicates that a male may attempt to
copulate more than one time.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 453

Morra.ity FAcToRS OF workers. Despite the fact that the entire colony is
quite concentrated in its living space and foraging range, there is substantial
mortality of workers. It is estimated that a four year old colony with 12,000
workers may have produced 26,000 workers during its life. The average rate
of removal is probably about one per day for the first year, five per day for the
second, ten per day for the third, and 20 per day for the fourth. This rate of
loss is not constant throughout the year. A three year old colony may lose
1,000 workers in two hours in a fight with an invading colony (Fig. 40).
During the dry season, predation of workers both on and off the shoot by
lizards and birds is much more frequent than during the rainy season, when
cther insect prey is available for these predators. The rate of loss increases as
the colony increases in size (age) because the larger the colony, the more
workers there are off the shoot; workers off the shoot are much more vulner-
able to predators. Further, the larger the colony, the more involved the colony
becomes in maintaining auxiliary-shoots; when the connection to an auxil-
iary-shoot is broken, or when a colony fights over an auxiliary-shoot, there is
substantial worker mortality. An age-specific survivorship curve for an even-

Fic. 40. A point of contact between a branch from a queen-shoot of Pseudomyrmex
ferruginea and a branch from an auxiliary-shoot from another colony of Pseudomyrmex
ferruginea. The branches were blown together and held by locking thorns. About one-half of
the workers in the photograph are from each colony. While much of the fighting starts when
a worker goes from one leaf to the other, many workers pass this area and do not begin the
fight until they are 5-50 cm from the point of contact. Most of the fighting workers fall to
the ground where they are often eaten by lizards. Photo early May 1964 in plot U.

454 Tue University SctENcE BULLETIN

aged worker cohort is a straight horizontal line until duties outside of the
thorn begin. Then the curve should turn downward at an increasing rate as
the workers become more aggressive with increasing age.

Fighting. Cleaning and patrolling workers do not hesitate to attack ants
from other colonies. When removing a founding queen from a thorn and
killing her, the worker is often not killed. When a single worker wanders
onto a shoot occupied by another colony, she rarely kills more than one for-
eign worker before being killed herself. The invading worker is normally
the less aggressive of the fighting pair. When the wind blows the branches
of two queen-shoots together each colony tries to invade the other (Fig. 40).
A group of 500-1000 workers from each colony gathers at the site of branch
contact. The leaves and branches are covered with balls of fighting ants, and
with workers that are chasing each other. Worker mortality on both sides is
very high and the ground is littered with dead, dying, and dismembered
workers. The queen of the overpowered colony is killed. How much of the
losing colony is incorporated into the winning colony is not known. The
same thing occurs when a colony tries to overpower an auxiliary-shoot pos-
sessed by another colony. At least 1,000 workers died in the merger of two
colonies which afterwards had 3,200 workers.

When a shoot containing a queen-unit colony is cut and laid on the
ground near a large queen-shoot, it is not long before workers from the cut
shoot find the live shoot. The attempted invasion is usually slow to start and
often disorganized with several trails leading to the upright shoot. If the
colony in the upright shoot has time to rest, a group of several hundred
workers cluster on the bark at the base of the trunk and attack each foreign
worker as it tries to go up the trunk. In contrast to the willing and disor-
ganized defense at branch contacts between two colonies, this organized
system results in the loss of few workers by the owner colony. The attempted
invasion usually stops in a short time.

There is also worker mortality during defense of the colony from other
organisms beside P. ferruginea. Workers that attack large animals are often
carried from the shoot. Occasionally, an insect that is chased from the shoot
leaves with a worker clinging to it. These workers do not normally find
their way back to the colony. When a normal shoot is invaded by a doryline
raiding swarm (Labidus praedator) or column (Nomamyrmex esenbeckt),
no disturbance reaction is observed; the workers outside of the thorn become
relatively motionless or evasive. The dorylines do not enter the thorns. On
shoots with experimentally clipped thorns, dorylines completely clean the
shoot of workers and brood, and many workers of P. ferruginea die fighting
the invading ants.

Wandering. There is a steady flow of a few workers away from the shoot
into the surrounding vegetation. It is these workers which are responsible for
the discovery of auxiliary-shoots. They are very susceptible to predation

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 455

(birds, spiders, lizards). In most cases, they probably starve if not eaten, or
are killed when invading a shoot occupied by another colony. If they find an
unoccupied shoot, then they must repeat the return trip to the parent shoot.
Single large workers are occasionally found wandering on shoots that have no
unit of brood or workers in then.

Products from the queen. An auxiliary-unit can maintain its organization
for at least five weeks without establishing contact with the queen-unit. Col-
ony disorganization following the loss of the queen seems closely related to
the maturation of larvae; this occurs when there are no new eggs. There is
no evidence of workers laying eggs. As the colony decreases in size, worker
mortality rates increase because the workers do not keep predators such as
spiders off the shoot. The number of workers present to care for the brood
decreases, with increasing appearance of starvation symptoms among the
larvae. With decreasing worker density, phytophagous insect damage to the
shoot increases, and this reduces the amount of Beltian bodies available for
the brood. In the final stages of colony degeneration, only moldy brood
remains.

Lizards. The lizard Sceloporus variabilis, skinks, and Anolis sp., all eat
workers of P. ferruginea when the workers are encountered on the ground.
During the dry season, from 2-44 workers have been found in the stomach
and intestine of adult S. variabilis. This lizard also occasionally eats ants in
the genus Solenopsis and other small insects. During the rainy season, this
species of lizard also eats P. ferruginea workers, but they constitute a much
smaller part of the total volume of insects eaten. S. variabilis and Anolis sp.
are not seen climbing on occupied shoots, but they are occasionally seen on
unoccupied shoots. Most of the workers that they eat are caught on the
ground. S. variabilis is often attracted to the ground under fighting colonies
where many workers have fallen. Since P. ferruginea is the commonest large
epigaeic ant in the newly cleared areas around Temascal, and since it is
present all year, it is undoubtedly very important in the diet of these lizards.

Birds. In view of the exposed position of workers on the foliage of A.
cornigera, it is a tribute to the aggressiveness of P. ferruginea that bird preda-
tion is relatively minor. The barred Ant-shrike, Thanmophilus doliatus, will
eat workers of P. ferruginea from the shoot in which it has its nest. Passerine
bird droppings often contain the head capsules of worker P. ferruginea. Small
warbler-like birds forage in occupied shoots during the cool season, when
there are low worker densities outside of the thorns. They eat 1-10 workers
per shoot. The Black-throated Oriole, Icterus gularis, feeds on the pulp
around the seeds in the legumes of A. cornigera. This feeding creates a strong
disturbance reaction among the ants, and many of the workers are eaten; the
gizzard and intestine of a male contained 332 head capsules of P. ferruginea.

Arthropod predators. Inasmuch as the colony is concentrated in the shoot
canopy, and patrols this heavily, there are very few arthropod predators that

456 Tue Universiry Science BuLLETIN

enter the canopy to take workers of P. ferruginea. Two species of jump-
ing spiders (Metaphidippus maxillosus, Corythalia sp.) take an occasional
worker during daylight hours but these are generally found only on weakly
occupied shoots. At night, an unidentified spider enters the canopy long
enough to catch one worker, then drops out of the canopy to feed on her.

Fire. Any fire hot enough to wilt all of the leaves of the shoot kills the
colony. No efforts to remove brood or exit from thorns by the workers are
seen at this ime. However, a fire may only scorch the lower bark and kill
the shoot but not the colony. The colony can then move into the new sucker
shoots from the root stock.

Drowning. A. cornigera is not found growing close enough to the water
so that the canopy will be submerged by the rising river during the rainy
season. The submerged shoots are not killed, but the ant colony is. This is
especially notable when the river subsides and drowned founding queens are
found in the young shoots growing in riverside vegetation. If there is a dry
year, the shoot may grow high enough to raise the canopy and its colony
above the high water mark, and then the shoot develops normally.

OTHER ACACIA-ANTS IN THE sTUDY AREA. Nearly all of the shoots in the
study area are occupied by Pseudomyrmex ferruginea. Pseudomyrmex nigro-
cincta and Pseudomyrmex gracilis mexicana are comparatively very rare. P.
nigrocincta is most readily separable from P. ferruginea on the basis of the
former’s paler color, smaller size, and habit of carrying the gaster bent for-
ward under the petiole. In respect to the total population of A. cornigera in
the Temascal area, P. nigrocincta plays a minor role in that there are about
250 colonies of P. ferruginea to each colony of P. nigrocincta. The same ratio
exists in respect to the number of founding queens of the two species. In
addition, the two species interact in a very similar manner with A. cornigera
and A. chiapensis, and the same ratio of founding queens is found in the
later species of acacia. P. ferruginea and P. nigrocincta appear indistinguish-
able in their behavior in harvesting Beltian bodies, collecting nectar, using
available thorns for brood, and interacting with other species of ants. They
are subjected to the same mortality factors. However, in comparison with P.
ferruginea, the colonies of P. nigrocincta 1) are less thorough and consistent
in their patrolling of the shoot, 2) have more timid workers, 3) must grow to
a larger size before they provide effective protection of the shoot tips, 4) are
more commonly established in shoots that are shaded, and 5) require more
time to become reorganized after moving into a new shoot.

There are some significant differences in reproductive biology between
these two species. A mature colony of P. ferruginea produces a quantity of
alates similar to the quantity produced by a colony of P. nigrocincta, but a
colony of P. ferruginea has its first alates 6-12 months before a colony of P.
nigrocincta. The mating flights of the two species take place around the same
high points throughout the year, but the males of P. nigrocincta fly slightly

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 457

earlier than those of P. ferruginea. Both sexes of P. nigrocincta have been
collected at white light about 30 minutes before the slightest visible trace of
the sunrise; P. ferruginea has not been taken at light. Both species of males
may be taken with one sweep of the net in a mating swarm, and mating pairs
of the two species have been observed within a few centimeters of each other;
it is very likely that the two species recognize their proper females by differ-
ences in the sex attractant. About one out of 20 of the new shoots (10-40 cm
height range) with founding queens in them has a founding queen of P.
nigrocincta in one of its thorns. This indicates that there is a mean number
of about 12.5 thorns per shoot at the time when one species or the other takes
over the shoot. Sufficient information is not available to postulate the reasons
why the relative density of the two species is so disproportionate in the study
area. As one moves south to Costa Rica, P. nigrocincta becomes much more
common.

P. ferruginea and P. nigrocincta are much more similar in behavior and
morphology than either is to Pseudomyrmex gracilis mexicana. While the
former two species can be regarded as obligate acacia-ants, the later must be
treated for the present as a facultative acacia-ant (Wheeler, 1942). In the
course of this study, evidence has been gathered that indicates that the appar-
ent intermediacy is due to the existence of several populations which are mor-
phologically indistinguishable but differ in behavior. This aspect of the bio-
nomics of P. gracilis mexicana will not be discussed further in this paper, but
is under study.

P. gracilis mexicana does not interact with any of the swollen-thorn acacias
in the same manner as P. ferruginea and P. nigrocincta. A mature colony of
P. gracilis mexicana rarely has more than 1,000 workers. When the colony is
living in A. cornigera, the thorns that contain brood are usually distributed
among 2-15 different shoots. Under natural circumstances, these shoots are
usually unoccupied and shaded shoots in the 15-60 cm height range. Taller
shoots of all heights are occasionally found with all or part of a colony of P.
gracilis mexicana in their thorns. These shoots are normally unoccupied by
P. ferruginea and in very poor condition in comparison with the neighboring
shoots that are occupied by P. ferruginea. The workers of P. gracilis mexicana
do not patrol the shoot surface and are only aggressive toward small insects;
even in this case, the initial approach to the insect is cautious, and an effort is
made to capture the insect. When living in a shoot of A. cornigera, a large
part of the diet is Beltian bodies. When harvesting them, the worker often
chews up the first one and incorporates it into the buccal pellet and then cuts
another to carry to the brood. A foraging worker may travel as far as 15 met-
ers between the shoot tip and the particular thorn to which she is returning.
The workers clean out green thorns in the same manner as do P. ferruginea,
but the entrance hole is larger. When the colony of P. gracilis mexicana is in

458 Tue University ScrENCE BULLETIN

hollow branches instead of swollen thorns, the bionomics of the colony is
like that of the species of Pseawdomyrmex not associated with living plants.

Along with ants of the genus Crematogaster, P. gracilis mexicana is the
common and usual inhabitant of shoots of A. cornigera that are unoccupied
by P. ferruginea. In areas where A. cornigera is an abundant plant, P. gracilis
mexicana is one of the most commonly observed ants foraging in second
growth vegetation 1-3 years old.

Discussion. The members of the genus Pseudomyrmex can be roughly
divided into three groups. Those of the larger group are not associated with
any particular species, or group of species, of living plants. They live for the
most part in hollow, dead stems and the colony is occasionally extended into
the pith canal of a living stem. The adults feed on nectar from plants (floral
and extra-floral nectaries) and/or the juices from dead insects and captured
living insects. Both living and dead insects are taken as prey by workers
foraging individually; this prey is used for food for the larvae. The members
of the smaller group of Pseuwdomyrmex regularly live in association with a
particular species, or group of species, of living plants. They live in spaces
that are in the living plant (e.g., hollowed pith canals) or directly attached to
it (e.g., swollen stipular thorns). They depend on modified structures of the
plant for part or all of their food (e.g., Beltian bodies), or rear coccids within
the hollowed pith canals.

Pseudomyrmex ferruginea, Pseudomyrmex nigrocincta, and Pseudomyr-
mex gracilis mexicana are the commonest members of the smaller group
within the study area. The former two species live only on/in swollen-thorn
acacias, and are quite distant from the larger group in their behavior and
ecology. P. gracilis mexicana may be intermediate between the two groups.

Like the other species of Pseudomyrmex associated with living plants, P.
ferruginea possesses a number of behavioral and ecological traits which ap-
pear to be associated with its interaction with its host plant. However, it also
has many behavioral and ecological traits in common with the members of
the genus that are not associated with living plants. To demonstrate this, the
following list is presented. Each trait is to some degree common to almost all
members of the genus.

1) The workers are fast runners, are extremely agile, hold themselves
close to the substrate on short legs, and rarely fall off the substrate.

2) With their large compound eyes, the workers are able to respond to
moving objects up to distances of at least 1 m. Up to a certain point, the faster
an object is moving the more likely it is that the worker will show a-response.

3) While the portion of the colony occupying a single shoot functions as
a unit due to the summation of the members’ activity, each individual worker
acts relatively independently of the other workers when outside of the thorns.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 459

Workers do not co-operate directly with each other in performing duties out-
side of the thorns and apparently act normally when separated from the other
workers.

4) The workers and alate females have a well developed barbless sting
with a barbed sting sheath; they do not hesitate to use it on any animal that
contacts them. Some of the Psewdomyrmex that are not associated with living
plants are somewhat reluctant to use their sting, or else the worker cannot
penetrate the skin or integument of the attacker.

5) The workers frequently lick the surfaces on which they are walking,
and what is licked up is incorporated into a buccal pellet; this is in turn fed
to a larva. There is no evidence that any solid food is eaten by the workers.

6) The prey (the Beltian body) is brought back to the swollen-thorn with-
out cutting it up or incorporating it onto the buccal pellet. It is then cut into
large pieces which are fed to the larvae without further preparation. (Some
of the larger species of Pseudomyrmex will incorporate small pieces of prey
directly into the buccal pellet).

7) When a source of nectar is found, the worker drinks until the gaster is
partially distended and then returns to the colony to regurgitate the fluid.
Apparently, no effort is made to lead other workers to the source of nectar. A
particular worker is quite capable of returning to the same food source if it
leaves voluntarily or is chased away. At times it uses an odor trail but at
others it appears to use visual orientation.

8) The entrance hole to the living space is cut out by the workers and
then masticated wood pulp is laid down by the workers on the walls of the
entrance antechamber; this narrows the opening to match the diameter of
the worker’s head.

9) In the advent that a nest structure is partially damaged or rots, the
workers seek a new one and move the contents of the old into it. The larvae
are carried one at a time, held straight out in front of the worker. Rarely two
workers will both carry a very large larva or pupa at the same time. Workers
are carried as well.

10) The living spaces are relatively tubular in form and carton partition-
ing structures are not produced. There is one entrance to the living space
(some species of Pseudomyrmex have more than one entrance to the living
space).

11) The colony is arboreal. (Some species of Pseudomyrmex live at
ground level and it is commonplace to find workers of some species of
Pseudomyrmex apparently foraging on the ground litter. However, they also
fall from the foliage and so presence on the ground does not demonstrate that
they are foraging there.

12) Except when small enough to fit into one living space, the colony is
distributed throughout a number of discrete units. In the case of P. ferruginea
and other Pseudomyrmex associated with living plants, these units are

460 Tue Universiry SciENCE BULLETIN

grouped at a second level, the individual shoot (e.g., the auxiliary-shoots and
queen-shoot). A single unit of a colony of a species not associated with living
plants often holds a larger part of the colony than does a swollen-thorn, for
example, but the units are generally more widely distributed (e.g., in several
different plants and vines). The units in all cases are connected by open air
paths which may or may not be odor trails.

13) Distinct morphological worker castes are not present and apparently
associated with this, each worker probably performs all of the duties in the
colony at some time in her life span.

14) The colony has one physogastric queen located in the same unit with
most of her eggs. She is agile, attempts to hide when the living space is
broken open, and even may attempt to sting. Founding queens are killed by
the workers that can catch them. (There are several species of Pseudomyrmex
which have more than one physogastric queen per colony.)

15) The founding queens use empty living spaces excavated by other ants
or insects in plant stems, or if these are not available, they excavate their own.
More than one queen cannot found a successful colony in the same nest cav-
ity, except when there is more than one physogastric queen per colony. The
founding queen forages food for herself and her brood, until at least several
mature workers have been produced.

The following list of traits of P. ferruginea contains those which are in
some way associated with the interaction of the ant with swollen-thorn aca-
cias. It is in the expression of these traits that P. ferruginea demonstrates the
greatest differences between it and the species of Psewdomyrmex that are not
associated with living plants in an obligate manner.

1) A large proportion of the workers in a mature colony are very aggres-
sive toward animals of all sizes that contact the substrate containing the living
space (a swollen-thorn acacia). These workers both bite and sting the animal,
and only a very few of the very small and soft bodied insects are captured as
prey. In contrast, those species of Pseudomyrmex that are not associated with
a particular species of plant are usually not aggressive towards animals on the
substrate containing the living space; even when the nest is broken open they
often flee without attacking.

2) Many of the workers in a large colony will maul any vegetation that
is not a part of a maturing swollen-thorn acacia, and contacts the shoot or
grows in a circular area under the shoot. There is no evidence that the work-
ers of species not associated with living plants chew on plant matter except
when excavating a nest site or in some cases, gathering food.

3) The workers of P. ferruginea have an especially well developed sting
which can penetrate more deeply into human skin than can that of any of the
other 29 species of Pseudomyrmex collected in the study area.

4) When the weather is clear and warm, the workers are active outside of
the thorns both day and night. The workers of species not associated with

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 461

living plants have not been collected outside of the nest after dark, and usually
they are not present during the early morning and late afternoon hours as
well. The workers of P. ferruginea are active outside of the nest (thorns)
during much more inclement weather than are those of species not associated
with living plants.

5) When the weather is clear and warm, from 2-25 percent of the workers
are active outside of the thorns, and they are for the most part concentrated
within the confines of the shoot. During the daylight hours, equivalent num-
bers of workers of species not associated with living plants may be outside of
the nest. However, they are not concentrated in the immediate area of the
nest. In addition, since the colony is often broken up into several widely dis-
persed units, the workers are even further dispersed. It is only when the liv-
ing spaces and the food sources are concentrated in a small area (the swollen-
thorn acacia), that large numbers of workers can be found outside of the nest
yet in its immediate vicinity.

6) On the surface of the shoot, workers often chase each other and only
those workers that are carrying brood or prey (Beltian body) dodge away
from these attacks. When a worker avoids or backs away from another
worker, it is often only after direct contact. The workers of species that are
not associated with living plans usually see other ants at distances of 2-15 cm
and often dodge away from them.

7) When on trails between shoots, the workers appear to be strongly de-
pendent upon odor trails for orientation. The workers of species not associ-
ated with living plants use both visual and odor orientation when traveling
between the parts of the nest.

8) The colony may attain a very large size (12,000-plus workers) and
when it is split up into auxiliary-units and a queen-unit, it is composed of
several large segments, each of which acts very much like a separate colony.
While the size of colonies not associated with living plants is very difficult to
determine due to the dispersed placement of the parts of the colony, it is cer-
tain that these colonies only rarely attain such a large size; they usually have

500-5,000 workers.

9) Almost the entire colony is confined to a small volume of space in re-
spect to foraging and the situation of the total nest. Workers of species not
associated with living plants will forage at least 30 meters from their fraction
of the nest, and the parts of a nest may be scattered over an area of at least
300 m?.

10) Not only is the colony supplied with a continuous food source (though
a fluctuating one), but it is also continually being supplied with new nest
sites, except during the dry season. The result is that the colony grows much
more rapidly than does a colony of a species that is not associated with living
plants.

462 Tue University SciENcE BULLETIN

11) At a given site, the presence of a colony of P. ferruginea is completely
dependent upon the presence of a living swollen-thorn acacia. Those species
not associated with living plants are not dependent upon the presence of any
particular species of plant for their existence. There are certain species of
plants which characteristically have hollow branches, but these species of
Pseudomyrmex may nest in any of them.

12) While the colony is dependent upon the presence of a swollen-thorn
acacia for its existence, the fact that a mature colony exists and produces
founding queens implies a high probability that there will be seedlings of the
swollen-thorn acacia present for the founding queen to settle in. The pres-
ence of a colony of a species not associated with living plants does not increase
the probability that there will be nest sites available for the founding queens
from that colony.

13) As a species, P. ferruginea has the opportunity to increase the prob-
ability that it will continue to exist by evolving behavioral traits that insure
the continued growth and reproduction of the swollen-thorn acacias on which
it lives. There is little that a species not associated with living plants can do
to insure the existence of sufficient nest sites and food.

14) In man-made disturbance sites, and to some degree in natural dis-
turbance sites, P. feruginea maintains a higher density of colonies, and a
higher density of individuals, per unit volume of plant community than does
any species of Pseudomyrmex not associated with living plants.

15) The founding queen does not have to forage more than a few centi-
meters from the thorn for food for her brood. Since the thorn is on a living
plant, she is insured that there will be food within a short radius. A found-
ing queen of a species not associated with living plants must forage long dis-
tances from her nest site, and is not insured of finding food.

16) Because of the concentration of the colony in a small volume, and the
aggressive behavior of most workers to intrusive organisms, the majority of
the members of the colony are not subject to predation. The workers of a
colony not associated with living plants are preyed upon whenever they leave
the nest.

It should be evident from the above comparisons of P. ferruginea with
other ants in the genus Pseudomyrmex that P. ferruginea is more than just
an incidental tenant in the swollen thorns of A. cornigera and the other
swollen-thorn acacias within its range. Behaviorally, P. ferruginea is highly
adapted to living on swollen-tthorn acacias—so much so that it is not found
elsewhere. However, this is not a one-sided interrelationship. A. cornigera
is likewise highly adapted to the presence of the ant; this aspect of the ant-
acacia interaction is presented in the third section of this paper. It should be
noted that in both the ant and the acacia populations, the most important
changes that are associated with the interaction are primarily behavioral and
physiological, rather than morphological. As such, they are often much less

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 463

clearly distinct from similar but different traits of the ants and acacias that do
not have an interaction with each other. The evolution of this system is dis-
cussed in more detail by Janzen (1966).

FACTORS INFLUENCING THE PARAMETERS OF
ACACIA CORNIGERA POPULATIONS

A swollen-thorn acacia is necessary for the development of a colony of
Pseudomyrmex ferruginea. This is evident from the previous information of
the bionomics of the ant. It is likewise evident that the behavior of P. ferru-
ginea is modified from those species of Pseudomyrmex that are not associated
with a living plant. This modification appears to be in a direction such as to
facilitate the association of P. ferruginea with a swollen-thorn acacia. Earlier
authors (Belt, 1874; Brown, 1960; Schwarz, 1916; Wasmann, 1915, 1916; Sak
ford, 1922; Schimper, 1888) expressed the belief that the ants protect the
acacia (in general reference to New World swollen-thorn acacias and Pseudo-
myrmex). Preliminary observations during the summer of 1962 at Campo
Cotaxtla suggested that some important population parameters (height, con-
dition) of Acacia cornigera are influenced in a positive manner by the pres-
ence of P. ferruginea. Often, this influence is not immediately apparent by
casual cbservation. Therefore, the majority of the experimental work in this
study dealt with attempts to gather data to reject the null hypothesis that the
absence of a colony of P. ferruginea has no effect on the population parame-
ters of A. cornigera.

Other authors (Skwarra, 1934a; Wheeler, 1913, 1942) have expressed the
view that the ants do not protect the acacia. Difficulty undoubtedly arises
over the word “protect.” Most of the confusion has arisen through the failure
to observe specific phytophagous animals and to record their effect on the
shoot, and the effects of the ants on them. An equal confusion has arisen
from the desire of most authors to develop a single hypothesis which covers
all interactions of an apparently mutualistic nature between ants and living
plants. Whether or not a mutualistic relationship exists between a given spe-
cies of ant and plant is completely dependent upon the behavioral and eco-
logical characteristics of the particular ant and plant species. Until many
more ant-plant interactions are investigated in detail, it is premature to at-
tempt general hypotheses that are concerned with the presence or absence of
a mutualistic relationship. Observations made external to the interaction be-
tween P. ferruginea and A. cornigera indicate that the interaction is at least
quantitatively, if not qualitatively, different from many other interactions be-
tween a species of Pseudomyrmex and a species of Acacia. Extreme care must
be used when using the data presented in this paper to develop hypotheses
bearing on the existence of a mutualistic relationship between other species
of ants and plants.

464 Tue Universiry ScIENCE BULLETIN

MATERIALs AND METHODS

Study area. ‘The major part of the experimental data in this paper is from the same site as
the data presented in the sections on Acacia cornigera and Pseudomyrmex ferruginea. The
experimental plots (1963-1964) on the land of Senor Eusavio Farfan (4 miles east of Temascal),
and Juan Torrealva (7 miles east of Temascal), are within 1,000 meters of the highway between
Temascal and La Granja (Fig. 7). This is lowland coastal plain 2-22 km east of the first foot-
hills of the Sierra Madre Oriental, at an elevation of 15-40 meters. Plot B is on the dam fill
and waste land along the west side of Temascal, and therefore in the foothills. The initial
experiments and observations were recorded from the lowland coastal plain immediately
surrounding Campo Cotaxtla. To aid in the comparison with further studies, the following
characteristics of the part of the study area containing the experimental plots are presented.

Soils. Unless otherwise indicated in the individual plot descriptions, all experimental plots
are on black soil with limestone outcrops, or on red to yellow laterite with quartz pebbles.
Nothing is known of the relation of the distribution of A. cornigera to these two soil types prior
to man’s intervention in the area. At present, 4. cornigera is common on all black soil areas
except where the land is burned and/or plowed annually, flooded annually, or covered with
forest that has not been disturbed for 9-20 years. The low laterite hills are covered with oak,
Byrsonima crassipes, Curatela americana, grasses, and other fire-tolerant plants. They are usually
burned every 1-3 years. A. cornigera is absent on these hills but once fires become less frequent
due to conversion of the site to a pasture or corn field, 4A. cornigera becomes common. There
are no obvious major differences between the density or distribution of A. cornigera that are
directly attributable to the difffferent physical or chemical properties of these soils, although
under the same disturbance regime, A. cornigera may not be as common on the drier laterite as
on black soils. However, the growth of other plants on these soils influences their use by man
and this does affect A. cornigera density and distribution. The two soil types have about equal
acreage in the area between Temascal and La Granja.

Precipitation. Rainfall data were obtained from unpublished and published (Comision del
Papaloapan, 1962) reports from the Comision del Papaloapan weather station that is situated
on a hill (elev. 80 m) overlooking Temascal. The amount of rainfall, and its periodicity, recorded
at this station closely approximately that observed in the plots. The same rain squalls that
passed over the plots usually passed over the weather station, although 10-120 minutes later.

At Temascal, the 12 year (1951-1962) mean annual precipitation was 282 cm (range 158 to
372 cm). In 1963, there was 198 cm of precipitation. The first heavy rain in 1963 was on 7
May (66.2 mm). From this date to 25 Sept. 1963, the date of the last heavy rain (25.3 mm),
there was 183 cm of precipitation. From 26 Sept. 1963 to 19 May 1964 there was 6 cm of pre-
cipitation. The first heavy rain was on 20 May 1964 (18.5 mm). From that date to 8 Aug.
1964, there was 129 cm of precipitation.

On the basis of these data, it is easy to define the beginning of the rainy season as the date
of the first heavy rain (7 May 1963 and 20 May 1964). In 1964, there was 3 cm of precipitation
in the 49 days preceeding 20 May. During the 49 days following 19 May, there was 94 cm.
Within two weeks after the beginning of the rainy season in 1964, the soil was soaked, and
patches of ground that had been bare during the latter part of the dry season were covered with
seedling plants. Mature plants were slower to show a reaction and only near the end of June
did most species have branches in full leaf. The general density of insects in the vegetation was
relatively slow to increase and did not appear to reach usual rainy season levels until the end
of June. There were, however, some species which made their appearance in large numbers 1-3
days after the 20th of May. Seeds of A. cornigera began to germinate almost immediately
after the first heavy rain and continued to germinate sporadically for at least two months.
Those mature shoots of A. cornigera that were in flower took as long as two months after the
first heavy rain to initiate lengthening vertical branches. Some sterile shoots initiated vertical
growth anytime after the cool season ended (early March) but most of the shoots initiated
vertical growth in the month preceding, and the six weeks following, the beginning of the
rainy season. Aside from those specific defoliators of A. cornigera which were present through-
out the dry season, (Coxina hadenoides, Aristotelia corallina), insects that eat A. cornigera did
not appear in numbers until the first week after the first heavy rain.

Defining the end of the rainy season (and the beginning of the dry season) is considerably
more difficult. The date of the last heavy rain in 1963 was 25 Sept. (3 cm). This is chosen
as the end of the 1963 rainy season. Thereafter untii 20 May 1964, rainy days were separated
by rainless periods of 1-25 days. When rain did fall there was rarely over 0.5 cm. The reduc-
tion of plant growth rates and insect density during the first part of the dry seacon was intensified
by three months of cold weather which began about 1 Dec. This is regarded as a cool season
(Table 3) and is included in the dry season. When temperatures rose in March, there were
few species of plants which initiated heavy growth. The point at which different plant species

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 465

drop their leaves ranges from early December to the middle of April. The largest portion of
deciduous plants drop their leaves shortly after the end of the cool season. However, this refers
only to those shoots that are over 6-10 months old. Younger regeneration shoots of many
species of trees and shrubs retain their leaves in large part; pastures cleared within the past 6-10
months stand out clearly against the surrounding brownish gray and leafless vegetation.

The shoots of A. cornigera are variable in respect to their reaction to the beginning of the
dry season. Those that are growing in areas where the water table is close to the surface
(swamps, river banks) lose their leaves and cease growth later than do those growing in better
drained sites. In some cases, A. cornigera is sometimes the only plant with green leaves except
for much larger trees. Flowering branches are often initiated during the dry season and
sometimes extended the height of the shoot by 10-30 cm. However, these branches do not
lengthen more than this; the new vertically lengthening sterile branches are initiated from a
thorn axil at the base of or below the flowering branch. Even when the shoot has an un-
damaged shoot tip, its rate of growth during the cool and dry season is somewhat lower than
during the rainy season.

Temperature. Temperature data were obtained from the same source as the rainfall data.
This station records the daily maxima and minima. The minima agreed very closely with those
occasionally recorded in the plots. The maxima were 1-4° C. lower than those recorded in the
plots on those days when the maxima were above about 32° C. It can be seen from Table 3
that there is a readily definable cool season that lasted approximately three months, December
through February. The lowest temperature recorded in the experimental plots during the cool
season was 10.5° C. at 4:00 a.m. on 26 Dec. 1963. The high number of days (40) with a
maximum of 24° C. or less during the cool season was of great significance to A. cornigera and
P. ferruginea. When the maxima are less than 24° C. there are almost no workers active on
the surface of the shoot throughout the entire 24-hour cycle.

Two types of cool weather occurred while the study was in progress. When storms blew
from the Gulf of Mexico, they brought wind and rain with diurnal temperature maxima in the
22-25° C. range. Nocturnal temperatures usually did not drop below 15° C. during these
nortes. During the cool season there were frequent storms with very light rain or drizzle and
overcast skies. During these periods, diurnal maxima ranged from 16 to 23° C. and nocturnal
lows ranged from 10.5-19° C.

On clear days during the warm season, temperatures at sunrise were usually 21-24° C. They
climbed rapidly after sunrise (e.g., from 22° C. at 6:00 a.m. to 30° C. at 9:00 a.m.) and fell
rapidly at sunset (Fig. 32, 33). On overcast days, maximum temperatures were sometimes less
than 4° C. higher than those at dawn. On clear days during August through November, a
heavy morning fog was often present until 7:00-9:00 am. This kept temperatures low until
the sun broke through.

Aside from the records of the weather stations of the Comicion del Papaloapan, all air tem-
peratures were recorded with a Yellow Springs Instrument Company Model 44TD_ telether-
mometer and probes (#402 Small Animal Probe). Over the range of the scale that was used,
these probes read within less than one-half degree C. of each other and a calibrated mercury
thermometer. Unqualified statements about air temperatures refer to that recorded 1 m above
open ground with the probe and lead in full sun and wind for about 30 seconds. When
measuring the temperature inside of small shaded sites (e.g., in thorns), at least 50 cm of the
lead immediately adjacent to the probe was also shaded. The probe was placed in the thorn
by clipping off the point lacking an entrance until the hole was just large enough to admit
the probe. In occupied thorns, the workers chewed the plastic coating off of the probe on
numerous occasions.

Land use. The La Granja-Temascal highway was built about 1950. Before this, the area
was used for milpa agriculture by Indians and contained several small holdings used as brushy
pasture (e.g., Sefior Farfan’s land). Upon opening to settlement and the relocation of eight
Indian villages evacuated from the lake site behind the Presa Miguel Aleman, farming became
more intensive, and grazing pressure by cattle increased. Every year, more land is converted to
sugar cane fields.

Plot B at the base of the earth-rock-fill dam is on sandy soil and in open, heavily grazed and
browsed vegetation. Throughout the study area, 4. cornigera is common on this type of land.

.The experimental plots on Seftor Farfan’s land (F-I, K, N-Q, S-V) are in second growth
natural vegetation which is subject to periodic grazing, clearing and/or burning. Whenever
funds are available, the owner has portions of the brushy pasture cleared. If the litter dries
it is sometimes burned. From 3-36 months may elapse between clearings and/or burnings.
This process results in a closed (lightly grazed) to open (heavily grazed) plant community
composed mostly of annuals and suckers from old perennial root stocks. The annuals set seed
every year but the suckers often require 2-4 years of growth before setting seed.

466 Tue UNiversiry ScIENCE BULLETIN

On Seftor Torrealva’s land (the remaining plots) the vegetation is subject to alteration by
farming and grazing. In general a sector is cut and/or burned in April or early May. It is then
planted to corn without being plowed; certain portions receive beans and squash as well. Fol-
lowing corn harvest, and the later squash and bean harvest, cattle are allowed to graze the area
until the following spring. This cycle is repeated at least three consecutive years and then the
land is left fallow for 2-5 years. This land receives light to very heavy cattle grazing while
fallow. During the fallow period, the vegetation may be cut and/or burned at irregular intervals
to provide new suckers for fodder.

At Campo Cotaxtla, almost all experimental work was with shoots over a year old in
short-grass pasture and brushy pasture.

Vegetation types. The coastal plain immediately east of Temascal is Tropical Tropical Dry
Forest based on climate and altitude (Holdridge, 1964) or Tropical Tropical Evergreen Forest
based on climatic factors and an appreciation of the vegetation itself (Leopold, 1959). The plot
B at Temascal was the only one within eight km of forest over 20 years old. The small amounts
of this older forest remaining are on the black soil and limestone hills, 150-300 m tall, west of
Temascal and bordering the lake behind the Presa Miguel Aleman. The vegetation in and
around all plots was cut and/or burned at least once every 10 years and usually more frequently.
It appears that before man, two plant formations may have been recognizable; one on the low
laterite hills, and the other on the high hills and bottomlands having black soils. No attempt
will be made to characterize these earlier formations. Man’s disturbance has been so great that
it is nearly impossible to say what plants lived in what areas and at what densities. Almost all
of the species of plants in the area appear to have their densities and distributions altered by
man’s disturbance.

At present, the low laterite hills have the following four types of vegetation. (1) Grassland
that is burned annually with herbaceous annuals and scattered fire-resistant trees. (2) Grass
with shrubs 1-2 m tall, and more dense trees that form a nearly closed canopy in some places.
These sites are burned every 2-3 years and are not used for intensive grazing or farming. (3)
Dense shrub layers of Compositae, Leguminosae, Melastomaceae, and other plants with root
stocks resistant to burning and plowing. These are tied together with vines which have
perennial root stocks, and are under scattered oaks, palms, Ateleia pterocarpa, Cochlospermum
vitifolium, Byrsonima crassipes and other trees. This vegetation is burned and cut periodically
for conversion to corn fields, with or without grazing. (4) A closed canopy of Quercus,
Leguminoase, Melastomaceae and other trees with woody vines in the canopy. This forest is
10-18 years old, gradually loses its undershrub layer and is not burned except by very light
ground fire.

These four patterns can be, and are, modified in all directions and to all extremes by
grazing, lumbering, and clearing para ver mas lejos (to improve the view). In type (3),
Acacia cornigera may become very common provided the land lies fallow without burning for
at least three consecutive years during each six years of use. A. cormigera is generally absent
in the other three and this appears to be associated with the frequent burning of types (1) and
(2) and the dense shade of type (4) during the rainy season. Plots D, J, and R are in the
successional stages of type (3).

The areas of black soil with limestone outcrops have received very different treatment in
past years due to their superior value for corn fields and pasture. Even under these conditions
of extreme disturbance there are certain plant species which are noticeably lacking from one
or the other soil types. Furthermore, it appears that a similar disturbance regime does not
always result in the same vegetation physiognomy on the two soils.

The disturbance regime of the vegetation on black soil is so complex and so compounded
by cutting, burning, grazing, and establishment of fallow lands, cornfields, and cane fields, in
every imaginable sequence, that a classification even as crude as that for the laterite soils is
unreasonable at this point.

In ungrazed regeneration following cutting and/or burning, the plants are characteristically
1) uniform in height within the species, 2) of 100 species per acre or more, 3) slender, 4)
sprouted from old rootstocks or are annual herbs and vines, 5) reactive to the dry period by
leaf-drop and cessation of vertical growth, 6) woody, 7) entangled by woody and annual vines,
and 8) with a species density and composition that changes from year to year (succession) and
from area to area (disturbance regime). A. cornigera is often common in such regeneration
as suckers from old root stock. Plots E, C, H, I, K, N, O, P contained vegetation of this type
that has been lightly to moderately grazed until the beginning of the 1964 rainy season when
the cattle were removed.

As grazing pressure increases, the vegetation structure deviates from this description. Re-
generation of natural vegetation during or immediately following heavy cattle or horse grazing
is characterized by 1) lack of uniformity in height within many of the species, 2) reduced

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 467

species composition (as low as 25 species per acre), 3) many plants laterally developed or with
stag-horned life form, 4) annuals or plants grown from old root stock, 5) asynchronous leaf
drop between species, 6) plants with thorny, bitter tasting, hairy, or urticaceous foliage, and/or
stinging ants, 7) being much less tangled with woody vines and almost lacking annual vines,
8) lacking a closed canopy above the grass level, 9) having many grasses and, 10) lacking a
great change in species composition on a yearly basis (succession) and area to area basis (dis-
turbance regime).

The degree of divergence of this type from the one previously described depends for the
most part on the time of year when the last clearing took place, intensity of grazing, and the
time of year the grazing occurred. In this last type, A. cornigera is often very common. Plots
L, M, Q, and S were of this type.

Sugar cane fields, owing to their annual cutting and burning, and the dense shade and root
competition produced by the tall grass formation, are relatively barren of anything other than
root stocks with suppressed shoots, and herbaceous annuals (e.g., Convolvulaceae). Shoot den-
sities of A. cornigera are very low and A. cornigera is usually absent in fields that have been
used for sugar cane for more than five years. When fallow they approximate ungrazed regenera-
tion following cutting and burning for a corn field.

An almost complete disappearance of natural vegetation occurs when the land is com-

pletely cleared and planted to a vigorous grass species. If such a pasture is not burned more
often than biennially, vigorous stands of 4. cornigera may develop and remain. The plots
established in a pasture of this type were accidentally destroyed.

Choice of plots. Plots were generally chosen in areas with moderate to high densities of
A. cornigera. See Figure 7 for the location of those retained in the study. Early in the study,
attempts were made to establish lines of shoots as linear plots. Owing to difficulties in marking
shoots, difficulties in explaining to land-holders what was desired, and a lack of understanding
of the auxiliary-shoot effect, these plots were found to be impractical at the time and were
abandoned. Rectangular plots were then used. Each plot has a single disturbance regime; in
some cases several plots were established in a field that had a uniform disturbance regime
throughout. The various portions of the plot that are altered, treated or used as controls are
designated as “subplots.” Rectangular subplots could be readily outlined, eliminating bias of
which shoots to treat since all A. cornigera in the experimental subplot could be treated equally.
They could be described more readily, and provided large samples without intricate mapping.
As the study progressed, difficulties with the placement and form of these subplots became
evident, such as mapping difficulties, inclusion of nonrepresentative shoots, and trails between
shoots across the subplot borders, but these problems can be avoided. Linear plots were also
used again after the auxiliary-shoot phenomenon was understood (plots B, G, I).

The presence of a population of A. cornigera of an age and form desired for treatment was
the first prerequisite for the choice of a plot. Of equal importance was permission for, and
acknowledgment of the plot’s presence by the owner. Some assurance had to be obtained that
the plot would not be cleared or burned before August 1964. This permission, acknowledgment,
and/or assurance was impossible to obtain for most land in the area between Temascal and La
Granja. In spite of having obtained these assurances, 20 plots were destroyed during the study;
they were not included in the plot description below. The futility of experimenting with single
shoots in small or marginal habitats was associated with these reasons as well as with the
high variation between the shoots in these habitats. However, with this study as background,
it would now be possible to experiment with the 4. cornigera growing in such habitats. The
above mentioned difficulties are in great part responsible for the lack of replication of sub-
plots. However, in most subplots, each shoot was a replicate in itself since a record was kept
of each individual for at least part of the history of the subplot.

Plots were chosen so that they contained approximately even-aged shoots of A. cornigera.
In special cases certain age or height groups were removed or not recorded in order to create an
even-aged stand. These cases are so indicated in the plot descriptions. Care was taken to
choose plots that included an area with only one history of disturbance. This history was deter-
mined by owner opinions, extrapolating from nearby vegetation with a known history, and
the number of annual growth rings in the trunks of 4. cornigera. The presence of fields,
pastures, and roadsides with uniform favorable histories of disturbances has resulted in much
larger areas with uniform populations of 4. cornigera and P. ferruginea than could have existed
before man’s presence, and this has simplified experimentation with these two organisms.

Treatments. Insecticide. At Campo Cotaxtla in July 1962, it was found that two methyl-
parathion sprays applied a month apart were effective in removing colonies of P. ferruginea from
2-4 year old queen-shoots; upon re-examination of these shoots in September 1963, it was
found that 18 of the 23 treated shoots completely lacked workers. The spray used was a 0.03
percent emulsion of 50 percent actual, liquid concentrate Beyer Leverkusen methyl-parathion.

468 Tue University ScIENCE BULLETIN

At Temascal, a 0.04 percent emulsion of the same insecticide was used. It was applied with
a back-pack sprayer with a 2 m extension boom until the foliage and bark were wet. Care
was taken to spray both sides of shoots. A “Spraying Systems C. Tee Jet TG-5”’ medium fine
spray nozzle was used. This application required approximately 35 cc per 30 cm shoot. 400 cc
per 160 cm shoot, and 1,000 cc per 400 cm shoot (determined by matching mean volumes
against mean heights in three subplots). At least 50 percent of the spray mixture fell directly
on the foliage of other plants but almost none fell directly on the ground, except when other
vegetation had been cleared. Spraying was done when the wind speed was 2 mph or less in
order to minimize drift onto control subplots and reduce foliar contamination of other plants.
Shoots were not sprayed when wet with rain or dew. This meant that spraying was almost
always done from 8:00-10:30 a.m. and 3:00-6:00 p.m.

The original schedule was two applications about a month apart on each treatment subplot
during the fall of 1963. The fisrt application was expected ‘to remove most of the adults and
larvae while the second would kill those workers that had hatched from unaffected pupae dur-
ing the intervening month. In order to avoid mortality to defoliators at the time when most
of the shoot growth occurs, no applications were planned for the 1964 rainy season. After this
schedule had been followed for most of the treatment subplots (through Dec.), it was decided
that a more satisfactory technique should be used. This was for the following reasons. 1)
Sprayed shoots were still available for founding queens and these would have small but active
new colonies by June of 1964. 2) Unoccupied shoots in the treatment subplots frequently were
invaded and became auxiliary-shoots of large colonies along the plot margins. 3) A few
workers (5-100) sometimes remained for a month or more after the last spray application and
could affect defoliating insects coming to the shoots. 4) In one case an entire queen-unit moved
into the treatment subplot and into a shoot emptied by methyl-parathion mortality. 5) Spray-
ing removed phytophagous insects for an undetermined period, and it was found that the
feeding of even one or two such insects could have a major effect on the population of A.
cornigera. 6) In some cases there was evidence that the methyl-parathion stimulated growth
of A. cornigera for a short period of time after each application.

After a shoot was sprayed, the workers were often seen turning in tight circles on the sur-
face and then falling off the shoot. These workers cease movement in 5-10 minutes. Larvae that
were dead due to contamination from workers were thrown from the shoot in large numbers.
Workers appeared not to die in the thorns but it is more likely that they were thrown from
the thorns like the dying larvae. The queen was usually killed within three days after the
first treatment. The dead queen was not pulled out of the thorn.

Two examples of the effectiveness of treatment with methyl-parathion are as follows. 1)
Nine days after the second spray in treatment subplot SA-1 (later abandoned), 2,105 type A
and B thorns from the area of brood thorns of 17 shoots were opened. Of these thorns, 43
sull had unstarved brood and workers, 351 had starved or moldy brood, and the remainder
had workers only or were empty. Before the first spraying there were probably brood in at least
90 percent of these thorns. By July 1964, the shoots along the north margin of subplot SA-1
appeared to be fully occupied as auxiliary-shoots of the colonies in control subplot SA-2. 2) On
21 Jan. 1964, 61 days after the second spraying of subplot U-1, 175 workers were counted on
the outside of the 86 shoots in the treatment subplot U-1 and 2,219 workers were counted on
the outside of the 93 shoots in the control subplot U-2.

Two days after a single spray application, ten unoccupied shoots (20-40 cm tall) had 21
percent of the thorns tenanted by founding queens. However, four days later, ten more shoots
from the same site had 87 percent of the thorns occupied by founding queens. This latter level
is about the usual frequency of founding queens. Since the first queen to occupy a thorn on an
unoccupied shoot is not necessarily the one to develop the owner colony, spraying to prevent
colony foundation is a useless procedure unless done so frequently that it would probably have
devastating effects on the other insect populations. In addition, such a treatment could have
major direct effects on the growth rates of the shoots.

The effect of the spray application on phytophagous insects could be very important but was
not of great concern for the following reasons. 1) The subplots were so small that insects
could rapidly reinvade them. 2) It was expected that the major part of the defoliation would
take place during the first three months of the 1964 rainy season rather than immediately fol-
lowing the removal of P. ferruginea at the end of the 1963 rainy season. 3) It was hoped that
methyl-parathion, being a compound with a short half-life, would be gone within a month after
the last application.

With the exception of Syssphinx mexicana, all of the insects listed in Table 1 are killed by
an application of methyl-parathion. This was determined by finding dead individuals on the
shoot and on the ground below the shoot. It is not known how long after an application the
residual affects the insects that land on the shoot. Dead insects are commonly found under

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 469

the sprayed shoots only during the first 2-4 days after the application. Severe defoliation may
take place on the first day following the application (and on any later date) but it is not known
how many of the defoliating agents are killed. Within two weeks after an application, the
numbers of phytophagous insects on the shoots were about as high as they were at any time.
In both control and treatment subplots, almost all phytophagous insects disappeared during
the Jan.-May cool and dry season. Of those insects that feed commonly on unoccupied A.
cornigera, only Coxina hadenoides, Coscinoptera mucida, Aristotelia corallina, and Mozena
tomentosa remained active at this time. No insecticides were used after the 15th of December.

Methyl-parathion stimulation. In plots with shoots 2-4 m tall, the second parathion appli-
cation was found to stimulate growth of the upper axillary buds. For example, for 2-6 weeks
following the second application (4 Dec. 1963), lengthening branches were evident from the
upper thorn axils in treatment subplot S-1. This was at a time (Dec. and Jan.) when the
shoots in the control subplot showed almost no new growth. Along the north side of S-1, the
treated shoots were less than 1 m from the untreated shoots. This new growth may have been
in part due to insecticide mortality of phytophagous insects. In the control subplot (S-2),
katydids (Tettigoniidae) and the larvae of Coxina hadenoides were responsible for the lack
of new growth (in addition to physiological slowing of growth due to the cool weather). These
insects are very mobile. Both of these insects were present at the time the new growth was
being produced. Due to the cool weather, there was little activity outside of the thorns at this
time and these insects were also present in the control subplots. Their characteristic feeding
damage was abundant on these new branches. It appears that the stimulation of growth was
sufficiently great that these insects were not able to suppress it for a 2-6 week period. Eventually
all of the new branches had their shoot tips eaten and no further stimulation took place. The
new growth was of a type not normally encountered on insolated shoots. The branches had long
internodes and thin thorns; there were 3-15 such branches per shoot. In treatment subplot S-1,
the stimulation caused an increase in mean height of 21.5 cm (difference between the growth
in the treatment and control subplot). It was in S-1 that the greatest stimulation took place.
No attempt has been made to extract this height increase from the mean values obtained for any
of the treatment subplots.

Clipping thorns. In December 1963, it was realized that the distal portions of a swollen thorn
are dead tissue and therefore can be removed without damaging the shoot or stimulating it
(Fig. 41a). There is ample evidence from natural damage that removal of the distal 4-%4 of
a green but stiff thorn does not affect its leaf. When clipping thorns with pruning shears, care
must be taken not to rupture the junction between the thorn and the bark. Preliminary trials
showed that if all the thorns on a shoot are clipped off about the one-half way between the point
and the base, almost the entire colony is destroyed or lost through disorganization. On these
shoots, the surviving workers and brood concentrated themselves in the thorns produced during
the next 2-4 weeks and could be clipped off. Searching queens do not establish colonies in the
clipped thorns. There is no leaf drop or color change on these shoots, and no stimulation of
growth in dormant shoots. These shoots are not acceptable as auxiliary-shoots until they have
grown new thorns. In order to avoid this problem, the acacia plants were removed whenever
possible around the new subplots for a distance of 2-3 m.

In four different cases, raiding columns of Nomamyrmex esenbecki went through the
subplots within a week after clipping and were partly responsible for removal of the ants. In
two cases, swarm raids of Labidus praedator did the same thing. When a thorn was clipped, the
brood was scattered on the ground, and Solenopsis geminata and other ants were generally
responsible for its loss. Rain drowned the workers and brood in the opened thorns.

While clipping thorns was time consuming, it was also very effective. A number of sub-
plots with clipped thorns were completely free of workers by June 1964.

Recolonization of treated plots. Ideally, the treatment subplots should contain only un-
occupied shoots and the control subplots only occupied shoots. In actuality, at the last recording
only treatment subplots C-1, E-2, H-1, K-1, L-1, M-1, N-1, O-1, and Q-1 contained no oc-
cupied shoots; thorn clipping had been used in all of these. The other treatment subplots had
various numbers of occupied shoots in them due to invasion from colonies outside the sub-
plot, or due to young colonies developed in situ. In the control plots, the percentage of oc-
cupied shoots ranged from 80-98 percent at the last recordings made between 26 Jul. and 8 Aug.
1964.

Most of the subplots were mapped and individual shoot records kept for most of the sub-
plots’ histories. This means that in addition to the possible comparisons between subplots, the
occupied shoots in the plot can be contrasted with the unoccupied shoots in the plot.

Invasion from outside of the treatment subplot can be prevented by removing the shoots of
A. cornigera for a distance of 5-10 m around the subplot. This cannot be done to control
subplots because of the likelihood of removing the queen-shoot that is associated with auxiliary-

Tue University SciENcE BULLETIN

*

“% ep

Fic. 41. a. The canopy of a shoot of Acacia cornigera in treatment subplot N-1. The thorns
were clipped on 3 Apr. 1964. Five males of Anomoea sp., probably A. rufifrons, can be seen
near the end of the right- hand branch. This shoot would be rated as a number 2.5. Photo 25
Jul. 1964. b. The canopy of a shoot of Acacia cormigera in treatment subplot S-1. The shoot has
not been occupied since late in 1963. This shoot would be rated as a number 2.5. Photo late
July 1964.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 471

shoots within the subplot. The fact that treatment subplots E-2, H-1, L-1, M-1l, and Q-1 were
free of workers at the last recording is due to this type of shoot removal. In treatment subplot
N-1I and O-1, in order to insure that the shoots with clipped thorns stayed free of workers,
all shoots less than 50 cm tall were removed. This removed the founding queens and small
colonies living in these suppressed shoots.

Removal of vegetation adjacent to A. cornigera. The amount of surrounding vegetation ad-
jacent to the shoot may influence the amount of phytophagous insect damage or cattle grazing
on A. cornigera. Several subplots were established in which all vegetation besides A. cornigera
was cut to ground and the shoots of A. cornigera that were less than 100 cm tall were re-
moved. The cut vegetation was not burned, but it is doubtful that it had a substantial effect
as a fertilizer during the experiments. This technique approximates normal conditions since
in vegetation that is being cleared by hand for pasture, the shoots of A. cormgera that are over
3 m tall are sometimes not cut because of the shower of ants which falls when the trunk is hit
with a machete.

Clearing of all vegetation. Several subplots were established by cutting all the vegetation
to ground level but not burning the litter. Where this vegetation was over | year old, the
stumps of A. cornigera were cut at 65 cm to facilitate finding them in the future regeneration.
Since the new suckers often started at some point above the base on the trunk, the lengths of
the shoots as well as their heights were recorded.

When all the vegetation is cleared and the shoots of A. cornigera contain large colonies, a
third type of treatment subplot is made. In the treatment subplot, the canopies of the 4A.
cornigera shoots are piled and burned. Any colonies that are to occupy the suckers from these
unoccupied stumps must develop in situ, or move in from outside the subplot. In the control
subplot, the canopy is placed across its cut stump. The workers patrol the stump and as the
new suckers appear (within 3-10 days) they are immediately occupied by the colony. Auxiliary-
units transfer into the new suckers in the same manner as do queen-units. However, if the
stump is very slow to produce new suckers, the auxiliary-unit may retract into the queen-unit
or move to a different stump.

Enclosure in barbed wire. Subplots enclosed in barbed wire were intended as contrasts to
subplots to which cattle had access. Three strands of wire were used. The enclosures were in
general too small (6 by 6 m or 6 by 12 m) since cattle could, and did reach in at least 1 m to
browse. Final recordings were not made in these plots since the cattle were removed from the
control subplots by their owner before the rainy season began.

Adding units. A mature colony may be added to an unoccupied shoot by cutting an isolated
queen-shoot with a large worker force and placing it in contact with the shoot. Cutting an
auxiliary-shoot and placing it on the unoccupied shoot will provide occupation for 1-2 months
but eventually the worker force 1s lost.

Multiple treatment. In several cases, it was necessary or of interest to use more than one
of the above treatments on a subplot. The effect of the presence or absence of a worker force
is of central interest in this study, rather than the effect of the treatment on the shoot. Because
of this, any treatment, or set of treatments that would insure the removal of the worker force
without damaging the shoot or phytophagous insect population was used.

Recording data. Shoot identification. The individual subplots were the original experimental
unit. However, it soon became evident that some method of individual shoot recognition was
necessary. This was because of the lack of subplot replicates, the variation in shoot height
among even-aged shoots within a subplot, the presence of occupied shoots in treatment sub-
plots, and the unoccupied shoots in control subplots. Placing individual numbers on the trunks
was too slow and tags were stolen. Mapping the location of the shoots proved very successful.
It simplified the location of the shoots and insured that all the shoots were counted in each
recording. The majority of the subplots were mapped by January or February. The consequence
of mapping is that each subplot can be treated as a set of replicates. In later analyses each
shoot can be regarded as an independent unit which is or is not occupied. Due to the presence of
auxiliary-units, all occupied shoots were actually not independent of each other, but time was
not available to determine which shoot had which kind of units.

Height. The height of the shoot is the distance above the ground of the highest node. This
was measured to the nearest cm in subplots with most of the shoots less than 2 m tall and
to the nearest 5 cm in subplots with most of the shoots over 2 m tall. When regeneration
from experimentally cut stumps was measured, the length of the tallest shoot from the stump
was recorded as well as the shoot height. When it was obvious that two or more major shoots
had grown from a single stump, only the tallest was recorded. Such paired shoots are always
occupied by the same colony.

Height of surrounding vegetation. Because of the detrimental effect that shading has on
A. cornigera, the height of the immediately surrounding vegetation was often recorded in

472 Tue Universiry ScrENcE BULLETIN

ungrazed or in lightly grazed subplots. In heavily grazed subplots, there was essentially no
shading of any shoots. When the height of the individual shoot was measured, the height of
the nearest other species of plant that was not damaged by P. ferruginea was recorded. The
mean of these measurements serves as an indicator of general canopy height in the subplot.
When this mean is subtracted from the mean height of the shoots of 4. cornigera, the amount
of emergence of 4. cornigera can be obtained.

Condition of the shoot. Since the degree of damage to the uppermost growing points or
highest nodes is a measure of potential height increment within a plot, the condition of these
points was recorded. The following numerical scheme was used, grading from a perfect terminal
to a dead shoot.

(4) The point to which the height is measured in an undamaged shoot tip, and the mature
foliage shows little or no animal or fungal damage. Fig. 42, 43, 44.

(3.5) There are two main shoots of subequal height from the same stump, one of which is
a #3 and the other a #4.

(3) The shoot tip at the uppermost node has been destroyed and any new shoot tip on the
shoot has not yet attained this height. These shoots have little or no damage to their mature
leaves and have growing points lower down on the shoot. Fig. 42, 43, 45, 46.

(2.5) Both the uppermost growing point and lateral growing points are destroyed, and the
mature foliage shows 25-50 percent removal. Fig. 40, 41, 47.

(2) The shoot is like #2.5, but 50-75 percent of the mature foliage has been removed.

(1.5) The shoot looks dead but still has a green layer under the bark and perhaps one or
two leaves. Fig. 48.

(1) The shoot is dead but still standing.

(0) The shoot is gone or at least has fallen over. Normal leaf drop is not regarded as
damage. While it would be difficult for an untrained observer using this scheme to arrive at
the same subplot numerical evaluation of condition as would the author, it is felt that evaluation
has been consistent throughout the study. All subplot evaluation was done by the author.

Causal agent of condition. The specific feeding damage of Coxina hadenoides, leaf rollers,
webbing larvae, Syssphinx mexicana, Mozena tomentosa, Pelidnota punctulata, Onicideres
poecila, Lochmaeocles cornuticeps, Chrysomelidae, cattle, rodents, and birds can be recognized
and at times during the spring of 1964 was recorded individually. The damage of Coxina
hadenoides, kathydids (Tettygoniidae), and some Chrysomelidae is easily confused and there-
fore was recorded as one.

Presence of foreign objects. The presence of vines and animals on the shoots was usually
recorded. They were not collected within the subplot since a single organism can have a great
effect on growth.

Presence of P. ferruginea. To obtain maximum assurance that shoots recorded as un-
occupied did not have workers in the thorns, the subplots were recorded only on days when
the maximum temperature was over 29° C. If possible, records were made when the tempera-
ture was over 32° C. The presence or absence of workers was recorded, or else the actual num-
ber on the shoot counted and recorded. During the counting process, the shoot was not dis-
turbed. The presence of entrance holes in the thorns does not insure that there is a worker
force present in the shoot. The entrances in new green thorns may be made by founding queens.
The presence of many terminal thorns without entrances usually indicates the lack of a worker
force. Since Beltian bodies were commonly removed by Pseudomyrmex gracilis mexicana and
Solenopsis geminata, the absence of these structures does not insure that there is a worker
force present. The larger entrance holes of P. gracilis mexicana can often be located by inspection
and if there are more than 2-3 workers of this ant seen on the shoot surface, there is not likely
to be a worker force of P. ferruginea in the thorns.

Ideal plot. Because of mistakes made in plot establishment, it seems advisable to clarify
what plan should be followed to establish plots of maximum usefulness. Some of the most
informative rectangular plots established in this study were those in which all of the vegetation
was cleared to ground level, the canopies of 4A. cornigera were piled and burned in the treat-
ment subplot, and the canopies were left next to the stump in the control subplot so that the
colony could invade the new suckers (e.g., plots A, E, H). This was done in plots that had
been regenerating for more than a year. To obtain the maximum amount of information this
plot should be split into at least three subplots. One subplot is the control, one has the canopies
removed, and one has the thorns of the new growth clipped as well as the canopies removed.
The ideal plot should be replicated during the period of the experiment. The subplots should be
at least 20 m apart and have at least 100 stumps in each. The subplots should be cut out of a
relatively uniform stand of vegetation, must be mapped, and must have all the 4. cornigera re-
moved to a distance of 10 m around each treatment subplot. This cannot be done around the
control subplot due to the danger of removing a queen-unit of one of the auxiliary-units in the

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 473

/
hada

al btiewe

Fic. 42. The upper left-hand shoot tip of Acacia cornigera is undamaged and would be
rated as a number 4. The lower right-hand shoot tip was damaged by Coxina hadenoides and
would be rated as a number 3. Photo late July in plot P.

subplot. The stump diameter should be recorded as an index to the size of the parent root
stock. It is best not to set up such a plot during the cool season because transfer of the colonies
in the control subplot is slowed at this time.

At least once every two months, the heights and lengths of the sucker shoots, the height
of the surrounding vegetation, the condition of the shoots, and the presence of other organisms
on the shoots should be recorded. The newly produced thorns in the second treatment subplot

474 Tue Universiry SciENCE BULLETIN

METRIC I 2 3 BP aoe,
Ly

Fic. 43. The left-hand shoot tip of Acacia cornigera was damaged by Pelidnota punctulata
and would be rated as a number 3. The right-hand shoot tip is undamaged and would be
rated as a number 4. Photo late July in plot M.

should be clipped about every four months. About every two months, and at each major change
in the weather, a nocutrnal and diurnal recording of the presence of phytophagous insects on
the shoots should be made. Specimens can be collected from one of the replicates for identifica-
tion. From this same replicate, samples of entire colonies can be made to determine colony
growth rates and sizes in the first treatment subplot. If possible, cattle should be allowed to
graze heavily in one of the replicates and be excluded from another. The more frequently and
closely the shoots are examined for phytophagous animals and their damage, the more informa-
tion will be obtained. Some insects are present only during very specific times of the day, and
for a very few days.

After about 1% days of such a schedule, nearly the maximum amount of information avail-
able, as to the effect of P. ferruginea on A. cornigera, will have been obtained. In the first
treatment subplot, those shoots that survived long enough to produce colonies will be growing
at the same rate as those in the control subplot. The shoots in the second treatment subplot
will probably all be dead.

In natural disturbance sites such as arroyos and river banks, the shoots of A. cornigera are
often highly dispersed. In these cases, linear plots yield the most information (e.g., plot B). Such
a plot consists of a long row of unevenly spaced shoots, some of which are treated and some
of which serve as controls. The treatment, mapping and recording of data should be as in the
rectangular plot discussed above. Such a linear plot may be several kms or more in length, and
pass through a large number of microhabitats. It seems preferable to alternate the treatments
from stump to stump rather than from one section to another of the plot. Therefore it is as if
the three subplots are superimposed. Rather than use a strictly random method to determine
what treatment an individual stump will receive, an attempt, where possible, should be made
to have an even alternation between controls and treatments throughout the length of the plot.

In linear plots, the decision as to what shoots can be used for controls, and what shoots
can be used for treatments is influenced by the presence of auxiliary-shoots. No treatment shoot
should be within 15 m of a control shoot, or a shoot not included within the plot (these latter
shoots can be completely removed). When the canopy of the treatment shoot is removed, it
should be totally destroyed or moved to a distance at least 20 m from any other treatment shoot.
The shoots within a cluster of shoots (e.g., 4 auxiliary-shoots and 1 queen-shoot), because of

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 475

f yr

Fic. 44. The canopy of an occupied shoot of 4. cornigera in control subplot N-2. This
shoot would be rated as a number 4. Photo late July 1964.

,

aS

Fic. 45. The canopy of an unoccupied shoot of A. cornigera in treatment subplot N-1.
This shoot would be rated as a number 3. Note that there are at least four number 3 shoot
tips in the top of the canopy. This damage was caused by Coxina hadenoides and adult tetti-
gonud grasshoppers. The ants were removed by parathion on 12 Oct. 1963. The shoot was a
number 4 at that time. Photo 28 Noy. 1963.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 477

Fic. 46. A shoot tip of Acacia cornigera during the cool season. The shoot’s colony of P.
ferruginea was removed with parathion in late 1963. This shoot tip was a number 4 until a
black sooty mold began to grow on the sticky material that covers the shoot tip. This mold
killed the shoot tip. It is now a number 3. The webbing is that of a spider from the right-
hand curled and stunted leaf. Photo 23 Dec. 1963 in treatment subplot N-1.

Fic. 47. A naturally unoccupied shoot of Acacia cornigera growing in plot B. This shoot
was apparently never occupied by P. ferruginea. It was occupied by a large colony of Cremato-
gaster sp. at the time the photograph was taken. Note the dead vines in the canopy and the

lack of mature leaves. This shoot would have been rated a number 2.5. Photograph taken in
late July 1964.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 479

)

BA So ac Ra Sea tea suse aca a Sete it ie Ss

t

Fic. 48. Two naturally unoccupied shoots of Acacia cornigera. They were rated as number
1.5. They were growing in treatment subplot S-1 and are of the size often eaten by Sigmodon
hispidus during the dry season. Photo late July 1964.

their proximity to each other, must all be controls or all be treated. However, each shoot can
be counted as a single shoot. The occupant ant colony must be identified (P. ferruginea or
P. nigrocincta in the Temascal area).

A linear plot of this type should include as many plants as time permits because 5-20 percent
of the shoots are invariably destroyed by someone using the cut canopies from the controls for
fencing, or invasion of the treatment shoots by an unnoticed colony, or a colony from an un-
destroyed but removed canopy.

Statistical treatment of data. The statistical tests in this study involved the contrasts of two

population means (X) for which a sample standard deviation (s.d.) or variance (s”) was

480) Tue University ScieNcE BULLETIN

calculated. In this case, the null hypothesis was that the mean of the first population (Ua) was
equal to the mean of the second population (Up), or Ho: Ua=Up. If accepted, this hypothesis
would indicate that P. ferruginea has no effect on A. cornigera. The sample means used were

those of the control subplot or the group of occupied shoots (Xa), and those of the treatment

subplot or the group of unoccupied shoots (X»). Xa was chosen to represent the occupied shoots
since it was expected that these means would be larger.

The statistics used was 7=(Xee— i) with the value for rejection of the null hypothesis

being Z greater than 1.645 (significant difference between the means p<0.05) and Z greater
than 2.326 (highly significant difference between the means p<0.01).

The use of the Z test of significance between the two means required that the variables
(individual height or length increment, and condition) be independent observations. This was
assumed for this study, although in reality the presence of the auxiliary-shoot effect means that
the height or length increment, and the condition value, of an individual shoot is to some degree
correlated with the number of ants on neighboring shoots. Queen-shoots usually have larger
colonies and consequently show large increments in height or length; large queen-units usually
have large auxiliary-units, which also have large increments in height or length.

The use of the Z test of significance did not require that the variables be distributed normally.
However, most of the sets of variables recorded as height or length increment, and condition,
were not too badly skewed. This was especially true when groups of continuously occupied
shoots were compared with groups of continuously unoccupied shoots.

The Z test of significance is usually used when the variances of the two populations are
known (they need not be equal). For the purposes of this study, it was assumed that the
sample variances obtained approximated the actual population variance. It is generally con-
sidered that this assumption is justified when the sample size is over 30 (Nissen-Meyer, personal
communication). In most of the comparisons in this study, this was the case. However, the Z
test was also used when the sample size was less than 30 for two reasons. One reason was for
consistency. The other was that, on the basis of many observations made inside and outside
of the experimental plots, it was felt that the variances recorded for most of the smaller sample
sizes would not have been greatly changed had the samples been larger.

Negative Z values were occasionally obtained in both the contrasts of height increment and
condition. Occasionally, these values were low enough to be significant. The reasons for these
significant negative Z values are the same as those for the low Z values that were not significant.
The major reason was the presence of occupied shoots in the treatment subplots, and occasional
unoccupied shoots in the control subplots. Of secondary importance was the production of short
vegetative or flowering branches following the stimulation of growth by parathion or insect
damage. These secondary effects were of only temporary importance and disappeared before
the 1964 rainy season began. The primary reason for significant negative Z values can hardly
be used as evidence against a positive effect of the ants on the acacia; it must be regarded as a
product of faulty experimental design.

It should be noted that of the Xo and Xine columns in Table 4 through 13, the Xen values

are those used in the tests of significance. The Xo column contains the mean heights at the
beginning of the interval over which the increment occurred.

PLOT DESCRIPTIONS

The following plot and subplot descriptions are presented as documentary materials, and to
enable comparisons of the experiments in this study with similar experiments in other studies.
The descriptions are arranged beginning at Temascal and progressing to the east. At each
general area they are arranged by the age of the occupied shoots of 4. cornigera at the time of
the first treatment. During the course of the experiments, 70 subplots were established and 20
of these were destroyed by burning, dynamiting, clearing, or changes in management plans by
the owner. Those remaining are plotted in figure 7. The descriptions of some subplots are
included even though they were destroyed since they aid in comparisons between other subplots.
All cattle were removed by the owner from plots C, E-I, K, N-Q, and S-V, after 3 May so
that the barbed wire cattle exclosures near these plots were rendered useless in respect to
examining the effect of cattle on A. cornigera. The descriptions of these exclosures are omitted.

The following data are presented for each plot: exact locality, photograph if available, purpose
of plot and subplots, history and size, vegetation physiognomy, plant species composition, repre-

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 481

sentativeness of the plot, characteristics of A. cornigera and P. ferruginea before treatment,
kind of treatment, characteristics of A. cornigera and P. ferruginea after treatment, animals
present in the plot, and comments.

B

Plot B was a linear plot extending 2.5 km south from the north end of the Presa Miguel
Aleman along its eastern base (Fig. 49). It was 2.5 km northwest of Temascal. The treatment
subplot (B-1) was superimposed on the control subplot (B-2) because the treated shoots alter-
nated with control shoots.

Plot B was established later than most plots (24 Apr.). It was intended as a plot of sucker
regeneration in which the individual treated shoots would be free from reinvasion. The site
was chosen because the open plant community insured that all shoots of A. cornigera would be
found, and because of the presence of widely spaced 2-5 year old shoots with large colonies.
When the shoots were cut, the canopies of the stumps that were to be ant-free were carried at
least 15 meters from their stumps. They should have been burned. The lack of living shoots in
the areas where the canopies were thrown caused the workers to wander very long distances
in search of A. cornigera. By 6 Aug., 19 of the 53 units removed from the treated stumps had
found their way back to the stump.

The site of plot B was freshly quarried limestone and sand in 1958. Examinations of annual
rings of the shoots of A. cornigera indicated that all the acacias on the site were seedlings and
the oldest germinated in 1959. The site had never been cleared except for random cutting of
fence poles. It was grazed by cattle and goats since 1958. The area from which treated and
control shoots were chosen was about 7 acres in size.

In general appearance, the vegetation formed an open plant community. Patches of trees
and shrubs 2-5 m in height and 8-15 m in diameter were scattered over relatively bare ground
with sparse grasses and broadleafed herbs. There were few woody vines in the patches of
trees but during the rainy season herbaceous vines covered large areas. Plot B was margined
on the west by piled bare limestone and on the east by dense second growth forest with a
canopy 5-15 m high.

Fic. 49. Looking north through plot B. The shoots were cut along both sides of the trail
in the center of the photograph. The tree on the far left with a single upper branch fork is
Tabebuia pentaphylla, the large tree to the right of it is Ateleza pterocarpa, and the tall tree on
the right-hand side of the trail is Acacia cornigera. The low tree below it is Muntingia calabura.
The A. cornigera is about 9 m tall and occupied by a colony of P. ferruginea with about 15,000
workers in it. Shoots of this type were cut to obtain the sucker regeneration in plot B. Photo
April 1964.

482 Tue University ScrENcE BULLETIN

The plants growing in plot B were characteristic of natural and man-made disturbance sites
in the Temascal area. The trees were Muntingia calabura (3-5 m), Cecropia obtusifolia (5-10
m), Ateleia pterocarpa (5-7 m), Pterocarpus sp. (5-7 m), Acacia cornigera (3-8 m) and Acacia
chiapensis (3-5 m). The shrubs were seedlings of the above species of trees plus Calliandra
houstoniana, Conostegia jalapensis, Bauhinia ungulata, woody Compositae, and Leguminosae.

Before any shoots were cut in plot B, there were four shoots with the thorns inhabited by
a colony of Crematogaster sp. and/or Pseudomyrmex gracilis mexicana; these shoots were not
included in the plot as they were nearly dead. Five shoots were occupied by large colonies of
Pseudomyrmex nigrocincta; these were treated as if they were colonies of P. ferruginea, even
though they were not quite as effective as was P. ferruginea in patrolling the shoot. The colonies
of P. ferruginea in the plot were large and all over two years old. The shoots of A. cornigera
were sufficiently scattered so that about one-half of the shoots had queen-units in them. Most
of the shoots of A. cornigera bore flowers or fruit. In many cases, the shoots were emergent
through the thin canopy of Muntingia calabura and the shade was sufficient to cause the sucker
shoots from the cut stump to take on a slightly attenuated life-form.

The shoots of A. cornigera were cut at 65 cm and the canopies of 53 removed and the
canopies of 47 left across their stumps. All of the 100 stumps initiated regenerating suckers.
During the following weeks of observation there were some occupied suckers in the treatment
subplot owing to reinvasion by the ants and some unoccupied suckers in the control subplot
owing to the death of one colony and the loss of several canopies to persons seeking fencing
material.

During the period of the experiment, cattle and goats grazed and browsed the area on a
daily basis. Except for the various species of Dzplotaxis, the insects usually found feeding on
occupied and unoccupied A. cornigera were all observed in plot B. Coxina hadenoides was
especially abundant and an unoccupied shoot rarely passed more than one night without having
the shoot tips eaten. Sigmodon hispidus damage was observed in this plot.

There was almost no seedling reproduction of A. cornigera in plot B. Occasional seedlings
were found under Muntingia calabura where they had been dropped by birds eating the fruits
of M. calabura. On one occasion in plot B a black squirrel was observed eating the undamaged
shoot tips of an unoccupied 2 m tall shoot.

KAR KeN Ome

These five plots were established in young regeneration in the brushy pasture 30-200 m
north of the Temascal-La Granja highway and 5-200 m west of El Mocho’s house on the land
of Sefior Eusavio Farfan. El Mocho’s house is 5 km east of Temascal. Within each plot, rec-
tangular treatment and control subplots were established.

It was originally anticipated that these five plots would be lightly grazed throughout the
experiment; however, the cattle were removed by 3 May 1964 and therefore the major portion
of the growth took place in the absence of cattle. Plots KA and K were intended for
examination of the growth from vegetation cleared to ground level but not burned; un-
fortunately it was not practical to map them with the amount of time available. Plots N and O
were intended as a contrast to plot P; in the former plot, the surrounding vegeation was
cleared but at first all shoots of A. cornigera were not cut (Fig. 50) and later only those that
were occupied (taller than 50 cm) were not cut. In plot P no vegetation was cut.

The regeneration in these five plots was part of a cycle of cutting and browsing and oc-
casional burning to produce more browse. It was last cut to the ground in January 1963 but it
was not burned at that time. The previous vegetation was cut in August 1962 and again not
burned. No cattle were present from June 1963 to 3 Dec. 1963; they were then removed on
3 May 1964. The land was cleared of forest in 1940. It had been occasionally used for corn
fields before 1956. The areas in square meters of the various subplots were as follows: KA-1,
107; KA-2, 90; KA-3, 108; K-1, 102; K-2, 108; N-1, 258; N-2, 212; O-1, 197; O-2, 183; P-1,
445; and P-2, 222.

In general appearance, the area containing the five plots was a dense stand of herbs, shrubs
and tree suckers from old rootstocks in September 1963. The canopy was about 75 cm above
the ground and densely interlaced with herbaceous and woody vines. Projecting through this
rather even canopy were scattered 1-2 m tall occupied shoots of A. cornigera. Below the canopy
were much more numerous heavily shaded and unoccupied shoots of 4. cornigera. During the
experimental period, the vegetation that had been cut in plots KA, K, N and O regenerated
a plant community with the same life form. During the period that cattle were present, nearly
all herbaceous vines were eaten and the shrubs were lightly browsed. Along with the leaf drop
during the dry season, this activity opened the canopy, but it filled again when the rains started
and the cattle were absent. By August 1964, the general canopy in plot P was about 180 cm

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 483

———— " 1s een

Fic. 50. General aspect of treatment subplot N-1 immediately after the general vegetation
was cleared in October 1963. The plant in front of the sheet is Acacia cornigera and is occupied
by a colony of P. ferruginea with about 1,500 workers in it. The smaller shoot to the left is
an auxiliary-shoot to the taller queen-shoot. The cluster of shoots in the right-hand side of the
photograph is a queen-shoot with four auxiliary-shoots.

tall and the heavily occupied shoots of A. cornigera were still emergent by 25-150 cm. There
were a few old trees 4-8 m tall in the five plots but they did not provide continual shade for
any portion of the plots.

The outstanding trees were Tabebuia pentaphylla, Cordia alliodora, Bursera simarouba,
Attalea cohune, and Parmentiera edulis. The tree suckers were those of the above species plus
Sapindus jabonera, Guazuma ulmifolia, Bombax elipticum, Ceiba pentandra, Lonchocarpus
longistylus, Cochlospermum vitifolum, Spondias mombin, Acacia cornigera, Inodes mexicana,
Casearia sp. and Terminalia sp. The shrubs formed the bulk of the vegetation and were all
regenerated from old root stock. They were Pithecolobium lanceolatum, Acacia .macracantha,
Bauhima unilateralis, Croton glabellus, Robinsonella lindeniana, Jacquinia pungens, Rauvolfia
heterophylla, Malphighia glabra, Coccoloba schiendeana, Coccoloba sp., Pisonta aculeata, Jatropha
urens, Tabernaemontana alba, Mimosa albida, Bixa orellana, Eupatorium odoratum, Triumfetta
semitriloba, and Bakeridesia galeottii, Some lower shrubs and annuals that disappear quite
early in the succession were Sida acuta, Croton muradorensis, Solanum torvum, Solanum
chloropetalum, Mimosa pudica, Melochia pyrmidata, Hyptis mutabilis, Iresine sp., and Salvia
sp. The woody vines were very quick to regenerate following cutting and were very abundant.
They were Tournefortia hirsutissima, Gouania lupuloides, Serjania cardiospermoides, Paullinia
tomentosa, Stigmaphyllon lindenianum, Turbina corymbosa, Bigonia unguis-cati, Smilax sp.,
and Asclepiadaceae. Annual and non-woody vines were Dioscorea spp., Convolvulaceae,
Leguminosae, Asclepiadaceae, Passifloraceae, and Marantaceae.

About 30 percent of the vegetation, in the area of Temascal to the east for 10 km, passes
through a seral stage similar to this in physiognomy and species composition at least once
every 10 years, if not more often. The largest variation between other similar areas and these
five plots is in species composition, not in physiognomy.

Throughout the experiments, the shoots in plot P remained in two size groups. About 75
percent of the shoots were in the 10-60 cm size class and were densely shaded during the
rainy season. They received light shade during the dry season. They were occupied by colonies

484 Tue University ScieNcE BULLETIN

of Pseudomyrmex gracilis mexicana, founding queens of Pseudomyrmex ferruginea, and rarely
by auxiliary-units from emergent queen-shoots or by new young colonies. These shoots grow
very slowly and are gradually eaten entirely by Sigmodon hispidus or die as a result of the
cumulative effect of continuous destruction of the shoot-tips by insects. About 25 percent of
the shoots were in the 100-250 cm size range and were emergent or canopy members. These
were occupied by large colonies of P. ferruginea and were not used as standards by the vines
in the canopy. During the dry season there was some abandonment of these shoots by auxiliary-
units but they were reinvaded when the rainy season started. The colonies of P. ferruginea
ranged from 500-10,000 workers; most queen-shoots had several auxiliary-shoots. There were
at least 150 colonies of P. ferruginea and five colonies of Pseudomyrmex nigrocincta within
these five plots before treatment.

The treatment schedules of the various subplots were as follows:

KA-1: Treatment subplot. Cut to the ground on 1 Oct. 1963. Grazed from 3 Dec. 1963

to 3 May 1964. All A. cornigera shoots sprayed with parathion on 26 Oct. and 21
Dec. 1963.

KA-2: Control subplot. As in KA-1, but no treatment.

KA-3: Control subplot. As in KA-2.
K-1: Treatment subplot. Cut to the ground on 1 Oct. 1963. Grazed from 3 Dec. 1963 to
3 May 1964. All thorns clipped on 19 Dec. 1963 and once during the dry season.

K-2: Control subplot. As K-1 except no treatment.
1: Treatment subplot. All vegetation except A. cornigera cleared to ground level on
10 Oct. 1963. All A. cornigera were sprayed with parathion on 26 Oct. 1963 and 14
Nov. 1963. On 3 Apr. 1964 all A. cornigera shoots less than 50 cm tall were removed
and the thorns of those remaining were clipped. This subplot was grazed from 3
Dec. 1963 to 3 May 1964.

-2: Control subplot. All vegetation except 4. cornigera cleared to ground level on 10

Octe19 63:

1: Treatment subplot. As in N-1.

-2: Control subplot. As in N-2.

1: Treatment subplot. Natural vegetation with all thorns of all 4. cornigera both above
and below the canopy clipped on 18 Dec. 1963. Grazed from 3 Dec. 1963 through
3 May 1964.

P-2: Control subplot. Natural vegetation.

In plot KA, the regeneration was apparently unaffected by the parathion treatment. In the
two control subplots, most of the shoots of A. cornigera remained stunted and gradually became
completely shaded. Those that were occupied by colonies invading from the cut shoots and those
that developed colonies in situ by the beginning of the rainy season (1964) maintained an
emergent or canopy level position. In the treatment subplot, the parathion had no noticeable
effect on the high density of founding queens and several colonies moved into the subplot after
the spraying from shoots cut outside of it.

In plot K-1, growth was relatively slow in both the treatment and control subplot until
shortly before the 1964 rainy season began. Clipping of thorns was extremely effective in
destroying the mature colonies remaining after the shoots were cut on 1 Oct. and in preventing
new colonies from developing in situ. When new growth was initiated at the beginning of the
rainy season, the shoots in the treatment subplot quickly became completely submerged in the
general vegetation while about 20 percent of those in the control subplot remained emergent.

In plots N and O growth was relatively slow in both the control and treatment subplots
until the beginning of the 1964 rainy season. Parathion was not effective in removing the
mature colonies from the treatment subplots because portions of colonies outside of the subplots
continued to move into the subplots to reoccupy the shoots vacated by the spray. However, the
parathion applications disrupted the occupation sufficiently to allow considerable insect damage
to the shoots. When the thorns were clipped and the shoots shorter than 50 cm tall were re-
moved, the occupied shoots outside of the treatment subplots were also removed when they
were not part of a control subplot. This produced two ant-free treatment subplots (N-1 and
O-1). In the control subplots the shoots less than 50 cm tall were not removed because this
might have disrupted the colonies occupying them as auxiliary-shoots. However, these short
shoots were not counted in the plot records.

In plot P, growth was relatively slow in both the control and treatment subplots until the
beginning of the rainy season. The thorn clipping was extremely effective in removing the
mature colonies and founding queens in the shoots. It was aided by the fact that it was the
dry season when some of the auxiliary-shoots were abandoned and the very hot sun killed
the exposed brood very quickly.

Sigmodon hispidus was present in all plots but did the heaviest damage in plot P and
adjacent areas of the same type; it appeared reluctant to forage into the open areas of the

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 485

other four plots until the rains started and then it did not feed on A. cornigera except rarely.
Practically all of the insects associated with A. cornigera in the Temascal area were found com-
monly in these five plots. While grazing was relatively light, it was present during most of the
cool and dry season; the cattle only occasionally browsed either occupied or unoccupied shoots.

A large part of the behavioral aspects of the bionomics of Pseudomyrmex ferruginea were
recorded in these five plots. Most of the mating flights that were observed were in the out-
standing trees, especially Attalea cohune.

Q

Plot O was a rectangular plot in the brushy pasture 100-250 meters north of El Mocho’s
house and 10-100 meters west of plots F, G, S and T.

This plot had one control subplot (Q-1) and one treatment subplot (Q-2). At the time
that the plot was established (8 Feb.) the area was very heavily grazed. However, the cattle
were removed from the area on 3 May and therefore the plot served only to examine the effect
of insects on unoccupied shoots (during the rainy season). Since thorn clipping was used to
remove the ants, the problem of parathion stimulation was avoided.

This plot was part of a cycle of cutting and burning with frequent heavy grazing. It was
last cut in October 1962 but it was not burned at that time. It was on the same 25 acres as
plots C, E-I, and S-V. The area of treatment subplot Q-1 was 962 m? and the area of control
subplot Q-2 was 650 m’.

In general appearance, plot Q was a very open stand of tree suckers, shrubs, and low herbs
with large patches of grass in between. During the dry season, this grass was beaten and grazed
down to ground level; during the rainy season, when there were no cattle present, the grass
grew to heights of 1-2 m. The tree suckers and shrubs were in the 1-2.5 m height range.
Acacia cornigera and Bixa orellana made up over 75 percent of the woody vegetation. Plot Q
had many fewer species of plants than any other plot around El Mocho’s house; plot L and M
were similar to Q in physiognomy and species composition.

The tree suckers were Ceiba pentandra, Bursera simarouba, Cordia alliodora, Parmentiera
edulis and Acacia corngiera. The shrubs were Bixa orellana, Eupatorium odoratum, Pisonia
aculeata, Jatropha urens, Tabernaemontana alba, Coccoloba schiedeana, Croton glabellus, and
Cassia bicapsularis. The annuals and low shrubs were Sida acuta, Cassia leptocarpa, Croton
muradorensis, Solanum chloropetalum, and Melochia pyrimidata. The woody vines were
Bignonia unguts-catt, Turbina corymbosa, Tournefortia hirsutissima, and Smilax sp. There
were a few annual vines in the Convolvulaceae and Passifloraceae and two major species of
grasses.

Nearly all of the land in the area between Temascal and La Granja that will become a
natural pure grass pasture passes through a stage like this one. If the root stocks of the woody
plants were removed at this time, and no grazing was allowed for a year, this site would
produce a nearly pure grass pasture. This type of plant community becomes increasingly com-
mon as cattle become more common.

The shoots of A. cornigera were not readily divided into two size groups. Apparently due
to the lack of shading of small unoccupied shoots, these shoots live longer in an open plant
community than in one with a dense canopy. They eventually become occupied and begin to
grow rapidly but the result is rather great variation in height among the shoots. With the
exception of a few young colonies developing in situ, the colonies of P. ferruginea were
mature colonies with 500-5000 workers in them; they often occupied more than one shoot. Be-
fore treatment, there were about 50 colonies in plot Q.

On 8 Jan. 1964 all of the thorns in treatment subplot Q-1 were clipped. This treatment was
very effective and within a month, all the shoots in the treatment subplot were unoccupied. It
was also necessary to remoye some of the marginal shoots around the treatment subplot in
order to prevent invasion of the clipped thorn shoots from these intact shoots.

The insects commonly found feeding on occupied and/or unoccupied shoots of A. cornigera
were abundant in plot Q-1. In the clumps of Bzxa orellana, Sigmodon hispidus was occasionally
responsible for the cutting of small unoccupied shoots. Since nearly all of the growth of the
plants in plot Q took place after the first rains, cattle were not present at the time of growth.

185 18, (G5 Sh IE
These five plots were established in a relatively uniform portion of the brushy pasture 200-
400 m north and 100 m west of El Mocho’s house.
The purpose of this series of plots was to establish the following types of plots in a single
stand of vegetation: 1) an undisturbed plot with the ants removed by parathion (plot S), 2)
a plot with all the vegetation except A. cornigera cut to ground level and the ants removed by

486 Tue UNiversity SCIENCE BULLETIN

parathion (plot T), 3) a plot with all of the vegetation cleared and the ants removed by
destroying the canopies of 4. cornigera (plot F and E), and 4) a plot in undisturbed vegetation
with the ants removed by destroying the canopies of 4. cornigera (plot G).

This series of plots was part of a cycle of cutting and burning with sporadic light grazing
and browsing. It was last cut in July 1962 but it was not burned at that time. During the ex-
periments, cattle were present only between 3 December 1963 and 3 May 1964. The areas in
square meters of the various subplots were as follows: S-1, 342; S-2, 448; T-1, 459; T-2, 405; F-1,
157> B23 1575 (G-1, 625;'G-2, 225; (G-35 1005 Blo 83> E25 i675 and 5, 202%

The general appearance of the area containing the five plots was a dense stand of tree suckers,
shrubs and herbs. Most of the vegetation was regenerated from old rootstocks. In September
1963 the general canopy was about 170 cm tall and bound together with a few vines. As the
weather became drier during the dry season, the leaf drop from woody plants and the wilting
of annuals opened up the vegetation and allowed the sunlight to reach the ground. Yet within
a month after the first heavy rains in late May, the vegetation took on again a very closed ap-
pearance and by August the canopy had risen to about 250 cm. There were abundant annuals
and vines present, apparently due to the absence of cattle which preferred them.

The tree sucker species were the same as in plots K, N, O, P, but they formed a larger portion
of the woody vegetation. The same species of shrubs, vines and annuals were present as in
plots K, N, O, and P but the proprotions were varied. This type of vegetation represented a
seral stage in the same progression of which the above plots and plots C, H, I, U and V were
a part, except that the plots S, T, F, G and E represented an intermediate stage between those
of plots K, N, O, and P and plots C, H, I, U, and V.

About 10 percent of the shoots of A. cornigera were in the 10-70 cm tall size range and
were severly shaded except during the peak of the dry season. S. Aispidus ate many of these
shoots during the dry season. About 85 percent of the shoots were in 100-400 cm size range
in November 1963. By August 1964 many of the shorter of these shoots had increased 50-100
cm in height. Most of the shoots in the small size range were tenanted by founding qucens of
P. ferruginea or colonies of Pseudomyrmex gracilis mexicana. The taller shoots were almost
entirely occupied by P. ferruginea during the rainy season but there was considerable abandon-
ment of auxiliary-shoots during the dry season, apparently due to the increased high density of
A. cornigera which resulted in one colony being spread over many shoots. Prior to treatment,
there were about 180 colonies of P. ferruginea in the five plots. There were also four colonies
of Pseudomyrmex nigrocincta occupying 11 shoots of A. cornigera. There were two tall shoots
that had been colonized by Pseudomyrmex gracilis mexicana but during the course of the
experiments, they were driven from their shoots by P. ferruginea. The colonies of P. ferruginea
were for the most part large, with 1000-15000 workers per colony.

The treatment schedules of the various subplots were as follows:

S-1: Treatment subplot. Natural vegetation, all shoots of A. cornigera sprayed with para-
thion on 4 Nov. and 4 Dec. 1963. Continuous light grazing and browsing until 3
May 1964.

: Control subplot. Natural vegetation and cattle as in S-1.

Nn
1
i)

T-1: Treatment subplot. All vegetation except 4. cornigera cut to ground on 28 Oct. 1963.
A. cornigera sprayed with parathion on 4 Nov. and 4 Dec. Cattle present as in S-1.

T-2: Control subplot for T-1. As in T-1 except no parathion used.

F-1: Treatment subplot. All vegetation cut and A. cornigera cut at 65 cm on 28 Oct.
Canopies of A. cornigera piled and burned. Cattle as in S-1.

F-2: Control subplot for F-1. As in F-1 except the canopies of A. cornigera left over the
stumps instead of burned.

G-1: Treatment subplot. Only A. cornigera cut at 65 cm, remainder of vegetation undis-
turbed, on Oct. 29. Canopies of A. cornigera piled and burned. Cattle as in S-1.

G-2: Treatment subplot. As in G-1.

G-3: Control subplot. As in G-1 except that the canopies of 4. cornigera were left over
the stumps instead of being destroyed.

E-1: Treatment subplot. All vegetation cut and A. cornigera cut at 65 cm on 18 Nov.
Canopies of 4. cornigera piled and burned. Cattle as in S-1.

E-2: Treatment subplot. As in E-1.

E-3: Control subplot for E-1 and E-2. As in E-1 except that canopies of A. cornigera left

over the stumps instead of being destroyed. On 23 Mar. added a queen-unit to each
unoccupied shoot (10 shoots).

In plot S, the parathion treatment was relatively effective in removing P. ferruginea but
due to the close proximity of colonies outside of the plot, there was a relatively rapid reinvasion
of these shoots. It was in this plot that parathion stimulation was most noticeable. The same
situation occured in plot T.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 487

In plot F, there were relatively few ant colonies around the margins and the treatment
subplots thus stayed relatively free of P. ferruginea. On 23 March a queen-unit in the form
of a cut queen-shoot was added to each of the stumps in the control subplot in which the
transfer of the unit from the old cut shoot had not been successful. All 10 of these queen-units
became established. Toward the end of the dry season there was a rapid invasion of the shoots
in the west side of treatment subplot E-1 from two large queen-shoots growing outside of the
subplot.

Fourteen adult and yearling cattle were present in all five of these plots until 3 May 1964.
Most of the animals browsed and grazed independently of the others; they were rather evenly
dispersed over the 25 acre area that includd plots F, G, S$, and T, E and C, H, I, U and V.
All the common insects that fed on Acacia cornigera were present in these plots; Coxina
hadenoides was especially abundant in plots F and G. S. Aispidus was abundant only in plots
S and G; this appeared to be associated with the high cover of these plots.

CHU Vi

These five plots were arranged along the sides of the water pipeline that ran northeast from
El Mocho’s house. They were 200-1000 m east of plot Q and 200-300 m north of the Temascal-
La Granja highway. Within each plot, rectangular subplots were established. Plot U is shown
in Figure 51.

The employment of these five plots was to use the various treatments (parathion, thorn
clipping, stump regeneration) in a thick stand of woody regeneration that was browsed by
cattle and had a high density of A. cornigera. The cattle were removed from the plots on 3
May so that the major portion of the growth that took place was free from cattle foraging.

The regeneration in these five plots before treatment was of the same type of cycle as in
plots K, N, O, P, E, F, G, S, and T, but was older. It was last cut to the ground in 1962 and
according to the owner was lightly burned at that time; it must have been a light fire since
there were many large colonies of P. ferruginea present in late 1963. The areas of the various

Fic. 51. General aspect of treatment subplot U-1 in January 1964. Leaf drop has begun on
the shoots of Acacia cornigera but the low sucker regeneration of many species of plants under
the A. cornigera is in full leaf though growing slowly. Before the area was sprayed, there
were about six colonies of P. ferruginea in the shoots immediately visible in the photograph.
Each colony was large with auxiliary-shoots.

488 Tue Universiry ScrENCcE BULLETIN

subplots in square meters were as follows: V-1, 1200; V-2, 1200; U-1, 750; U-2, 720; H-1,

2,025; H-2, 2,005; H-3, 1,895; 1-1, 300; I-2, 325; I-3, 250; C-1, 500; C-2, 460; and C-3, 620.
In general appearance before treatment, these plots contained a very dense stand of woody

regeneration that was interlaced with woody and annual vines. There were relatively few low

shrubs. As the dry season progressed, the stand opened to such an extent that one could walk

through it but by August 1964 it was nearly impassible except along old cattle trails. 4.

cormigera formed such a large part of the canopy that from a distance of 500 m, it looked like

a pure stand. The canopy was 2-3.5 m high and cast a very heavy shade during the rainy

season. There were occasional old trees 30-50 meters tall that were remnants of the old forest

that was on the land 20 years earlier. Though the land was relatively flat, there were occasional
limestone knolls with a very distinctive xerophytic vegetation; A. cornigera did not grow on
these and no knolls were contained in the subplots.

The scattered old trees were Ceiba pentandra, Sapindus saponaria, Enterolobium cyclo-
carpum, Bursera stmarouba, Tabebua pentaphylla, and other species. The limestone knolls
were covered with Bombax ellipticum, Inodes mexicana, climbing cacti, cycads, Marantaceae,
Acacanthaceae, succulent leaved xerophytic herbs, Agavaceae, and terrestrial bromeliads. On the
level ground, there were tree suckers of the trees listed above, except Bombax ellipticum, plus
Piscidia communis, Cochlospermum vitifolium, Cordia alliodora, Spondias mombin, Attalea
cohune, and Acacia cornigera. In one of the arroyos there were a few Acacia macracantha and
Acacia chiapensis. The common shrubs in the plots were Solanum umbellatum, Croton glabellus,
Jatropha urens, Casimiroa edulis, Cassia bicapsularis, Esenbeckia berlandiert, Buettneria aculeata,
Eupatorium odoratum, Jacquinia pungens, and other unidentified species. The woody vines
were Turbina corymbosa, Tournefortia hirsutissima, Bignonia unguts-cati, Gouania lupuloides,
Serjania sp., Cissus sp., and various species of Leguminosae, Asclepiadaceae, and Convolvulaceae.
The annual vines were Dioscorea spp., Leguminosae, Convolvulaceae, and Asclepiadaceae. Small
openings had several species of grasses and herbs growing in them.

The stand in which these five plots were established represents an older seral stage in the
progression and cycle represented by plots K, N, O, P, E, F, G, S, and T. After another year
of growth this stand of vegetation would have begun to open up underneath as had happened
in the case in plots D and R.

Better than 95 percent of the shoots of A. cornigera were canopy members or emergent at
the time the plots were established. When the vegetation was cleared in some of the plots a
few old stumps produced new sucker shoots. The shoots were old enough to respond strongly
to the cool and dry season by cessation of most growth except the production of axillary tufts
of leaves and flowering branches. The colonies of P. ferruginea had 1000-15,000 workers in
them and occupied 3-15 shoots. There were no colonies of Pseudomyrmex nigrocincta; two
nearly dead tall shoots had colonies of Pseudomyrmex gracilis mexicana. Most of the P. gracilis
mexicana colonies in these plots lived in the hollow stems of Buettneria aculeata. During the
dry season there was a 10-30 percent abandonment of auxiliary-shoots by P. ferruginea. There
were approximately 120 colonies of P. ferruginea in the five plots before treatment.

The treatment schedules for the various subplots were as follows:

V-1: Treatment subplot. Natural vegetation, all A. cornigera sprayed with parathion on
30 Oct. and 21 Nov. Light grazing and browsing untll 3 May 1964 when they were
removed by the owner.

V-2: Control subplot for V-1. As in V-1 except no parathion used.

U-1: Treatment subplot. All vegetation except A. cornigera cut to ground on 20 Oct. and
all shoots of A. cornigera sprayed with parathion on 31 Oct. and 21 Nov. Cattle as in
V-1.

U-2: Control for U-1. As in U-1 except no parathion used.

H-1: Treatment subplot. All vegetation cut on 17 Oct., A. cornigera cut at 65 cm and
canopies piled and burned. A. cornigera sprayed with parathion on 30 Oct. and thorns
clipped on 13 Mar. Cattle as in V-1.

-2: Treatment subplot. As in H-1 except that thorns not clipped.

3: Control for H-1 and H-2. As in H-1 except that canopies of A. cornigera left over

their stumps and no parathion or thorn clipping.

I-1: Treatment subplot. A cornigera cut at 65 cm and canopies piled and burned on 18 Oct.

Cattle as in V-1.

I-2: Treatment subplot. As in J-1.

I-3: Control subplot. As in I-1 except canopies left over their stumps.

C-1: Treatment subplot. A. cornigera cut at 65 cm and canopies piled and burned on 5

Jan. Thorns clipped on 14 Mar. Cattle as in V-1.

C-2: Treatment subplot. As in C-1 except thorns not clipped.

C-3: Control subplot for C-1 and C-2. As in C-1 except thorns not clipped and what ap-
peared to be a queen-unit added to each shoot on 14 Mar.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 489

In plot V the parathion spray was only moderately effective but still reduced the colony size
of most colonies. There was a rapid initial reinvasion of the treatment subplot from large
colonies outside of the subplot but, judging by the amount of insect damage that took place in
the 1964 rainy season in the treatment subplot, the numbers of workers per shoot still remained
much lower than normal. If the acacias outside of the treatment subplot had been removed be-
fore the treatment, it appeared that the parathion treatment could have been permanently
effective.

In plot U the same thing happened as in plot V except that the reinvasion was slower but
by the 1964 rainy season the numbers of workers on the shoots were back to normal.

In plot H and I the treatment was very effective in removing the colonies of P. ferruginea.
The parathion spray used in plot H was not necessary but it seemed to be so at the time because
of the few workers left when the shoot was cut. The thorn clipping in subplot H-1 prevented
the development of young colonies in situ; in plots H-2, 30 of these colonies had developed
by 5 Aug. The long thin subplots in plot I were very close to A. cornigera that were not cut.
Therefore, abandonment to, and reinvasion from these plants considerably complicated the
degree of occupation of shoots.

In plot C, the treatment was effective. However, the damage sustained by the unoccupied
shoots in the control subplot before the queen-units were added on 14 Mar. caused a number
of these to grow very poorly or not at all even after they were occupied. A further complication
was that apparently about one-half of the supposed queen-units were auxiliary-units and there-
fore did not persist on the shoots.

All of the common insects that fed on Acacia cornigera were found in these plots. When
cattle were present, they browsed and grazed in a highly dispersed pattern. They fed very
rarely on A. cornigera, irrespective of whether it was occupied or not. However, it was in this
area that the cattle were frequently observed browsing A. cornmigera during the cool season
when it was unoccupied due to lowered worker activity at the time. S. Azspidus was occasionally
seen in these plots but there were almost no shoots present that were small enough for it to
feed on.

J

Plot ] was a rectangular plot in the middle of a corn field 800 m south of Senor Torrealva’s
house. His house was on the north side of the Temascal-La Granja highway at Parada Pochota
which was 7 km east of Temascal.

The site of plot J was chosen because it was on red laterite soil and because of the very
young vegetation that was present following the ue as a corn field. The plot was established
after the soil had begun to dry out and the cool season had set in. While this soil had never
been plowed, the vegetation had been burned severely during the 1963 dry season and was
cut very close to the ground in July of 1963. It had been cleared for a corn field in 1960. The
land-use schedule was that cattle were excluded and corn planted when the first rains fell.
By November, the cattle were allowed to graze in the regeneration until the end of the fol-
lowing dry season. The total area of the corn field was about 15 acres and plot J was roughly
in its center. Treatment subplot J-1 had an area of 200 m®° and control subplot J-2 had an
area of 180 m°.

In December 1963 the general aspect of the field was one of nearly bare red laterite littered
with corn stalks, and a very sparse cover of shrub suckers and vines. There were a few
scattered trees. The field was surrounded on three sides by extremely dense regeneration from
previous corn fields; this vegtation was about 2.5 m tall and composed of Malvaceae, Legumi-
nosae, and Sterculiaceae. There were very few A. cornigera in it. In plot J the A. cornigera
were 5-35 cm tall and none were occupied; nearly all of the new thorns had founding queens
in them.

The trees in and around plot J were Acromia mexicana, Inodes mexicana, and Attalea
cohune. There were suckers of the above three species plus Acacia cornigera, Byrsonima cras-
sipes, Curatella americana and Quercus sp. The shrub species were Croton glabellus, Bauhinia
ungulata, Mimosa albida, Calliandra houstoniana, Cassia bicapsularis, Conostegia jalapensis, and
Coccoloba sp. There were a number of prostrate herbs and non-woody vines. During the
winter, grasses were not evident but during the rainy season they sprouted in great abundance.

When the 1964 rainy season began, the entire corn field was fenced off. It was cleared and
planted except for the plot. The vegetation in plot J had grown very thick and lush by the
end of July. Most of the vegetation consisted of grasses and herbaceous annuals. The woody
shrub and tree suckers constituted a minor part of the vegetation. The vegetation was so dense
that there was standing rain water on the soil under it even on the hottest and driest days. All
shoots on A. cornigera, except for three that had young colonies of P. ferruginea that had
developed in situ, were well below the canopy and received almost no sunlight.

490 Tue University ScleENcE BULLETIN

The treatment schedule of the two subplots was as follows:

J-1: Treatment subplot. Undisturbed vegetation with all A. cornigera sprayed with para-

thion on | Dec. thorns clipped on 15 Jan. Cattle present until 22 May.

J-2: Control subplot. As in L-1 except that no parathion or thorn clipping.

The parathion had no noticeable effect on the founding queen population. The thorn
clipping was used to keep new colonies from developing in the shoots. When the final record-
ing was made, the thorns in the control subplot were opened. Many of them contained young
colomies with 5-20 workers but only three had a colony large enough to occupy the outside of
the shoot.

The insects commonly associated with A. cornigera were present in plot J but at very low
densities. The cattle browsed and trampled the area severely during the dry season but rarely
ate foliage of A. cornigera. Sigmodon hispidus was not present in the plot during the dry
season but was extremely common there during the rainy season. It did not feed on A. cornigera
at that time. ‘

L, M

These two plots were established in the brushy pasture 50-100 m south of the Temascal-La
Granja highway across from Sefior Torrealya’s house and 50 m southeast of the largest Cetba
pentandra in the area (La Pochota).

As can be seen in Figure 52, these two plots had very open vegetation. The site was un-
complicated by the presence of large colonies remaining after the last clearing; the site was
severely burned in May 1963. It received heavy grazing throughout the rainy season but the
cattle were removed about the time that the experiments started (Oct.); picketted Indian
horses remained, however. The result of the heavy grazing during the rainy season was that
there were a large number of unoccupied but living shoots in the 30-60 cm size range in
October; in a neighboring area that had been burned at the same time but not grazed, nearly
all of the shoots were severely stunted and nearly dead. During the course of the experiment,
young colonies developed in most of the shoots in the two plots while none developed in the
adjacent area.

+ ae wie & pe i

Fic. 52. General aspect of treatment subplot M-1 in November 1963. The shoot of Acacia
cornigera directly in the center of the sheet is occupied by a small colony of P. ferruginea that
had developed in situ. It has an undamaged uppermost shoot tip. The shoot in front of the
right-hand portion of the sheet contains only founding queens of P. ferruginea. It also has an
undamaged uppermost shoot tip but is festooned with vines and has produced much less foliage
than the occupied shoot.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 49]

The area included in the two plots was used as a corn field in 1962 and the two preceeding
years. It had a species composition and physiognomy rather similar to plot J. The areas of the
four subplots in square meters was as follows: M-1, 600; M-2, 600; L-1, 300; L-2, 525.

The general aspect of the plots was one of a very open canopy, with woody shrub and
tree suckers scattered on low grass. By August 1964 the grass was 40-100 cm deep and
there was considerable herbaceous growth. It was not until several months after the A. cornigera
became occupied that they began to appear as emergents above the heights of other species of
woody plants.

The tree suckers were Acromia mexicana, Inodes mexicana, Ateleia pterocarpa, Coccoloba
schiedeana, Acacia cornigera, Guazuma ulmifolia, and Cordia alliodora. The annual and woody
shrubs were Helicteres guazumaefolia, Croton glabellus, Pithecellobium lanceolatum, Jatropha
urens, Cassia bicapsularis, Bauhinia ungulata, Bauhinia unilateralis, Solanum torvum, Cordia
ferruginea, Mimosa albida, Mimosa pudica, Tabernaemontana alba, and Jaquinia pungens. Once
the cattle had been removed, the following vines became common: Passiflora foetida, Rhabda-
demia paludosa, Stigmaphyllum lindenianum, and Ipomoea sp. There were a number of
species of grasses and sedges.

These two plots were representative of the second year of the seral progression of which
plot J represented the first year. The land-use pattern, cornfield to fallow and back again to
cornfield, is very common in the Temascal area. These two plots on a soil intergrade zone
between the black soils and red laterites.

At the beginning of the experiment, the 4. cornigera were occupied by founding queens
and an occasional small colony with 20-50 workers. By the end of July 1964, most of the
untreated shoots had developed a young colony in situ with 100-1600 workers in it. In
several cases, there were auxiliary-shoots present as well. At the beginning of the experiment
there were four shoots along the margins of the area that were occupied by Pseudomyrmex
gracilis mexicana. By late July 1964 three of these colonies had been removed by Pseudomyrmex
ferruginea and one by Pseudomyrmex nigrocincta.

The treatment schedules for the various subplots were as follows:

M-1: Treatment subplot. Natural vegetation, all 4. cornigera sprayed with parathion on
28 Oct. and on 6 Dec. Thorns clipped on 20 Apr.

M-2: Control subplot for M-1. As in M-1 but no parathion or thorn clipping.

L-1: Treatment subplot. Natural vegetation, all A. cornigera thorns clipped on 11 Dec.

L-2: Control subplot for L-1. As in L-1 but no thorn clipping.

All of the insects commonly found associated with A. cornigera were very common in this
plot. Dziplotaxis denigrata was especially common. Cattle were not allowed free browsing in
the two plots, but Indian horses and burros were occasionally tethered on the acacias. Sigmodon
hispidus cut a few shoots in these plots during the dry season.

A, D,R
These three rectangular plots were established in the five year old regeneration 50-300 m
south of Seftor Torrealva’s house and 100-150 m west of the plots L and M.

Before treatment, these three plots contained a five year old stand of fallow regeneration
after use for a cornfield in 1959. It was lightly grazed and browsed throughout this regenera-
tion. There was a closed canopy of young trees at about 3-4 m above ground level. While this
canopy was interlaced with woody vines, the understory was sufficiently clear to walk through
with ease. It was hoped that the plots would clarify some aspects of the effects of shading on
A. cornigera. The areas in square meters of the various subplots were as follows: R-1, 800;
R-2, 600; D-1, 120; D-2, 120; D-3, 110; A-1, 225; and A-2, 550.

There were a few large trees remaining from the forest that covered the site before it was
cleared in the late 1940's. The canopy at 3-4 m height cast a very heavy shade during the rainy
season and the plants growing at ground level were either shade tolerant species or ones such
as A. cornigera which occupy open sites and show abnormal growth in heavy shade. The
canopy member or emergent A. cornigera were fully developed shoots that had maintained
their position in the canopy while the regeneration of the vegetation occurred. There were
a few shoots of A. cornigera that developed later and grew up into vacancies in the canopy once
they became occupied; these were usually auxiliary-shoots to canopy members or emergent
queen-shoots.

The large trees were Ateleia pterocarpa, Casearia sylvestris, Attalea cohune, Tabebuia penta-
phylla, Cecropia obtusifelia, and Quercus sp. The main canopy was made up of Cordia ferrugi-
nea, Vitex pyramidata, Bauhinia ungulata, Acacia cornigera, Guazuma ulmifolia, Helicteres
guazumaefolia, Cochlospermum vitifolium, and Bakeridesia galeottii. The woody vines were
in the Leguminosae, Asclepiadaceae, Sapindaceae, and Bignoniaceae.

49? Tue Universiry SctENcE BULLETIN

Two size classes of A. cornigera were present. In plot R the canopy member and emergent
shoots were omitted in the plot consideration because of the impossibility of spraying these
5-9 m tall shoots with the apparatus available. The members of the smaller size class were
those used in the experiments; these plants probably would never have reached maturity unless
the vegetation was cleared. These smaller plants were occupied by founding queens of
Pseudomyrmex ferruginea, P. nigrocincta and P. gracilis mexicana. There was an occasional
small colony and some of the shoots were occupied as auxiliary-shoots to the canopy member
or emergent shoots. In the other two plots, the stumps of the cut taller shoots were used as
the experimental shoots. The canopies of the taller shoots had colonies of P. ferruginea in them
that ranged from 1,000-15,000 workers.

The treatment schedule for the various subplots was as follows:

R-1: Treatment subplot. Natural vegetation, all A. cornigera below the canopy sprayed
with parathion on 9 Dec. 1963. Site very lightly grazed by cattle throughout the
experiment.

R-2: Control subplot for R-1. As in R-1 except no parathion used.

D-1: Treatment subplot. Natural vegetation, all canopy member or emergent shoots of
A. cornigera cut at 65 cm and the canopies piled and burned on 10 Dec. Cattle as in
R-1.

D-2: Treatment subplot. As in D-1.

D-3: Control subplot for D-1 and D-2. As in D-1 except that canopies of A. cornigera
left over their stumps.

A-1: Treatment subplot. All vegetation cut on 25 May 1964 and A. cornigera cut at 65
cm. Canopies of 4. cornigera plied and burned. Cattle browsed and grazed the area
slightly more heavily than plots R and D.

A-2: Control subplot. As in A-1 except that canopies of A. cornigera left over their stumps.

The spraying in plot R was relatively ineffective since founding queens moved back into the

shoots and the canopy member and emergent shoots replaced the destroyed auxiliary-units.
However, the low growth rates in these two subplots helped to elucidate the effect of shade
on A. cornigera.

The regeneration in plot D became very complicated because of the close proximity of uncut

shoots to the cut shoots; auxiliary-units from these colonies moved into a number of the new
regeneration from the stumps.

In plot A, the tall shoots around the treatment subplot were removed when the treatment
subplot was established; this prevented movement into the area by auxiliary-units. In the
control subplot the large units moved into new shoots and there was very effective occupation.

All of the insects commonly associated with A. cornigera were seen in plot D and A;
since the emergent and canopy member 4. cornigera in plot R were not removed, the canopy
remained intact and the shade dense and this appeared to reduce the numbers of insects found
on A. cornigera. Their damage was however frequently present. Cattle did not browse J.
cornigera except when the tall shoots were cut. Sigmodon hispidus was not seen in the plots
nor were shoots cut by it in the plots.

Data FRoM suspLots. When comparing the growth and development of
unoccupied and occupied shoots of Acacia cornigera, the following parame-
ters were recorded in the subplots: mortality, height increment, emergence,
invasion by vines, condition, presence of phytophagous animals, thorn produc-
tion, leaf production, and total biomass production. All of these parameters
were recorded during the study for at least one pair of treatment and control
subplots. The data from these recordings is examined in this section. In the
22 plots containing 50 subplots, the height increments and condition of the
shoots were recorded at intervals of a month or more. The plots were estab-
lished between October 1963 and May 1964. The majority were followed dur-
ing the 1964 dry season and the first two months of the 1964 rainy season. The
data from these plots are examined by comparisons between subplots (Tab.
4-9, 14-21). Since most of the subplots were mapped, and the individual
shoots numbered, and since the presence of workers on the surface of the

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 493

shoot was recorded in each case, it was possible to extract a second set of
height increment and condition data from the subplots. This consists of two
groups of shoots, one of which was unoccupied during the interval between
two recordings, and the other occupied. The data from the groups are exam-
ined by comparisons within the plots taken as a whole (Tab. 10-13, 22-26).
In these comparisons, both the subplot pairs or triplets, and the groups of oc-
cupied and unoccupied shoots are arranged as groups of suckers from stumps
that were cut as part of the experimental program (Tab. 4-6, 10-11, 14-17, 22-
24) and groups of naturally existing shoots (Tab. 7-9, 12-13, 18-21, 25-26).

Height increment. Significance. The height increment is one of many
possible indicators of the amount of growth that has taken place over a period
of time. A. cornigera is a plant of disturbed sites. In order to obtain sufficient
sunlight for growth, it must maintain its position in or above a rapidly rising
canopy.

A positive height increment can only occur when the uppermost shoot tip
is undamaged. A few days to several months are required for a new vertically
lengthening shoot tip to surpass a damaged terminal shoot tip. Therefore,
the height increment is also a function of the amount of time that the shoot
tip remains undamaged. For example, if a shoot has a mean height incre-
ment of 2.5 cm per day for a two month period, it is very unlikely that the
uppermost shoot apex has been damaged by any agent during the two month
period.

Negative height increments occurred on some heavily occupied shoots
during the later portion of the rainy season. In this case, the terminal 30-150
cm of the main axis was not able to support the weight of the swollen-thorns
and became bent. Figure 19b shows a section of such a central axis. Shoots
are also shortened by having the terminal 3-10 cm of the succulent shoot tip
eaten off by insects such as Pelidnota punctulata, Coxina hadenoides, and
tettigoniid grasshoppers. Large negative height increments occur on unoccu-
pied shoots, when the cerambycid Onicideres poecila cuts off the shoot, or the
rodent Sigmodon hispidus eats the shoot. The weight of the flowers and
flowering branches produced on heavily defoliated unoccupied shoots some-
times causes the central axis to bend and thereby shortens the height by a few
centimeters.

Because of the frequently curved or slanted life form of the shoot, the
length of shoots over 100 cm tall is often greater than the height of the shoot.
Suckers from cut stumps in only slightly shaded sites grow vertically without
bending during the first 9-12 months of regeneration. In these cases, the
length rather than the height of the shoot is of importance and is recorded;
this removes the variation associated with the height on the stump at which
the sucker originated. 7

Even among a group of even-aged shoots on the same site, there is con-
siderable individual variation in height (or length) increment rate. This can

494 Tue University ScreNcE BULLETIN

Tasre 3. The air temperatures recorded at he Temascal weather station of the

Comicion del Papaloapan, during the period | Jul. 1963 through 31 Jul. 1964 (un-

published). All values are reported in degrees C. The three sets of underlined
values represent the location of the cool season.

Jul. Aug. Sept. Oct. Nov. Dec. Jan. Feb. Mar. Apr. May Jun. Jul.

X daily

maxima

@ 1G) S145 3210 808 29:2 28:4. 24 242) 255 294 eo Sm UE
X daily

minima

(©) ee 2 Pty PP INE NOS NGS 54/7 i740) 17) PP) 238 aeG 225
Days with

maxima 24°

Cor less 0 0 0) 0 3 16 15 9 3 1 0 1 0

,

be readily seen in the high standard deviation values in the tables for height
increment. The major portion of this variation is associated with the activity
of the occupying ant colony (or fraction thereof), the frequency of attack of
the plant by phytophagous animals, the size of the parent root stock, and the
time of year (temperature and rainfall effect on physiological growth rates).
Individual inherent variation in growth rate is present but plays a minor
role compared to the above four factors. It appears that there is little variation
in growth rates of undamaged shoot tips of vertically lengthening branches
on occupied shoots over 100 cm tall at a given time of year.

Subplot length increment records for suckers and stumps. In table 4, the
length increments are recorded for nine plots that contained 24 subplots. The
shoots were all suckers regenerating from stumps cut as part of the experi-
mental program. The length increment is that which accrued during the first
part of the 1964 rainy season. They were cut before the cool season in 1963
(plots H, I, F, G, and E), during the cool season 1963-1964 (plots D and C),
or during the 1964 dry season (plots A and B). Parathion was used only
during October and/or November in the treatment subplots H and J. Thorn
clipping was used in treatment suplots H-1 and C-1 to insure unoccupied
shoots. The original removal of the canopies was the only treatment used in
subplots I-2, G-1, E-1, E-2, C-2, B-1, and A-1.

Of the 17 treatment-control contrasts between subplots in table 4, 11 of the
pairs have a highly significant difference in mean height increment, one is
significantly different, and five are not significantly different. On each of the
comparisons, significant Z values pertain to higher increment values in the
control subplot. High negative Z values are the result of the same factors
responsible for negative Z values that are not significant at either the 5 or 1

Tasre 4. The length increment data for suckers from stumps treated as subplots.
The majority of the days in the time intervals occur in the 1964 rainy season. In
the “ants removed by” column, R=removal of the colony, P=spraying with para-
thion, T=clipping the thorns, and C=control subplot. Where more than one treat-
ment was used, the order of the letters indicates the order of treatment. The frac-
tion in the “Begin” and “End” column has as its numerator the number of
occupied shoots in the subplot and as its denominator the total number of shoots
in the subplot. The plots in Tables 4 through 28 were recorded during daylight
hours. “Begin” signifies the count at the beginning of the time interval and

“End” signifies that at the end of the interval. The X, and s.d., columns contain
the mean length, and its standard deviation, of the shoots in the subplots at the

beginning of the interval. The Xine and s.d.jn¢ columns contain the mean length

increment and its standard deviation over the time interval. The Z values are

computed between the mean growth increments in the treatment and control

subplots. Significant Z values are indicated by an asterisk, and highly significant

Z values are indicated by a double asterisk. Large negative Z values are not
asterisked for reasons explained in the text.

Occupied '
far Total _Length Increment

Sub- removed Time Xo Gyalies XGirres Shing Z
plot by interval Begin End (cm) (cm) (cm) (cm) values
A-l R 25 May- 0/42 0/42 0.00 0.0 6.23 5.6 7 56"™
A-2 c 16 Jun. 29/29 26/29 0.00 OOP S096 WA
A-1 R 16 Jun.- 0/42 1/42 6.23 Om LOSS 9.8 8.808**
A-2 C 3 Aug. 26/29 29/29 30.96 17.1 72.86 37.4

B-1 R 23 May- 3/53 7/53 Iall5 66 BSill5S 257 4.250**
B-2 C 17 Jun. AN AAA A/a tes On el ESD FO Same LOL2

B-1 R 17 Jun.- WSS IOS B30 232 447s Wy Soils
B-2 G 7 Aug. 44/47 45/47 70.19 26.4 92.19 69.6
C-1 Rew 27 May- 2/20 6/20 3525 40Y PAoeil0) Sai E22
C-2 R 2/19 Syl) SSNS BE WEN} 0) 0.995
C-3 Cc 6 Aug. 13/34 Gi eS Zell Or So: Side Ome OO:
D-1 R 25 May- 3/9 3/9 30.11 29.5 36.33 53.9: —0.058
D-2 R 6/12 WIND D3 DI BIG 532 07
D-3 C 28 Jul. 4/19 3/19 47.52 53.9 3494 403
E-] R 10 Jun.- 4/29 G20 Shes 4b Wy “vo 3pl>**
E-2 R 2/24 A AEN Byh AKG) Shilo) AIS 7e7 = LOA **
E-3 C 7 Aug. Pil PifPil GES GOS news) — AvaI
F-1 R 24 Apr.- 19/42 26/42 30.38 24.6 30.57 49.6 0.884
F-2 C 6 Jul. 25/40 29/40 32.37 23.4 39.32 39.6

G-l R 24 Apr.- 1/18 (Sil) asA7/ Alfa 7.00 23.4 4.703**
G-2 R 5/26 VA 3320 Als 207 soy 1.982*
G-3 c 6 Jul. IO OVA Bex) Ais Silay aos,

H-1 R,P,T 10 Jun.- 0/43 (Q/4's} Sail} 14! 0.47 Tale L0%344""
rI-2 Roe 2/66 32/66" 25:12" 20.6 8.16 19.0 8.803**
H-3 Cc 5 Aug. Ti 2° 64/72 8348) AC, 974. 63a0 6019

I-] R 10 Jun.- ISS IBY} sisal) IG SI7/SH) GBD 3.593**
I-2 R 8/32 BiB SNS SG OIA Axo 5.459**
I-3 © 5 Aug. 27/39 27/39 52.79 45.9 83.00 66.0

496

Tue University Science BULLETIN

Tasre 5. As in Table 4 except that the majority of the days in the time intervals

occur in the dry season.

Occupied h
ACR Tome _Lengt _Increment

Sub- removed Time Xo Sols Sire aelving Di
plot by interval Begin End (cm) (cm) (cm) (cm) values
B-1 R 29 Apr.- 0/53 3/53 0.00 0.0 5.15 6.1 4.898**
B-2 Cc 23 May 47/47 = 41/47 0.00 OOF 4365 se liles

C-1 RT 14 Mar.- 2/20 2/20) 16:25), 1528) 9 19100 eee 0.387
C-2 R 1/19 2/19 05 13.4 20.10 24.2 0.236
C-3 Cc 27 May 1/31 13/31 LOM67 1435) 211-49 eS 0G
D-1 R 10 Dec. 0/9 3/9 0.00 010) S095 1.101
D-2 R 0/12 6/12 0.00 OL0 28825 eZ el 1.347
D-3 (Cy 25 May 19/19 15/19 0.00 0.00 47.52 53.9
E-1 R 23 Mar.- 0/29 4/29 16.72 132 Aes3 eovee 2.887**
E-2 R 0/24 224 N837 154 187922 6.958**
E-3 (G 10 Jun. U2 21/20 2666 225/69 Oe
F-] R 6 Jan.- 11/42 19/42 22.00 13.5 8.38 17.4 1.664*
F-2 (C 24 Apr. 27/40 25/40 17.60 13.6 14.77 17.4
G-1 R 6 Jan.- 2/18 1/18 11.16 14.3 HAM 7.1 —1.114
G-2 R 10/26 D260) 25.00N D4 9.90 14.6 —1.818
G-3 (@; 24 Apr. 15/24 10/24 29.41 14.3 3:79) al 2a
H-1 R,P,T 11 Mar.- 0/43 0/43 15.34 13.0 —0.16 9.3 9.508**
H-2 RS Ie 0/66 2/66 24.12 16.6 1.00 13.6 9.144**
H-3 G 10 Jun. 65/72 67/72 43.80 21.1 39.68 33.5

I-1 R 13 Mar.- 12/38 14/38 12.63 19.0 8.89 15.2 0.003
I-2 R 0/32 7/32 13.84 16.8 153 11.3 2.585**
I-3 c 18 Apr. 33/39" 29/39) 21-71 234 8.90 12.7

I-] R 18 Apr.- 14/38 4 Soe 22 30 eG 18.8 2.060*
I-2 R 7/32 8/32 15:37 2128 9 22078) S26 — CLO o
I-3 Cc 10 Jun. 29/39) 27/39) 30.61) S06 228eess

percent level. In treatment subplot G-2 and F-1, the large length increment
(small Z value) was due to the presence of several vigorous shoots occupied
by large ant colonies. In plot D and C, there were several heavily occupied
shoots in each treatment subplot and a number of dead stumps in each control
subplot. A major portion of the variance of the length increments in all of the
subplots in table 4 was due to 1) the presence of unoccupied shoots in treat-
ment subplots, 2) unoccupied shoots in control subplots, 3) dead shoots that
were killed by Chrysobothris sp., near multistigmosa and continuous defolia-
tion by phytophagous insects, 4) damage to shoot tips by Coxina hadenoides
and Pelidnota punctulata, and 5) variable occupation of shoots in the control
subplots during the period of the height increment.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 497

In table 5, the length increments are recorded for eight of the nine plots
in table 4. The length increment was for the most part that which accrued
during the 1964 dry season. Of the 16 treatment-control contrasts, six of the
pairs have a highly significant difference in mean height increment, two
were found to be significantly different, and eight were not significantly dif-
ferent. The low Z values obtained in the contrasts in plot F and I were asso-
ciated with the presence of 1) lightly occupied shoots in the control subplots,
2) occupied shoots in the treatment subplots, 3) apparently lowered replace-
ment rates of the shoot tips destroyed by Coxina hadenoides (due to the dry
season), and 4) dead shoots in the control subplots. The low Z values ob-
tained in the eight contrasts that did not show significant differences were
associated with the same factors, but the presence of occupied shoots in treat-
ment subplots, and the lowered rate of shoot tip replacement by all shoots,
contributed the most to the reduction of differences between the control and
treatment subplots.

In table 6, the length increments are recorded for 7 of the nine plots dis-
cussed in table 4. The length increment was for the most part that which
accrued during the cool season, and the period associated with the end of the
1963 rainy season and the beginning of the 1963-1964 dry season. The 24
contrasts must be divided into two groups. The majority (18) are con-
trasts of treatment subplots with control subplots in the same manner as
presented in tables 4 and 5; they are possible because individual records were
kept for each shoot. Four of these contrasts have highly significant differ-
ences in mean length increment, one has a significant difference, and 13 do
not show significant differences. In the case of the other six contrasts, the
individual length increment can not be associated with the individual shoot.
The mean lengths of the shoots in the subplots can be contrasted at the start
of the interval and at the end of the interval. Four of these pairs of contrasts
(plot H) have highly significant differences both at the beginning and end of
the interval, and show increases in the Z value over the interval. The differ-
ences in mean length increment among the subplots in plot H appear great
enoufih during the period 19 Nov.-26 Dec., so that significantly different
length increments probably occurred. During the period 26 Dec.-I1 Mar., it
is doubtful that significantly different length increments occurred. In the
contrast of subplot I-1 with 1-3, the Z value changed from a negative number
to a significant value over the time interval. In the contrast of subplot I-2
and I-3, the negative Z value increased in magnitude during the time interval;
this was associated with the presence of occupied shoots in treatment subplot
1-2 and unoccupied shoots in control subplot I-3. Therefore, table 6 presents
six highly significant contrasts, two significant contrasts, and 16 contrasts of
means that are not significantly different.

During the rainy season, a higher frequency of significant Z values among
the subplot contrasts was recorded than during the dry or cool season (Tab.

Taste 6. As in Table 4 except that the majority of the days in the time intervals
occur in the cool season or before. Where two Z values are given on a single line,
the first represents a comparison of the shoot lengths at the beginning of the in-
terval and the second, the shoot lengths at the end of the interval. This was neces-
sary because the earlier records were not kept on an individual plant basis and
thus a standard deviation for the increment could not be calculated.

Occupied

ke Total 2 Length a Increment
Sub- removed Time Xo s.d.o Xine S.d.ine Z
plot by interval Begin End (cm) (cm) (cm) (cm) values
C-l Roe 5 Jan.- 0/20 2/20 0.00 00 1625 13.8 —1.381
C-2 R 0/19 1/19 0.00 0.0 15.05 13.4 —1.086
C-3 C 14 Mar. 0/31 L/St 0:00) 00) 106745
E-1 R 18 Nov.- 0/29 0/29 0.00 00 331 39 2.220*
E-2 R 0/24 70/247 {O00 0.0 S25) a0 1.221
E-3 (e: 26 Nov. 21/20) Af 21 O00) 10:0) 9 7.95) 90
E-1 R 26 Nov.- 0/29' 0/29" ssl 39) 1342s 1.152
E-2 R 0/24 ~~ 1/24 5.255" 5.0 “1Wsa2” 150 1.111
E-3 (e 23 Mar. M/Z 10/219 7.95 SSO is: 71ass
F-1 R 28 Oct.- 0/42) 20/42 0100) 0:0) 1292) 73 —0.187
F-2 (¢: 28 Nov. 40/40 26/40 0.00 0.0 12.15 25.0
F-1 R 28 Nov.- 20/42 11/42 12.92 7.3 9.08 9.5 —1.762
F-2 Cc 6 Jan. 26/40 27/40 12.15 25.0 545 92
G-1 R 28 Oct.- 0/18 7/18 0.00 0.0 14.72 10.4 0.977
G-2 R 0/26 6/26 000 00 1465 69 1.343
G-3 C 29 Nov. 24/24 16/24 0.00 0.0 17.70 89
G-1l R 29 Nov.- 7/18 2/18 .14.72 10.4 —3.56 13.9 3.688**
G-2 R 6/26 10/26 14.65 69 8.65 10.7 0.930
G-3 G 6 Jan. 16/24 15/24, 17.70 289 Wil 124
H-1 Rope 17 Oct.- 0/48 16/48 0.00 00 814 8.7 6.745**
H-2 R, P 0/69 18/69 0.00 0.0 12.47 8.6 4.746**
H-3 (e 19 Nov. 61/61 61/61 0.00 090.0 20.35 10.2
H-1 R,P,T 19 Nov.- 16/48 0/46 8.14 8.7 7.57 X 6.745** 8:850**
H-2 R, P 18/69 3/69 12.47 86 11.24 X 4.746** 6:015**
_H-3 GC 26 Dec. 61/61 48/68 2035 102 22.25 Xx
H-1 R,P,T 26 Dec. 0/46 0/43 15.34 13.4 —0.37 X 8.850** 8.949**
H-2 R, P 3/69 0/66 23.71 16.7 0.41 X 6.015** 6.101*
H-3 c 11 Mar. 48/68 65/72 42.60 19.9 1.20 XxX
I-1 R 18 Oct.- 0/38 0/38 0.00 0.0 11.89 10.6 —1.283
I-2 R 0/31. 0/52)" 0:00 saiOy oss 264 —0.197
I-3 GC 20 Nov. 39/39 13/37 0.00 00 9.08 83
I-] R 20 Nov.- 0/38 3/38 11.89 10.6 —1.68 X —1.283 2.066*
I-2 R 0/82 0/821 69.434) 46449 82h X —0.197 —0.225
I-3 CG 26 Dec. 13/37 ees /59 ) poLUS A homeos ».4
I-] R 26 Dec.- 3/38) 12/88 1000 16s wee 15-6 0.782
I-2 R 0/32 0/32 17.09 145 —3.25 6.0 2.766**
I-3 G 13 Mar. 3/39 33/39 16:30—15.7 541 184

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 499

14). This was due to several major factors. The treatment subplots in plots
A and B were established in such a manner that only a low incidence of
shoots remained occupied in spite of the treatments. Occupied shoots in all
control subplots had sufficient water and insolation for high growth rates,
which thereby accentuated the differences between them and the unoccupied
shoots in the treatment subplots. Phytophagous insects were much more
abundant during the rainy season than during any part of the preceding nine
months, and their damage to the unoccupied shoots was much more severe.
In the control subplots, a number of the colonies moved out into nearby shoots
to establish auxiliary-units, and thereby occupied a number of shoots that
were unoccupied or lightly occupied during part or all of the preceding nine
months. This also increased the variation within the treatment subplots be:
cause the ants sometimes moved into the treatment subplots. In all of the
treatment subplots except B-1 and A-1, many of the stumps that had been
unoccupied for 4-9 months were dead and had no height increment during
the rainy season. They produced a negative increment when they died.

During the dry season, cool season, and period before the cool season, the
lower frequency of significant Z values among the subplots was for the most
part associated with 1) lowered potential replacement rates of shoot tips, 2)
lowered incidence of phytophagous insects, 3) less thorough and intensive
occupation of auxiliary-shoots in the control subplots, 4) the dominant role of
Coxina hadenoides as a defoliator that is relatively insensitive to attack by
Pseudomyrmex ferruginea, 5) the ability of severely defoliated unoccupied
suckers, and associated stumps, to live for 1-6 months and continue to produce
occasional new shoots, and 6) the lack of shade from the surrounding leafless
vegetation in the dry season so that slow growing shoots were not shaded.

With the exception of the period 26 Dec.-11 Mar. (for the most part dur-
ing the cool season), plot H showed the highest and most consistent Z
values throughout the year. If plot A and B had been established at the same
time as plot H, it is very likely that they would have shown the same high Z
values. Two plots, C and D never showed significant differences in mean
length increment when treated as entire subplots. This was associated most
strongly with the small sample size and presence of large numbers of occu-
pied shoots in the treatment subplots, and unoccupied shoots in the control
subplots.

Subplot height increment records for existing shoots. In table 7, the
height increments are recorded for 13 plots which contain 26 subplots. The
shoots are all shoots that were in existence before the beginning of the experi-
ment. The height increment is that accrued for the most part during the
rainy season; plots U and V showed little height increment during the cool
and dry seasons. In the treatment subplots in plots K, Q, L, and P, thorn
clipping was used to remove the ant colony. In plots R, T, U, and V, para-
thion was used to remove the ant colonies and no applications were made

500 Tue University SciENcE BULLETIN

after 9 Dec. 1963. In the remaining plots (J, M, N, and O), the fall spray
applications were followed during the cool or dry season by thorn clipping to
remove the colonies that had reinvaded the thorns.

Of the 13 treatment-control subplot contrasts presented in table 7, 11 are
based on mapped subplots in which individual records were kept. Of these
11, seven have a highly significant difference in mean height increment, three
have a positive but not significant Z value, and only one has a negative Z
value but again not significant. In one unmapped plot (K), the Z value
changed from a negative value to a highly significant value which indicates
that there was a significant difference between the mean height increments
over the interval. In the other unmapped plot (V) the highly significant Z
value changed from 4.207** to 6.087** over the 10 month interval which indi-
cates a possible significant height increment during this period.

Since only three shoots in subplot J-2 were occupied by the end of the
rainy season, the low Z value from the contrast of J-1 with J-2 was expected.
The shoots in plot R were heavily shaded and the treatment and control sub-
plots contained nearly equal numbers of occupied shoots. A very large por-
tion of the shoots in the treatment subplots T-1 and U-1 were reoccupied by
auxiliary-units from outside of the subplots by beginning of the rainy season
so that these treatment and control subplots constitute a poor contrast of un-
occupied and occupied subplots. While six of the time intervals for the 13
contrasts include all or a portion of the dry season, the major portion of the
growth in all of the plots in table 7 took place during the 1964 rainy season.

In table 8, the height increments are recorded for 11 of the 13 plots in
table 7. These 11 plots contained 22 subplots. Of the seven mapped plots, two
have a highly significant difference in mean height increment. The shoots
in plot J were unoccupied during the interval. In plot M the growth of the
treatment subplot was not inhibited due to a lack of phytophagous insects
during the dry season, and the presence of small unanticipated colonies of P.
ferruginea that had developed in situ before the thorns were clipped. In plot
Q, the greater height increment in the treatment subplot was associated with
the production of flowering branches and the lack of defoliating insects. The
situation in plot R was the same as during the rainy season; there was shading
and a high number of occupied shoots in the treatment subplot.

Of the four unmapped plots, K and O showed changes in Z values over
the interval which indicate a significant difference in mean height increment
between the treatment and control subplots. The lack of an indication of a
significant difference in the mean height increment between the treatment
and control subplots in plot N was associated with the usual agents that tend
to equalize height increment rates in treatment and control subplots during
the 8 months preceding the rainy season. The considerable decrease in the Z
value for plot P from the beginning to the end of the interval was strongly

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT [NHABITANT 501

Tasre 7. As in Table 4 (rainy season) except that the values are height, and

height increment instead of length and length increment. This is because these

values are for the existing shoots, rather than for suckers from stumps cut during
the treatment schedule.

Occupied ;

Ree aE = Height a Increment
Sub- removed Time Xo s.d.o Xine s$.d.ine i,
plot by interval Begin End (cm) (cm) (cm) (cm) values
J-1 1240 25 Apr.- 0/29 OP) PAT il Sallsy BMS 1.073
J-2 Cc 23 Jul. 0/38 Byes! GEG 7) A) Nee
K-1 I 4 Apr.- 0/205 0/266 35.76 11.5 —2.09 X —2.530 4.839**
K-2 C lAug. 48/138 49/149 32.06 14.5 12.81 X
L-l at 26 May- 0/44 0/44 69.04 28.6 —2.23 17.3 3) feb shee
L-2 ¢c 31 Jul. 28/52 38/52. 85:03 323° 13.07 216
M-1 Peal 26 May- 0/79 0/79 116.75 61.4 —1.35 23.1 9 .066**
M-2 Cc 31 Jul. 59/88 59/88 87.81 39.2 26.03 21.7
N-1 Pali 3 Apr.- 0/81 0/81 100.92 372 447 10.8 a7
N-2 & 29 Jul. 43/56 53/56 93.76 42.7 80.22 48.6
O-1 Pa 3 Apr.- 0/38 OSS > SONS Or ators elated 10.083**
O-2 ¢c 29 Jul. 30/34." 34/34 102-73 385) 92.79. 49.1
P-1 ae 24 Mar.- 7/61 0/61 111.40 34.3 —4.76 19.2 092"
P-2 C 2 Aug. 28/46 34/46 100.26 64.8 61.82 74.9
Q-1 aw 30 May- 0/149 0/149 101.96 46.1 —0.74 3.1 8.660**
Q-2 (e 6 Aug. 80/128 90/128 118.29 40.8 52.63 53.6
R-1] 1p 28May- 14/42 14/42 116.02 112.5 —9.65 23.0 1.370
R-2 C 27 Jul. 11/48 12/48 58.91 73.4 —2.32 36.3
S-1 P 6May- 55/211 95/211.166.54 91.7 10.41 34.8 5.864 **
S-2 © 4Aug- 175/251 195/251 15863 769° 32.05 “44-4
T-1 P 8May- 50/116 102/116 195.75 94.1 12.87 40.7 1.439
T-2 Cc 29 Jul. 78/104 90/104 201.79 94.9 20.30 35.8
U-1 P 21Jan.- 39/86 78/86 363.18 103.2 11.45 28.2 —1.073
U-2 C 5 Jul. 73/93. 185/93) 335.05— 927, , 2.90 8749
v-1 P 8 Oct.- 0/170 133/187 250.44 128.9 11.88 X 4.207** 6.087**
V-2 C 14Jul. 191/192 192/205 308.93 135.5 39.22 Xx

associated with the lack of any kind of growth in the control subplot and the
production of flowering branches as a reaction to insect damage in the treat-
ment subplot.

In table 9, the height increments are recorded for eight of the 13 plots in
table 7. These contained 16 subplots. The one mapped plot (J) does not
show a significant difference in height increment; none of its shoots were
occupied. Of the unmapped plots, only the contrasts of heights in plots N and
O during October and November show an indication of a significant differ-

502 Tue University ScteENcE BULLETIN

Taste 8. As in Table 5 (dry season) except for the changes necessary to account
for the fact that these are existing shoots rather than suckers from experimentally

cut stumps.
Occupied : .
ners Total " Height = Increment
Sub- removed Time Xo s.d.o Xine S.d.ine Z
plot by interval Begin End (cm) (cm) (cm) (cm) values
J-1 Pel 15 Jan.- 0/29 O¥29 5 22724) 28:9" 03s —1.010
J-2 c 25 Apr. 0/38 0/38 25.26 7.9 —050 4.5
K-1 Gc 22 Dec.- 0/178 9/205 38.84 15.0 —3.06 X —4.329 —2.530
K-2 c 4Apr. 42/153 48/138 32.00 13.8 0.06 xX
L-1l alt 18 Jan.- 0/44 0/44 65.00 263 4.04 29.2 Suse
L-2 o. 26 May 20/52 27/52 64.36 20.3 20.67 21.4
M-1 PT 18 Jan.- 6/79 0/79 96.60 34.3 20.15 44.0 0.933
M-2 c 26 May 59/88 59/88 61.78 25.6 26.03 37.2
N-1 Pale 20 Dec.- 20/81 49/81 105.92 62.2 —5.00 X —1.033 —1.015
N-2 © 3 Apr. 51/56 43/56 97.50 32.4 —3.74 x
O-1 Pa 20 Dec.- 5/38 0/38 88.42 40.8 —2.53 X —0.573 1.898*
0-2 ( 3 Apr. 32/34 32/34 81.02 649 21.71 ».4
P-] ar 28 Dec.- 0/86 8/61 102.13 384 9.27 X 0.424 —1.060
P-2 c 24Mar. 26/47 28/46 105.74 50.9 —5.48 xX
Q-1 ae 16 Mar.- 7/149 0/149 93.28 49.0 8.68 9.9 —1.216
Q-2 Cc 30 May 94/128 80/128 113.10 43.4 5.19 31.2
R-1] P 9Dec.- 18/42 14/42 121.83 100.1 —5.81 17.9 —1.960
R-2 Cc 28 May 3/48 11/48 73.54 70.8—14.63 24.6
S-1 I 11 Jan.- 22/211 55/211 164.62 107.9 —1.92 41.7 2.480**
S-2 C 6May 131/251 175/251 149.71 90.2 8.92 52.1
T-1l Ie 6Jan.- 17/116 50/116 194.87 93.7 0.88 5.6 0.038
T-2 ec 8 May 82/104 78/104 200.87 94.8 0.92 9.1

ence in the mean height increment over the interval. It should be empha-
sized that the height increments in plots N and O that were presented in
table 7 and 8 only represent the shoots in these plots that were taller than 50
cm on 3 Apr. 1964. The sudden burst of growth in subplot M-1 was very
likely associated with parathion stimulation and possibly with insecticide
mortality to the general defoliators in the plot. The lack of significant dif-
ferences in the contrasts presented in table 9 was for the most part associated
with the lack of growth among all shoots during this time of year, the pres-
ence of unoccupied shoots in control subplots, and the presence of occupied
shoots in treatment subplots.

Height or length increments of occupied shoots contrasted to unoccupied
shoots. In the mapped plots, it was possible to determine which shoots re-

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 503

Taste 9. As in Table 6 (cool season) except for the changes necessary to account
for the fact that these are existing shoots rather than suckers from experimentally

cut stumps.
Occupied

Rats eee _Height _Increment
Sub- removed Time Xo s.d.o Xine S.d.inc Vd
plot by interval Begin End (cm) (cm) (cm) (cm) values
J-1 Pe 12 Dec.- 0/29 0/29 23.93 7.8 —1.69 3.4 1.357
J-2 c 15 Jan. 0/38 0/38 25.81 7.8 —055 3.4
L-1 ae 12 Dec.- 0/44 0/44 68.97 27.8 —3.97 X —0.383 —0.131
L-2 Cc 18 Jan. 7/69 20/52 66.95 264 —2.59 xX
M-1 PAE 28 Oct.- 7/97 27/85 61.30 23.7 36.46 X 5.466**—7.553
M-2 Cc 6 Dec. 7/72 54/101 85.68 31.9 —21.08 xX
M-1 ee 6 Dec. 27/85 7/79 97.76 31.1 —1.16 X —7.553 —7.361
M-2 Cc 18 Jan. 54/101 59/88 64.60 28.2 2:32) ex
N-1 js at 12 Oct.- ?/157 32/368 54.75 33.1.—15.98 X—3.517 1.202
N-2 Cc 14 Nov. P/136) 53/243) 42:65) 25.7 —ll> &
N-1 PE 14.Nov.- 32/368 35/342 38.77 38.2 9.00 X 1.202 —1.187
N-2 (e 20 Dec. 53/243 44/240 41.50 14.9 V/s .6
O-1 jpak 12 Oct.- 2/83 8/146 39.60 33.2 —058 XK—0.905 0.664
O-2 Cc 14 Nov. ?/179 40/145 35.66 31.9 465 xX
O-1 Pat 14 Nov. 8/146 9/183 39.02 13.9 427 X 0.664 —0.577
O-2 (G 20 Dec. 40/145 31/181 40.31 18.9 104° xX
S-l P 28 Oct.- 2/149 22/211 179.52 128.8 —14.95 X —1.636 —1.594
S-2 C 11 Jan. ?/226 131/251 160.37 74.7 —10.66 xX
T-1l 2 25 Oct.- 0/93 17/116 194.05 90.2 0:82 xe L275) 10472
6 ey € 6 Jan. 93/94 82/104 211.11 91.9 —10.24 xX
U-1 iP 28 Oct.- ?/99 =39/86 356.36 105.6 6.82 XxX —1.162 —1.913
U-2 (© 21 Jan. ?/98 78/93 340.20 89.2 —5.15 xX

mained occupied and unoccupied for one or more intervals. By selecting the
continuously occupied and the continuously unoccupied shoots from the en-
tire plot (i.e., from both the treatment and control subplots), it was possible
to remove a large part of the variation in length and height increments
associated with the presence of occupied shoots in treatment plots, and unoc-
cupied shoots in control plots. This selection also removes the variation asso-
ciated with the abandonment of auxiliary-shoots during the dry season, and
reinvasion of these shoots during the rainy season. The continuously occu-
pied shoots contained queen-units and large auxiliary-units. The continu-
ously unoccupied shoots were those treated shoots which were not reinvaded,
or in which colonies had not yet developed in situ. There were a few shoots in
the control subplots which were naturally unoccupied due to failure of a

504 Tue UNiversiry ScrENCE BULLETIN

colony to develop in situ, abandonment of the shoot by the colony during the
dry season, or destruction of the queen-unit in the neighboring treatment plot.

The samples of occupied shoots used in tables 10-13 were recorded as occu-
pied or unoccupied consistently at each recording, from the earliest date
listed for the particular plot, until the last date for the particular plot. For
example, for a shoot to be used in any of the length increment means for
plot G, it must have been recorded as occupied or unoccupied on 28 Oct., 29
Noy., 6 Jan., 24 Apr., and 6 Jul. For any given plot, as many intervals are
presented in tables 10-13 as allows a reasonably large sample size; the more
intervals, the fewer consistently occupied or unoccupied shoots that can be
pooled from a single plot. The results based on this segregation of samples
by ant presence or absence are presented under the following subtitles. They
are also tabulated in table 14.

1. Length increment of suckers. Table 10 contains the length increment
data for nine plots and 10 intervals during the 1964 rainy season. In the 10
contrasts between the mean length increments of occupied and unoccupied
shoots, all of the Z values are highly significant. The 10 intervals yielded 329
length increments for occupied shoots and 433 length increments for unoccu-
pied shoots. There is a total increase of 31,566 cm for the 329 occupied incre-
ments and a total increase of 1,487 cm for the 433 unoccupied increments.
The maximum daily mean length increment was 2.079 cm and was recorded
in plot I for 36 occupied shoots over 57 days. This value is very close to that
of 2.50 cm per day that was recorded for nine occupied shoots from 10-28
Jul.; these nine shoots were known to be undamaged over the interval. The
minimum daily mean length increment of —0.026 cm was also recorded in
plot I; it was for 50 unoccupied shoots over a 57 day period.

In Table 10, the high increment rates of occupied shoots, in contrast to the
low increment rates of unoccupied shoots, were definitely correlated with the
lack of insect damage and shading of the shoot tips, both before and during
the interval. The low increment rates were definitely correlated with shad-
ing and severe insect defoliation of the unoccupied shoots, both before and
during the interval. Many of the shoots that had been unoccupied for two
or more months were dead.

Table 11 contains the length increment data for seven plots and 10 inter-
vals during the nine months preceding the 1964 rainy season. Of the 10 con-
trasts, nine of the Z values are highly significant. The 10 intervals yield 261
length increments for occupied shoots and 379 length increments for unoccu-
pied shoots. There is a total increase of 10,211 cm for the 261 occupied incre-
ments and a total increase of 1,880 cm for the 379 unoccupied increments.
The maximum mean daily length increment of 1.54 cm was recorded in plot
E for 21 occupied shoots over 79 days. The minimum daily length increment
of -0.0146 cm was recorded in plot F for 26 unoccupied shoots over a 91 day

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 505

Taste 10. Length increment data for suckers from stumps, treated as occupied
and unoccupied groups rather than by subplots as in Table 4. Time interval
during the rainy season. Terminology as in Table 4.

Number Length __Increment
Time of Xo s.d.o Xine s.d.ine Z
Plot interval Occupation shoots (cm) (cm) (cm) (cm) values
A 25 May- unoccupied 41 0.00 0.0 5.78 4.8 Wtlip
16 Jun. occupied 25 0.00 0.0 33.00 17.1
A 16 Jun.- unoccupied 41 5.78 4.8 11.78 9.4 8.397**
3 Aug. occupied 25 33.00 17.1 77.28 58.3
B 23 May- unoccupied 36 4.13 ell 22a 21.6 S981 ="
17 Jun. occupied 42 16.52 10.0 60.76 15.3
B 17 Jun.- unoccupied 36 26.27 6.4 7.24 il7/ 14.321**
7 Aug. occupied 42 77.28 14.2 117.28 19.4
Cc 27 May- unoccupied 47 25.31 28.5 0.68 20.3 0/22"
6 Aug. occupied 16 62.80 SU) AS) 52.4
D 25 May- unoccupied 13} 21.76 24.3 —0.61 Ziles DLW io
28 Jul. occupied 22 52.18 405 56.40 4371
F 24 Apr.- —_ unoccupied 26 18.03 13.9 4.69 10.8 8.870**
6 Jul. occupied 26 46.23 11.8 68.91 35.3
G 24 Apr.- unoccupied 25 NS 14.1 5.50 11.8 Oo =
6 Jul. occupied 15 42.46 15.0 83.66 775)
H 10 Jun.- unoccupied 79 13.50 15.5 —0.97 11.2 13.679**
5 Aug. occupied 59 96.05 41.1 114.62 64.2
I 10 Jun.- unoccupied 50 5.36 10.2 —1.30 Dell 20.807**
6 Aug. occupied 36 80.44 40.2 118.55 34.3

period. The single contrast with a very low Z value (plot G) is from a plot
in which there was a consistent removal of shoot tips by Coxina hadenoides
from both occupied and unoccupied shoots.

The length increments recorded for occupied shoots in the cool and dry
season (Tab. 11) were lower than those recorded for occupied shoots in the
rainy season (Tab. 10). During the cool and dry season there was a lowered
replacement rate of shoot tips so that when a shoot tip was destroyed (e.g., by
Coxina hadenoides), more time was required for a positive height increment
to begin. During the cool season, the nights were often cold enough so that
shoots that were recorded as occupied during the day did not have workers on
their outer surface at night. During the dry season, the number of workers
active on the surface of the shoot was generally reduced even at high tempera-
tures; occasionally an insect could eat all or part of a shoot tip without com-
ing in contact with an aggressive worker.

506 Tue University ScreENcE BULLETIN

Tasce 11. As in table 10 (rainy season) except time interval during the dry

season.
Nees Length __Increment
Time of Xo s.d.o Xine s.d.ine Z
Plot interval Occupation shoots (cm) (cm) (cm) (cm) values
c 17 Mar.- unoccupied 47 15.19 125 10.12 19.8 4.181**
27 May occupied 16 WES 7, 18.7 SetS Sei
D 10 Dec.- unoccupied 13 0.00 0.0 21.76 24.3 DAS Se
25 May occupied 22 0.00 0.0 52.18 49.3
E 23 Mar.- unoccupied 39 14.87 EA 7A 7/Al 26.3 133565"
10 Jun. occupied 21 26.66 22 22200 30.2
F 6 Jan.- unoccupied 26 17.70 11.3 —0.33 95 4.298**
24 Apr. occupied 26 27.00 12.8 1923 PN)
G 28 Oct.- unoccupied 25 0.00 0.0 11.80 3D 4.077**
29 Nov. occupied 15 0.00 0.0 21.26 79
G 29 Nov.- — unoccupied DS, 11.80 55 —0.11 8.1 ».325"*
6 Jan. occupied 15 21.26 Ws) 154 95
G 6 Jan.- unoccupied 25 11.69 10.2 5.46 10.3 0.059
24 Apr. occupied 15 36.80 13.6 5.66 10.2
H 11 Mar.- —_ unoccupied 79 14.83 18.1 —1.33 15, LO0i435"*
10 Jun. occupied 59 49.69 18.1 46.36 32.6
I 13 Mar.- —_- unoccupied 50 4.34 3.1 —0.30 3.0 8.836**
18 Apr. occupied 36 30.25 21.0 17.61 12.2
I 18 Apr.- unoccupied 50 4.04 ks 1.22 8.2 O22
10 Jun. occupied 36 47.86 S053 32.58 29.2

The length increments recorded for unoccupied shoots during the dry
season (Tab. 11) are higher than those recorded for unoccupied shoots dur-
ing the rainy season (Tab. 10). During the cool and dry season there was a
severe reduction in general insect density in the plots and associated with this,
an unoccupied shoot was occasionally recorded as having an intact terminal
shoot tip for as long as a month. Since it required 1-9 months for an unoccu-
pied shoot to sustain enough damage to kill it, many of the unoccupied
shoots included in table 11 were not dead. Many more of them were dead
during the interval in the rainy season (Tab. 10). These dead shoots are in-
cluded because they are dead owing to the lack of ants to keep them alive.

Both the experimental plots and observations of suckers from naturally
occurring stumps indicated that the data presented in tables 10-11 are rep-
resentative of length increment rates in both naturally occurring and man-
made disturbance sites in the area between Temascal and La Granja. At the
present, at least 99 per cent of the shoots that reach maturity in this area are
occupied sucker shoots from stumps that were cut or burned. At any partic-

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 507

ular season, the differences between the increment rates of occupied and un-
occupied shoots are due to the relatively severe insect damage to unoccupied
shoots, and the inability of unoccupied shoots to maintain an emergent posi-
tion in the general vegetation canopy.

2. Height increments of existing shoots. Table 12 contains the height in-
crement data for occupied and unoccupied shoots for seven plots and seven
intervals during the 1964 rainy season. In all of the seven contrasts, the Z
values are highly significant. The seven intervals yielded 413 height incre-
ments for the occupied shoots and 716 height increments for the unoccupied
shoots. There was a total increase of 38,056 cm for the 413 occupied incre-
ments and —1,324 cm for the 716 unoccupied increments. The maximum
daily mean height increment was 1.58 cm and was recorded in plots N and O
combined for 95 occupied shoots over a period of 119 days. The minimum
daily mean height increment was —0.26 cm and was recorded in plot R for
57 unoccupied shoots over a period of 59 days.

The variation in height increment of occupied shoots between plots, dur-
ing the rainy season, was primarily due to local differences in abundance of
those phytophagous insects that can feed in the presence of P. ferruginea.

Table 13 contains the height increment data for occupied and unoccupied
shoots in four plots and four intervals during the 1964 dry season. In three of
the four contrasts, the Z values are highly significant. The fourth contrast
(Q) does not have a significant Z value. The four intervals yielded 267 height
increments for the occupied shoots and 466 height increments for the unoccu-
pied shoots. There was a total increase of 3,634 cm for the 267 occupied
shoot increments and 1,621 cm for the 466 unoccupied shoot increments. The
maximum mean daily height increment was 0.29 cm and was recorded in
plot M for 44 shoots over a period of 128 days. The minimum daily mean
height increment was —0.05 cm and was recorded in plot S for 181 shoots
over a period of 114 days.

The variation in height increment of occupied shoots during the dry sea-
son was primarily due to local differences in abundance of Coxina hadenoides
(Q), and increasing dry season dormancy with increasing age of the shoot
(S). In addition to the above two major factors, the variation in height incre-
ment of unoccupied shoots during the dry season was due to the production of
short vertical flowering branches (Q and P) and cutting of the shorter
shoots by Sigmodon hispidus (S).

Discussion. It is evident from the data presented in tables 4-13 that the
presence of the workers of a colony of Pseudomyrmex ferruginea on a shoot
of Acacia cornigera has an effect on the length increment of the shoot (Fig.
53-59). That the presence or absence of this effect, and its magnitude, varies
with the time of year and the size of the shoot, can be clearly seen in table
14. The time of year affects the abundance of phytophagous insects, the
physiological growth rates of the shoot, and the shading effects of the sur-

508

Tue University ScreNcE BULLETIN

Taste 12. Height increment data for existing shoots treated as occupied and
unoccupied groups rather than by subplots as in Table 7. Time interval during
the rainy season.

Nariner = = Height _ Increment
Time of Xo S.d.o Xine s.d.ine Z
Plot interval Occupation shoots (cm) (cm) (cm) (cm) values
i 26 May- unoccupied 49 68.27 27.1 —2.30 15.5 S 6llgze
31 Jul. occupied 12. 103.66 3357, 20.16 20.1
M 26 May- unoccupied 77 ~—«107.94 61.2 —0.38 14.6 8.990 **
31 Jul. occupied 44 — 109.06 17.4 45.11 40.0
N, O 3 Apr.- unoccupied 122 95.98 Sail 5.47 15a! 28.534**
31 Jul. occupied 95 100.68 43.5 188.66 61.2
P 24 Mar.- —_ unoccupied 71 103.91 34.4 —9.78 29 193344" *
2 Aug. occupied 34 B02 Gg MISE! Sf
Q 30 May- unoccupied 159 96.87 47.2 —1.10 il 1516272 *
6 Aug. occupied 63°) L445 Bayi 79.85 421
R 28 May- unoccupied Sil 58.82 88.8 —15.68 53.4 S25 0ee
26 Jul. occupied I SEL BG 23.00 39.2
S 5 May- unoccupied 181 122.99 78.9 —0.48 24.1 11590**
3 Aug. occupied 1435 200536 78.8 43.47 43.2
Taste 13. As in Table 12 (rainy season) except time interval during the dry
season.
Naabes Height __Increrment
Time of Xo 5.d.o Xine S:d-ine Z
Plot interval Occupation shoots (cm) (cm) (cm) (cm) values
iW, 18 Jan.- unoccupied 49 62.90 24.8 5.37 28.0 3.934**
26 May occupied 12 74.50 24.6 29.16 OFS
M 18 Jan.- unoccupied 77 93.16 35.8 14.78 Z5R2 Se
26 May occupied 44 72.36 21.6 36.70 18.2
P 19 Mar.- — unoccupied 159 89.27 29.5 7.60 31.0 0.437
30 May occupied 68 132.48 41.6 8.97 16.0
S 11 Jan.- unoccupied 181 128.45 70.2 —5.46 47.2 3:193%*
5 May occupied 143 lO 2295 T&A TAL 11.2

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 509

Fic. 53 (left). A 65 cm tall cut stump of Acacia cornigera in control subplot H-3. The
regenerating branches are occupied by P. ferruginea and the stump has been cut for about 35
days. There are three number 4 shoot tips on the stump. This is a representative shoot. Photo
late November 1963.

Fic. 54 (right). Two 65 em tall cut stumps of Acacia cornigera in treatment subplot H-1.
The regenerating branches are unoccupied and the stumps have been cut for about 35 days.
This stump would receive a rating of 1.5 because it is still alive. The unoccupied foliage was
eaten by adult tettigoniid grasshoppers. These are representative shoots. Photo same date as
in Fig. 53.

rounding vegetation. In those places where Sigmodon hispidus was present,
the season influenced the amount of damage done by this rodent since it fed
on Acacia cornigera for the most part only during the dry season. The size
of the shoot affected its ability to withstand both insect and rodent damage.
The size of the shoot also affected its responsiveness to changes in weather
(e.g., from the dry season to the rainy season) which in turn affected the ants’
opportunity to influence the shoots’ height increment.

The lower Z values in the subplot contrasts were for the most part associ-
ated with the presence of unoccupied shoots in the control subplots, and of
occupied shoots in the treatment subplots. These shoots were present because
of the gradual reinvasion of the shoots in the treatment subplots by large colo-
nies, the development of colonies in situ in the treatment subplots, and the
partial or total abandonment of auxiliary-shoots in the control subplots during
the cool and dry seasons. When groups of consistently occupied or unoccu-

510 Tue University SctENcE BULLETIN

Fic. 55. A representative sucker shoot from a cut stump of Acacia cornigera in control sub-
plot H-3. The shoot is occupied and has a number four shoot tip. The stump was cut on
17 Oct. 1963. Photo 21 Dec. 1963. This is the same stump as the one in Fig. 57.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 511

Fic. 56. Three stumps of Acacia cornigera in treatment subplot I-2. The two shoots on the
left were reinvaded by auxiliary-units from shoots outside of the plot. The shoots on the right
remained unoccupied until the photograph was taken. Shortly after this photograph, it became
occupied by a unit from the shoots on the left; by July 1964 it was over 200 cm tall. Most
of the unoccupied shoots in plot I continued to look like the one on the right throughout the
entire experiment. Photo early January 1964.

512 Tue University SciENCE BULLETIN

Fic. 57 (left). The same shoot as the one shown in Fig. 55. A heavily occupied shoot in
control subplot H-3, it was 305 cm tall at the time the photograph was taken. Note the
slender pole-like life form, despite the fact that the surrounding vegetation was only about 70
cm tall. This shoot is representative of most shoots in control subplot H-1, I-3, E-3, and F-2, al-
though many were more spreading as in Figure 58. The stump was cut on 17 Oct. 1963. Photo
5 Aug. 1964.

Fic. 58 (right). A heavily occupied shoot in control subplot E-3. The shoot was 365 cm tall
when the photograph was taken on 6 Aug. 1964. This shoot is representative of occupied
shoots regenerating from large stumps in the control subplots. This shoot was cut on 18
Nov. 1963.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT

D1

*

$l
Fic. 59. A representative unoccupied shoot in treatment subplot E-3. The stump is the

same size as the stumps of the shoots in Fig. 57 and 58. This shoot would be rated as a

number 2.5. All damage is phytophagous insect damage. It is representative of the unoccupied

shoots in treatment subplots H-1, H-2, C-1, C-2

5-2, E-1, E-2, F-1, G-1, G-2, I-1, and 1-2. The
stump was cut on 18 Nov. 1963. Photo 6 Aug. 1964.

514 Tue UNiversiry SciENCE BULLETIN

Taste 14. The distribution of significant Z values among contrasts of length or

height increment of Acacia cornigera made at different times of the year. Note

that the highest frequencies of Z values are associated with recordings made

during the rainy season of subplots, or groups, of suckers from stumps. Those

contrasts where the standard deviation of the increment was not available are
omitted from this table.

Number Number
Time Number highly Number not Type
of of significant significant significant — of
Table year contrasts contrasts contrasts contrasts shoots
(treated as
subplots )
4 rainy 17 11 1 5 suckers
3 dry 16 6 2 8 suckers
6 cool & 18 4 1 13 suckers
before
if rainy 11 7 0 4 existing shoots
8 dry 7 2 0 > existing shoots
9 cool 1 0 0 1 existing shoots
(treated
as groups
of shoots)
10 rainy 10 10 0 0 suckers
11 dry 10 9 0 1 suckers
12 rainy 7 7 0 0 existing shoots
13 dry 4 3 0 1 existing shoots

pied shoots were compared (Tab. 10-13), these sources of variation were ex-
cluded. A relatively important source of variation in height increment among
occupied shoots that was not removed by grouping occupied and unoccupied
shoots was that associated with the size of the colony; the larger the number
of workers active outside of the thorns, the larger the number of intact shoot
tips on the shoot. A contrast between consistently occupied or unoccupied
shoots is primarily one between healthy undamaged shoots that are that way
because they contain a large worker force, and damaged and dying shoots
that are that way because they have no worker force.

During the rainy season, the high density of phytophagous insects, and
the high growth rate of undamaged shoots, tends to emphasize the differences
between the occupied and unoccupied shoots. Better than 90 percent of the
height increment of Acacia cornigera occurs during the five months between
1 Jun. and 1 Dec. It is likewise at this time when the highest densities of
Pseudomyrmex ferruginea on occupied shoots are recorded. Superficial ob-
servation of populations of A. cornigera in other parts of the study area indi-
cates that the length and height increment parameters presented in tables 4-13

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 515

are representative of those throughout the study area, provided that the shoots
are of comparable size and growing in similar plant communities.

Condition of the shoot. Significance. The mean condition of the shoots in
a subplot, or in a group of occupied or unoccupied shoots, is a measure of the
amount of growth that is taking place at the time. For the shoot to produce
new leaves and branches, it has to maintain undamaged shoot tips for a cer-
tain period. Even when the shoot tips are damaged, and few new leaves are
being produced (during the cool and dry seasons), the condition is a measure
of the amount of damage sustained by the mature leaves and branches. In
treatment subplots, and among groups of unoccupied shoots, the condition of
the shoots serves as a relatively instantaneous measure of the activity of insects
that feed on A. cornigera. Since condition records do not cover an interval,
they can be more readily segregated into those that are recorded during the
various seasons. In control plots and groups of occupied shoots, condition
records serve as a measure of the degree of occupation of the shoots, and of
the density of insects that are relatively unaffected by the presence of P. fer-
ruginea. The condition of the shoots is also a cumulative measure of the
impact of phytophagous insects on the shoots. It often takes a month or more
for a shoot to be sufficiently damaged to receive a condition rating below 3.

Height increment and condition are not independent variables. Height
increment is a cumulative function of the condition of the uppermost shoot
tip of the shoot over an interval of time. However, the condition of the upper-
most shoot tip has a major influence only on the value of the condition rating
for values 3, 3.5, and 4. The rating of a shoot could decrease from 3-1 without
a change in shoot height. When the shoot is dying, and the condition rating
falls below 2, a negative height increment is often recorded because of the loss
of branches.

The condition of the unoccupied shoots on a site is often a very sensitive
indicator of rapid changes in the density of phytophagous insects that attack
A. cornigera. It was often the case in Oct. and Nov. 1963 that within five
days after removing the ant colony from a shoot with parathion, the previ-
ously undamaged shoot tips were all eaten by Coleoptera, Orthoptera, and/or
Lepidoptera. At the relatively sudden onset of the cool season about 1 Dec.
1963, there was a sharp reduction in the number of intact shoot tips on shoots
that were occupied during the day. At night the temperatures were so low
that the shoots were unoccupied, and adult tettigoniid grasshoppers and the
larvae of Coxina hadenoides fed unhindered on the shoot tips. When the
first rains fell at the beginning of the 1964 rainy season (20-23 May) there was
a sudden hatch of Diplotaxis denigrata. These nocturnal feeders destroyed
almost every previously intact shoot tip on unoccupied shoots in plots L and
M within three days.

Subplot records of condition for suckers from stumps. The condition
records for nine plots and 17 sets of independent recordings from sucker

516 Tue University SciENCE BULLETIN

growth during the rainy season are given in table 15. There are 28 possible
contrasts between control and treatment subplots. Of these 28 contrasts, 17
have a highly significant value, four have a significant Z value, and seven do
not have a significant Z value. The low value in plot B on 17 June was most
strongly associated with a very high density of Coxina hadenoides which low-
ered the condition value for the control subplot, and a low density of those
insects which feed on unoccupied Acacia cornigera. This latter factor was
apparently associated with the relatively barren aspect of the plant commu-
nity in plot B. In plot D and F, the low Z values were associated with high
numbers of occupied shoots in the treatment subplots. The negative values in
plot C were associated with the fact that by the beginning of the rainy season
almost every unoccupied shoot in control subplot C-3 was dead or nearly dead
and thus had a condition rating of 0-1.

The condition records for six plots and seven sets of independent recordings
during the dry season are given in table 14. There are 13 possible contrasts
and of these, seven have a highly significant Z value, one has a significant Z
value, and five do not have a significant Z value. The low Z values in all con-
trasts in table 16 are primarily associated with the general lack of insects that
feed on unoccupied shoots of A. cornigera during the dry season. In plot F
and G, the situation was further complicated by the presence of occupied
shoots in the control subplots. Plot C is particularly instructive. The shoots
were cut on 5 Jan. 1964 and the canopies removed from the stumps in the
control as well as in the treatment subplot. The canopies were not placed on
the stumps in the control subplot until 14 Mar. 1964. Therefore, the control
subplot was essentially unoccupied during the period preceding the record of
condition on 14 Mar.

The condition records for four plots and four sets of independent record-
ings during the cool season are given in table 17. There are seven possible con-
trasts and of these, four have a highly significant Z value while three do not
have significant Z values. The three low Z values are associated with the
presence of occupied shoots in treatment plots and a reduction of phytopha-
gous insect density during the cool season. Plot I shows the effect of previous
occupation on condition values. Nearly all of the occupied shoots in control
subplot I-3 were occupied as small auxiliary-units. When the weather turned
cold about 1 Dec., most of them were merged with the queen-unit and this
left only three occupied shoots out of 39. Since the weather was cold there
was little damage that occurred immediately to the vacated shoots. The
shoots in the two treatment subplots had low condition values because of the
cumulative effect of a month of insect damage in November.

The condition records of three plots and three sets of independent record-
ings during the end of the rainy season and shortly before the cool season are
given in table 18. Of the five possible contrasts between control and treatment
subplots, one has a highly significant Z value, one has a significant Z value,

Taste 15. The condition values of the sucker shoots for which length increments
were recorded and treated by subplots (Table 4-6). All recordings made during
the rainy season.

Number
shoots

Sub- RRS x 2
plot Date total Condition a sel= value
A-l 16 Jun. 0/42 3.309 0.731 3.514**
A-2 16 Jun. 26/29 3.896 0.310
A-l 3 Aug. 0/42 2.654 0.628 7.625"*
A-2 3 Aug. 29/29 3.706 0.116
B-1 23 May 3/53 2.839 1.372 6.236**
B-2 23 May 41/47 3.893 0.375
B-1 17 Jun. 7/53 3.245 0.871 2.187*
B-2 17 Jun. 44/47 3.595 0.441
B-1 7 Aug. 19/53 2.641 1.639 SAV:
B-2 7 Aug. 45/47 3.744 0.626
C-1 27 May 2/20 3.100 0.384 —2.126
C-2 27 May 2/19 3.289 0.592 —2.645
C-3 27 May 13/31 2.564 1.379
C-l 6 Aug. 6/20 2.150 1.488 —0.788
C-2 6 Aug. 5/19 2.184 1.811 —0(.789
C-3 6 Aug. 14/31 1.838 2.998
D-1 25 May 3/9 2.666 1.187 1.320
D-2 25 May 6/12 2.833 0.879 1.128
D-3 25 May 15/19 3.210 0.731
D-1 28 Jul. 3/9 2.444 1.528 ee
D-2 28 Jul. 7/12 2.291 1.657 2.294*
D-3 28 Jul. 16/19 3.289 0.981
E-] 10 Jun. 4/29 3.120 0.529 Byefoil its}
E-2 10 Jun. 2/24 2.479 1.249 6.321**
E-3 10 Jun. 21/21 3.952 0.048
E-1 7 Aug. 6/29 2.310 | fa5y7/ Os
E-2 7 Aug. 4/24 1.333 1.666 OVE 3
E-3 7 Aug. 21/21 3.976 0.012
F-] 6 Jul. 26/42 2.726 0.966 1.900*
F-2 6 Jul. 29/40 3.087 0.524
G-1 6 Jul. 6/18 2.166 0.647 A2NSe*
G-2 6 Jul. 9/26 3.019 0.350 1.236
G-3 6 Jul. 19/24 3.291 0.846
H-1 10 Jun. 0/43 1.988 il Sly 7.843**
H-2 10 Jun. 2/66 2.393 1.088 8.027**
H-3 10 Jun. 67/72 3.541 0.294
H-1 5 Aug. 0/43 1.058 1.276 SSSA
H-2 5 Aug. 32/66 2.166 2.327 HADEN)
H-3 5 Aug. 64/72 3.576 0.349
I-] 10 Jun. 13/38 2.302 2.020 2.605 **
I-2 10 Jun. 8/32 2.306 1.678 2.609**
I-3 10 Jun. 27/39 3.128 1.851
I-1 5 Aug. 15/38 Soy 3.308 2.:769**
I-2 5 Aug. 8/32 1.419 2.901 3.058**

I-3 5 Aug. 27/39 2.679 3.072

518 Tue Universiry SciENcE BULLETIN

Taste 16. As in Table 15 (rainy season) except that the recordings were made
during the dry season.

Number
Shoots

occupied —
Sub- eee xX 7E
plot Date total Condition Gale value
C-1 14 Mar. 2/20 2.550 1.260 —0.113
C-2 14 Mar. 1/19 2.947 0.636 —1.752
C-3 14 Mar. 1/31 2.516 0.917
E-1 23 Mar. 0/29 2.948 oilils) BAYS)
E2" 23 Mar. 0/24 2.729 0.716 SiO
E-3 23 Mar. 11/21 3.442 0.417
F-] 24 Apr. 19/42 3.238 0.746 1.383
F-2 24 Apr. 25/40 Sola 0.865
G-] 24 Apr. 1/18 jf Ppt 1.389 11,.335)3)
G-2 24 Apr. 5/26 3.423 0.254 —1.278
G-3 24 Apr. 10/24 3.166 0.732
H-1 11 Mar. 0/43 2.139 1.039 ao (a
H-2 11 Mar. 0/66 2.750 0.294 030%
H-3 11 Mar. 65/72 3.166 0.671
1-1 13 Mar. 12/38 1.907 3.106 SOS
[-2 13 Mar. 0/32 1.859 ZAUsi 300mm
I-3 13 Mar. 33/39 3.089 1)
I-1 18 Apr. 14/38 2.105 3.488 Sees
I-2 18 Apr. 7/32 1.859 2.987 3.798**
I-3 18 Apr. 29/39 3333}3) 2.230

Taste 17. As in Table 15 (rainy season) except that the recordings were made
during the cool season.

Number
shoots

eis occupied = -
plot Date total Condition s.d. value
F-1 6 Jan. 11/42 3.261 0.332 — 1.800
F-2 6 Jan. 27/40 3.000 0.530

G-l 6 Jan. 2/18 2.916 0.654 1.442
G-2 6 Jan. 10/26 3.269 0.204 —).294
G-3 6 Jan. 15/24 3.229 0.260

H-1 26 Dec. 0/46 2.902 1.118 4.109**
H-2 26 Dec. 3/69 1.934 0.678 13S 5am
H-3 26 Dec. 48/68 3.617 0.418

I-1 26 Dec. 3/38 2.684 1.681 SSA
[-2 26 Dec. 0/32 2.781 1.321 allislO) 2

1-3 26 Dec. 3/39 3.487 0.309

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 519

Tase 18. As in Table 15 (rainy season) except that recordings made during the
late end of the 1963 rainy season immediately before the cool season.

Number
shoots

occupied —
Sub- : x Li
plot Date total Condition sds value
E-] 26 Nov. 0/29 2D 1.046 0.967
E-2 26 Nov. 0/24 3.062 0.767 0.030
E-3 26 Novy. 11/21 3.071 1.207
F-1 28 Nov. 20/42 Split 0.335 —1.739
F-2 28 Nov. 26/40 35137 1.538
G-l 29 Nov. 7/18 3.361 0.808 Dota
G-2 29 Nov. 6/26 3.596 0.576 PAW e
G-3 29 Nov. 16/24 3.916 0.080

and three do not have a significant Z value. The fact that all of the Z values
are relatively low was very strongly associated with the fact that newly cut
unoccupied stumps tend to continue to produce new shoots for a month or
more despite the continued defoliation by insects. In addition, phytophagous
insects were much less abundant than they were during the first three months
of the rainy season, and control subplot E-3 and F-2 had a number of unoccu-
pied shoots in them.

Plot A, B and H yielded highly significant Z values throughout their re-
cordings. Plot C yielded no significant Z values. These two extremes were
associated, in the first case, with treatment subplots which had relatively few
occupied shoots, and in the second case, with a control subplot which had
many unoccupied shoots in it.

Subplot records of condition for existing shoots. The condition records for
12 plots and 16 independent recordings during the rainy season are given in
table 19. Of the 16 possible contrasts between the control and treatment sub-
plots, 11 have a highly significant Z value, and five do not have a significant
Z value. The low Z value in plot J was associated with the fact that the con-
trol plot only had three occupied shoots in it; these three shoots were the
three tallest shoots in the plot, had intact shoot tips, and were emergent. The
high condition value in plot Q-1 was the result of flowering branches pro-
duced in thorn axils in the upper part of the shoot; the low condition value in
plot Q-2 was the result of a general slowness in this plot to initiate vertically
lengthening branches at the end of the dry season. The low Z values recorded
for plot R were due to the large number of occupied shoots in the treatment
subplot; the low condition values for both subplots were due to the heavy
shade in plot R. Heavily shaded shoots were generally very slow to replace
damaged shoot tips. The low Z value in plot U was definitely due to the high
percentage of occupation of the shoots in the treatment subplot.

520 Tue University SciENcE BULLETIN

Taste 19. The condition values of the existing shoots for which height incre-
ments were recorded and treated by subplots (Table 7-9). All recordings made
during the rainy season.

Number
shoots
occupied =

Sub- —— XxX HA,
plot Date total Condition sid- value
J-l 23 Jul. 0/29 2.603 0.471 —1.202
J-2 23 Jul. 3/38 2.342 1.185

K-1 1 Aug. 0/266 2.537 0.338 5.416**
K-2 1 Aug. 49/149 2.875 0.411

L-1 26 May 0/44 3.011 0.029 7.868**
L-2 26 May 28/52 3.550 0.218

L-1 31 Jul. 0/44 2.284 0.272 iolo**
L-2 31 Jul. 38/52 3.096 0.323

M-1 26 May 0/79 2.892 0.725 5.7/9**
M-2 26 May 59/88 3.522 0.246

M-1 31 Jul. 0/79 2.329 0.755 PMVAV
M-2 Sie 59/88 B12 0.523

N-1 29 Jul. 0/81 2.456 0.207 10.205**
N-2 29 Jul. 33/56 3.650 0.248

O-1 29 Jul. 0/38 2.421 0.331 9.601**
0-2 29 Jul. 34/34 3.602 0.223

P-1 2 Aug. 0/61 2.327 0.182 3.15 0"*
P-2 2 Aug. 34/46 3.000 1.988

Q-1 30 May 0/149 3.328 0.120 0.000
Q-2 30 May 80/128 3.328 0.205

Q-1 6 Aug. 0/149 2.318 0.142 12-4 77=*
Q-2 6 Aug. 90/128 3.285 0.660

R-1] 28 May 14/42 2.976 0.658 —2.482
R-2 28 May 11/48 2.395 1.882

R-1 27 Jul. 14/42 A Sy/\\ 1.586 —1.895
R-2 27 Jul. 12/48 2.020 2.265

S-1 4 Aug. 95/211 2.445 1.302 6.114e*
S-2 4 Aug. 195/251 3.087 12232

T-1 29 Jul. 102/116 2.672 1.291 3. 101**
a2 29 Jul. 90/104 3.100 0.826

U-1 5 Jul. 78/86 3.505 0.397 —1.701

U-2 5 Jul. 85/93 Seullil 0.787

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 521

Tasve 20. As in Table 19 (rainy season) except that recordings made during the

dry season.
Number
shoots
occupied _

Sub- ooo Xx Z
plot Date total Condition s.d2 value
J-l 25 Apr. 0/29 B2/p 0.207 5 Sam
J-2 25 Apr. 0/38 3.342 0.285

K-1 4 Apr. 0/205 3.214 0.230 4.865**
K-2 4 Apr. 48/138 3.467 0.223

N-1 3 Apr. 0/81 3.339 0.205 Srl S ie
N-2 3 Apr. 43/56 3.714 0.162

O-1 3 Apr. 0/38 3.500 0.229 0.126
O-2 3 Apr. 32/34 3.514 0.219

P-1 24 Mar. 8/61 3.040 0.035 —0.188
P-2 24 Mar. 28/46 3.032 0.060

Q-1 16 Mar. 7/149 3.057 0.001 —0.041
Q-2 16 Mar. 94/128 3.058 0.088

S-1 6 May 55/211 3.063 0.623 0.747
S-2 6 May May250 ZY 0.683

T-1l 8 May 50/116 3.112 0.730 1.539
T-2 8 May 78/104 3.269 0.431

The condition records for eight plots and eight independent recordings
during the cool season are given in table 20. Of the eight possible contrasts
between the control and treatment subplots, three have highly significant Z
values, while five other values are not significant. The high values for treat-
ment subplots were associated in part with the production of flowering
branches and in part with the lack of phytophagous insects during the dry
season. In the control subplots, most of the shoots were old enough to react
to the dry season by not producing vertically lengthening branches and new
leaves so that their condition value was depressed.

The condition ratings for 11 plots and 13 independent recordings during
the cool season are given in table 21. Of the 13 possible contrasts between the
control and treatment subplots, one has a highly significant Z value and 12
do not have significant Z values. Of the 12 low Z values, 10 are minus values.
This was associated with the parathion stimulation, the presence of occupied
shoots in treatment subplots, the reduction in phytophagous insect activity by
the cool weather, the reduction of worker activity outside of the thorns during
cool weather, and the general lowering of plant growth rates by cool weather.

The condition records for two plots and two independent recordings dur-
ing November 1963 are given in table 22. Both contrasts have large negative

522

Tue Universtry ScteENcE BULLETIN

Tasie 21. As in Table 19 (rainy season) except that recordings made during the

cool season.

Number
shoots

oe occupied = ;
plot Date total Condition Side value
J-l 2iDec: 0/29 3.379 0.297 —1.408
J-2 12 Dec: 0/38 3.210 0.157

K-1 22ADEGs 0/178 3.016 0.016 2.950**
K-2 22 Wece 42/143 3.075 0.067

L-l 12 1D lee. 0/44 2.977 0.115 0.243
L-2 12 Dec. 7/69 2.992 0.088

L-1 18 Jan. 0/44 3.090 0.293 —1.925
L-2 18 Jan. 20/52 2.884 0.251

M-1 6 Dec. 27/85 3.294 0.198 —2.615
M-2 6 Dec. 54/101 3.133 0.160

M-1 18 Dec. 6/79 3.506 0.253 —7.122
M-2 18 Dec. 59/88 2.954 0.247

N-1 20 Dec. 35/342 3.061 0.059 —0.550
N-2 20 Dec. 44/240 3.050 0.079

O-1 20 Dec. 9/183 3.090 0.100 —2.958
O-2 20 Dec. 31/181 3.019 0.023

P-] 28 Dec. 0/86 3.244 0.295 0.355
P-2 28 Dec. 26/47 3.276 0.221

R-1 ONDeEc: 18/42 3.000 0.268 —7.014
R-2 9 Dec. 3/48 2.906 0.570

S-1 1] Jan. 22/211 3.431 0.236 —7.648
S-2 11 Jan. ISiljADil 53511335) 0.118

T-l 6 Jan. 17/116 3.258 0.219 —4.801
T-2 6 Jan. 82/104 3.028 0.052

U-1 21 Jan. 39/86 3.093 0.079 —3.595
U-2 21 Jan. 78/93 2.924 0.130

Z values. The magnitude of the Z value was strongly associated with the
large sample sizes. The larger values in the treatment subplots were appar-
ently associated with parathion stimulation.

Contrasts of the condition of occupied shoots with unoccupied shoots. The
condition records in tables 23 through 27 are for the same groups of continu-
ously occupied or unoccupied shoots that were used in the comparisons of
height increment for occupied and unoccupied shoots in tables 11 through 14.

1. Suckers. Table 23 presents the condition ratings for suckers that were
recorded during the rainy season. All 17 of the contrasts have a highly sig-

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 523

Taste 22. As in Table 19 (rainy season) except that recordings made before the
cool season (in the end of the 1963 rainy season).

Number
shoots

occupied —
Sub- a X Z
plot Date total Condition Stale value
M-1 14 Nov. 32/368 3.266 0.148 —260.000
M-2 14 Nov. 52/243 3.006 ° 0.054
O-1 14 Nov. 8/146 3.102 0.085 —27.000
0-2 14 Nov. 40/145 3.075 0.046

nificant Z value. The contrasts for plot A on 16 Jun. and B on 23 May are
presumed to be highly significant since other contrasts with similar differ-
ences have a highly significant Z value. It was impossible to apply a Z test
since the s.d. is 0.0 in the control subplots.

Table 24 presents the condition records for four plots during the dry sea-
son; six of the seven possible contrasts have a highly significant Z value. The
single contrasts with a significant Z value from plot C was associated with the
lower condition value in the control subplot (from which most of the occu-
pied shoots came). This value was lower because the shoots had been unoccu-
pied since 5 Jan.; the only reason why the difference between the two means
is great enough to yield a significant Z value is because there were four shoots
in the plot which were occupied since 5 Jan. by auxiliary-units which had
moved into the plot on their own. These four shoots all had a value of four.

Table 25 presents the condition records for three plots during the cool
season; two of the three possible contrasts have a highly significant Z value
and one has a significant Z value. The relatively low Z values for each con-
trast were associated with the small sample sizes, the relative newness of the
stumps, and the high incidence of Coxina hadenoides in both of the plots.

Existing shoots. Table 26 presents the condition records for seven plots
during the rainy season; nine of the ten contrasts have a highly significant Z
value. The low Z value obtained from the contrast from plot Q was associ-
ated with the productien of flowering branches on unoccupied shoots (a char-
acteristic damage reaction), and the slow response of the occupied shoots to
the beginning of the rainy season.

Table 27 presents the condition records for five plots during the cool and
dry season; two of the six contrasts have a highly significant Z value. In plot
L the low condition records among the occupied shoots was associated with
the inactivity of the workers during the cool weather; since these were young
colonies that had very recently developed in situ, they did not have many
workers outside of the thorns. The same situation existed in plot M, but in

524 Tue UNiversiry SciENCE BULLETIN

Tasie 23. The condition values of the sucker shoots for which length increments
were recorded and contrasted as groups of occupied and unoccupied shoots (Tables
10-11). All recordings made during the rainy season.

Number x 1s

Plot Date shoots Condition s.d.? value
A un 16 Jun. 41 3.309 0.731 eae
A occ 16 Jun. 25 4.000 0.000

A un 3 Aug. 41 2.695 0.423 7.888**
A occ 3 Aug. 25 3.760 0.731

B un 23 May 36 2.500 1.383 ss
B occ 23 May 42 4.000 0.000

B un 17 Jun. 36 2.861 0.937 4.792 **
B occ 17 Jun. 42 3.714 0.245

B un 7 Aug. 36 1.861 0.680 125352
B occ 7 Aug. 42 3.904 0.352

C un 27 May 47 2.627 0.940 DJA
C occ 27 May 16 3.687 0.229

C un 6 Aug. 47 1.234 1.226 1225/9
C occ 6 Aug. 16 3.687 0.196

D un 25 May 13 2.076 0.340 7.804**
D occ 25 May 22 3.590 0.253

D un 28 Jul. 13 1.653 0.975 6.491**
D occ 28 Jul. 22 3.568 0.269

E un 10 Jun. 39 2.692 0.878 50s"
E occ 10 Jun. 2) 3.052 0.693

E un 7 Aug. 39 1.435 0.897 16.607**
E occ 7 Aug. ZA 3.976 0.012

F un 6 Jul. 26 PMN TS) 0.759 6.605**
F occ 6 Jul. 26 3.461 0.238

G un 6 Jul. 25 2.080 0.868 7.688**
G occ 6 Jul. 15 3.733 0.174

H un 10 Jun. 79 1.734 2.130 8/4 75*
H occ 10 Jun. 59 3.652 0.200

H un 5 Aug. 79 0.987 1.126 l6552**
H occ 5 Aug. 59 3.652 0.760

I un 10 Jun. 50 1.480 1.208 16.064**
I occ 10 Jun. 36 3.986 0.007

I un 6 Aug. 50 0.340 0.739 24.431**

I occ 6 Aug. 36 3.736 0.164

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 525

Tasre 24. As in Table 23 (rainy season) except that recordings made during
the dry season.

Number Xx ZL

Plot Date shoots Condition s.d.” value
C un 17 Mar. 47 2.552 0.828 0.794
C occ 17 Mar. 16 2.781 1.032

E un 23 Mar. 39 2.820 0.611 3.308**
E occ 23 Mar. 21 3.442 0.417

F un 24 Apr. 26 2.903 1.320 3.734**
F occ 24 Apr. 26 3.788 0.145

G un 24 Apr. 25 2.980 0.760 Brls 9"
G occ 24 Apr. 15 3.633 0.195

H un 11 Mar. 79 DD, 0.921 9.032**
H occ 1] Mar. 59 3.338 0.210

I un 13 Mar. 50 1.130 1.967 12.262**
I occ 13 Mar. 36 3.791 0.276

I un 18 Apr. 50 0.740 1.686 14.740**
I occ 18 Apr. 36 3.916 0.078

Taste 25. As in Table 23 (rainy season) except that recordings made during the
cool season and before.

Number x Z

Plot Date shoots Condition s.d.7 value
F un 6 Jan. 26 2.903 0.900 2.221*
F occ 6 Jan. 26 3.365 0.231

Gun 29 Nov. DS: 3.380 0.651 2.627**
G occ 29 Nov. 15 3.866 0.124

G un 6 Jan. 25 2.880 0.443 2:475**
G occ 6 Jan. 15 3.333 0.238

this plot there was an exceptional lack of phytophagous insects in the treat-
ment subplot, in which most of the unoccupied shoots were located. This
may have been due to one of the earlier insecticide applications. In plot P the
high condition value for unoccupied shoots was associated with the produc-
tion of flowering shoots. The similar condition values for the occupied and
unoccupied shoots in plot S are a reflection of the production of new shoot
tips by the parathion stimulation, and the lack of ants during the cool nights
on the surface of shoots that are occupied during the warmer diurnal hours.

Discussion. It is evident from the data presented in tables 15-27 that the
presence of the workers of a colony of Psewdomyrmex ferruginea on a shoot

526 Tue UNiversiry ScIENCE BULLETIN

Taste 26. The condition values of the existing shoots for which height increments
were recorded and treated by groups of occupied and unoccupied shoots (Tables
12-13). All recordings made during the rainy season.

Number xX iL

Plot Date shoots Condition Sidea value

L un 26 May 49 3.061 0.058 MMF
L oce 26 May 12 3.833 0.106

L oun 31 Jul. 49 2.295 0.290 Aollteisy
L occ 31 Jul. 12 3.416 0.220

M un 26 May 77 DS)IS 0.779 6.766**
M occ 26 May 44 ST 0.191

M un 31 Jul. 77 2324 57 8.598**
M occ 31 Jul. 44 3.545 0.230

N,O un 31 Jul. 122 BASS) 0.176 19:450**
N, O occ Sila 95 3.626 0.215

P un 2 Aug. 7/1 PMA 0.658 12.951**
P occ 2 Aug. 34 3.705 0.198

Q un 30 May 159 35 0.347 1329
Q oce 30 May 68 54D 0.203

Q un 6 Aug. 159 2.267 0.223 16.077**
Q occ 6 Aug. 68 3.669 0.422

R un 28 May 47 2.776 0.454 10.948**
R occ 28 May 7/ 3.441 0.246

R un 26 Jul. 47 DNA 1.281 8.204**
R occ 26 Jul. 17 3/35 0.191

S un 3 Aug. 143 Ieyetet 1.118 19386."
S occ 3 Aug. 181 Sl 0.206

Tasie 27. As in Table 26 (rainy season) except that recordings made during the
dry season or cool season.

Number xX id

Plot Date shoots Condition sul value

L un 18 Jan. 49 2.979 0.426 —0.210
mace 18 Jan. 12 2.958 0.019
M un 18 Jan. 77 3.467 1.245 —3.296
M occ 18 Jan. 44 3.045 0.015

P un 24 Mar. 7M 3.070 0.048 —1.400
P occ 24 Mar. 34 3.014 0.057

Q un 19 Mar. 159 2.990 0.102 2.476**
Q occ 19 Mar. 68 3.095 0.084

San 11 Jan. 143 3.325 0.207 —3.115

S occ 11 Jan. 181 3.182 0.138

S un 5 May 143 2.772 1.240 SAVES
S occ 5 May 181 S57, 0.208

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 527

Taste 28. The distribution of significant Z values among contrasts of condition
values of Acacia cornigera made at different times of the year. Note that the
highest frequencies of Z values are associated with recordings made during the

rainy season of subplots, or groups, of suckers from stumps.

Number Number

Time Number highly Number not Type
of of significant significant significant of
Table year contrasts contrasts contrasts contrasts shoots
(treated as
subplots)
15 rainy 28 17 4 7 suckers
16 dry 13 Hl ] 5 suckers
17 cool If 4 0 3 suckers
18 rainy 1963 5) 1 1 3 suckers
19 rainy 16 11 0 5 existing
20 dry 8 3 0 5 existing
21 cool 13 1 0 12 existing
22 rainy 1963 2 0 0 2 existing
23 rainy ily) 17 0 0 suckers
Da dry i 6 1 0 suckers
25 cool 3 Z 1 0 suckers
26 rainy 10 9 0 ] existing
aif cool & dry 6 2 0 4 existing

of Acacia cornigera has an effect on the condition of the shoot. In view of the
dependent relation between height increment and the condition of the shoot,
this result is expected. It seems clear that the same variables in the environ-
ment that affected the height increments of occupied and unoccupied shoots
also affected the condition of these shoots. This is clear from table 28. The
two tables are quite similar in respect to the temporal and bionomic location
of the significant Z values; the highest numbers of significant Z values are
yielded by contrasts of recordings made during the rainy season and among
suckers from stumps cut during the experimental program. In addition,
tables 14 and 28 show that the treatment of shoots as unoccupied and occupied
groups in contrast to subplot comparisons, show more clearly the effect of the
ants on the shoot.

The lowered frequency of significant Z values in subplot contrasts is due
to the same factors as is the case with subplot contrasts of height increment.
It appears that the contrast of condition values from groups of consistently
occupied and unoccupied shoots is just as sensitive as the contrast for height
increment values in showing the influence of P. ferruginea. In fact, if the
condition records are made at the beginning of a period of growth, they may
show significant differences in the condition between groups of occupied and
unoccupied shoots before significant height increments can accrue.

528 Tue University Science BULLETIN

Taste 29. Mortality of occupied and unoccupied stumps and their suckers of

Acacia cornigera in plots H, I and E during 1963-1964. These data are grouped

by subplots and this is associated with the high mortality shown in control sub-

plot I-1 before the 1964 rainy season began; those stumps which died in [-1 were
almost all unoccupied.

Treatment (1) Treatment (2) Control (3)

Sub- number number number number number number

plot Date alive dead alive dead alive dead

H-1, H-2, 17 Oct. 43 0 66 0) We 0
H-3 11 Mar. 35 8 61 5) 71 1
10 Jun. 27 16 53 13 71 1
5 Aug. 23 20 48 18 71 1
licils Te. 18 Oct. 38 0 31 0 39 0
1-3 13 Mar. 22 16 7) 10 34 5
10 Jun. 18 20 17 14 28 11
6 Aug. 17 Dil 13 18 28 11
E-1, E-2, 18 Nov. 29 0 24 0 21 0
E-3 23 Mar. Di 2 23 1 Dil 0
10 Jun. Dik 2 19 5 21 0
7 Aug. 24 5 14 10 Dil 0

Ratings of condition of heavily occupied shoots during the rainy season
furnish a very good indicator of the density of the two insects which are rela-
tively insensitive to P. ferruginea, and eat shoot tips: Coxina hadenoides and
Pelidnota punctulata. It appears that if these two insects are not present dur-
ing the rainy season, nearly every occupied shoot will maintain an undam-
aged terminal shoot tip (a 4 condition rating), and show a resultant high
height increment. The condition ratings of completely unoccupied shoots ap-
pear to be a good indicator of the density of insects that feed on A. cornigera
at all times in the year; the rating values increase during the dry season, and
then drop very rapidly when the rainy season starts.

Mortality of A. cornigera. Greater mortality occurred among sucker
shoots from stumps cut during the study, than among existing shoots. While
this was in part due to a greater susceptibility of sucker shoots, and the stump
from which they came, to cumulative insect damage, it was also associated
with the problem that none of the treatment subplots of existing shoots were
free from occupation until the 1964 dry season. On the other hand, plot H, I,
and E had treatment subplots that were relatively free of ants from their time
of cutting in Oct.-Nov. 1963 to the end of the experiment in August 1964.
The mortality of shoots in these plots is presented in table 29.

It should be emphasized that when a shoot is recorded as dead in table 29,
it means that the stump as well as the sucker is dead. In most cases, the root
system appeared to be dead as well. The death of the shoots in these three

INTERACTION OF THE BULL’s-HOoRN ACACIA WITH AN ANT INHABITANT 529

plots appeared to be associated entirely with the cumulative effect of contin-
uous insect damage to the inner bark and foliage. In all three plots, every
stump produced at least one new branch and usually many more before it
died. The larvae of Chrysobothris sp. near C. multistigmosa were found in
nearly every dead stump during the August recordings.

It was often the case that a newly cut stump produced a cluster of 5-15 new
branches within six weeks after being cut. If the shoot is occupied, one of
these usually becomes the main axis of the shoot and grows rapidly upward.
If the shoot is unoccupied, the shoot tips of these branches are usually eaten
within a few days to a week after their appearance. The stump continues to
produce new shoots but at a rapidly decreasing rate and within several months
becomes nearly dormant. From this stage, it is killed after a variable number
of months by the internal girdling activity of Chrysobothris sp., near C. multr-
stigmosa. However, in plot B there was a cluster of five unoccupied stumps
in the relatively dormant stage following defoliation, which became occupied
by a large colony from a large cut shoot in May. By August, these stumps
had main sucker shoots 50-85 cm in length. It appears that occupied stumps
with vigorous suckers are not always killed by the larvae of Chrysobothrts sp.,
near C. multistigmosa, since these five stumps had been infested.

The data presented in table 29 for three plots can be considered as repre-
sentative of the fate of occupied and unoccupied stumps. However, shoots
from which the ant colonies are removed by clipping the thorns or spraying,
still have their crop of mature leaves and therefore are more slow to die; these
leaves are only slowly removed by phytophagous insects. In addition, the
ability of an unoccupied shoot to continue to produce new shoot tips, even
when they are being rapidly destroyed, appears to be stronger among exist-
ing shoots than among suckers from stumps. For example, between 3 Apr.
and 29 Jul., the 81 unoccupied shoots in treatment subplot N-1 produced 328
new vertically lengthening branches which could have become central axes.
Every one of them had its shoot tip eaten off before it was more than 25 cm
long. It is very unlikely that any equal sized group of cut stumps that had
been unoccupied for the length of time that those in N-I were (116 days),
would have produced anywhere near this number of new shoots. It should
be noted that while occupied suckers from stumps tend to grow almost con-
tinuously for the first six months to a year after the stumps are cut, existing
shoots often respond to the cool and dry season by not producing new growth.
Therefore, new growth is not present to be destroyed by phytophagous in-
sects if the shoot is abandoned by an auxiliary-unit.

One aspect of the mortality of existing shoots is due to the feeding be-
havior of the rodent Sigmodon hispidus. In no case was an occupied shoot
found that had been cut and fed on by S. Aispidus; during the later part of the
dry season, it was a relatively common case to find a deep notch cut by this
rodent in the trunk of an occupied shoot a few cm above the ground, While

530 Tue University SCIENCE BULLETIN

workers of P. ferruginea were never observed to turn one of these rodents
away, it appears that this is what occurred. S. hispidus only cuts shoots of a
basal diameter of about 1.5 cm or less. Thus, it confined its activity to the
small unoccupied shoots; these shoots were usually below the canopy of the
general vegetation. Once a shoot became occupied and began vigorous
growth, its basal diameter became so great that S. Aispidus generally did not
try to cut it even if it was unoccupied. Occasionally, S. hispidus climbed into
the canopy of 1.5-2 m tall unoccupied shoots and cut off the short new
branches which were then eaten on the ground. The damage of the rodent to
unoccupied shoots became increasingly severe as the dry season progressed
but stopped almost entirely after the first rains caused new growth among
other plants.

In plots such as P and S, S. hispidus was responsible for the removal of
better than 50 per cent of the short unoccupied shoots during the dry season.
New shoots often grew from these stumps when the rains started or shortly
before, but they were so heavily shaded that they had almost no height incre-
ment. These shoots were also heavily damaged by the insects which com-
monly fed on unoccupied A. cornigera. It should be re-emphasized that al-
though S. Aispidus was the only common rodent in the area of the rodent
damage, an individual was never collected while feeding on A. cornigera,
though three were seen while feeding on it. The damage described above may
have been caused by some other species of rodent which is very similar to S.
hispidus in appearance. However, specimens of S. hispidus fed rapidly on A.
cornigera in the laboratory.

There was no evidence of direct biotic mortality factors of A. cornigera in
natural disturbance sites that were absent in man-made disturbance sites. The
opposite situation was also the case. However, the density of an animal such
as S. hispidus was definitely higher in man-made disturbance sites than in any
natural disturbance sites in the area between Temascal and La Granja.

Leaf production by A. cornigera. The number of leaves on a shoot of A.
cornigera is a function of how many are produced and how many are re-
moved. Unoccupied suckers from stumps and existing shoots produce fewer
leaves and have more removed by phytophagous animals than do occupied
shoots. They produce fewer leaves because the continual defoliation of the
shoot weakens it. The phytophagous animals in the experimental plots were
insects and occasionally the rodent Sigmodon hispidus. In those plots, and at
those times when cattle were present, their direct defoliating activity was al-
most nonexistent. This latter statement applies to both unoccupied and occu-
pied shoots. A leaf had to be at least one-half intact to be counted in the fol-
lowing four examples recorded in July and August 1964.

(1) On 29 Jul. the 81 unoccupied shoots in treatment subplot N-1 had a
mean number of 88.59 leaves per shoot (s.d=45.4 leaves). On the same date,
the 69 occupied shoots in control subplot N-2 and O-2 had a mean number of

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 531

183.23 leaves per shoot (s.d.=245.3 leaves). However, control subplot N-2 con-
tained one exceptional shoot with 2,000 leaves on it. If this shoot is excluded,
then the mean number per occupied shoot becomes 156.51 leaves (s.d=105.3
leaves). A Z test of the mean of 68.59 contrasted with 156.51 yields a highly
significant Z value (8.108). The high s.d. of the number of leaves on the occu-
pied shoots was in great part associated with the presence of six shoots which
had just recently had their leaf crop eaten off by the larvae of Syssphinx mext-
cana. Virtually all of the leaves counted in the above subplots were produced
by the shoots after 31 Dec. Most of them were produced after 1 May. In the
treatment subplots, nearly all of them were from the axils of swollen thorns
produced when the shoots were occupied in 1963. In the control subplots, most
of the leaves were borne on new swollen thorns on lateral branches produced
in 1964.

(2) The stumps in plot A were cut on 25 May 1964. By Aug. 1964 the 42
unoccupied shoots in treatment subplot A-1 had produced 1,458 leaves and
1,016 swollen thorns. The 29 occupied shoots in control subplot A-2 had pro-
duced 4,818 leaves and 3,774 swollen thorns. Less than 10 per cent of the
leaves and swollen thorns on the unoccupied shoots were produced after 1
July; at least 50 per cent of the leaves and swollen thorns on the occupied
shoots were produced after 1 July.

(3) The stumps in plot H were cut on 17 Oct. 1963. By 5 Aug. 1964, the
66 shoots in treatment subplot H-2 had produced 3,460 leaves and 2,596 swol-
len thorns. The 72 shoots in control subplot H-3 had produced 7,785 leaves and
7,483 swollen thorns. The high number of leaves produced in the treatment
subplot H-2 was in part associated with the presence of 30 new colonies which
had developed in situ, and which occupied the shoots during the month of
July. It was also associated with some growth that occurred during the last
month of the dry season when there were almost no active phytophagous in-
sects in the plant community.

(4) The stumps in plot D were cut on 10 Dec. 1963. Due to the proximity
of uncut shoots with large colonies of P. ferruginea, a number of the shoots in
the treatment subplots became occupied. Because of this, the shoots were seg-
regated in the recording and the leaves were counted as 1) those on occupied
shoots and 2) those on unoccupied shoots when the final recording was made
on 28 Jul. 1964. The 14 unoccupied shoots had produced a total of 321 leaves
and 128 swollen thorns. The 26 occupied shoots had produced a total of 2,391
leaves and 1,578 thorns. In comparing the leaf and swollen thorn production
in this plot with that in plot B the effect of the dry season, even on stumps,
can be seen. The plants in plot D were nearly six months older than those in
plot A but plot D still had a lower mean number of leaves and swollen thorns
per shoot. Most of the six month lead was during the dry season. This effect
is also in part due to the fact that the shoots in plot D were partially shaded
while those in plot A were receiving full sunlight.

532 Tue University ScreNcE BULLETIN

On the basis of the above 4 examples, it is clear that a colony of P. ferru-
ginea has a definite effect on the number of leaves borne by a shoot of A.
cornigera. In addition, the colony has a much better opportunity for growth
on an occupied shoot, than on an unoccupied shoot. On an occupied shoot,
it has both more living space and more food available to it since each new leat
means a new crop of Beltian bodies, another source of nectar, and often a new
swollen thorn.

Biomass production of A. cornigera. In the plots with sucker regeneration
from cut stumps, it was possible to harvest the crop of shoots and weigh them.
An insignificant error was introduced by the weight of the ants living in the
thorns. The shoots were weighed within 24 hours after being cut; the shoots
from the various groups within a plot were all weighed after the same time
interval. The original stumps were not weighed. The following five exam-
ples of biomass production by occupied and unoccupied shoots are representa-
tive of that which occurred throughout the experimental plots.

(1) The stumps in plot A were cut on 25 May. By 13 Aug., the 42 shoots
in treatment subplot A-1 weighed 850 gm; the 29 shoots in the control subplot
A-2 weighed 7,750 gm. (2) The stumps in plot B were cut on 29 Apr. By
7 Aug., the 36 unoccupied shoots weighed 450 gm; the 64 occupied shoots
weighed 43,650 gm. (3) The stumps in plot D were cut on 10 Dec. 1963. By
28 Jul. the 14 unoccupied shoots weighed 365 gm; the 26 occupied shoots
weighed 4,810 gm. (4) The stumps in plot E were cut on 26 Nov. 1963. By
7 Aug. the 29 shoots in treatment subplot E-1 weighed 3,690 gm., the 24
shoots in treatment subplot E-2 weighed 11,240 gm., and the 21 shoots in con-
trol subplot E-3 weighed 31,212 gm. (5) The stumps in plot H were cut on
17 Oct. 1963. By 5 Aug. the 66 shoots in treatment subplot H-2 weighed 2,900
gm; the 72 shoots in control subplot H-3 weighed 41,750 gm (weights
for treatment subplot H-1 were not recorded because the thorns had been
clipped).

Comparisons of the weight of occupied shoots with unoccupied shoots
show stronger differences than comparisons of the number of leaves or swol-
len thorns. This is due for the most part to several factors. The leaves on
occupied shoots are usually large and completely intact in contrast to the
leaves on unoccupied shoots which are usually smaller and damaged to some
degree. The thorns produced by occupied shoots are fully developed while
the thorns on unoccupied shoots are often partly eaten while still green. The
leaves on occupied shoots are for the most part produced during branch elon-
gation and bear very large swollen thorns; the leaves on unoccupied shoots
are often from the axils of older thorns and have minute stipules. The num-
ber of meters of branch length is much greater on occupied shoots in contrast
to unoccupied shoots. Those branches that are present on unoccupied shoots
are very thin in contrast to the thick branches on most of the occupied shoots.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 533

Observations of the plots with cut stumps in them showed that a single
heavily occupied stump of A. cornigera produces far more woody vegetation
and height increment in the first year of regeneration than does a single
stump of any other species of plant in the plots. This was true in respect to
both maximum and mean values. However, some plants such as Bixa orellana
and Croton glabellus produced a higher amount of woody vegetation per
hectare because of their very high density of shoots. On the other hand, it
also appears that a single unoccupied stump of A. cornigera produces less
vegetation in the first and following year than does a single stump of any
other tree species in the plots. This was true in respect to both maximum
and mean values. This marked superiority of growth rate enjoyed by occu-
pied A. cornigera slowly diminishes after the end of the first year of regenera-
tion, although the effects of this early burst of growth are indicated by the
emergent position of the shoot for nearly its entire life.

Presence of vines. While annual and perennial vines constituted a major
part of the regeneration in many of the experimental plots, it was extremely
rare to find one using a shoot of A. cornigera for a standard (Fig. 60, 61). If
a vine was well established on A. cornigera, the shoot was an unoccupied one
or the ants had recently invaded the shoot as an auxiliary-unit, or as a colony
developed in situ. The tips of the vines are usually killed by the ants as soon
as they contact occupied shoots (Fig. 34, 37b); when growing up through a
vine mat, the shoot is often ringed by blackened vine ends.

In the experimental plots, it was common to find undamaged vines grow-
ing in the canopies of unoccupied shoots of A. cornigera. The number of
vines and the degree to which they pulled the shoot over, or shaded it, was
in great part associated with the type of plant community. Unoccupied shoots
in areas browsed by cattle were often free of vines because the cattle removed
them from the shoots. When the canopy of A. cornigera was part of a dense
general canopy, vines often grew completely over it when the ants were re-
moved. Since this acacia is so intolerant of shading, this is an obvious handi-
cap. The following four examples give a representative indication of the inci-
dence of vines on occupied and unoccupied shoots.

(1) On 31 Jul. 1964, the 88 shoots in control subplot M-2 (59 occupied
shoots) had 34 vines on 10 of the shoots; none of these 10 shoots were occu-
pied. In treatment subplot M-1, the 79 unoccupied shoots had a total of 158
vines on 52 of the shoots. In both subplots, most of these vines were in the
families Passifloraceae, Convolvulaceae, and Leguminosae. These vines had
been accumulating on the unoccupied shoots since late in 1963. In some cases
(Convolvulaceae, Leguminosae), they had flowered and set seed during this
period. Many of the unoccupied shoots were completely covered with vines
and the thinner shoots were often bent by the weight of the vines.

(2) On 24 Mar. 1964, the 46 shoots in control subplot P-2 (28 occupied)
had no vines on them. On the same date, the 61 shoots in treatment subplot

WI
Oo
pa

Tue University ScteNcE BULLETIN

-

=
»

%
Pe
> soe

‘

6
> *

‘

Fic. 60. The usual freedom from vines exhibited by occupied shoots of A. cornigera. The
3.5 m tall shoot in the center of the photograph is about 3 years old and growing in a roadside
mat of Ipomoea sp. The white flowers of this vine are scattered over the canopy. This shoot of
A. cornigera was ringed by at least 50 dead vine leaders that had been killed by P. ferruginea.
Photo 10 km north of Tierra Blanca, Veracruz, Mexico in November 1963.

P-1 (7 occupied) had one vine on them. By 2 Aug., the 46 shoots in control
subplot P-2 (34 occupied) still had no vines on them. On this date, the 61
shoots in treatment subplot P-1 (0 occupied) had a total of 129 vines on 35
shoots. Most of these vines were in the Asclepiadaceae, Sapindaceae, Meni-
spermaceae, Bignoniaceae, Convolvulaceae, and Leguminosae. They were all
present in large numbers in the control subplot but their branches had been
killed when they started to enter the canopies of the occupied shoots of .A.
cornigera. In the treatment subplot, the vines had formed a solid mat over the
tops of many of the occupied shoots; the result was that under casual observa-
tion the density of 4. cornigera in the treatment subplot appeared to be very
low. The lack of vines on unoccupied A. cornigera during the dry season was
due to the general lack of growth of most vegetation in the plot, plus the fact
that the shoots in the treatment subplot were occupied until the thorns were
clipped.

INTERACTION OF THE BULL’s-HoRN AcAcIA WITH AN ANT INHABITANT 535

a large colony of P. ferruginea while the right-
es of the
as been bent over by the weight of the vines. In
damaged by insect attack. The vines have
seen by the large number
Photo 2 Aug. 1964 along roadside

Fic. 61. The left-hand shoot is occupied by
hand shoot is unoccupied. The unoccupied shoot is festooned with a number of vin
Convolvulaceae; the upper part of the shoot h
addition, the foliage of this shoot was very badly
attempted to enter the occupied shoot a number of times as can be
of vertical and dead, bare tendrils among the lower branches.

east of Temascal.

536 Tue Universiry ScteNcE BULLETIN

(3) On 3 Apr. 1964, both the occupied and unoccupied shoots in plots N
and O were completely free of vines; this was due to the low amount of
growth in the plots during the preceding eight months plus the fact that the
shoots were for the most part occupied during the previous rainy season. By
29 Jul. 1964, the 119 unoccupied shoots in the two treatment subplots had a
total of 153 vines on 54 shoots. There were no vines on the 90 shoots in the
two control subplots (N-2 and O-1). The taller unoccupied shoots were
lightly draped with vines, but some of the shorter shoots were completely tied
into the general vegetation by the vines. These vines were in the Convolvu-
laceae, Asclepiadaceae, and Leguminosae.

(+) On 4 Aug. 1964, there were 290 occupied shoots among the 462 shoots
of A. cornigera in plot S. None of the occupied shoots had vines in their
canopies. There was a total of 150 vines on 61 of the unoccupied shoots. The
relatively low number of vines was associated with the slow vegetation
growth during the dry season, and the relatively low number of vines in the
vegetation in plot S in comparison with other similar plots.

The vines in the experimental plots covered unoccupied A. cornigera to a
much greater extent than other species of woody plants; the vines tended to
form larger clusters on the emergent unoccupied A. cornigera and bind them
into the general canopy (Fig. 61). This was due to the prevention of further
growth of the unoccupied shoots and their relatively stiff woody form with
thorny branches that served as an excellent support for the vines. Further-
more, the emergent shoots of A. cornigera that became unoccupied for some
reason, were usually the tallest plants in regeneration less than four years old,
and therefore were the most suitable for climbers by virtue of their position in
the vegetation.

Presence of basal circles. While normally occupied shoots are nearly always
free of foliate intrusive vines or branches of other plants, the circular area
around the base of occupied shoots (Fig. 35) is much more variable in respect
to the degree of damage to the vegetation by P. ferruginea. This variation
in the size and cleanliness of the basal circle is associated with the size of the
colony, type of vegetation under the A. cornigera, length of time that the
colony has occupied the shoot, and the height of the shoot canopy above the
ground. In addition to these four factors, there is variation that appears to be
due to the individual behavior of the colony itself.

Large colonies often have well defined, large, and clean basal circles under
their shoots. A basal circle is usually not evident until the portion of the
colony in the shoot has 500 or more workers. The larger the colony, the
more workers patrol and/or clean on the soil under the shoot. The longer it
has been in the shoot, the more likely it is that the ants have had time to pro-
duce a very clean and well defined basal circle.

The type of vegetation under the shoot to some extent dictates how long
is required for the ants to kill the vegetation, or stop it from producing new

INTERACTION OF THE BULL’s-HoRN ACACIA WITH AN ANT INHABITANT 537

leaves. Seedlings are killed almost immediately after appearing above the
soil, while on existing woody stems, the new leaves have to be repeatedly de-
stroyed. As the shoot grows, the canopy is carried upward, and the colony is
located further and further above the ground. As this occurs, the density of
workers in the basal circle decreases. When a canopy is over 5-7 m above the
ground, there is usually little trace of the basal circle. However, vines at-
tempting to enter the canopy are attacked with the usual aggressiveness.

The cleanest and largest basal circles are found under shoots that are in
their second to fourth year of occupied growth, and are growing over grass,
herbs or low leafy shrubs. In a site such as that which contained plots C, H, I,
U and V, the basal circles varied from 30-150 cm in diameter around single
shoots and up to 4 m in diameter where the basal circles of several shoots
overlapped. These basal circles were characteristically bare dirt with a light
to heavy even litter of pinnules, thorns and twigs from A. cornigera.

There are a number of facts that indicate that the basal circle is the result
of the workers’ mauling activity of growing vegetation, and not the result of
a toxicant released by A. cornigera. 1) In the treatment subplots, after the
first two months of rainy season growth had taken place, there was total
obliteration of the basal circles under the shoots that had been rid of their ants
by parathion or other experimental means. This rapid loss of basal circles was
due to the sprouting of both monocotyledonous and dicotyledonous seedlings,
and the leafing out of the branches of low perennial shrubs and vines. 2) Not
all occupied shoots have basal circles. Even when the colony is large, has been
in the shoot for a year or more, is close to the ground, and is over a grass
ground layer, the basal circle is not always present. 3) When a large shoot is
cut and the canopy left across the stump, the colony disorganisation in the
following 3-6 months is sufficiently great that the old basal circle usually be-
comes obliterated by seedlings and vine sprouts. 4) Finally, the workers are
very often observed mauling young leaves or seedlings in the basal circle.

Seeds were planted in basal circles on two occasions to examine experimen-
tally the activity of the ants. On 3 Aug. 1962, near Campo Cotaxtla, Veracruz,
a 250 cm isolated queen-shoot with a large colony was chosen in a lightly
grazed grass pasture. It had a roughly circular basal circle about 110 cm in
diameter. Three rows of seeds (one each of corn, beans, and radishes) were
planted, radiating from the trunk to the margin of the basal circle, and ex-
tending 20 cm into the surrounding grasses. On 13 Aug. 1962 there were no
seedlings left alive within 6 cm of the trunk. From this point outward the
beans increased gradually in height to the margin of the basal circle; from the
margin to the end of the row they were 8-10 cm tall. On this date, the row of
corn was in discontinuous segments with seedlings 1-3 cm tall out to the mar-
gin of the circle; the last 20 cm of the row had plants 10-14 cm tall. Only
about 14 of the radish row had live plants, and they were 1-2 cm tall; outside
the basal circle they were 3-5 cm tall. The mauling damage to the growing

538 Tue University ScreNceE BULLETIN

points of all three species of plant was clearly evident, and the workers were
seen chewing on the shoots. On 27 Aug. 1962, the beans were completely ab-
sent to a distance of 25 cm from the trunk, the corn row to 35 cm, and the
radish row to 50 cm. Those plants growing outside the basal circle were all
over 10 cm tall and in excellent condition.

During the 1964 rainy season, in the vegetation along the west margin of
plot C, this type of experiment was repeated. The seeds of Bixa orca
Bauhinia ungulata, Acacia chiapensis, Acacia cornigera, Mucuna pruriens,
Cassta bicapsularis and Ipomoea sp. were planted in seven radiating rows
from the base of each of eight large queen-shoots with basal circles ranging
from 75-150 cm in camer The seeds of B. ungulata, A. chiapensts and A.
cornigera were scarified by cutting off a piece of the seed coat. All of the seeds
germinated in large numbers except B. orellana. The trunks of four of the
shoots were banded with a 10 cm band of tree banding compound which
stopped ant traffic to and from the ground. In the basal circles of the un-
banded shoots, the sprouting seedlings were killed within eight days of the
time that they appeared. Most were killed almost as soon as they appeared
above the ground. The sprouts of Mucuna pruriens and Cassia bicapsularis
were very tough and lived longer though they stopped growing upward as
soon as the shoot apex was destroyed; it is the shoot apex which is usually
attacked first by the workers. Outside of the basal circles, the seedlings sur-
vived and grew well. In the basal circles of the banded shoots, about three-
fourths of the seedlings were damaged or killed by the mauling of workers
that had fallen from the shoot, and that came to the base of the queen-shoot
from nearby auxiliary-shoots. However, it took a month (July) for the basal
circle to be completely cleaned of the seedlings; in one basal circle, a grass
stem fell against the trunk above the band and the workers used this as a
path to the ground. An unexpected result of this experiment was the discov-
ery that the seedlings of 4. cornigera and A. chiapensis were destroyed. Yet,
small shoots of 4. cornigera and A. chiapensis that have attained the usual
characteristics of older swollen-thorn acacias, are treated like the parent shoot.

The significance of the basal circle to either the ant or the shoot is difficult
to state clearly. The removal of vines and branches of other plants from the
canopy of 4. cornigera by the ants serves to avoid shading of the shoot, avoid
mechanical bending of it or its branches, and remove possible highways of
approach for phytophagous insects. None of these factors apply as clearly in
the case of the basal circle. However, to a lesser degree the production of the
basal circle furthers these purposes. The production of the basal circle by the
workers can also be regarded as merely a downward extension of the work-
ers’ activity in mauling intrusive foliage in the canopy of A. cornigera. How-
ever, a big difference between the two activities is that when mauling a vine,
the worker either maintains direct contact with the shoot or at best, is only a
few centimeters from it. It does not range more than 15 to 20 cm past the

INTERACTION OF THE BuLv’s-HoRN ACACIA WitH AN Ant INnuaABitaNt 539

shoot canopy in this activity. However, when the worker is mauling vegeta-
tion on the ground, it may walk out as far as 75 cm from the nearest part of
the shoot. The possibility that the worker is obtaining something of nutritive
value from the plant cannot be examined with the data at hand; but if this is
to be the case, why does the worker stop its mauling activity when only a
relatively short distance from the host shoot ?

During the dry season, there is some evidence that the presence of a well
developed basal circle can affect survival of the shoot and/or colony during
fires (Fig. 22, 23). When the shoot is part of a dense stand of dry vegetation,
a fire is generally strong enough to completely destroy, or at least to kill, both
the colony and the shoot. However, when the surrounding vegetation is 2 m
or less tall, there is a well developed basal circle, and the shoot of A. cornt-
gera is sufficiently emergent so that most of the canopy is above the surround:
ing canopy, the heat is often not intense enough to kill the ant colony. It
usually kills the shoot of A. cornigera by scorching the lower trunk, rather
than by actually burning the canopy. In some cases, the fire does not even get
hot enough to kill the shoot. In most of these cases of shoot or colony sur-
vival, there is a very large and clean basal circle in which there has been no
fire. It is likely that the complete lack of inflammable materials directly un-
der the shoot lowers the temperature in the canopy sufficiently so that the
ant colony is not killed.

The survival of the ant colony, or portion of the colony, in the scorched
shoot is of very great importance to the total plant; the entire ant colony is
available to move into the new sucker shoots that appear within a week after
the fire. An example of this is as follows. On 19 May 1964, about 15 acres of
15 month old ungrazed regeneration were burned along the east side of the
plots C, H, and I. On 11 Jul., an area 100 by 40 m was marked out in the
center of this burn. There were 365 standing tall shoots in this area; the
small shoots had been completely consumed by the fire. Of these 365 shoots,
96 were dead and had no new suckers from the base. There were 26 shoots
that had not been killed; all of these had canopies in full leaf and were occu-
pied. Nearly all of them had well developed basal circles. The remaining 243
shoots had been killed by scorching but new suckers were growing from the
old root stocks. Of these, 123 were occupied by portions of mature colonies
that had survived the fire. The sucker shoots from the root stocks of these 123
shoots had a mean height of 59.66 cm (s.d.=24.9 cm). The single tallest sucker
from each root stalk was measured. The mean condition rating of these 123
shoots was 3.731 (s.d.=0.383) (Fig. 22). The remainder of the 243 shoots
(120 shoots) were not occupied, but some of the swollen thorns had found-
ing queens in them. These unoccupied shoots had a mean height of 18.08 cm
(s.d=11.2 cm). The mean condition of these 120 shoots was 2.533 (s.d=0.750)
(Fig. 23).

540 Tue University ScteNce BULLETIN

All of the damage to the unoccupied shoots was from phytophagous in-
sects. While the first vegetation in this area following the fire was a dense
stand of grasses, the various perennial species’ roots were beginning to pro-
duce a dense stand of suckers by the middle of July. Almost every unoccupied
shoot was in the dense shade of this 50-75 cm tall vegetation while at least the
shoot tip of nearly all the occupied shoots was a member of the canopy or was
emergent. The vigorous growth of the occupied shoots, in contrast to that of
the unoccupied shoots, can be attributed to the fact that there were usually
100-500 workers on the surface of a shoot that had at most three shoot tips and
20 mature leaves. That there were large numbers of workers available to
patrol the shoots is in great part due to the emergent position of the occupied
canopies and large basal circle before the fire. See Janzen (1967b) for
further discussion of this topic.

ADDITIONAL OBSERVATIONS. Development of colonies in situ. When a plant
community was burned by a sufficiently hot fire to kill all the colonies of P.
ferruginea in the canopies of A. cornigera, the new ant colonies on the site
must either develop in the new regeneration, or invade from the margins of
the burn. When the colonies invade, the resultant effect on the biology of A.
cornigera is little different from the case where the vegetation is only cut and
the colonies not destroyed. However, if the colonies develop in the shoots (in
situ) the pattern of regeneration of A. cornigera is quite different.

One of the abandoned experimental plots simulated the case of a burned
site in which most of the colonies developed in situ. In plot KA the vegeta-
tion was cut to ground level on 1 Oct. 1963. In treatment suplot KA-1, the
new shoots of 4. cornigera were sprayed with parathion on 26 Oct. and 21
Dec., but due to the high density of founding queens and the invasion of two
colonies from outside of the subplot, the applications had little effect on the
interaction of the ants with A. cornigera. In each of the two control subplots
(KA-2, KA-3) there were large numbers (200 plus) of new shoots from old
smal] stumps exposed by the cutting, but only 22 were occupied by the four
large colonies that had survived the cutting. These four colonies stayed in
these 22 shoots and did not move out to occupy more shoots during the
experiments.

During the period from 1 Oct.-1 Aug. 1964, the vegetation in the three
subplots rose gradually until there was a very dense undulating canopy of
grasses, herbs and shrubs 75-120 cm in height. Part of the slowness of growth
of the general vegetation was associated with the very heavy grazing and
trampling by cattle until 3 May 1964. On this date, the general vegetation
had a mean height of about 20 cm. Likewise, on this date the mean height of
the unoccupied shoots of A. cornigera was about 20 cm. The mean height of
the shoots that had been occupied by the mature colonies since late 1963 had
a mean height of about 55 cm. There were some shrubs such as Tournefortia

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 54]

hirsutissima, Bixa orellana and Croton glabellus that grew to 50-100 cm be-
fore the cattle left; like A. cornigera, the cattle almost never browsed these
three species of shrubs.

By 1 Aug., the population of A. cornigera in the three subplots was very
easily segregated into two groups, occupied and unoccupied. There were 99
occupied shoots. Of these 99 shoots, 28 were occupied by the six ant colonies
which had been in the 3 subplots since late 1963. These colonies had moved
into the shoots as large, well developed colonies and began patrolling the
shoots immediately. The remaining 71 shoots were occupied by 53 small
queen-units that had developed in situ. Each colony had 50-250 workers iin it.
There were 18 of the shoots occupied as auxiliary-shoots. At this time, there
were 253 unoccupied shoots in the plot KA. The occupied shoots ranged
from 63-189 cm in height with a mean height of 95.9 cm (s.d=23.6 cm); they
were all canopy members or emergents (Fig. 62). The unoccupied shoots
ranged from 6-100 cm in height with a mean height of 42.58 cm (s.d=18.9
cm); nearly all of them were below the canopy and many were in the dense
shade under the canopy. The occupied shoots had a mean condition rating
of 3.611 (s.d=0.515) and the unoccupied shoots had a mean condition rating
of 2.185 (s.d.=0.877). The 99 occupied shoots weighed 10,790 gm and the 253
unoccupied shoots weighed 5,410 gm.

The young queen-units which had developed in situ began to have work-
ers patrolling the shoot tips about the middle of May. During the dry season,
their biggest contribution was to remove the larvae of the gelichiid web spin-
ner Aristotelia corallina from the shoot tips. When the rains started, the sur-
rounding vegetation began to grow upward and the density of phytophagous
insects increased rapidly. The workers at this time were effective in keeping
these insects away except for the larvae of Coxina hadenoides. Since the shoot
tips remained intact, the occupied shoots were able to keep up with the rising
general canopy, and in some cases, to rise above it. The young colonies were
very small and there were only 25-75 workers on the shoot. However, the
shoots were also very small and therefore a few workers could patrol them
quite effectively.

Based on observations made of other similar regenerating vegetation, prob-
ably less than 10 percent of the shaded unoccupied shoots would ever have
become canopy member or emergent shoots. If the vegetation were to be cut
again, these shoots, or the shoots from their roots, would have had another
chance to become occupied and experience vigorous growth. It could be seen
from plot KA that if an area were cleared every 1-4 years (burned and/or
cut), the occupation by P. ferruginea, and the consequent opportunity to grow
a vigorous shoot, would be rotated among the various root stocks of A. corni-
gera. It appears that often the same root stocks bear the occupied shoots after
each cutting, since the cut canopy is usually left across the cut stump, and the
root stocks that have recently borne an occupied shoot are probably the health-

542 Tue University ScteNcE BULLETIN

“a
: “
rs, by
*) ¢Y a
a y
~<
|
» 4
Pa aN % te,
a al
y
. oe . =
|

a

Fic. 62. A shoot of Acacia cornigera from control subplot KA-3. It is occupied by a small
colony that developed in situ during the preceding nine months. The cluster of leafless branches
in the lower part of the photograph was below the canopy and grew before the colony began
to patrol the shoot. The long branch extending up to the left is the branch that grew after the
workers began to patrol the shoot surface. The loss of mature leaves and shoot tip was due to
nocturnally active phytophagous insects; young colonies tend to decrease their meager patrolling
after dark. Photo 7 Aug. 1964.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 543

iest. It appears that the healthier the root stock, the longer it can continue to
maintain new unoccupied shoots, and the more likely it will be that the new
shoots will have ample thorns, Beltian bodies and foliar nectaries to provide a
founding queen with the food and thorns necessary to rear her first brood
rapidly.

There was no indication in this study that the young colonies moved
around in the plot until they found the tallest shoots. Queen-units appear to
be very stable in respect to staying with the original queen-shoot. Shoot aban-
donment and recolonization is associated almost entirely with auxiliary-units
except when the shoots have been cut. The number of tall shoots, in areas
such as plot KA, that lack old invader colonies matches very closely the num-
ber of new young colonies and new auxiliary-units.

The situation described above in plot KA was approximately replicated in
treatment subplot H-2. In this subplot, the thorns of the unoccupied sucker
shoots from the cut stumps were not clipped. By about the beginning of the
rainy season, 30 young colonies had developed in situ on these stumps. The
effect on the morphology of the shoot was striking. The top of each stump
was enclosed in a cluster of nearly dead and badly stunted short branches that
had been produced before the beginning of the rainy season. Out of the cen-
ter of the cluster of branches there was usually a single well developed branch
with intact leaves and shoot tip (Fig. 62). This branch was 30-100 cm long by
5 Aug. and was patrolled by 20-75 small workers from each of the young colo-
nies. Those acacias that did not have new colonies in them did not show this
new growth. Two of the shoots in this subplot were invaded by an auxiliary-
unit from a large colony in control subplot H-3 about the middle of June.
These two shoots developed in the same manner as those with young colonies
in them.

Auxiliary-shoot effects. Among the emergent occupied shoots of A. corni-
gera in a single stand, there is often strong variation in the heights of the
shoots. A major portion of this variation appears to be associated with various
biotic factors in the environment rather than inherent physiological differ-
ences in growth rate. The sporadic damage of the rutelline scarab Pelrdnota
punctulata causes great variation in height as well as variation in life form;
when it feeds on a shoot, it usually eats all of the undamaged shoot tips, and
it is therefore several weeks or more before a positive height increment occurs.
Since the incidence of feeding damage by Coxina hadenoides is influenced by
the size of the colony on the shoot, height variation is also in part associated
with the size of the colony.

In addition to the direct action of Pelidnota punctulata and Coxina hade-
noides, the auxiliary-shoot phenomenon has a strong effect on height varia-
tion. What appears to happen in many cases is that a single colony occupies a
shoot as a queen-shoot. This shoot grows vigorously and the colony likewise
grows. As the colony becomes larger, it moves out to neighboring small and

544 Tue Universiry SctENcE BULLETIN

unoccupied shoots. When they become occupied, they have a positive height
increment. As the rainy season progresses and the colonies grow, they incor-
porate more and more shoots into their individual groups of auxiliary-shoots.
Since each successively occupied shoot started its vigorous upward growth
later in the year, there is great variation in the individual heights of the shoots.

The auxiliary shoot effect was particularly noticeable in plots N, O, P, L,
and M. These were plots with moderate shoot density, in which the original
sizes of the colonies were relatively small, due to the fact that they were young
colonies developed in situ or had their numbers severely reduced when the
vegetation was last cut.

When the dry season starts in those areas that have high shoot density, and
the new leaf production rate is lowered, there is a gradual abandonment of
10-50 percent of the auxiliary-shoots. By the end of the dry season, a much
larger proportion of the occupied shoots are queen-shoots than is the case at
the beginning of the dry season. When the rainy season begins, and colony
expansion (both in size and area) begins again, the same progressive develop-
ment of height increment began. As shoots become occupied by auxiliary-
units, and as the auxiliary-units grow in size, the height increment rate of the
shoots correspondingly increases. This took place in plots U and V.

Where the shoots are very close together over a large area, the acacias are
not as free of insect damage as is expected. This is apparently due to the fact
that in such a case, many of the shoots (as high as 50) have contacting
branches. Due to the aggressiveness of a colony of P. ferruginea, this series of
contacting shoots has to be occupied by one colony. However, it is not often
that one colony is large enough to occupy effectively such a large series of
shoots. Therefore, there is a higher incidence of insect damage than would
occur if a colony of equal size was isolated in only 5-10 shoots.

Reproductive biology. As is the case with most of the woody plants grow-
ing in the experimental plots, A. cornigera reproduces both by seed and by
suckers from root stocks. However, it does not possess the inherent physio-
logical and morphological properties which are apparently responsible for the
relative freedom from insect attack that is enjoyed by the majority of plant
species in the area. The trait which insures freedom from serious insect dam-
age, namely the occupation by a colony of P. ferruginea, is a trait which has to
be acquired through development of a colony in situ, or invasion by a queen-
unit.

The age at which a shoot becomes occupied is of utmost importance. Ob-
servations of seedlings on newly cleared sites show that the shoots rarely grow
over 25 cm tall, or produce more than five swollen thorns and 25 leaves during
their first rainy season and the following dry season. During this year of
growth, in about half of the shoots that live to be a year old, at least one queen
develops a colony with 10-50 workers in it. If the newly occupied seedling is
growing at a site where it receives at least some direct sunlight, it grows up-

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 545

ward to become a canopy member or emergent. If the seedling is not occupied
by the beginning of the second rainy season, it is usually unable to grow more
than another 10-20 cm until it becomes occupied. However, the longer it re-
mains unoccupied, the higher is the probability that it will become too heavily
shaded for growth, or will be eaten during the dry season by a rodent such as
Sigmodon hispidus. Observations of seedlings of known age indicate that
unoccupied seedlings rarely live more than two years after germination, On
the other hand, young shoots that become occupied in sufficient time for them
to become canopy members or emergents, generally live for 8-10 years in un-
grazed regeneration. They set their first heavy seed crop at the end of the
fifth dry season after the seeds germinate.

The suckers from stumps are generally faster than seedlings in developing
a colony and becoming a canopy member or emergent. At the outset, these
shoots have a swollen thorn at nearly every node, large numbers of Beltian
bodies on the leaves, and functional nectaries on the petiole. They often grow
to 30-50 cm in height while unoccupied. However, as with seedlings, there is
still moderate to high mortality among unoccupied sucker shoots. While it is
usually a year or more before a colony develops in a seedling, a colony with
10-100 workers is usually present in a sucker in less than nine months. Those
suckers which are occupied at the beginning of their second rainy season
grow vigorously, become canopy members or emergents, and live for 8-10
years. They set their first heavy seed crop at the end of the fourth dry season
after the original shoot was cut or burned.

Frequently, in both natural and man-made disturbance sites, a colony does
not develop in situ in suckers from stumps; the mature colony moves directly
from the dead shoot into the new suckers. The shoot is occupied from the
first day of growth. These shoots suffer almost no mortality during their first
five years of growth. The first heavy seed crop is produced at the end of the
third dry season after the original shoot was cut or scorched.

The age attained by a shoot is in part a function of the type of plant com-
munity. The figure of 8-10 years cited above is in reference to shoots growing
in ungrazed regeneration. In old established pastures in the vicinity of
Campo Cotaxtla, occupied shoots are commonly found with 10-15 annual
rings; these shoots are growing in open grassland and are rarely over 5 m
tall. On river banks and in old roadside drainage ditches throughout the
study area, shoots of this age were occasionally encountered. In the area
around Temascal the shoots in pastures and fallow corn fields rarely live over
5 years because of the sporadic cutting and burning of these sites.

The physiognomy of the vegetation at the time a seedling or sucker ap-
pears is also of importance. Seedlings and suckers in heavy shade usually die
within 2-3 years, irrespective of the presence of a colony of P. ferruginea. A
seedling or sucker growing on a site which is cleared but not burned, is in
much greater danger of being shaded before it becomes occupied than a seed-

546 Tue Universiry SctENcE BULLETIN

ling or sucker on a site that has been burned. During the first year after a
fire, the vegetation is much more irregular and lower than the regeneration
following a cutting. Seedlings and suckers in brushy pastures are in less
danger of being shaded than those in ungrazed vegetation; the cattle eat most
of the vines and open the general canopy with their trails and browsing. The
cattle generally ignore the young unoccupied A. cornigera.

The reproductive features described above are further complicated by the
time of year at which the seed germinates, or the sucker shoots appear. Seeds
usually germinate during the first two months of the rainy season, and are
almost immediately in danger of being heavily shaded. During the dry sea-
son, the vegetation opens up somewhat, and allows sunlight to reach the lower
strata. If the seedlings survive until this time, they are still very likely to be
eaten by Sigmodon hispidus, and to have any new shoot tips eaten by the lar-
vae of the gelichiid moth Aristotelia corallina. If the seeds germinate late in
the rainy season, the seedlings generally do not grow a root system deep
enough to be able to survive the drying of the upper layers of soil during the
dry season.

If a shoot is cut or burned during the rainy season, the suckers are sub-
jected to severe defoliation by the abundant insects. The new branches from
the stumps are often stripped of all green vegetation, and at times, all of the
green thorns are eaten; in such a case, a new colony cannot possibly develop in
situ. By the time that insect populations are lowered by the cool and diy sea-
son, the root stock is often either dead or so weakened that it produces very
few leaves and swollen thorns. If a shoot is cut or burned during the dry
season, the new suckers are often vigorous and relatively free from insect
damage. However, if such a shoot does not become occupied by the begin-
ning of the following rainy season, the high height increment rate and condi-
tion value are abruptly lowered as the shoot is subjected to severe defoliation
by insects.

Of the approximately 60,000 seeds that are produced by a shoot of A.
cornigera in its lifetime in the Temascal area, somewhat less than one percent
have the opportunity to be dispersed to sites for germination. The other 99
percent are destroyed by bruchids while in the legumes hanging on the shoots.
Under natural conditions, dispersal by water and birds is further complicated
by the relative paucity of newly created disturbance sites in which the new
seedling can grow well. Dispersal to these sites is, however, aided by the habi-
tat preferences of the birds concerned (newly disturbed sites) and the usual
deposition site of the floating legumes (areas cleared by the action of the cur-
rent). In areas of man’s activities, there are usually many acres of recently
disturbed vegetation within a very short distance of the seed-bearing shoot.
In this case, birds are quite effective in distributing seeds to suitable areas.
Furthermore, in the soil under any old stand of regeneration, there is un-

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 547

doubtedly a reservoir of viable seeds that have not germinated. Such seeds
remain viable at least three years, and probably much longer.

Reproduction by seed is complicated by another factor. In man-made dis-
turbance sites, the regeneration is frequently cut or burned when 1-5 years
old. The shorter the time required for a shoot to produce its first heavy seed
crop (about 22,000 seeds), the higher the probability that it will be able to set
seed before it is killed. As was stated above, seedlings require five years, suck-
ers with a colony developed in situ require four years, and suckers with a
mature colony from the start require three years, to produce the first heavy
seed crop. It is of importance to the shoot that the root system be able to
continue to grow new shoots after each cutting and/or burning until a shoot
is left long enough to set a seed crop. It should be remembered that the shoot
has to survive at least nine months after the first heavy flower crop in order
for the seeds to mature. It is likewise of importance that the ant colony is
able to continue to move from the cut or scorched shoots into the new sucker
shoots, rather than dying with the shoot. The habit of establishing auxiliary-
units in neighboring unoccupied shoots is beneficial in that one mature colony
can insure normal growth for more than one shoot. However, if the shoot
density becomes so great that there is too much shoot surface area to patrol
effectively, the growth rates of all of the shoots except the queen-shoot will be
reduced. In natural disturbance sites, these factors are perhaps of slightly less
importance since the vegetation is less frequently removed from these sites,
and the shoots are more highly dispersed.

Naturally unoccupied shoots. The naturally unoccupied shoots in the
study area can be roughly arranged into three groups: 1) seedlings and young
suckers in which colonies are developing in situ, or that are later invaded by
auxiliary-units, 2) well developed shoots that are abandoned by auxiliary-units
during the dry season, but are later re-occupied, and 3) well developed shoots
that were never occupied. Members of all of these groups are found in both
natural and man-made disturbance sites. The first two have been discussed
in sufficient detail in previous sections; the shoots in these two groups almost
invariably do not contribute to the sexual reproduction of A. cornigera unless
they are occupied for the major portion of their life spans.

Well developed shoots that were never occupied are extremely rare in the
study area. This is in part due to the great ability of founding queens and
auxiliary-units to find unoccupied shoots. This makes it very difficult to de-
termine on an observational basis how much damage would be received by A.
cornigera if P. ferruginea were not present. However, the experimental plots,
the observations of unoccupied seedlings and young suckers, and the observa-
tions of the very rare unoccupied well developed shoots, show that the damage
is very substantial. In spite of this, there are exceptional micro-habitats in
which unoccupied shoots can grow; the presence of these few shoots indicates

548 Tue University SciENCE BULLETIN

that A. cornigera probably would not become extinct in the study area were
P. ferruginea (and P. nigrocincta) to become abruptly extinct.

All of the well developed unoccupied shoots of A. cornigera found in this
study were growing in relatively barren habitats. The shoot in figure 47 was
growing on barren open ground at the base of the dam at Temascal; it had
17 legumes on it and had produced several flowers during the 1964 dry season.
It had never been occupied by P. ferruginea. Near Tierra Blanca, Veracruz,
an isolated single unoccupied shoot about 2 m tall was found growing on a
large sewage outwash delta below a drainage culvert; during most of the dry
season its uppermost growing point was undamaged. In Temascal, an unoc-
cupied shoot about 150 cm tall was found growing in a garden; during the
dry season and the first month of the rainy season it was free from phytopha-
gous insect damage. On a large gravel bar in the Rio Tonto east of Temascal,
a twisted and water-washed 2 m tall unoccupied shoot was found with four
legumes and several flowers; it was apparently able to survive the rainy season
floods. The first three shoots mentioned above were tenanted by a colony of
Pseudomyrmex gracilis mexicana. They were growing in sites where there
were no other A. cornigera within 200 m, and phytophagous insects were un-
doubtedly very low in numbers in the immediate area of these three shoots.

Unlike the obligate dependence association of P. ferruginea on a swollen-
thorn acacia, A. cornigera is not totally dependent upon P. ferruginea to grow.
Not only have a few unoccupied shoots been found that are well developed,
but it is also possible that there are sufficiently barren habitats such that unoc-
cupied shoots can produce a seed crop. It is impossible to determine by obser-
vation if there are members of the occupied portion of the population of A.
cormigera that retain the chemical characteristics that keep other species of
Mimosaceae in the Temascal area relatively free from insect damage. Cer-
tainly no such shoots were among the approximately 2500 unoccupied shoots
in the experimental plots.

Economics. From conversations with land owners, and by observations of
pastures, it was evident that in the study area, A. cornigera is an economic
weed of some importance. On sites that are burned on a regular annual basis
(permanent corn fields, cane fields, and some grass pastures), A. cornigera is
present only as a very minor element of the vegetation; the few shoots are
usually less than 50 cm tall, stunted, nearly dead, and unoccupied. On sites
that are cut and/or burned on a relatively haphazard schedule, A. cornigera is
often a common and vigorous plant. In brushy pastures it shares a dominant
position with such woody plants as Croton glabellus, Bixa orellana, Eupato-
rium odoratum, Parmentiera edulis, Bauhinia ungulata, and Jatropha urens;
none of these species are eaten by cattle. In well established grass pastures, 4.
cornigera occasionally comprises more than 90 per cent of the woody vegeta-
tion. At times it became so dense in patches that it is impossible to walk be-
tween the plants without being stuck by the thorns.

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 549

The production of dense vigorous stands of A. cornigera is primarily due
to the way in which land is cleared. In both brushy and grass pastures, it is a
common practice to cut the woody vegetation with a machete at intervals of
0.5-3 years; often this is not followed by burning. The cut shoots are left lying
over the stumps and the old mature colonies move into the new sucker shoots.
This cutting appears to cause almost no mortality among the root systems of
A. cornigera, while it appears to gradually eliminate the other species of
woody plants. In fact, when the shoots of A. cornigera are cut, the colonies
often move in part into neighboring stunted shoots that were previously un-
occupied; thus the ratio of occupied to unoccupied shoots is increased by this
type of cutting. In some areas around La Granja, the practice of cutting with-
out burning has been followed for at least 15 years with no decrease in the
density of vigorous shoots of A. cornigera.

Cutting is occasionally followed by burning in brushy and grass pastures.
Since the cut shoot with its ant colony is lying at ground level, it is consumed
by the fire. This burning usually occurs during the dry season. During the
following year it is necessary for a colony to develop in situ in the sucker
shoots before the ant colony develops, but generally the root stock is strong
enough to survive without a vigorous shoot until a colony is large enough to
occupy the shoot. The critical factor in this type of shoot and colony destruc-
tion is that this burning does not occur every year. It appears that as long as
the root system bears an occupied vigorous shoot for at least 1-2 years, it is
then able to survive for the six months to a year following a fire while a
colony develops in situ. It should be emphasized that in the pastures exam-
ined, it is insect damage that stunts the unoccupied suckers of A. cornigera,
and not the browsing of cattle.

A further complication arises when newly cut brushy pastures are not
burned evenly. It is often the case that the fires burn in strips and patches,
and the cut shoots of A. cornigera are occasionally missed. If the unburned
cut shoot contained a queen-unit, the colony then moves into an available
sucker shoot. If it contains an auxiliary-unit, it also moves into a sucker shoot,
but the unit gradually dies over a period of one to two months. However, the
shoot is occupied during that period and often grows to a height of 50-100 cm
above that of the unoccupied suckers in the pasture.

In grass pastures, the basal circle has to be taken into consideration. Dur-
ing the dry season, the grasses are often trampled and grazed to ground level
by the cattle. If such a pasture is burned, the fire usually passes rapidly. As
long as the shoots are not cut, the colony of P. ferruginea is often high enough
so that the heat does not kill it. On some occasions, the lower trunk is not
scorched due to the complete lack of imflammable materials in the basal circle.
It should be emphasized that this is usually the case only in grass pastures
where the dry grasses are lying very close to the ground. In taller vegetation,
the trunk is usually scorched, but the ant colony often survives.

550 Tue University SciENcE BULLETIN

It appears that A. cernigera can be removed from pastures with relatively
minor changes in management plans. In established corn and cane fields it is
not present, or if present, is a very minor part of the vegetation; this appears
to be almost entirely due to the annual burning of these fields. If pastures
could be burned on an annual basis, the same effect would undoubtedly be
achieved. However, regular annual burning is regarded by the land owners
as detrimental to the soils, and to many of the grasses which are being pro-
moted in the pastures. Since regular annual burning cannot be used, an
equivalent practice should be effective. If the cut shoots are piled and burned,
or just piled, 50 or more meters from any A. cornigera stumps, the new suck-
ers from the cut stumps will be unoccupied. After this has been done an-
nually for 2-4 years, it is very unlikely that A. cornigera will form more than
a very minor part of the woody vegetation. There is the continuous threat of
new colonies developing in situ in seedlings, but as larger and larger areas
are cleared of their vigorous stands of A. cornigera, the frequency of both
seedlings and founding queens will be greatly reduced.

DISCUSSION

At both the individual and population level, Acacia cornigera has charac-
teristics which are associated with its interaction with Pseudomyrmex ferru-
ginea. These characteristics fall into two groups. The less controversial char-
acteristics are those which are necessary for the maintenance of a colony of the
ant: swollen thorns, Beltian bodies, and foliar nectaries. These features are
discussed at length in the first section of this paper. In view of the interde-
pendent relationship that has been demonstrated between the ant and the
acacia, it is certain that the present-day selective pressure, which makes the
energy expended in the production of these three morphological structures
profitable to the acacia, is the phytophagous insect damage with subsequent
shading that is experienced by the acacia when it is not occupied by P. ferru-
ginea (or P. nigrocincta). The characteristics that require close observation
under experimental conditions for verification, are ecological and physiolog-
ical, and the shoot requires occupation by P. ferruginea for their normal ex-
pression. In this latter group are the acacia’s (1) high growth rate, (2) rela-
tive freedom from insect and mamal damage, (3) freedom from use by vines
as a standard, (4) emergent or canopy member position, (5) abundance in
man-made disturbance sites, and (6) attainment of sexual maturity. These
interdependent points are discussed below.

Observations in both natural and man-made disturbance sites in the study
area show that 4. cornigera has a height increment rate which is superior to
that of most woody plants during the first 8-12 years of secondary succession.
This is evidenced by its position as a canopy member or emergent, and by the
recorded rate of main-axis elongation of 2.50 cm per 24-hour day. However,

INTERACTION OF THE BuLL’s-HoRN ACACIA WITH AN ANT INHABITANT 551

in the experimental plots, it is clearly shown that if the colony of P. ferruginea
is removed from the shoot, this high height increment rate is abruptly low-
ered. This is likewise the case for the high rate of leaf and thorn produc-
tion, excellent condition of the shoot, and high total biomass production.
In the experimental plots, this reduction is a result of the feeding activities of
phytophagous insects and rodents, which are normally prevented by P. ferrz-
ginea from feeding on the shoot. In addition, once the unoccupied shoot
looses its ability to keep up with the surrounding vegetation, it becomes
shaded, and this further stunts its growth. There is no doubt that this reduc-
tion in growth is due to these factors, rather than to some detrimental effect
of the techniques used to remove the ants in the treatment subplots. In addi-
tion to damage of the unoccupied shoots, there is high mortality of the shoots
due to the cumulative effect of shading and repeated damage. The slowing of
growth rates, and direct mortality through the damage of unoccupied shoots,
is of obvious significance in natural selection for shoots with the properties
necessary to maintain a large colony of P. ferruginea (swollen thorns, Beltian
bodies, nectaries).

The differences in growth rate and mortality between occupied and unoc-
cupied shoots are of course affected by some major variables: the time of year,
the structure of the plant community, and the sizes of the occupant ant colony.
During the dry season, growth rates of all the plants in the community are
reduced or stopped. Associated with this, there is a great lowering of phy-
tophagous insect activity. However, growth rates of A. cornigera are only
partly lowered; continued leaf production and maintenance during the dry
season are necessary for survival of the ant colony. But since phytophagous
insect activity is greatly lowered and the shade of the general surrounding
vegetation is greatly reduced, unoccupied shoots are able to grow more at this
time of year than at any other. This factor makes the natural abandonment
of some shoots by auxiliary-units during the dry season of much less serious
consequence than abandonment during the rainy season (a very rare occur-
rence). It should be noted that this abandonment occurs as a consequence
of partial or total failure of the shoot to produce sufhcient food for the unit of
ants. However, as long as sufficient shoots are present in the acacia population
with nectaries and new leaves with Beltian bodies, to support the colonies
present in a given area, abandonment during the dry season is of little imme-
diate consequence to P. ferruginea. Thus, certain variation in reaction to the
dry season by leaf drop and growth cessation within the population of A.
cornigera, where dense, is not inimical to the existence of the ant-acacia inter-
action; when the rainy season begins, the ants move back into the abandoned
shoots. Since phytophagous insect activity increases greatly during the rainy
season, unoccupied shoots have almost no growth, while the occupied shoots
grow vigorously. It should be remembered that in those areas where A. corni-
gera has a low population density, it is less likely that there will be a nearby

Sb Tue Universiry SciENcE BULLETIN

shoot with food for a starving colony to move into if it abandons its unpro-
ductive shoot. In this case, there should be severe selection for shoots with the
ability to continue to produce food for the ant colony throughout the dry sea-
son. Relatively low densities of A. cornigera are the rule in natural disturb-
ance sites.

While the structure of the plant community has a relatively small effect
on the quantity of phytophagous insect damage (it is nearly always severe),
it has a large effect on the degree of shading of shoots, and consequent shoot
mortality. In the dense vegetation of ungrazed secondary succession, unoccu-
pied shoots quickly become heavily shaded due to their inability to keep above
the rising vegetation, that is in turn due to the lowered growth rate. Even if
a shoot becomes occupied while it is heavily shaded, it is normally unable to
reach the canopy and become sexually reproductive. However, in heavily
grazed and browsed pastures, many of the shading shrubs and vines are
thinned out; the insolated unoccupied shoots produce more leaves and swol-
len thorns and appear to be more tolerant of insect damage. Due to this, a
colony of P. ferruginea has a longer time to develop in situ before the shoot is
killed by cumulative insect or rodent damage. In this sense, it can be said that
cattle promote the survival of unoccupied acacias. Grazing and browsing
likewise reduce the selective pressure against shoots which are slow to acquire
an ant colony, or have low height increment rates. Those very few shoots
which reach sexual maturity without being occupied by P. ferruginea are
virtually always growing on sites that have been almost completely cleaned
of other vegetation (e.g., rock quarries), or for some other reason have very
low phytophagous insect populations.

Even though the shoot is occupied by a queen-unit or auxiliary-unit of P.
ferruginea, it is of course not guaranteed to have a high growth rate. The
larger the unit in proportion to the size of the shoot, the less insect damage
and shading occurs. In view of this, it is expected that there is strong selection
for shoot characteristics that promote the rapid growth of the colony—such as
heavy nectar flow, high and continuous rate of Beltian body production, large
and durable swollen thorns, and the growth of new suckers from cut stumps
with the foliar characteristics of a mature shoot. Not only does the colony
need to be large for efficient patrolling of the shoot, but the weather conditions
must be suitable for normal worker activity outside of the thorn. P. ferra-
ginea has developed the ability to be active under much more inclement
weather than those Psewdomyrmex not associated with living plants. Associ-
ated with this, it is not surprising that P. ferruginea has developed the be-
havioral trait of being active outside of the thorns throughout the 24-hour day
(Fig. 32 and 33), quite in contrast to those species of Pseudomyrmex which
are not obligatorily associated with living plants. These latter are strictly
diurnal in their foraging behavior. It should be noted that variation in the

INTERACTION OF THE BULL’s-HoRN ACACIA WITH AN ANT INHABITANT 553

size of the ant colony and in weather conditions, were the two factors of most
importance in the promotion of variation in height increment rate and condi-
tion of A. cornigera in the experimental plots.

Like the surrounding vegetation in the plant community, heavily occupied
shoots of A. cornigera show little obvious insect damage to the leaves and
shoot tips. It should be noted however, that upon close examination most
species of plants in the plant community demonstrate minor to major damage
to growing points that is not immediately obvious. Further, damaged plant
parts are not always obvious because they have been removed by the insect.
This is of course not the case with unoccupied shoots of A. cornigera. Natu-
rally unoccupied shoots, such as very young suckers and seedlings, have a
very high frequency of shoot tip destruction and their mature leaves are usu-
ally badly damaged. This damage is however, normal for the reproductive
biology of A. cornigera, and some of these plants survive until they become
occupied. It is noteworthy that young unoccupied shoots of A. cornigera are
able to live for periods of 1-2 years with only 4-10 mature leaves; there is not,
however, any evidence that seedlings or young suckers are any less palatable
to phytophagous insects than mature shoots. In the experimental plots, it has
been clearly shown that the frequency of insect damage is greatly increased
if the ant colony is removed (Tab. 15-27). This removal of the ants results in
a great lowering of growth rate (Tab. 4-13) by removal of the shoot tips by
phytophagous insects. However, as the mature leaves are destroyed, and
more and more shoot tips are eaten, the plant is weakened and its physiolog-
ical ability to have a high production rate is lowered as well. It is probable
that in this weakened state, the shoot is much more susceptible to attacks by
the wood-boring buprestid larvae of Chrysobothris sp. near C. multistigmosa.
In addition, if the branches are cut off of such weakened shoots, the stump
shows very poor production of new foliage. It should be remembered that the
direct consequence of failure to grow, shading by the rising surrounding
vegetation results in lowering of the physiological growth rate as well.

With the exception of those insects that have developed the ability to with-
stand or avoid the attack of P. ferruginea, occupied shoots are virtually free
of phytophagous insects. This has been shown quite clearly by the censuses
of insects on occupied and unoccupied shoots (Tab. 1). That this difference
is due to the activity of the ants, and not some property of the shoots, is shown
by the several hundred observations in this study of these same species of in-
sects being attacked and removed from the shoot by the ants. Thus, while it
is obvious that plants have many characteristics that allow only those insects
adapted to those characteristics to feed on them, it appears that 4. cornigera
has lost some of these characteristics (bitter chemicals in the new foliage,
fibrous tissues in new foliage, canopy member or submergent). However, 1n
return it has gained a characteristic that serves the same purpose, namely P.

--

554 Tue University SciENcE BULLETIN

ferruginea. Occupation by the ant requires an expenditure of energy on the
part of the shoot in the form of swollen thorns, Beltian bodies, foliar nec-
taries, and new leaves during the dry season. In return, the shoot is effectively
relieved from the phytophagous insect damage which was shown by this
study to be inflicted on unoccupied shoots. A. cornigera (and other swollen-
thorn acacias) is unique among plants in that one of its major deterrent
characteristics to insect attack can be experimentally removed without di-
rectly altering the morphology or physiology of the plant. Associated with
this, the new seedling or sucker does not inherently possess the characteristic
that allows A. cornigera relative freedom from insect attack; a colony of P.
ferruginea must develop in situ or invade from another shoot.

Unfortunately, the role that P. ferruginea plays in deterring mammals
could not be thoroughly documented. In this study, occupied shoots showed
almost no mammal damage. There is little doubt that the rodent Sigmodon
hispidus is deterred by P. ferruginea from eating young shoots if the shoots
are occupied. At present, cattle are the only large browsing mammals that
are abundant enough to detrimentally affect 4. cornigera. In view of the will-
ingness of the Brocket Deer to eat A. cornigera foliage, and the active brows-
ing of cut shoots by cattle, the reluctance of cattle to browse standing A.
cornigera in their pastures is probably not due to a bad taste in the plant
foliage. Likewise, the thorns do not have an absolute deterrent effect on
either the Brocket Deer or cattle. In fact, they did not appear to have any
effect on the feeding of these two mammals. However, it is reasonable to
expect that the thorns might cause the plant to be placed somewhat lower on
the host plant preference list of large browsing mammals. In view of the
Brocket Deer’s reaction to P. ferruginea, it is likely that under natural circum-
stances it would not feed on occupied shoots. As far as cattle are concerned,
it is still only indirect evidence that indicates that they are turned away from
occupied shoots by the ants, but it is difficult to imagine for what other rea-
son they do not browse A. cornigera freely.

As is the case with most plants, there are a few insects that have evolved
to feed on A. cornigera despite its characteristics. These are the few insects
that are not efficiently removed by the ants, such as Pelidnota punctulata,
Coxina hadenoides, Syssphinx mexicana, Rosema dentifera, Acanthoscelides
oblongoguttatus, Mimosestes sp. However, A. cornigera shows good growth
in spite of the damage by these insects. These insects may be regarded in the
same light as other insects in the environment which, though they feed on
their host plants, have evolved to a state of balance whereby they do not de-
stroy their host population, yet they harvest enough of it to survive. There
are a few insects such as Mozena tomentosa and Artstotelia corallina that
normally are found only on naturally unoccupied shoots; they are in large
part responsible for the very slow growth rate of unoccupied young suckers

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 555

and seedlings. No one of the eight species listed above, nor the cumulative
sum of their feeding, is serious in comparison to the sum effect of those in-
sects that feed on the plant when the ants are removed.

It must be noted that a major part of the usefulness of P. ferruginea to A.
cornigera depends on the presence of phytophagous insects in the environ-
ment that will feed on A. cornigera when the ants are removed. In the
Temascal area, there are at least 40 species of insects that feed on unoccupied
shoots of A. cornigera. While each individual insect does relatively little
damage in comparison to that done by an individual of an insect such as
Pelidnota punctulata, the cumulative effect of these insects feeding on the
shoots night and day, and in relatively large numbers, is almost entirely re-
sponsible for the lack of growth and poor condition of unoccupied shoots in
the experimental plots. As one would expect, the normal primary host plants
of these insects are the numerous species in mimosaceous genera that occur in
the same plant community with A. cornigera, e.g., Calliandra houstoniana,
Leucanea glauca, Acacia farnesiana, Acacia macracantha, Acacia angustatis-
sima, Enterolobium cyclocarpum, Mimosa albida, Inga sp., and others. Each
of these plant species has a number of insect species that feed heavily on it,
and occasionally on other plants. In other words, the insects which feed on
unoccupied A. cornigera have host preference lists, at the top of which is
some other species of plant (often a mimosaceous genus) and at the bottom
of the list is occupied A. cornigera. Removal of the ants moves A. cornigera
very close to the tip of the host preference list.

It is clear from the experimental plots, observation of naturally unoccupied
shoots, and the reaction of workers of P. ferruginea to foreign vegetation, that
the ant prevents vines and lateral branches from growing into the canopy of
occupied A. cornigera. Structurally, an occupied shoot is ideal for the support
of vines. It has a stiff and often emergent central axis with strong lateral
branches extending out into, and over, the general canopy. The numerous
swollen thorns are excellent sites for tendril attachment. However, occupied
shoots are almost invariably free of living vines due to the ant’s habit of maul-
ing foreign vegetation. It should be noted that the other emergent woody
plants in the first several years of succession have structural properties which
lower the frequency of vines using them for standards (e.g., Brxa orellana,
Croton glabellus, Bauhinia ungulata). These plants have slightly drooping
leaves with long petioles and very smooth upper surfaces; as they blow in the
wind, it is very difficult for vine runners to become attached to them. They
also have very smooth bark. In addition, these plants do not develop many
lateral branches, and this removes another potential vine support. Associated
with this, many of the vines that use unoccupied shoots of A. cornigera as
standards normally have their primary growth in a lateral direction through
the general canopy rather than piling up on the emergent species of plants.

556 Tue Universiry SctENcE BULLETIN

When a canopy member, or emergent, shoot of A. cornigera is deprived
of its ant colony in ungrazed vegetation, the vines frequently form a heavy
mat over the shoot, or use it as a standard to rise above the general canopy.
The emergent unoccupied shoots often accumulate very heavy masses of vines
during the rainy season, and as a consequence are bent over and firmly bound
into the general canopy. However, the thing that makes the unoccupied shoot
most severely susceptible to vine damage is that there is also such severe insect
damage to the shoot tips that the shoot cannot grow up through the vine mat.
Virtually all of the mature foliage present at the time the ants are removed be-
comes shaded by the vines. Thus not only does the ant alleviate the potential
danger of the vines by mauling them, but it also is responsible for protecting
the rapidly elongating main shoot tip which is the only protection against
vines that the shoot could have were the ants to be removed.

As a vertical downward extension of the workers’ activity in keeping for-
eign vegetation out of the canopy of A. cornigera, the workers often produce
a bare basal circle under the shoot that is free of living foliate parts of other
plants. If the workers gain directly from their chewing of vines in the canopy
(i.e., by getting plant juices), they very likely profit in the same direct manner
when mauling vegetation in the basal circle. However, it is more likely that
the use of the basal circle to P. ferruginea is indirect. When the basal circle is
large, fires passing through the general vegetation during the dry season are
less likely to destroy the ant colony. This is of great significance to A. corni-
gera in that while the shoot is often killed by fire, suckers from the roots are
immediately occupied by the mature ant colony. In addition it is likely that
the bare basal circle lowers the incidence of phytophagous insects reaching
the shoot. The removal of foreign vegetation from the canopy of A. cornigera
may serve the same purpose.

In view of the definite effect that P. ferruginea has on the growth rate of
shoots of A. cornigera, it is reasonable to consider the ant-acacia interaction in
explaining the abundance of A. cornigera in man made disturbance sites in
contrast to other species of acacia. In the area between Temascal and La
Granja, there are at least five native species of acacia: A. cornigera, A. mac-
racantha, A, chiapensis, A. farnesiana, and A. angustatissima. All of these
species grow in natural disturbance sites such as river banks and arroyos;
while A. cornigera is often the most common, none of the species have an out-
standing dominance in respect to numbers of individuals. While all five spe-
cies have invaded man’s fields and pastures, only A. cornigera has become
widespread and very abundant. In some fields and pastures, it is an economic
weed of considerable importance. Initially, this successful population expan-
sion was due to the efficient dispersal of seeds by birds both to the newly
cleared sites, and between areas under continuous disturbance. The seeds of
the other four species of acacia are distributed by gravity and water. Once a
plant of A. cornigera has become established, it grows rapidly, and is very

INTERACTION OF THE BULL’s-HORN ACACIA WITH AN ANT INHABITANT 557

hardy even when the shoot is repeatedly cut or burned. However, in the case
of both of these factors (seed dispersal and shoot hardiness) P. ferruginea
must be considered. Unoccupied shoots do not normally survive long enough
to produce seeds which can then be dispersed. Secondly, even if the roots of
an established plant produce new suckers after the shoot has been killed, an
ant colony must be present to occupy the shoot, in order for successful growth
to occur. Thus it is that fields that are annually burned lack A. cornigera as
maturing shoots. Associated with this is the significant fact that the sooner a
shoot becomes occupied, the sooner it produces its first heavy seed crop. This
becomes very important in an area where the vegetation is cut frequently (on
a 2-5 year cycle). Finally, it is very doubtful that A. cornigera would have
become so common in man-made disturbance sites had P. ferruginea been un-
able to move out of the natural disturbance sites with it, and had P. ferruginea
been unable to survive under the various shoot and colony destruction regimes
that man’s activities produce.

Based on the data presented and discussed above, the following conclusion
appears justified. Pseudomyrmex ferruginea is dependent upon a swollen-
thorn acacia for survival, and Acacia cornigera is dependent on an obligate
acacia-ant for normal development; the interaction between the two can there-
fore be properly termed one of obligatory mutualism. Were all the colonies of
P. ferruginea (and P. nigrocincta) in the study area to abruptly disappear,
the density, distribution, and condition of the population of A. cornigera
would be drastically lowered. Individual shoots would survive in relatively
isolated sites, and under unusual circumstances. Presumably, the population
would be able to evolve the characteristic resistance to insects and mammal
damage that is displayed by the species of Acacia that are not associated with
ants.

Thus, as was stated in the introduction, the opinions of earlier authors who
felt that the ant “protects” the acacia have been clearly substantiated by this
study. This does not apply of course to the other ant and plant relationships
that these authors also discussed. The bearing of the demonstrated obligate
mutualism between P. ferruginea and A. cornigera on other ant-plant interac-
tions will be discussed in a later paper. However, it should be noted that in-
terpretation of the facts associated with these other interactions is made much
simpler by the present understanding of the interaction between P. ferruginea
and A. cornigera.

LIFERADURE €lRED

AvFaro, A. 1935. Investigaciones cientificas. San Jose de Costa Rica, Trejos Hnos. 317 pp.

Bett, T. 1874. The naturalist in Nicaragua. London, E. Bumpas. 403 pp.

Brown, W. L., Jr. 1960. Ants, acacias and browsing mammals, Ecology 41:587-592.

CoMISION DEL PapaLoaPaNn. 1962. Boletin hidrologico No. 14. Mexico, D. F.

Darwin, C. 1877. Cross and self-fertilization in the vegetable kingdom. Appleton, London.
482 pp.

558 Tue University ScreNcE BuLLerIn

Darwin, F. 1877. On the glandular bodies on Acacia sphaerocephala and Cecropia peltata
serving as food bodies. Jour. Linn. Soc. London, Bot. 15:398-409.

Devpino, F. 1886-1889. Funzione mirmecofila nel regno vegetabile. Prodromo d’una mono-
grafia delle piante formicarie. Mem. Acc. Bologna 7:215-323 (1886); 8:691-659
(1887): 10:115-147 (1889).

Emery, C. 1890. Studi sulle formiche della fauna neotropica. Bull. Soc. Ent. Ital. 22:38-80.

Hovprince, L. R. 1964. Life zone ecology. Tropical Science Center, San Jose, Costa Rica.
124 pp.

InERING, H. von. 1907. Die Cecropien und ihre Schutzameisen. Engler’s Bot. Jahrb. 39:666-
7V4.

Janzen, D. H. 1966. Coevolution of mutualism between ants and acacias in Central America.

Evolution 20:249-275.

. 1967a. Obligate acacia-ants of Central America. ms.

.1967b. Fire, vegetation structure, and the ant X acacia interaction in Central America.

Ecology, in press.

Leopotp, S. 1959. Wildlife of Mexico. Univ. Calif. Press. 556 pp.

Rettic, E. 1904. Ameisenpflanzen-Pflanzenameisen. Beih. z. Bot. Centralbl. 17:89-122.

Rupp, V. E. 1964. Nomenclatural problerns in the Acacia cornigera complex. Madrono 17:
198-201.

Sarrorp, W. H. 1922. Ant acacias and acacia ants of Mexico and Central America. Smith-
sonian Report for 1921:381-394.

ScHimpER, A. F. W. 1888. Die Wechselbeziehungen zwischen Pflanzen und Ameisen im
tropishen Amerika. Jena. 95 pp.

ScHwarz, E. A. 1917. Ants protecting acacia trees in Central America. Proc. Ent. Soc. Wash.
18(1916):211-212.

Skwarra, E. 1930. Ameisen und Ameisenpflanzen im Staate Veracruz (Mexiko). 4. Wander-

vers. Deutsch. Ent. Kiel: 160-170.

.1934a. Oekologie der Lebensgemeinschaften mexikanischer Ameisenpflanzen. Zeitschr.

Morph. Oekel. Tiere 29:306-373.

. 1934b. Oekologische Studien uber Ameisen und Ameisenpflanzen im Mexiko. Konigs-

berg Pr. 153 pp.

SmitH, F. 1877. Descriptions of new species of the genera Pseudomyrma and Tetraponera
belonging to the family Myrmicidae. Trans. Roy. Ent. Soc. London 25:57-72.
STANDLEY, P. C. 1928. Flora of the Panama Canal Zone. Contrib. U. S. National Herbarium

27:1-416.

Ute, E. 1905. Wechselbeziehungen zwischen Ameisen und Pflanzen. Flora 94:491-497.

. 1906. Ameisenpflanzen. Engler’s Bot. Jahrb. 37:335-352.

UpnuorF, J. C. T. 1942. Ecological relations of plants with ants and termites. Bot. Rev. 8:563-
598.

WasMANN, E. S. J. 1915. Eine neue Pseudomyrma aus der Ochsendornakazie in Mexiko.

Tijdschr. v. Entom. 58:296-352.

. 1916. Nachtrag zu “Eine neue Pseudomyrma aus der Ochsendornakazie in Mexiko.”’

Tijdschr. v. Entom., Suppl. 58:125-131.

WueEELER, G. C. and J. WHEELER. 1956. The ant larvae of the subfamily Pseudomyrmecinae
(Hymenoptera:Formicidae). Ann. Ent. Soc. America 49:374-398.

WHEELER, W. M. 1913. Observations on the Central American Acacia ants. Trans.. 2d.
Intern. Entom. Congr. Oxford (1912) 2:109-139. s

——. 1942. Studies of neotropical ant-plants and their ants. Bull. Mus. Comp. Zool.
Harvard 90:1-262.

—-—, and I. W. Battey. 1920. The feeding habits of Pseudomyrmine and other ants. Trans.
Amer. Phil. Soc., (N.S.) 22:235-279.

Witson, E. O. 1953. The origin and evolution of polymorphism in ants. Quart. Rev. Biol.
28:136-156.

fs Me?) lt Ao ee | ey
ih Ni i es 3a"

4c 4
7 La bs _ ‘< Bb } " i ?
q F ur ily fh Ny
‘ Wve) & re. *
al ‘ y A eer

ro!
Bien: wr
i ed) tk
i

Ta Aa vi, i He PAE, -

‘ | ao Ai UE
: : a —

thi

; i |

_% iF

Nf 2

wt

é ae
snp

& i sae st 1 (

——-

fer £
I % 2° f

“>
+ + Ae

Tt sac aatel
f ve
ai ‘Oy 2
j & %,
' me

mo.
ed
IN:

ue

SMITHSONIAN INSTITUTION LIBRA

WNW LI ANNO

9088 01301 3636