Ok (
be COS4F
; Contributions from the \o Y
GRAY nS
HERBARIUM

1978 No. 209
SEP 296 1978

zerald V. Dahling

Contributions from the
GRAY
HERBARIUM

1978
Gerald V. Dahling SYSTEMATICS AND EVOLUTION OF GARRYA
GB JA 900%

EDITED BY Reed C. Rollins
Kathryn Roby

PUBLISHED BY
THE GRAY HERBARIUM OF HARVARD UNIVERSITY

I eee ea

SYSTEMATICS AND EVOLUTION OF GARRYA
GERALD V. DAHLING!

History

For many centuries the medicinal properties, charm, and aesthetic
beauty of ‘Esawana’ were known to the Indians of southeastern Mexico.
However, the plant was not discovered by white explorers until 1826
when David Douglas, an early explorer-naturalist, found Garrya in
the Pacific Northwest. Seeds and cuttings were sent to the London
Horticultural Society for propagation and within a few years it was
flourishing in the Botanical Garden. Lindley (1834) used these plants,
along with the Douglas collection, to first describe the order Garrya-
ceae, named in honor of Nicholas Garry Esq., the first secretary of
the Hudson Bay Company. The single species represented was Garrya
elliptica and was designated the type of the genus. Thus the rich
and varied taxonomic history of Garrya began (Table 1). The group
was first allied with several families of the Amentiflorae because
of the highly reduced nature of the flowers and the bracteate inflores-
cences. Included were the Piperaceae, Cupuliferae, Chloranthaceae,
Urticaceae, Euphorbiaceae, and the Mensloviaceae. Endlicher (1836-
1841), in his Genera Plantarum, also classified Garrya in a similar
fashion. Considering Garryaceae and other amentiferous families to
be primitive, he placed them directly after the gymnosperms. Garrya,
in his treatment, was specifically related to the cupuliferous and
chloranthaceous members of the class Juliflorae.

A much broader concept of Garryaceae became possible after the
Mexican expedition of Theodor Hartweg in 1836. Several new species
were discovered and subsequently described by Bentham in Plantae
Hartwegiana.

Endlicher (1847) constructed a segregate genus Fadyenia for these
hew species but retained the original genus Garrya for G. elliptica.
His familial division was based on the differences in the inflorescences
and floral bracts.

Lindley revised his original classification in 1847, and established
the order Garryales for the Garryaceae and Helwingeaceae. He also
Suggested that a progression from the Gnetales to the Amentales,
Urticales, and Euphorbiales was so obvious that no explanation was
required. He believed that the Garryales and Quernales represented
neighboring groups which were epigenous forms of the Euphorbiales
alliance. This Garryales-Euphorbiales relationship was strengthened
by the putative approach of Garryaceae to Helwingiaceae, another
Euphorbiales-related group included in the Garryales. To make the

"Present address: Macalester College, Department of Biology, St. Paul, Minn. 55105, U.S.A.

3

4 GERALD V. DAHLING

TABLE 1. TAXONOMIC HISTORY

Lindley (1834, 1847) Founder of the genus, family, and order; related Garrya
to Euphorbiales, Urticales, Quernales, Menisperma-
les, and Gnetales.

Endlicher (1837, 1847) Placed Garryaceae within the Apetalae; related the ©
group to Cupuliferae and Chloranthaceae; construct-
ed Fadyenia as a segregate genus.

Bentham & Hooker (1867) Lowered the taxonomic rank and placed Garrya within —
the Cornaceae.
DeCandolle (1869) Placed Garryaceae near Salicaceae within the Amenti-

ferae; sections Eugarrya and Fadyenia with nine and
one species, respectively, were recognized.
Related G toC ted an Aucuba- —

Baillon (1879)

oS

Garrya affinit

Harms (1898) Established the subfamily Garryoideae within the —
Cornaceae.

Wangerin (1906) Rejected the Baillon classification and placed Garrya
among the Amentiferae.

Wangerin (1910) Monographed and considered Garrya an aberrant

member of the Cornaceae; sixteen species were listed.

Faure (1924) Anatomically compared Garryaceae with the Cornaceae
and : Cee | : ae Ae on 1 a rs

J allydalts é
to Salicales. '
Engler & Gilg (1924) Garryales was related to the Amentiferae and placed ©
near Salicales.
Maintained the order Garryales and provisionally posi- —
tioned it near the Cornaceae. :
Hallock (1930) Based on morphology, Garryaceae was placed in the
Umbelliflorae next to the Cornaceae. 4
Anatomically related Garryaceae with the Cornaceae.
Suggested an affinity between Garrya and the Corna-
ceous genera Aucuba and Griselina. :
Family rank maintained; Garryaceae placed next to the |
Cornaceae; subgenera Garrya and Fadyenia with six ©
and eight species, respectively, are recognized. j

Bacigalupi (1924)

Moseley & Beeks (1955)
Eyde (1964)

Present revision (1974)

trend complete, in Lindley’s system Garrya was also related to the
Menispermales which followed the Garryales. Thus Garrya represent-
ed a single step in the ‘obvious’ progression from Gymnosperms to
Dicots. ;
Lindley’s position was attacked by Bentham and Hooker (1867),
who lowered the taxonomic rank, dropped the segregate genus Fa-
dyenia, and placed Garrya in the Cornaceae. Within two years,
however, this was disputed by Alphonse DeCandolle (1869) and Garrya
was again given a family rank and placed among the amentiferous
groups. On the basis of stylar characters, Garrya was divided into

sections Eugarrya and F adyenia. This was an extremely artificial
division of the genus and res

two parts, but without formal recognition. Although the names are
similar, sections Eugarrya and F adyenia of DeCandolle do not corre-_

BN

SYSTEMATICS OF GARRYA 5

spond and are less satisfactory than the genera Garrya and Fadyenia
of Endlicher (1847). The treatment of DeCandolle is completely
artificial, without merit, and justifiably has never been followed.

Baillon (1874, 1879) studied living Garrya flowers from the Botanical
Garden of Paris and concluded that an alliance with the Cornaceae
was indicated. His opinion was based on morphological study and
on the successful grafting of Garrya and Aucuba japonica of the
Cornaceae. This relationship was maintained by Harms (1898) who
constructed the subfamily Garryoideae within the Cornaceae.

Wangerin (1906), in an involved discussion, rejected Baillon’s
classification and included Garrya in the Amentaceae. This was rather
short-lived, for in his monograph a few years later he reversed himself
and placed it in the Cornaceae, but as an extremely aberrant form
(Das Pflanzenreich, 1910). This conclusion failed to win support from
Faure (1924) in his study of the Garryaceae and Cornaceae. After
a detailed morphological investigation, a relationship with the Corna-
ceae was rejected in favor of an alliance with the Amentiferae. The
order Garryales was maintained and placed next to the Salicales because
both were interpreted as having superior ovaries. A similar conclusion
was drawn by Engler and Gilg (1924), who also considered Garrya
to represent a separate order near the Salicales. This was disputed
in the same year, however, when Bacigalupi (1924) provisionally placed
the order near the Cornaceae.

Modern studies and investigations have repeatedly demonstrated
a Garryaceae-Cornaceae relationship. Hallock (1930) studied the
reproductive structures of Garrya and clarified the morphology of
the fruits and seeds. Careful developmental analysis, stressing the
cytological events associated with the pre- and _post-fertilization
periods, also supported the separation of Garrya from the amentiferous
groups. In addition, many anatomical features relating Garrya to the
Cornaceae were discovered. As a result, Hallock placed Garryaceae
in the Umbelliflorae next to the Cornaceae.

The most thorough phylogenetic investigation of Garrya was com-
pleted by Moseley and Beeks (1955). Eleven species were studied
and compared to 23 families which were at one time considered to
be related to Garrya. Concentrating on wood anatomy, they demon-
strated profound differences between Garryaceae and the amentiferous
families. Significant differences in wood anatomy were found in almost
all the families except the Cornaceae. An analysis of other charac-
teristics, including the phloem, and the nodal, floral, and pollen
anatomy, also demonstrated trends which are common in the Cornaceae
but are lacking in the other families compared. Developmental anat-
omy, cytology, paleobotany, and phytochemistry also were considered,
allowing for a broadly based phylogenetic interpretation. The main

6 GERALD V. DAHLING

conclusion drawn from this was that the cumulative comparative
evidence favored a relationship with the Cornaceae. It was also shown
that, where similar structures are lacking, sequences of specialization
from the Cornaceae to the Garryaceae are evident. Moreover, the
Garryaceae was hypothesized to have differentiated by reduction from
pro-umbelliflorean ancestral members which were cornaceous in
nature.

The relationship between Garrya and the Cornaceae was supported
and expanded by a critical examination of the ovary by Eyde (1964).
Previously the ovary was variously interpreted, accounting for the
vacillation in the placement of Garryaceae between the Amentiferae
and the Cornaceae. Eyde suggested that a floristic relationship exists
between Garrya and the Cornaceae, especially Aucuba and Griselina.
While an alliance with Aucuba was previously recognized (Baillon,
1879), the suggestion of a Garrya-Griselina affinity was novel. Consid-
ering Griselina closest to the primitive condition, he further suggested
the possible derivation of Garrya, Aucuba, and Griselina from common
ancestral stock via reduction.

The present treatment maintains the rank of family for the Garryaceae
and places it next to the Cornaceae. The subgenera Garrya and
Fadyenia, with six and eight species respectively, are also recognized.

DISTRIBUTION

The genus Garrya is confined to the New World and ranges from
the Pacific Northwest of North America to the volcanic peaks of |
Panama. As shown in Figure 1, subgenus Garrya generally occurs
in northern areas ranging from Washington to Baja California and
eastward to Sonora in Mexico, Arizona, Utah, and Nevada. A major
exception to this otherwise close-knit and geographically related group
is G. corvorum, an isolated representative found in central Guatemala.
Subgenus Fadyenia is comprised of closely related species and gener-
ally occurs in more southerly areas. The species of this group range

from Arizona, New Mexico, and Texas to western Panama and the
Greater Antilles. :

TAXONOMIC LIMITS

morphology to warrant the recognition OF

it as a distinct family. The bracteate and amentiferous nature of the

SYSTEMATICS OF GARRYA

ome |

“Tih

Mart!

Fics, 1-2. FW tetethiacat fek 1
ributi Ya

2, Dist

8 GERALD V. DAHLING

inflorescences, in addition to the much reduced flowers, serve to
remove Garrya from the Cornaceae, supporting the recognition of
it as a separate family. Moreover, while the Garryaceae and Cornaceae
share certain glycosides and flavonoids, their phytochemical relation-
ship, as outlined in a later section, is not close, thus helping to justify
the retention of family status. The species form two infrageneric groups
which are here recognized and designated as subgenus Garrya and
subgenus Fadyenia. These subgenera can be readily characterized
by their morphology (Table 2).

The taxa of subgenus Garrya are characterized by flowers borne
in groups of three and with unbranched, condensed, and compact
female inflorescences. The floral bracts are fused laterally and at the
base forming cup-like structures which house and protect the floral
clusters. The successive cup-like bracts are not leafy but overlap and
appear imbricated. The ovaries are nude or occasionally bear small
paired perianth remnants which alternate with, and arise near, the
base of the paired styles. The smooth, terete, and elongated styles
often equal, and frequently exceed, the length of the ovary in young
material. Members of subgenus Garrya are usually shrubs, but small
trees have also been observed.

Taxa of subgenus Fadyenia are characterized by bearing solitary
flowers and the female inflorescences are branched, lax, and loosely
spreading. Floral bracts are free, or fused only at the base, and appeal.
expanded and distinctly leaf-like. The successive floral bracts are free
and not overlapping or imbricated. Ovaries appear nude or sometimes
bear minute bracteate appendages which are variably adnate and arise
opposite and near the base of the paired styles. Being rough, thick, ;
and short, the styles appear less delicate than those of subgenus Garry4. i

TABLE 2. COMPARIS

N OF Ls s>E NERA

Garrya Fadyenia

cs a
Inflorescences not branched Inflorescences branched, at least at the
Female inflorescen Be Sy :
ces condensed, com- Female inflorescences lax and loosely _

spreading :
Successive floral bracts separate, free, and
: not over i
Paired floral bracts fused at base and _ Paired Sols al free or fused only at
ase

Female floral bracts expanded and dis-
tinctly leaf-lik
—— in groups of three eS waiaeg dee.
ee bi persing pened perianth Ovary nude or bearing a minute pair of
ase of styles adnate bracts
phan bead ace on from 2-20 feet tall Large trees and shrubs from 3-40 feet tall :
rr y to the far western United Mostly throughout Mexic Central

0,
America, and the Greater Antilles

SYSTEMATICS OF GARRYA 9

Large trees are common in subgenus F adyenia although some species
are distinctly shrubby. While a few species are found in the south-
western United States, most of them are geographically located in
Mexico, Central America and the Greater Antilles.

MORPHOLOGY

Habit. The plants of Garrya have numerous crown sprouts, in size
they range from small shrubs to large trees, and appear distinctly
bushy and clumped. In arboreal members the crown sprouts ring
the trunk and are clearly secondary to the main stem axis, contrasting
with the more shrubby members where the crown sprouts approach
or sometimes exceed the main stem, giving a bushy appearance.

Within subgenus Garrya, the growth habit is variable even within
given species. Garrya elliptica is a low shrub when growing near
the coast but it often becomes arboreal inland, attaining heights of
over 20 feet. Garrya veatchii is predominantly shrubby but occasionally
small gnarled tree-like forms are observed. The latter has also been
seen in G. fremontii, which sometimes also takes on the bushy character
of G. buxifolia. The habit of G. flavescens is influenced by the
availability of water and ranges from bushes to 20 foot trees. The
Guatemalan G. corvorum is almost always arboreal.

Most species of subgenus Fadyenia are arboreal but Garrya wrightii,
G. grisea, G. ovata ssp. ovata, and G. ovata ssp. goldmanii are
exceptions. The latter is a small 2-5 ft. shrub and shows the most
reduced habit of the genus. The other taxa mentioned are clumped
and bushy, ranging from 6 to 15 feet. Species with an arboreal habit
include G. laurifolia, G. longifolia, G. salicifolia, G. glaberrima, G.
fadyenii, G. ovata ssp. mexicana, and G. ovata ssp. lindheimeri. The
latter is rather shrubby in the northern part of its range and contrasts
with G. laurifolia which often exceeds 35 feet in height. In general,
an arboreal habit is most prevalent in subgenus Fadyenia while a
shrubby habit is most frequent in subgenus Garrya.

Root. Root morphology in Garrya is quite uniform with little
significant variation (Faure, 1924). Rows of cork along with a less-
developed phelloderm, a narrow cortex, and a wide pericycle are present
in all species. Circular vessel elements are prominent in the xylem
as are sclerenchymatous rays. Calcium oxalate crystals and secretory
cells are also common in the root parenchyma.

Although Garrya flavescens shows minor variation in the pericycle
width, the general systematic and evolutionary importance of compar-
ative root morphology appears to be very limited. This is especially
true since the narrow pericycle of G. fla isan isolat
that fails to correlate with other characters.

10 GERALD V. DAHLING

Stem. While the shape, color and pubescence of Garrya stems are
important, the internal morphology yields little systematic information.
Faure’s comparative anatomical study (1924) demonstrated a general
uniformity in stem structure although some variation was noted. The
species examined contained cork, an inner and outer cortex, and a
continuous or incomplete cylinder of fibres which borders the wood.
As in the roots, calcium oxalate crystals are abundant in the parenchy-

Growth rings in the secondary wood are usually distinct but rather
narrow, indicating slow growth. This is not surprising since Garrya
inhabits areas which are somewhat arid and generally unfavorable
for rapid plant growth.

Wood and phloem anatomy has been studied in detail (Moseley
& Beeks (1955); Metcalfe & Chalk, 1950) and is summarized here.

mperforate elements, such as tracheids and fiber tracheids, are
common in Garrya wood and range from 210 to 1470 wm. in length.
The average is 762 ~m. which is moderately short as defined by
the Committee on the Standardization of Terms of Cell Size (1937).
Also, the wood is mostly ring porous and shows a solitary distribution
of vessels although small clusters and chains are sometimes observed.

The average vessel diameter for Garrya is 34.2 m., or very small
as defined by Chalk (1938). Their perforation plates are scalariform
showing from 1 to 13 bars and the end walls are quite oblique, ranging
from 50-70° off the vertical. Pitting is opposite to alternate or transi-
tional in most species. The average vessel length in Garrya is 562
1m. or medium-sized, as defined by the Committee on the Standard-
ization of Terms of Cell Size (1937).

Both uniseriate and multiseriate rays are present in Garrya and
while compound rays are not found, a definite trend toward aggregation
of rays is evident except in G. elliptica and G. veatchii. This may
be systematically significant since both G. elliptica and G. veatchii
are closely related. In contrast, no significance should be attached
to ray and wood parenchyma type since both are highly variable
even within taxa. :

Phloem in Garrya is divided into wedge-shaped masses and is
composed of sieve tubes, companion cells, phloem rays, and phloem
fibres. The sieve tubes are arranged in radial rows and range from
120 to 400 pm. in length and have extremely oblique walis with
scalariform compound sieve plates. In most species the phloem fibres
are arranged in bands but in G. laurifolia and G. ovata ssp. lindheimert
fibres are solitary or sometimes absent. This may be significant

ecause of their close systematic relationship. Large druses are some-
times found associated with the phloem rays, but are not common
in the genus as a whole. : :

SYSTEMATICS OF GARRYA i fa

Vascular anatomy traditionally has been useful in phylogenetic
studies, but, at the level of species its systematic importance is doubtful
or at least diminished. This is also the case in Garrya, where exterior
stem characters are more important and diagnostic than the internal
morphology. It should be emphasized that this is only the case when
questions of phylogeny are not considered.

Leaf. The leaves in Garrya are thick, persistent, and uniformly
coriaceous, ranging from oblong, oval, ovate, to obovate in shape.
On the exterior they appear opposite but, as shown by Moseley and
Beeks (1955), the trilacunar vascular traces for one leaf become distinct
from the stele before the traces for the other. Therefore, the arrangement
is actually sub-opposite even though the leaves appear to occupy
opposite positions at the nodes. The bases of the petioles are connate
and adnate to the stem, which is characteristic of the genus. The
angle of fusion with the stem is distinct and shows considerable
variation among taxa (Plate 1). Leaf surfaces, variably covered with
unicellular trichomes of differing shapes and arrangements, appear
reticulate-rugose, especially within subgenus Fadyenia.

The internal anatomy of Garrya leaves was reported by Paliwal
and Kakkar (1970) and is summarized here followed by my own
observations. The upper epidermis, which is composed of a single
layer of rectangular cells, is covered by a well-developed cuticle. The
lower epidermal surface is double, with the inner layer often becoming
sclerotic. A prominent hypodermis is found beneath the upper epider-
mis and is composed of large circular cells which contrast with the
2-3 layers of palisade parenchyma which are below. While usually
retaining their nuclei and cytoplasm, they sometimes sclerify as do
the isodiametric parenchyma cells in the spongy layer. In addition
to collenchyma, brachysclereids are also found near the midrib and
major veins. Sclereids are therefore produced in almost all areas of
the leaf, accounting for their frequency and wide distribution.

The venation pattern of Garrya is distinctive and easily observed
in cleared specimens. In the present study, leaf material was cleared
in the usual way (Foster, 1949), mounted, and examined. Three vascular
traces enter the leaf, with the outer two branching to form the midrib
at the base of the lamina. A series of lateral or primary veins are
Produced and are directed toward, but never reach, the leaf margins.
Secondary veins from the laterals connect neighboring veins, forming
smooth loops which are quite prominent. Anastomoses of the secon-
dari s prod ] hich lose terminal vein endings.
As reported earlier by Paliwal and Kakkar (1970) filiform, stellate-
Polymorphic, and brachysclereids are common in the areoles, veins,
and terminal vein endings. Filiform sclereids have narrow lumens
and thick striated walls. The stellate-polymorphic sclereids are found

GERALD V. DAHLING

1. Node, petiole, and leaf differences in Gar ya

es wrightii, Dahling 543; B, fremontit,
buxifolia, Dahling 1081: D. G. laurifo

x Dike eee? hling
laurifolia, Dahling 329; E, G. fremontii, Dahlins

PLAT
Dahling 1007: C. G
1007

specie
lia ssp.

near the midrib, vein endings

and leaf margins but are also widely
scattered throughout the

mesophyll. They are vertically oriented an
have thick striated walls with a nucleus and cytoplasm. Brachysclere-
ids, isodiametric cells with thick striated walls and small lumens;
are chiefly confined to the midrib and lateral vein areas.

SYSTEMATICS OF GARRYA 3

. 6,

PLATE 2. EF pidermal and trichome differences in Garrya species: A, G. flavescens ssp pallida, — —
'50x; B, G. ovata ssp. lindheimeri, Dahling 113, 75x; C, G. grisea, Wiggins and emaree 4903 si ae
: wright Dahling 1208, 1350x; E, G. ove a ssp. lindheimeri, Dahling 113, 2000> G. elliptica, Dahling 369,

E pidermal characters in Garrya are important sy stematic indicators
and are quite significant (Plate 2A, B, C). Stomata are uniformly
Paracytic and confined to the lower leaf surface in all taxa. Their
size and shape varies along subgeneric lines (Table 3) with the larger
stomata being found in subgenus Garrya. Garrya elliptica, G. fremontii,
and G. flavescens show the largest and most pronounce -d stomata

14 GERALD V. DAHLING

TABLE 3. STOMATAL APPARATUS

Subsidiary cells undulate Subsidiary cells not undulate or wavy
G. elliptica (least wavy)* G. corvorum G. grisea
G. veatchii* G. ovata ssp. ovata G. glaberrima
G. fremontii® G. ovata ssp. lindheimeri G. fadyenii
G. flavescens ssp. flavescens® G. ovata ssp. goldmanii G. longifolia
G. flavescens ssp. pallida® G. ovata ssp. mexicana G. laurifolia
G. wrightii

* Subgenus Garrya

while those of G. corvorum and G. veatchii are somewhat smaller
but exceed those of subgenus Fadyenia. The stomata of subgenus
Fadyenia are smooth, not undulate, and rather small. They are clearly
wavy in G. elliptica, but less undulate than others in the subgenus.
Subsidiary cells of G. corvorum are smooth and appear very similar
to those of subgenus Fadyenia, indicating a possible distant relation-
ship between G. corvorum and the other members of subgenus Garrya.
Species of subgenus Faydenia have subsidiary cells which are smooth,
uniformly not undulate, and not Wavy.

Epidermal surfaces range from papillate, as in most Garrya species,
to muricate in G. fremontii. The papillae are most dense in G. corvorum,
G. ovata ssp. ovata, G. ovata ssp. mexicana, G. longifolia, and G.
laurifolia, While not as dense in other Garrya taxa, they are usually
prominent and highly developed. Papillae in G. fremontii are smaller
and less numerous than those of other taxa. Variation in the epidermal
surfaces appears greatest in subgenus Garrya, since both maximum
and minimum papillae development are represented in G. corvorum
and G. fremontii respectively. Size variation in papillae crosses sub-
generic lines. The largest are found in G. flavescens, G. corvorum, —
G. wrightii, G. grisea, G. salicifolia, and GC. longifolia. The remaining
Garrya species show prominent but smaller papillae with the smallest
being found in G. fremontii.

Leaf sizes are quite variable in Garrya (Tables 4 and 5). Petiole
length (PLX), blade length (BLX), total length (TLX), and width (WX) |
were measured and expressed as averages for natural populations. —
The figures were also expressed as proportions for calculating the
petiole length:blade length (PL/BL), petiole length:total length
te fein width:blade length (W/BL), and width:total length (W/TL)

Garrya corvorum has uniformly small leaves, shows little intraspeci-
fic leaf variation and is easily distinguished by the leaf length and
width from the other members of subgenus Garrya (Table 4). The
small amount of intraspecific leaf variation within G. corvorum is

SYSTEMATICS OF GARRYA 15

TABLE 4. AVERAGES OF LEAF DIMENSION (cm.) AND PROPORTION FOR SUBGENUS GARRYA
Species PA" “BUX WK YEA” Pee P/L W/BL W/TL
G. elliptica 0.632 5.239 3.034 5.871 0.120 0.107 0.579 0.517
G. veatchii 0.530 4.242 1.959 4.773 0.125 0.111 0.461 0.410
G. buxifolia 0.603 3.568 1.791 4.172 0.169 0.144 0.502 0.429
G. fremontii 1.061 4.943 2.330 6.004 0.214 0.176 0.471 0.388
G. flavescens
ssp. flavescens 0.891 4.890 2.491 5.781 0.182 0.154 0.509 0.430
G. flav
ssp. pallida 0.916 4.494 3.324 5.410 0.203 0.169 0.517 0.429
G. flavescens
ssp. congdonii 0.538 5.259 1.940 5.787 0.102 0.092 0.368 0.335
G. corvorum 0.362 3.070 1.266 3.433 0.118 0.148 0.412 0.368

TABLE 5. AVERAGES OF LEAF DIMENSION (cm.,) AND PROPORTION FOR SUBGENUS FADYENIA

; Species PX. __BLX . WX TLX P/BL P/TL W/BL W/TL
G. ovata ssp.

Ovata 0.825 4.441 2.458 5.266 0.185 0.156 0.553 0.466
G. ovata ssp

goldmanii 0.630 3.485 1.768 4.115 0.180 0.153 0.507 0.429
G. ov :

lindheimeri 13083 5.250 2.616 6.334 0.206 0.171 0.498 0.413
G. ovata ssp

mexicana 1466 8.328 3.717 9.795 0.176 0.149 0.446 0.379
G. wrightii 0.563 3.679 1.748 4.243 0.153 0.132 0.475 0.411
G. grisea 0.600 3.766 1.550 4.366 0.159 0.137 0.411 0.354
G. salicifolia 0.739 5.609 1.763 6.348 0.131 0.116 0.314 0.277
G. glaberrima 1.003 6.718 2.937 7.722 0.149 0.129 0.422 0.367
G. fadyenii 0.812. 6.601. 2.057... 6:893)..0.135 -0:119:...0:342 . 0,301
G. longifolia 1.210. 10567. 4:09! 41.777 0.114 0102 0387 -.6.347
G. laurifolia

ssp. laurifolia 1.218 8.698 3.573 9.916 0.140 0.122 0.410 0.360
G. laurifolia

SSP. macrophylla «1.339 +=: 9.673 5.122 11.012 0.138 0.121 0.529 0.465
G. laurifolia

SSP. racemosa 1.925 13.684 4.312 15.609 0.140 0.123 0.315 0.276
G. laurifolia

SSP. quichensis 1.202, 8.502 3.030 0704. O44), 6123. 0.357 .0,318

Not surprising since it grows in a restricted area where presumably
the temperature and the precipitation are relatively constant from one
Site to the next. The other larger-leaved members of subgenus Garrya
do, however, show considerable inter- and intraspecific variation in
leaf size, Although leaves of G. elliptica, G. fremontii, and G. flavescens
are larger than those of G. veatchii and G. buxifolia, the leaf size
within a species frequently varies extensively, depending on moisture,
and the edaphic and general climatic conditions for growth. Intra-
SPecific variation in the petiole length was also noted. On the average,
the longest petioles are found in G. fremontii and G. flavescens.

16 GERALD V. DAHLING

Variation in the leaf size ratios (Table 4) for members of subgenus
Garrya is not extensive but there are some differences among the
taxa. The smallest petiole length:blade length ratios are found in G.
elliptica, G. veatchii, G. corvorum, and G. flavescens ssp. congdonii.
Their ratios are quite similar, ranging from .102 to .125. Garrya
fremontii, G. buxifolia, G. flavescens ssp. flavescens and G. flavescens
ssp. pallida have larger and less uniform petiole:blade length ratios
which range from .169 to .214. In addition, these taxa also have larger
petiole length:total length ratios than do G. elliptica, G. veatchii,
and G. flavescens ssp. congdonii. Within subgenus Garrya the
width:blade length ratios are quite similar with the exception of G.
flavescens ssp. congdonii, G. corvorum, and G. elliptica. The two
former taxa show small ratios of .368 and .412 respectively. The latter
shows a ratio of .579 which is the largest width:blade length ratio
within Garrya. The width:total length ratios generally parallel the
width:blade length ratios and will not be discussed further.

The largest leaves are found within subgenus F adyenia (Table 5).
The average petiole lengths of Garrya laurifolia, G. longifolia, G. —
glaberrima, G. ovata ssp. mexicana, and G. ovata ssp. lindeheimeri
contrast with the shorter lengths of the remaining taxa. Also, the average
leaf blade lengths and widths of the large-leaved G. laurifolia, G.
longifolia, and G. ovata Ssp. mexicana stand out from the smaller
leaved species. The latter include G. wrightii, G. grisea, and G. ovata
ssp. goldmanii with small leaves; and G. ovata ssp. ovata, G. ovata
ssp. lindheimeri, G. salicifolia, G. glaberrima, and G. fadyenii with
leaves of intermediate size. |

phylla; the smallest are found in G. salicifolia and G. laurifolia ssp.
racemosa. Thus very large and very small ratios are found within
the same species. Because interspecific leaf variation is nearly continu- |
ous, data of this sort must be cautiously applied if valid systematic
conclusions are to be drawn.

Trichomes. The diagnostic value of trichomes in identifying Garry4
Species was recognized early. Taxonomic keys (Eastwood, 1903;
relied heavily on the trichome
From my own observations i?

structure yield information of consider

iets able systematic and evolutionary
significance. a

SYSTEMATICS OF GARRYA 17

Trichomes of Garrya are unicellular and are found in all species
except G. glaberrima which, as the name implies, completely lacks
an indumentum. The trichomes are not confined to the leaves, but
commonly occur on the stems, flowers, and fruits as well. However,
the SEM investigations concentrated on the leaf trichomes because
of their accessibility and ease of preparation.

Method: Mature leaves appearing normal in size, shape, and color
were selected and labeled. Small square leaf portions were excised
with a sharp razor blade and then transferred to aluminum pedestals
coated with silver paint. The leaves, with the adaxial surfaces down,
were placed in the paint and allowed to dry in a flat position. After
20 minutes, additional silver paint was applied to the specimen corners,
insuring a firm attachment to the pedestal. Next, the specimens were
evacuated to 10-4 — 10-5 torr. in a Varian Vacuum Evaporator VE
10, coated with carbon and then gold-palladium. They were then
placed in an AMR Model 900 scanning electron microscope and
examined. The results were recorded photographically.

Discussion: The trichomes of Garrya are delicate and quite sus-
ceptible to electron beam damage (Plate 2D). They are roughened
by ridges and furrows which are oriented in a counterclockwise
direction, In some species, the ridges bear small protuberances which
stand out in high relief (Plate 2E). In species lacking the protuberances,
the smoother trichomes bear only ridges and furrows. The rough-ridged
trichomes appear on all species of subgenus Garrya except G. corvorum
(Figure 2). Garrya corvorum has smooth-ridged trichomes. which are
characteristic of subgenus Fadyenia. This extremely interesting finding
indicates a possible early evolutionary divergence from the ancestral
stock of subgenus Garrya. The peculiar isolated Guatemalan distribu-
tion of G. corvorum also supports this view. With other species confined
to the far western United States, G. corvorum is spatially isolated,
making a close relationship with the western members difficult to
explain. Flavonoid investigations show only loose affinities with the
other taxa and G. corvorum appears chemically intermediate between
the two subgroups, also suggesting an early divergence from ancestral
types common to both subgenera.

Only Garrya ovata and G. wrightii of subgenus Fadyenia show
the Tough-ridged trichomes. Since both taxa are northern and show
moderate Specialization within the subgenus, the development of
trichomes of a different sort is not unexpected.

Within subgenus Garrya, trichome variation in both arrangement
and detail is striking. Trichomes of Garrya elliptica (Plate 2F) densely
Cover the abaxial leaf surfaces and appear short, curly, and intertwined.
The counterclockwise distribution of ridges and furrows is prominent
aS are the small protuberances which appear to be confined to the
tops of the ridges (Plate 3A). Garrya veatchii (Plate 3B) has trichomes

GERALD V. DAHLING

PLATE 3 ote Sx: differences in Gar rya species: A, G Por a, Dah ng 1369, 150x; B, C,G eugene: Da
5 ~3370%; D, G - uxifolia, Da iling 1512 ; E, G. frer
,»G ul ‘ 7:
B os soni f Yah =. fi emontii, Dahling 1078, 75x; F, . fla ve

which are extremely curly and interwoven, forming a dense mat-like
covering on the underside of the leaf. They are shorter and much
more dense and curled than those of G. elliptica. In addition to the
characteristic ridges and furrows, the trichome surface shows a series
of rather large vertical valleys (Plate 3C). Thus, two systems of ridges

SYSTEMATICS OF GARRYA 19

and furrows running in opposite directions are present on the tri-
chomes. Trichomes of G. buxifolia (Plate 3D) are wavy to straight
and upwardly appressed forming a dense silky covering of the leaf
abaxial side; the surfaces are covered with the customary ridges and
furrows. Again protuberances are numerous and are located on the
ridges. Garrya fremontii (Plate 3E) normally shows few trichomes
on its leaves, but those present are striking. In addition to the ridges
and furrows, the entire trichome is twisted in a counterclockwise
direction. Thus in this taxon, the trichomes have taken on a compound
counterclockwise orientation. Numerous protuberances are present,
chiefly occurring on the ridges but occasionally protuberances in the
furrows have been observed. Trichomes of G. flavescens ssp. flavescens
are more or less straight and upwardly appressed to the lower leaf
surface (Plate 3F). They form a dense and silky white covering that
appears soft and fine. Upon closer examination, several trichomes
reveal a compound counterclockwise orientation similar to that of
G. fremontii. This compound orientation is especially pronounced
in G. flavescens ssp. pallida (Plate 4A). These trichomes appear
upwardly appressed but are wavy, coarse, and only sparsely cover
the underside of the leaf. Ridges and furrows (Plate 4B) are oriented
in a counterclockwise direction with protuberances confined chiefly
to the ridges. They appear similar in size and shape to those of G.
flavescens ssp. flavescens (Plate 4C). The size and design of the
stomatal apparatus of G. fremontii (Plate 3E), and both subspecies
of G. flavescens, are similar and distinctive, suggesting a close
relationship. Garrya corvorum (Plate 4D) has trichomes which are
long, straight, and generally upwardly appressed. They are white and
Soft, giving the leaf undersides a light silky appearance, and they
glisten in the sun. Lacking protuberances (Plate 4E), the trichomes
are smoother and more uniform than those of other members of the
subgenus. The usual ridges and furrows are present (Plate 4F) but
the protuberances so characteristic of subgenus Garrya are not.
Members of subgenus Fadyenia show a high degree of trichome
variation. All subspecies of Garrya ovata have a dense trichome cover
on the leaf undersurface. Trichomes of G. ovata ssp. ovata are
moderately appressed to the surface and appear somewhat tangled
and curly, Garry ovata ssp. goldmanii (Plate 5A) has short, curly,
crinkly, grossly entangled leaf trichomes, forming a dense mat on
the leaf undersurface and, in younger material, on the adaxial surface
aS well. Ridges and furrows, bearing extremely large protuberances,
‘re oriented in a counterclockwise direction (Plate 5B). The protuber-
ances are much larger than those of other species and stand out in
high relief, Leaves of G. ovata ssp. lindheimeri appear woolly, due
to their short, highly coiled, crinkly, curly trichomes (Plate 5C). Also,

GERALD V. DAHLING

a system of shallow ridges and furrows (Plate 5D) bearing large
protuberances is present. The protuberances are numerous and cove!
the trichome surface almost to the tip (Plate 2E). Trichomes of G
ovata ssp. mexicana show similar ridges, furrows, and protuberances
but are less curly (Plate 5E). T hey also appear appressed and entangled
but less numerous. Trichomes of G. wrightii are usually sparse

SYSTEMATICS OF GARRYA
ne
° e, +>. ~ 4 i

fies S y's
.@) ve ? ‘tk ‘é 7

ie !

PLAT

P E5. Trichome differences in Garrya species: A, B, G. ovata ssp. goldmanii, Dahling 1414, A
2, D :

25x:

, A-75X, B-1k
4p i Bn ahling 126,
: » G. ovata ssp. lindheimeri, Dahling 333, C-25%, D-1350X; E, G. ovata ssp. mexicana, Dahling |
FG. wri 150X

ightii, Dahling 1208, 150

absent, but those present are short, straight, and upwardly appressed
(Plate SF). The system of ridges and furrows, with numerous protuber-
ances, is gently oriented in a counterclockwise direction (Plate 6A).
The protuberances are confined to the ridges and are smaller than
those of G. ovata. Leaves of G. grisea are densely covered with short,
Straight, and upwardly appressed trichomes on both the abaxial and

ye GERALD V. DAHLING

es
“ec
€

Eo nero 2 ge =
e¢ . ° :
ett Se.

% SE ns
a >

ey tag
<

PLATE 6. Trichome differences i é
a wi etek species: A, G. wrightii, Dahling, 1208, 1350x; B, C, G. grisea, Wiggins
‘ set Baer -75X, C-1350x; D, G. salicifoli 7 ‘ a ASICS SiS te rie Sea A
Rehder 2-10-1903. F-75x. = fresinons G. salicifolia, Nelson and Goldman 7458, 150x; E, F, G. fadyent

Sea surfaces (Plate 6B). The ridges and furrows are coarse, 45
ae Arby sein Bleed eh protuberances are essentially lacking (Plate
of ewelliz ‘ =) re rceptible and are represented by minor areas
shades et on the tops of ridges. Garrya salicifolia is the only membe!
bik th SiGe ee The trichomes are extremely sparse

e structural twisting in those present is evident (Plate 6D):

SYSTEMATICS OF GARRYA pat’

x, B-350x, C-1350x;

+4 y Pe? 2 919 75
PLATE 7. Trichome differences in Garrya species: A, B, C, G. longifolia, Dabling 2121, odes
, D-75x, E

DEF,G sess
EF, G. laurifolia, Dahling 3838, D-75x, E-350x, F-1350x.

Ridges and furrows line the trichome surface but there are no protuber-
ances. Garrya fadyenii shows abaxial leaf trichomes which are dense,
Curly, tangled, and more or less erect (Plate 6E). They are cogcieigepeee
long and form an intertwining mat on younger leaves. Ridges an

furrows line the trichome surfaces and appear widely spaced and
rather coarse (Plate 6F). No protuberances have been observed. The

24 GERALD V. DAHLING

leaf trichomes of G. longifolia are short, curly, and more or less
appressed (Plate 7A). Some have a compound counterclockwise orien-
tation (Plate 7B), but the structural twisting appears gentle and not
as severe as that seen in G. salicifolia. The ridges and furrows are
coarse and completely lack protuberances (Plate 7C). The leaf under-
sides of G. laurifolia are covered by moderately long, curly, tangled
and more or less erect trichomes (Plate 7D). They form a woolly
mat in G. laurifolia ssp. laurifolia and G. laurifolia ssp. macrophylla
but the trichomes are rapidly lost in G. laurifolia ssp. racemosa and

G. laurifolia ssp. quichensis as the leaves mature. The trichomes appear |

relatively smooth even though ridges and furrows are present because —
of the absence of protuberances (Plate 7E). They are oriented in the ©

usual fashion and appear fine and more delicate than those of G.
longifolia (Plate 5F).
Inflorescence. The morphology of the inflorescence (Reeve, 1943)

varies according to subgeneric lines and is an important source of —
systematic characters. The inflorescence is essentially an axis bearing |
oppositely arranged bracts which subtend highly reduced flowers —

(Plates 8C, D). Frequently the weight of the inflorescence causes
itto droop and resemble an ament, accounting for the earlier placement
of Garrya among amentiferous groups (Table 1).

Within subgenus Garrya, both the male and female inflorescences —

are unbranched and of variable lengths. They are linear and compact —

with reduced decussate bracts closely spaced, overlapping, and imbri- _

cate. The paired bracts are connate both at the base and laterally,

forming cup-like structures which house ternate flowers. The fascicu- —

late inflorescences are borne at the ends of the branches and are

usually pendulous, although in Garrya fremontii they may also be —
semierect. At maturity the inflorescences are sometimes less compact, —
especially the male inflorescences, where internodes between the floral —

bracts are often visible.

The inflorescences within subgenus Fadyenia are pendulous,
branched, bear solitary flowers, and are borne in clusters at the ends —
of the branches. The female inflorescences are lax, with the decussate —
floral bracts connate only at the base, and not imbricate as in subgenus —

Garrya (Plates 8A, F). The bracts are expanded and distinctly leaf-like —
in G. laurifolia (Plates 9A, C). They are reduced and much smaller —
than the foliage leaves in other species of the subgenus (Plate 9B). —
While most of the flowers are subtended by a single bract, the terminal —

flowers are subtended by a partially adnate bract pair.

A strong trend toward reduction in lateral branching of the inflores-_
cence is evident within subgenus Fadyenia. Branching is common —
in Garrya laurifolia and G. longifolia, but is restricted and reduced —

in G. wrightii. Garrya laurifolia has long branches but those of G

SYSTEMATICS OF GARRYA 25

B, G. ovata
lf

when 8. Inflorescence and floral diffesences in Garrya species: A, G. longifolia, Dahling 8190; 7 :
Ri lindheimeri, Dahling 331; C, G. flavescens ssp. flavescens, Dahling 781; D, G. fremontii, Dahling
“ G. flavescens, Dahling 918; F, G. ovata ssp. lindheimeri, Dahling 325.

wrightii are short and have only terminal flowers. These are generally
subtended by a pair of bracts, making their recognition easy and
definite. Short lateral branches with several flowers are also present
in G. wrightii but are confined to the bases of the inflorescences,
as is seen in G. ovata. The number, position, and length of the lateral
branches in the former species have become supressed until only
terminal flowers are represented in branches above the base. A
“ontinuation of the evolutionary trend leads to the complete suppres-
‘10n of branching, as is found in the inflorescences of subgenus Garrya.

The pendulous male inflorescences are shorter and more condensed
than their female counterparts. The inflorescence branches are short
and only bear a few flowers although branching is common near
the base. Terminal male flowers are also subtended by a pair of bracts,

26 GERALD V. DAHLING

PLATE 9. Floral and fn
wrightii, Dahling 1206: C.
longifolia, Dahling 2130. F,

1it differences in Garrya species:
we

; ‘ G
A, G. laurifolia ssp. racemosa, Dahling 118; . C
=. laurifolia ssp. macrophylla, Dahling 280; D, G. buxifolia, Dahling 1050; is
G. elliptica, Dahling 1361: C 1635.

G, G. flavescens ssp. flavescens, Dahling

and are easily distinguished from the lateral flowers. The bracts are
uniformly reduced and are less expanded or leaf-like than the female
structures,

The male inflorescence of Garrya longifolia is an important tax”
nomic feature because of the prominent branching and lax charactel
The flowers are not crowded and appear widely spaced, as they ai

SYSTEMATICS OF GARRYA ys

in the female inflorescences. Hence, G. longifolia may serve as the
starting point in a series of reductions that lead to the condensed
male inflorescences which characterize other members of the genus
(Plate 8A).

Flowers. The reduced nature of Garrya flowers has been responsible
for the phylogenetic confusion surrounding the group. The female
flowers have been variously interpreted as having superior or inferior
ovaries, dramatically affecting the classification of the genus (Table
1

The minute female flowers are represented by an ovary subtended
by a single bract or sometimes by a pair (Plate 9). The ovary is
bicarpellate (rarely tricarpellate) with two persistent styles, and pro-
duces a pair of subterminal anatropous ovules. The perianth is reduced
or lacking in older material making the ovary appear superior. However,
careful examination sometimes reveals small paired appendages at
the base of the styles. Typically one pair of appendages, but sometimes
two, are present and situated at about the same level. They are variably
adnate to the ovary and show considerable variation in size.

Paired ovarian appendages are found in both subgenus Garrya and
subgenus Fadyenia but their orientations are quite different, an
observation first recorded by Eyde (1964). The appendages of subgenus
Garrya are alternate, while those of subgenus Fadyenia are opposite
the styles. In addition, the ovarian appendages of subgenus Fadyenia
are restricted to the terminal flowers of the inflorescences or inflores-
cence branches. Pronounced differences in vasculature suggest that
the appendages are not homologous and therefore not comparable.

Each carpel contains a dorsal and two ventral or marginal vascular
bundles (Eyde, 1964). At lower levels in the ovary the lateral bundles
of adjacent carpels fuse, forming two large compound ventral bundles
which are oriented opposite each other. Near the top of the ovary
the compound ventral bundles bifurcate and proceed, with the dorsals,
to the base of the styles and then terminate. The styles are not
vascularized in either subgenus. Each ovule is supplied with a single
vascular strand which connects to the compound ventral at about
the level of bifurcation. If the vascular connection is above the point
of bifurcation, two strands diverge from each ventral bundle, later
fusing to form the single ovular supply.

Within subgenus Garrya, the paired ovarian appendages sometimes
lack a vascular supply. However, when present it consists of a single
vascular bundle which is fused, at least procambially, with the
“ompound ventral bundle. Ovarian appendages of subgenus F adyenia
“re variably situated on the ovary and are also not always vascularized.
_ 1€ appendages borne near the styles are less leaf-like than those
n'a lower Position, which show a prominent midrib and reticulate

28 GERALD V. DAHLING

venation. When present, the vascular supply consists of a single strand —
which, although aligned with the dorsal bundle, does not unite with .
it but rather extends independently into the pedicel. Therefore, the
paired ovarian appendages in subgenus Fadyenia are leaf-like in

tips, forming slits or windows for the alternately arranged anthers
(Plate 8B). They often appear yellow-green to red and brown. While

of subgenus Garrya. The perianth seg-
ments are variably pubescent, especially near the connate tips, but

are essentially glabrous on the adaxial side. Normally the flowers-
of G. glaberrima are entirely glabrous, although sometimes a single
row of stiff trichomes is found on the perianth borders. The anthers
are oblong-elliptic, introrse, open by longitudinal slits, and are basi-
fixed to extremely short filaments which barely exceed the length
of the anthers. The presence of a degenerate or obsolete ovary within

the male flowers has been periodically reported, but my own observa”
tions have not verified this.

The male flowers in Garr

ya are fairly uniform in size, but those
of G. wrightii and CG. grise

a

SYSTEMATICS OF GARRYA 29

to arid and semiarid environments. The length of the anthers and
the filaments of several taxa were measured, but because significant
results were not obtained, this approach was abandoned. As with
female flowers of Garrya, the general reduction, simplicity of design,
and uniformity of male flowers severely limits their significance and
taxonomic usefulness.

Fruit. Garrya fruits are two-seeded berries which persist on the
parent plants for long periods of time. When young, the fruits appear
green and fleshy (Plates 9D, E, F) but this condition gradually gives
way to the dry and brittle form seen at maturity (Plate 9G). Pairs
of persistent styles grace the tops of the nearly globose fruits. The
color ranges from dark blue to brown or white, but is not of taxonomic
importance.

Hallock’s investigation (1930) of the internal anatomy of the fruit
revealed several distinct cell layers and my own observations in both
the field and laboratory supplement these findings. In longitudinal
section, the outside of the ovary wall forms a large circle with two
bulges representing the bases of the styles and the pedicel. The fruit
wall is composed of nine cell rows making up three distinct layers.
Cells of the four outer rows are rounded and more or less irregular,
with the outside row being heavily cutinized. The fifth or middle
row is made up of deeply pitted stone cells, which demarcate the
remaining inner rows which contain flattened cells of variable lengths.
The row of stone cells diverts outward, forming a semicircle around
the fibrovascular bundles which frequent the area. This displaces
the outer cells, forcing them to curve outward forming the veins which
are visible on the fruit exterior. Two, or rarely three, pendulous ovules
with parietal placentation normally occur in a subterminal position
and partially fill the ovarian chamber. Much of the flesh of the fruit
is derived from the outer layer of the single massive integument.
In drying, this flesh gradually withdraws, forming a thin papery
Covering around the seeds. At maturity, two-thirds of the fruit remains
hollow with the seeds occupying only the terminal portion.

Herbarium material was used to examine the size and shape of
Garrya fruits. A calibrated dissecting microscope was used and both
the length and the width of the fruits were measured. A comparison
of dimension averages (in millimeters) is given in Table 6. The fruit
Shape is expressed as the width/length (W/L) proportion. If the
quotient is less than 1.00, the fruits are longer than they are broad,
“nd if the quotient is greater than 1.00 the fruits are broader than
Pe: An exact value of 1.00 indicates a perfectly globose fruit with

© width and length being equal.

_tuit size within shanna is relatively uniform and _
little information of systematic interest. Garrya elliptica has the larges

30 GERALD V. DAHLING

TABLE 6. FRUIT DIMENSIONS (mm.), AVERAGES,

Species LX wx W/L
G. elliptica 8.20 7.16 0.873
G. veatchii 7.24 6.32 0.852
G. buxifolia 6.74 5.72 0.849
G. fremontii 7.08 6.30 0.900
G. flavescens ssp. flavescens 6.39 4.98 0.778
G. flavescens ssp. pallida 7.67 6.53 0.851
G. flavescens ssp. congdonii 7.70 6.25 0.810
G. corvorum 7.41 6.58 0.887

fruits but this is of questionable significance when the total range
of fruit dimensions of the other taxa is considered. Even G. corvorum,
highly isolated geographically and strongly divergent in other charac-
ters, shows fruit dimensions similar to the other species of the subgenus.
Fruit shape is also rather uniform, with the W/L ratios varying from
0.77 to 0.90. All fruits appear subglobose and are longer than they
are wide. Garrya flavescens ssp. flavescens is the most elongated, —
showing a ratio value of 0.77 while G. fremontii is nearly round
with a ratio value of 0.90. Because of the almost continuous distribution
of shapes between these minimum and maximum values, the taxonomic —
importance of fruit shape is, at best, minimal.
The fruit dimensions within subgenus Fadyenia (Table 7) are more
variable than those of subgenus Garrya. Garrya laurifolia ssp. racemo- —
sa has the largest fruits of the entire genus, with an average length
of 10.30 and width of 8.38 mm. This contrasts with the small fruits
of G. salicifolia and G. grisea, with widths of 5.50 mm. and lengths ©
of 5.91 and 6.30 mm. respectively. The fruit dimensions of other
taxa in the subgroup are intermediate (Table 7). Fruit shape is of
greater interest because of the definite differences among taxa. Garryd
ovata, G. wrightii, G. grisea, G. salicifolia, and G. fadyenii have

TABLE 7. FRUIT DIMENSIONS (mm.), AVERAGES, AND SHAPES IN SUBGENUS FADYENIA

Species LX wx W/L

es Sl
G. ovata ssp. ovata 8.07 7.03 0.870
G. ovata ssp. goldmanii 6.84 6.34 0.927
G. ovata ssp. lindheimeri args 7.24 0.935
G wrightii 6.21 6.08 0.979
: grisea 6.30 5.50 0.873
salicifolia 5.91 5.50 0.929
. aba 6.70 6.11 0.911
- longifolia 8.70 6.94 0.797
G. laurifolia ssp. laurifolia 8.38 6.92 0.825
G. laurifolia ssp. macrophylla 8.26 6.76 0.818
G. laurifolia ssp. racemosa 10.30 8.38 0.806

SYSTEMATICS OF GARRYA ol

nearly globose fruits with shape ratios ranging from 0.87 to 0.97.
Garrya glaberrima, G. longifolia, and all subspecies of G. laurifolia
have more elongated fruits, with shape ratios ranging from 0.79 to
0.82. Of these species, G. laurifolia and G. longifolia appear closely
related in other respects. Garrya glaberrima is isolated within the
genus, but shows an affinity with G. laurifolia in pollen and leaf
dimension. Within this group of species, G. ovata, G. wrightii, and
G. grisea appear related, but the inclusion of G. salicifolia and G.
fadyenii is more difficult. They show an affinity with G. laurifolia,
but they are each geographically well isolated. Garrya salicifolia occurs
in southern Baja California and G. fadyenii is found in the Greater
Antilles. It seems likely that these taxa originated from isolated colonial
or peripheral populations of laurifolian ancestors. The chance fixation
of smaller and more globose fruits in such populations would not
prove difficult because the gene pool would be constricted.

Garrya fruits are indehiscent and have no abscission layer separating
them from the parent plants. The action of the wind and rain opens
them and gradually wears holes in the walls, exposing the seeds.
Insects also burrow into the walls and weaken or actually crack the
brittle fruits in half. The seeds are quite hard and do not appear
to suffer from such insect intrusion. However, this is not always
the case since some fruits show a small hole, lack seeds, and are
filled with insect larvae. The wood rat, Neotoma fuscipes annecteur
Elliot (Hallock, 1930), also eats Garrya fruits.

eeds. Garrya fruits produce from one to three seeds. However,
three seeds are found only in plants with tricarpellate ovaries, which
are rare. The seeds are subglobose and are borne in a subterminal
Position on parietal placentae. During development, one seed is pushed
down and to the side, making the pair appear somewhat asymmetric.
The embr yology of Garrya has been investigated by Kapil and Moshang
(1966) and also by Rao (1963). The minute embryos of Garrya are
embedded in a copious green endosperm which gradually becomes
Purple and hard. The arilloid structure (Hallock, 1930) develops om
with the outer cells of the integument absorbing water, bulging
utward, and forming a silvery transparent layer which divides oon
clinally. After the first division, the inner layer divides periclinally
again Producing a third layer. Cells of the outer layer do not divide
but become greatly elongated, cubical, and 10 to 15 times their original
Size. The translucent arilloid structure becomes purple and quite fleshy
during seed development, but dries out at maturity and appeal thin,
wrinkled and membranous. However, this is reversible since, when
Wet, the arilloid structure regains its thickness and again covers the
seed. Within subgenera Garrya and Fadyenia, the seed size and shape
ate relatively uniform. Minor variation in seed size and coat appears

32 GERALD V. DAHLING

PLATE 10. Pollen differences in Garrya

flavescens (acetolysed), Cottam 67! 0, 672:

— al rescens SSP

species: A, B, unacetolysed Garrya pollen, 675; C, G. on
756 ie

ys 5x; D, G. ovata ssp. goldmanii (acetolysed), Johnson whe plier

E, G. glaberrima (acetolysed), Dahling 38, 6725~x: F. G. laurifolia ssp. : tolysed), Dahling 111,

to reflect growth conditions rather than genetic differences betwee?
taxa. The seeds have fairly large diameters, ranging from 2 to 3 —
but the dimensions fail to show significant differences that can
used for taxonomic purposes.
Palynology. Garrya pollen is copious and is yellow-green to orang
in color. It is small to medium in size, and is small enough to a
effectively transported by wind. The shape of the pollen is subsphero

SYSTEMATICS OF GARRYA

‘ Polle 7 see : a : :
SSP. mexi af Cibtetonces ts Garrya species: A, G. ovata ssp. lindheimeri, Warnock 46012, 1350x; B, G
a *icana, Dahling 511, 675x: C 350: D, G. laurifolia
87,

he, Oe p
C, G. laurifolia ssp. laurifolia, Dahling
|

Pik thin 11,
1350 » Dahling 16, 1350x; E, G. laurifolia ssp. racemosa, Dahling 45, 1350; F, G. wrightii, Peebles

d: -

Dretaning from oblate-spheroidal to prolate-spheroidal. This inter-

suboblate t. aS from that of Erdtman (1952) who describes it as

te to the : errs spheroidal. The differences are probably attributa-

a larger eS: different mounting media and to my having examined

¥ haus imber of samples. The exine is thick and quite roughened
€ processes. The sexine is thinner than the nexine and

34 GERALD V. DAHLING

reticulate with muri simpli- or duplibaculate. Garrya pollen is tri-
colporate with the colpi being rather short (Plate 11D).

Pollen data are often important in the resolution of systematic
questions, and thus a survey of Garrya pollen was undertaken.
Comparisons were made of pollen size, shape and the polar area
index for the various Garrya taxa.

Method: Pollen samples were obtained from herbarium material.
Small amounts were placed in the center of microscope slides and
then washed five times with ethyl alcohol. This was done by applying
one or two drops of alcohol to the sample and allowing it to dry.
The oils and resins were removed with a small piece of gauze after
each application. Then a small quantity of diaphane mounting medium
was placed on the sample, and, with the aid of a dissecting needle,
the polleniferous material was evenly spread over the slide. After
positioning a cover glass, the samples were examined. A light micro-
scope was calibrated allowing for the expression of equatorial and
polar diameters in microns. The Polar Area Index (Kapp, 1969), defined
as the ratio of the distance between adjacent furrows and the equatorial
diameter, was calculated for all taxa. Pollen shape was examined an
measured as the ratio of polar to equatorial diameter (P/E).

Discussion: With few exceptions, pollen size of Garrya species is
rather uniform. Equatorial and polar diameters range from 21-38 and
22-39 um. respectively. The polar furrow distance varies from 15-30
ym. Measurement averages (Table 8) afford a convenient method of
comparing taxa. With the exception of G. flavescens and G. laurifolia,

TABLE 8, AVERAGE POLLEN SIZE (um.)

Equatorial Pola

r Polar Furrow

Garrya Diameter Diameter __ Distance
buxifolia 26.689 27.423 19.770
elliptica 29.380 29.400 21.687
veatchii 27.985 29.184 20.913
fremontii 28.444 30.057 21.375
flavescens ssp. flavescens 31.702 33.355 24.187
flavescens ssp. pallida 30.084 31.756 22.833
corvorum 96.611 29.162 20.076
salicifolia 28.350 29.137 22.398
wrightii 27.612 29.164 20.160
grisea 27.918 28.464 20.650
ovata ssp. ovata 26.268 27.232 19.627
ovata ssp. lindheimeri 27.749 28.723 20.846
eran ssp. mexicana 25.258 26.616 18.892
eae 25.987 6.334 18.985
ost cals 26.535 26.550 20.713
auri olia ssp. laurifolia 31.224 31.741 26.621
laurifolia ssp. racemosa 32.4 33.300 27.450 :
laurifolia ssp. macrophylla 30.776 30.746 25.082 i

SYSTEMATICS OF GARRYA 35

most equatorial and polar diameter averages are similar. Species within
subgenera Garrya and Fadyenia are not distinguishable by their pollen.
Taxa showing size variation belong to both groups, including G.
flavescens and G. laurifolia, which show equatorial and polar diameter
averages of 30 and 33 um. respectively. The remaining Garrya taxa
have smaller equatorial and polar diameters ranging from 25-29 an
26-29 wm. Polar furrow distance averages also show a general unifor-
mity (Table 8) with the exception of G. flavescens ssp. flavescens
and G. laurifolia. Both of these taxa have higher polar furrow distances
which correspond to their large equatorial and polar diameters. Since
these species are not closely related to each other, the significance
of their increased pollen size is not evident.

Variation in polar area index (PAI) is less continuous than other
dimensions of Garrya pollen. Both the largest and smallest PAI values
(Table 9) are found in subgenus Fadyenia. Garrya longifolia, G.
sdalicifolia, and G. laurifolia show large PAI averages ranging from
0.79 to 0.85. This contrasts sharply with G. grisea of the same subgenus,
which has a PAI average of 0.68, the smallest of the family. Remaining
taxa of both subgenera show intermediate PAI averages ranging from
0.72 to 0.76. It is interesting to note that taxa sharing large PAI values
also appear morphologically related in other ways and may represent
a separate line of specialization within the genus.

ere is no significant variation in the shape of Garrya pollen,
which is almost identical in all species (Table 9). What variation
there is (P/E ratios), is extremely limited, ranging from 0.99 to 1.12,
and is not of taxonomic importance.

TABLE 9. POLAR AREA INDEX A

Garrya is —
buxifolia 0.7407 : rie
elliptica Rae
yenenis 0.7473 ; sei

remontii 0.7301 .
flavescens ssp. flavescens 0.7629 set
flavescens ssp. pallida pi 1.094
corv f
Salicifolia rea mpl
wrightii pices sce
Ovata ssp. ovata 0.7277 er
Ovata ssp. lindheimeri 0.7513 a
Ovata ss i O70 seh

P. Mexicana 1.0133

fadyenii 0.7305 1 0005
longifolia 0. 7803 1 0165
laurifolia ssp. laurifolia 0.8521 ots
laurifoli papntehbe “
Pei olia ssp. racemosa 0.9989
aurifolia ssp. macrophylla 0.8015

36 GERALD V. DAHLING

SEM Pollen Morphology. Examination of pollen by means of the
scanning electron microscope (SEM) was carried out to survey the
fine structure of the exine.

Method: Polleniferous material was obtained from herbarium speci- —
mens and divided into two samples. One was acetolyzed and the —
other was not. The acetolysis procedure was as follows: polleniferous
material was placed in 40 ml. glass centrifuge tubes, covered for
two minutes with 100% glacial acetic acid, and then centrifuged for
three minutes. After decanting, 20 mls. of acetolysis mixture, composed
of nine parts of electronic grade acetic anhydride and one part
concentrated sulfuric acid, was added to the samples. The sample
tubes were then placed in a water bath, heated to 100° C, and stirred
for three minutes. Following centrifugation and decanting, the material
was put through a 100% and 50% glacial acetic acid series. After
the final centrifugation, 20 ml. of distilled water was added and the
samples were centrifuged and decanted once more. This procedure
was repeated four times to insure the removal of all traces of acid.
The final polliniferous sediments were resuspended in 15 ml. of
distilled water and stored in small vials. Samples were then placed —
on aluminum pedestals applying double stick scotch tape and then
several drops of pollen suspension. Next, the pedestals were transferred
to a slide warmer and allowed to dry for 12 hours. Untreated pollen
was applied to the taped pedestal surfaces dry. Pollen samples were
evacuated in a Varian Vacuum Evaporator VE 10 to 10-*—10~° torr.
and then coated with both carbon and gold-palladium. The samples
deteriorated after coating, so the specimens were immediately placed |
in an AMR Model 900 SEM and examined. The maximum resolution
of the SEM employed was 100-200 A.

Discussion: Most untreated Garrya pollen tends to be distorted of
completely collapsed,. making observation and photographing under —
high magnification almost impossible (Plates 10A, B). Untreated pollen
also tends to shatter when exposed to the electron beam. These problems _
are almost eliminated when acetolyzed material is used (Plate 10C);
and although chemical treatment expands the individual grains t0_
a degree, the pollen becomes much more workable. The increase i?
pollen size is more than compensated for by the expansion of collapsed
grains and the removal of surface depressions. |

The size of Garrya pollen shows great uniformity, but there ar
several interesting variations in the fine structure of the exine. All
species bear minute processes which are borne on the reticulate
sculptural elements, but the frequency of these processes is variable.
Within subgenus Fadyenia, G. ovata displays the greatest frequency
showing many processes on the surface of the grains (Plate 10D).
Garrya glaberrima follows with a lesser, but impressive, number 0

SYSTEMATICS OF GARRYA aot

processes (Plate 10E). Garrya laurifolia shows only moderate sculp-
tural development and resembles G. fadyenii, G. salicifolia, G. grisea,
and G. wrightii (Plate 10F), in this respect contrasting with G.
longifolia, which shows the least number of processes. In subgenus
Garrya, the pollen of G. fremontii, G. flavescens, and G. corvorum
is most densely covered with the minute processes. Because of their
ability to hybridize in nature, the similarity between G. fremontii
and G. flavescens is especially noteworthy. The pollen of G. corvorum
also approaches the former taxa, but that species is isolated and not
closely related. Although the remaining species of the subgenus, G.
buxifolia, G. elliptica, and G. veatchii, display a low frequency of
sculptural processes, only G. elliptica and G. veatchii appear to be
closely related on other grounds.

The size and shape of the lacunae are variable and are important
taxonomic indicators for some species. In subgenus Fadyenia, Garrya
ovata ssp. lindheimeri (Plate 11A) and G. ovata ssp. mexicana (Plate
11B) show the largest lacunae. They appear almost square and .are
of diagnostic importance because identification can be made from
pollen alone. Pollen of G. laurifolia (Plates 11C, 11D, and 11E),
G. salicifolia, G. fadyenii, and G. wrightii (Plate 11F) display lacunae
which, though moderately large, are smaller than those of G. ovata.
In addition, they are less angular and more rounded making them
easily distinguishable. An intermediate condition between G. ovata
and G. laurifolia is demonstrated by G. glaberrima (Plate 12A) where
the relatively angular lacunae are larger than those of G. laurifolia,
but smaller than G. ovata. Apparently pollen of G. glaberrima is
not as different from the Garrya mainstream as is that of G. ovata.
This is peculiar since other evidence (Figure 4) suggests that G.
glaberrima is strongly isolated from other members of the genus.
Pollen of G. grisea approaches G. wrightii in size, and is distinguished
from the other taxa by the presence of elongated lacunae (Plate 12B).
Lacunae of G. longifolia are small and mostly round (Plate 12C),
and so reduced that in some areas the muri appear confluent.

Within subgenus Garrya, pollen lacunae range from very large and
Square, as in G. elliptica, to small and round, as in G. veatchii (Plate
12D). The intermediate condition is shown in G. flavescens ssp. pallida
and G. fremontii which have smaller and angular to subround lacunae
(Plate 12E). Pollen of G. flavescens ssp. flavescens differs from G.

avescens ssp. pallida by the smaller more rounded lacunae, not
unlike those of G. corvorum. Lacunae of G. buxifolia are similar
'0 G. veatchii in size but appear more rounded (Plate 12F).

Compared with other families (Moseley & Beeks, 1955), the pollen
of Garrya strongly resembles that of the Cornaceae. Aucuba japonica
(Erdtman, 1952), long reputed to be an ancestor of Garrya, has

38 GERALD V. DAHLING

tricolporate pollen, which is prolate spheroidal with an average polar
and equatorial diameter of 43 and 39 pm. respectively. The shape
of the Aucuba pollen is similar to that of Garrya, but is larger and
has a pilate sexine which is thinner than the exine. Griselina litoralis,
another putative ancestor (Eyde, 1964), has pollen less similar to Garrya
than does Aucuba. It has tricolporate pollen, which is prolate in shape,
with an average polar and equatorial diameter of 31.0 and 20.5 pm.
respectively. The sexine is the same thickness as the exine and is
tegillate (Erdtman, 1952). The basic features of cornaceous pollen,
as seen in Aucuba and to a lesser extent in Griselina, do not controvert
a relationship with Garrya, although the pollen is not similar in every
aspect.

Breeding System. Generally the barriers between gene pools in
Garrya may be easily breached, although isolation in some species
is definitive. The speciation is of the geographic type (Grant, 1971)
and is associated with only weakly formed reproductive barriers. In
addition, both subg Garrya and subg Fadyenia have a uniform
chromosome number of n = 11 (Turner, 1960), thus eliminating some
of the cytogenetic aberrations which result from interspecific crosses
where chromosome numbers are different. With a few notable excep-
tions, speciation and gene pool integrity in Garrya are dependent
on geographic separation rather than on specific chromosome or genetic
incompatibilities.

Garrya is well-adapted for wind pollination, a beneficial strategy

in arid regions. Moreover, the pendulous, cryptic, and catkin-like
inflorescences are ideally suited for anemophily. The structures are
blown by the wind and clouds of yellow pollen are visibly dispersed.

Once caught by the air currents, the pollen is easily transferred to

the receptive stigmas of the female inflorescences, which are pendulous
at the ends of the shoots and are readily exposed to the pollen bearing

wind.

The male floral morphology is also well-adapted for wind pollination. —
Since the stamens are opposite the four slits or windows of the partially
connate perianth, the anthers are directly exposed. When air passes
through the flowers, the pollen is blown out of the slit and into .

the air stream. In addition, the passage of the wind over the subtending

and partially enveloping bract causes suction which forcibly removes —
the pollen from the perianth chamber. Formed from the terminally
fused perianth segments (Plate 8B), the perianth chamber is 0 |
considerable significance. Anther dehiscence is introrse by means of |

longitudinal slits so that during anthesis the pollen is unable to fall

os the ground. When there is no wind, it is stored in the shaded
c oer and protected from excessive desiccation which otherwiS*
might occur in the semiarid Garrya habitats. When the air curren’

SYSTEMATICS OF GARRYA 39

return, the pollen in the chambers is quickly sucked out and freed.

The breeding system of Garrya is designed for maximum outcrossing
and genetic variability. Its dioecious condition prevents selfing and
insures a gene pool with sufficient variability for successful divergence
under arid conditions (Stebbins, 1950, 1952). In addition, variability
is also stimulated by anemophily which allows for gene flow between
spatially isolated populations. This accounts for the blurring of
infraspecific and specific taxa within Garrya, especially when the
weak interspecific reproductive barriers characteristic of the genus
as a whole are considered. Therefore, many infraspecific taxa reach
the level of semispecies but fail to achieve complete isolation and
attain specific status.

Such a condition may indicate a relatively recent evolutionary origin
for the group, and it may be supposed that complete divergence will
eventually occur. However, it might be argued also that such systems
themselves may be selected for in order to maintain the genetic
variability that is advantageous in harsh environments. Although the
question is still open, with respect to Garrya, the latter is the most
convincing.

The problem of delineating the intergrading Garrya species was
recognized early by several authors (Eastwood, 1903; Bacigalupi, 1924).
Interspecific hybridization does occur in Garrya, but not to the extent
that would be the case if the presently recognized infraspecific taxa
were given specific rank. The early taxonomic history of Garrya was
characterized by the splitting and erection of “species” which naturally
intergraded with other “species.” Examples of these problematic taxa
include G. rigida, G. pallida, G. congdonii, G. mollis, G. lindheimeri,
G. goldmanii, G. macrophylla, G. oblonga, G. racemosa, and G. gracilis.
Such taxa as are represented by these names are better considered
to be infraspecific taxa of polytypic species. Intergradation, where
it occurs, is thus accounted for in a more satisfactory manner.

The most extensive hybridization in subgenus Garrya is between
G. fremontii and G. flavescens ssp. pallida. They intergrade in leaf,
trichome and epidermal characters, making their distinction sometimes
difficult. Evidence for hybridization and introgression are also provid-
ed by chemical data (Figure 4). While delineation is problematic in
local areas, these taxa are sufficiently distinct in most regions to justify
their retention in specific rank. Since intermediates are occasionally
observed, it is also likely that G. flavescens and G. veatchii hybridize,
but not extensively.

Hybridization in subgenus Fadyenia —".
Garrya. Garrya ovata ssp. mexicana and G. laurifolia hybridize in
northern Mexico where their ranges overlap, and morphologically
intermediate specimens are common in this area. Garrya ovata ssp.

exceeds that of subgenus

40 . GERALD V. DAHLING

ovata and G. ovata ssp. lindheimeri also hybridize with G. laurifolia
but less frequently. Garrya glaberrima is of particular interest because,
although sympatric with both G. ovata and G. laurifolia ssp. macro-
phylla, no intermediates or putative hybrids have been found. The
assumption is that breeding barriers are present which provide for
genetic isolation. The distinctiveness of G. glaberrima is supported
by phytochemical data (Figure 4) which suggest that it is not closely
related to other taxa. Genetic isolation is rare in Garrya which is
a tightly knit group that generally shows reproductive flexibility.
While appearing distinct where geographically isolated, Garrya
longifolia and G. laurifolia ssp. racemosa hybridize in areas where
their ranges meet. They are closely related, with only weak isolating
barriers. Intermediates are numerous at the contact points, but the
degree of swamping is insufficient to warrant changing their taxonomic
status. Most of the populations of G. longifolia are biochemically
and morphologically distinct, lending additional support for the
retention of their specific status (Figure 6). Interspecific hybridization
does not affect G. fadyenii because of its complete geographical
isolation in the Greater Antilles. The same is true for G. salicifolia
and G. grisea of southern and northern Baja California respectively.
Phytochemistry. Discovery of the chromogenic glycoside aucubin
(Herissey & Lebas, 1910) launched the chemical investigation of
Garrya. The same glycoside (Figure 5) also occurs in Aucuba japonica
(Bouquelot & Herissey, 1910; Lebas, 1911), often resulting in the
conclusion that a Garrya-Aucuba alliance exists. Aucubin is not
restricted to either subgenus and has been reported in the seeds, stems,
leaves, and fruits of G. elliptica, CG. laurifolia ssp. macrophylla, and
in the hybrid G. elliptica X G. laurifolia (Herissey & Lebas, 1910).

These occurrences provide little information about species affinity,
and are mainly citeda ti id

PI geviden ce for a Garryaceae-Cornaceae
relationship. Since aucubin is rarely found in other than gamopetalous
families (Swain, 1963), the isolated occurrence in Garrya and Aucuba
may prove important.

Dipterpenoid alkaloids are characteristic of several Garrya species
(Oneto, 1946; Wiesner et al., 1952. Moselting, 1961). Leaf extracts
of G. fremontii and G. buxifolia contain amorphous basic nitrogenous
residues which react positively with general alkaloid reagents (Table
10). The crystalline salts of alkaloids have also been isolated and

constituents, garryine, veatchine, garryfoline, isogarryfoline, cuau-

been characterized. The |

SYSTEMATICS OF GARRYA 41

TABLE 10. DISTRIBUTION OF ALKALOIDS

Species Leaves Bark Alkaloids Present

G. fremontii ne Amorphous basic nitrogenous residues reacting posi-
tively with alkaloid reagents. No crystalline salts
produced.

G. buxifolia + Amorphous basic nitrogenous residues reacting posi-
tively with alkaloid reagents. No crystalline salts
produced.

G. flavescens = — No alkaloidal material found.

G. elliptica “+ Garryine and veatchine. Both isolated as crystalline
salts.

G. veatchii + + Garryine and veatchine. Both isolated as crystalline
salts. Highly concentrated in the bark.

G. wrightii + + Garryine and veatchine. Both isolated as crystalline
salts. Slightly more concentrated in leaves.

G. laurifolia + Garryfoline, isogarryfoline, cuauchichicine, and iso-

cuauchichicine.

latter four were isolated from G. laurifolia (Djerassi et al., 1955) and
bear a close isomeric relationship to each other and.to veatchine and
garryine. Oneto (1946) reports that G. flavescens completely lacks
alkaloidal constituents but this has not been verified by other inves-
tigators. The Garrya alkaloids bear a structural resemblance to atisine
(Pelletier, 1960) and are probably derived from the phytol part of
chlorophyll (Robinson, 1967).

The present work gives a brief comparative survey of the flavonoid
content within Garrya and selected genera of the Cornaceae in order
to determine the biochemical affinity between Garrya species and
to investigate suggested phylogenetic relationships to the Cornaceae
(Hallock, 1930: Moseley & Beeks, 1955; Eyde, 1964, 1967).

Flavonoids consist of two phenol units linked together by a three
carbon chain and are one of the most common constituents of vascular
Plants (Alston & Turner, 1963; Hegnauer, 1966; Gibbs, 1974). While
the biosynthetic pathways for specific flavonoids remain obscure,
much is known of the biogenesis of flavonoid compounds in general
(Geissman, 1969; Robinson, 1967; Seshadri, 1962; Griseback &
Patschke, 1960). Although data suggesting a physiological importance

ave been presented, the functional role of flavonoids in plants is
still the subject of much speculation. Flavonoids may be involved
in the formation of reproductive barriers (Kuhn & Low, 1949) and
also in the regulation of plant growth and development (Furuya et
al. 1962; Mumford et al., 1961; Hendershott & Walker, 1959). In
addition, since some flavonoids are synthesized in response to injury
ine, asitic attack, a protective role has also been suggested (Hadwiger,

).

Method: Plant material was collected in the spring and summer

GERALD V. DAHLING

¢ f one A iat + : ndegee
differences in Garrya species: A, G. glaberrima, Dahling 38, 1350: B, G Branc yf
: 7 > fremont,
+. longifolia, Dahling 2233, 1350x: D, GC. eatchii, Dahling 1674, 1350; >. freme
FC

G. buxifolia, Henderson 5716. 1350

: ry a . : Be ove from
of 1972 and 1973 respectively. Approximately twenty-five leaves fron
individuals of a population were ground in
2)

—

a Waring blender for
3 minutes, until powdered. In every case mature nonsenescent leaves
from the previous season’s growth were used and unusual or distorted
leaves were eliminated. A small] portion (about 0.3 grams) was extracted
in .1M HCl in methanol at room temperature for 24 hours. The extracts
were immediately applied to a Watmann No. 3 n
paper as 2

am. chromatogTaa a
iy : - *hicn
cm. streaks. Electronic grade solvents were used, whic

SYSTEMATICS OF GARRYA 43

included benzene-acetic acid-water (4:4:2 v/v) and isopropanol-ethyl
acetate-water (7:1:2 v/v).

The latter was the most satisfactory, especially when preceded by
a benzene prerun which eliminated streaking and tailing. When the
solvents reached a height of 35.5 cm. the chromatograms were air-dried
and examined in the presence of ammonia, first in natural and then
in ultraviolet light. The chromatograms were done in triplicate and
independently scored to eliminate error or variation. Once separated,
compounds were assigned reference numbers. Those compounds
appearing identical in color and position were given the same number.
n this way the flavonoid content of different species could be
compared

A combined total of 82 flavonoid compounds were found in Garrya.
The number per species varied from 8 to 20, with an average of
15. All the species were compared by calculating the coefficient of
association for species pairs. This was done by dividing the number
of common compounds by the total number of compounds in the
species pair. A score of 1.00 indicates complete compound corre-
spondence while 0 indicates no agreement and presumably no bio-
chemical relationship. The Paired Affinity Index (PAI) (Ellison et

FLAVONOID wisi malty: wen me celle!
——_—_

buxifolia ee 08 8 28 e000 ° e

elliptica Ceee eeeceeeee0 eo oe eee °
veatchii Peeeeeceeces ce @ ° oF .
flavescens
Sspflavescens © @00@@0O0C8® e800 ° e oo
flovescens
$8. pallida ©@ eeeeeeeeecesce se0e Rh dh eatin 4
Corvorum ee ef eee oo @ ° ° ee ¢
fremontii ee ce eeece ° coco
wrightii © 08 eeeccccces ooo 8 o6ete, EF oR ed caer x
ovata
SSp. lindheimeri © ee cece e , et Se 8 eeeceeee 0
ovata
SSp-goldmannii @ ee ee eee ae eeeeee ©
ovate
SSDP. Mexicana e068 & « ° eeeeee ©
Ovota
it © ee ee eee ° fe gerreeere iy
— @ @8 e080 ee « oe
Salicifolia ‘ice. ao eo ° eos ee °
Glaberrimg . be ce ° e ° ee sooo @
@e0

fodyennj ee ee ef « ° .

gifo ® 8 ef ete eeee Be. Oreo? ©
lauritolig

°

S8D. laurifolig ce ee ee ee esoeeee Oe ee0e° °
‘ourifotig ee escee Ly
ton maerophya ° ef ee ee . anata

ifolig ° e

"Neem 6 @ 6 88 668 oe 8 es eae ge wife Oe
lQurifolig ° coo o
85 ee

P- Guichen: e ce ee ' r \ i

: Me f
Black dots—presence of individual flavonoids; white dots—trace amounts or questionable presence o
a flavonoids. Lower scale indicates flavonoids by number.

44 GERALD V. DAHLING

al., 1962) was obtained by expressing the coefficient of association
as a percentage—that is, multiplying the value by 100. These indices
can be expressed as polygonal graphs (Ellison et all., 1962) allowing
for a rapid quantitative comparison of species. The center of the circular
graphs (Figure 4) represents a PAI value of 0, while the periphery
represents a value of 100, with intermediate values along the radii.
Taxa are assigned different radii so that each graph compares one
taxon with the others. At least one PAI value per graph will equal
100 because one radius represents the taxon in question. It is also
possible to quantitatively express the relationship of one taxon to
the others as a group. The Group Affinity Index (GAI) (Ellison et
al., 1962) is obtained by adding the various PAI values of a taxon.
Thus, each PAI graph takes on a numerical value indicating the general
group relationship. For the 27 taxa considered, the maximum possible
GAI is 2700, but only if one taxon completely corresponds to all
other taxa. Conversely, if a taxon failed to correspond in any way
to the other taxa, the GAI would be 100, the lowest value possible.

Consideration of unique compounds (those occurring in only one
taxon) is also valuable. The Isolation Value (Ellison et al., 1962) is
found by dividing the number of unique compounds by the total
number of compounds of the taxa under consideration and expressing
the value as a percentage. A large isolation value indicates extensive
biochemical isolation and may also suggest an early evolutionary
divergence. Smaller values probably result from adaptation to specific
habitats or environments.

of subgenus Garrya are numbered and located in the upper left while
those of subgenus Fadyenia are in the lower right of the graph. Note
that several compounds are common to both subgenera and _ that
compound number 12 is universal, occurring in all species. Since

the taxa share a similar genetic background, the occurrence of common
compounds is expected.

teristics with members of subgenus Fa-
d quite early from ancestral stock of
followed an independent evolutionary

Garrya but also contai

ns compounds of its own subgenus. This 15 |

Fic
eliptica

a

SYSTEMATICS OF GARRYA

4. Paired beg “gray (coefficients of association expressed as percentages).

; a ey eRe G. flavescens ssp. flavesc
wren, 7; G. pier ss mane ri, 8; G. ovata ssp. goldmanii, 9; G. orighti,
8laberrima, 13: G. ‘sleia ie C longifolia, 15; G. laurifolia ssp. sch oe

Garrya buxifolia, 1; G.

ens, 5; G. ne lavescens ssp. pallida,
G. grisea

‘ee
, 11; G. eatichiotta,

46 GERALD V. DAHLING

peculiar, since morphologically G. wrightii does not appear to be
intermediate, but rather a derived member of the Fadyenia line.

An examination of PAI’s (Figure 4) reveals a closer chemical
relationship within rather than between subgenera. Garrya buxifolia
and G. veatchii appear chemically distinct, while G. fremontii shows
a relationship to G. elliptica and G. flavescens. It is not surprising
that the two subspecies of G. flavescens show a close affinity, but
the approach of G. flavescens ssp. pallida to G. fremonitii is of interest.
The two taxa intergrade morphologically, indicating hybridization and
introgression, and their similar flavonoid constituents add support
for this supposition.

Several species within subgenus Fadyenia show chemical affinities.
All subspecies of Garrya ovata share similar flavonoid constituents
and closely approach each other. In addition, a chemical relationship
with G. fadyenii, G. wrightii, and G. laurifolia is indicated. Populations
of G. wrightii vary chemically depending on their location. In the
extreme eastern parts of the range, the flavonoid content is similar
to G. ovata while in the south and west populations approach the
flavonoid content of G. grisea. Garrya salicifoli pp losely related
to G. laurifolia and G. fadyenii while the latter approaches G. ovata,
G. longifolia, and G. laurifolia. Garrya glaberrima is of special interest
because chemically it does not show a strong affinity towards other
species. It is endemic to northern Mexico and appears to be well
isolated reproductively. Although it is sympatric with G. ovata and
G. laurifolia, no intermediates or hybrids have been found. Thus,
G. glaberrima, because it maintains a unique gene pool, could follow
an independent evolutionary path resulting in the development of
flavonoids not shared by other members of the genus. Garrya glabertt —
ma is not only isolated from the species of subgenus Garrya, but —
from members of its own subgenus as well. This is not the case,
however, with G. longifolia, which appears chemically related to both —
G. fadyenii and G. laurifolia. While all G. laurifolia subspecies bea! _
high PAI’s with each other, they also chemically approach G. fadyenit,
G. longifolia, and G. ovata. In general, most of the species show
strong chemical relationships with the other species of the genus.
This is not always the rule, however, as is demonstrated by G. buxifolia
and G. glaberrima (Figure 4).

Ate Gal untiies the latonhip of one taxon the 0%
. Indices of Garrya taxa (Table 11) range from
approximately 1300-860. With the exception of G. corvorum and ©
glaberrima, lower GAI’s generally are found in subgenus Garry4
ree ono, chemically intermediate between subgenera Garry?
Bie Piet td, has a high GAI of 1272, showing a close relationship
arrya species as a whole. The reverse situation, indicatiné |

|
|

SYSTEMATICS OF GARRYA 47

TABLE 11. GROUP AFFINITY INDEX

Species GAI Species GAI
Garrya wrightii 1303.7 G. flavescens ssp. pallida 1110.5
G. laurifolia ssp. racemosa 1302.6 G. fremontii 1108.2
G. fadyenii 1292.4 G li 1089.7
G. ovata ssp. mexicana 1272.9 G. buxifolia 1083.5
G. corvorum 1272.1 G. elliptica 1060.9
G. laurifolia ssp. laurifolia 1262.4 G. gl 0864.7
G. ovata ssp. lindheimeri 1245.2 Griselina scandens 0579.0
G. ovata ssp. ovata 1245.1 G. ruscifolia 0577.7
G. grisea 1244.5 Aucuba japonica 0569.7
G. ovata ssp. goldmanii 12125 Griselina racemosa 0543.0
G. salicifolia 1179.6 Aucuba chinensis 0428.7
G. laurifolia ssp. macrophylla 1163.4 Griselina litoralis 0388.5
G. longifolia 1158.9 Macrocarpum officinalis 0161.2

G. flavescens ssp. flavescens 1114.8

biochemical divergence from the group is demonstrated in G. glaberri-
ma with a GAI of only 864. Indices of the other Garrya species
are larger and show a continuous distribution which makes the gap
of G. glaberrima especially striking.

Some flavonoids are found in only one species, indicating a bio-
chemical divergence. Only 12 taxa show unique flavonoid constituents
and these are mostly members of subgenus Fadyenia (Table 12). Garrya
glaberrima tains 26.6% of unique constituents, again demonstrating

iochemical di g d singularity. A smaller percentage of unique
compounds, 11.7%, is found in G. fadyenii. It is isolated geographically
So the occurrence of unique compounds is not particularly surprising.
The remaining species, probably as the result of adaptation to specific
habitats or environmental conditions, developed only 5.0-8.7% unique
constituents.

Eyde (1964) suggested that Garrya is closely related to Aucuba
and Griselina of the Cornaceae. A phylogenetic relationship between

TABLE 12. UNIQUE FLAVONOID CONSTITUENTS

fi RE UES ee Species Percent

G. buxifolia ue
G. elliptica .

G. flavescens ssp. pallida a
G. salicifolia 3
G. fadyenii ee
G. ovata ssp. ovata co
G. ovata ssp. lindheimeri a
G. longifolia one
G. glaberrima 57
G. laurifolia ssp. laurifolia ee
G. laurifolia ssp. racemosa ag
G. laurifolia ssp. macrophylla .

48 GERALD V. DAHLING

Garryaceae and Cornaceae was suggested (Bentham, 1867), based on
the morphology and the presence of aucubin in Garrya and Aucuba.
Current investigations of flavonoid constituents of Garrya, Aucuba,
and Griselina show only small degrees of similarity. Only herbarium
material of the cornaceous genera was available. Leaf material from
approximately ten individuals of a particular cornaceous species was
ground, extracted, chromatographed, and numbered in the manner
already described. The results indicate a close chemical relationship
between Aucuba japonica and A. chinensis. Griselina ruscifolia and
G. scandens also have similar flavonoid constituents and appear
chemically close. Aucuba and Griselina species share few common
compounds with Macrocarpium officinalis, indicating a more distant
relationship. The emerging pattern suggests a loose but definite
chemical affinity among the cornaceous taxa examined (Figure 5).
The flavonoids of Garrya, in comparison with those of the Cornaceae,
do not reveal a striking chemical affinity. Of the combined total of
135 compounds discovered in Garrya and the Cornaceae, only two
are present in both groups (Figure 5). Macrocarpium does not share
any flavonoid compounds with Garrya and appears more closely allied
with Aucuba and especially with Griselina. Since some flavonoids
are ubiquitous and are found in widely separated plant groups, the
occurrence of only two common constituents in Garrya and the
Cornaceae scarcely suggests a genuine affinity. Also, the PAI’s of
Garrya and members of the Cornaceae are very low (Table 13), and
with less than 1.5% of the total compounds found in both groups,
relationships based on flavonoid constituents are not significant.

FLAVONOID CONSTITUENTS WITHIN THE CORNACEAE
<= GARRYA COMPOUNDS ——>
Aucuba 1 MS I

.
chinensis t ® eve ee ese00

Griselina

litoralis a eo @ 20e seececeee .
Griselina °
racemosa a a e ee eeecee

Griselina .
ruscifolig =s 8 2 ee eee 2 se00 .
Griselina

scandens a a ee e@ eee e e coe

Macrocarpium

officinals

“es ° e eeeecese of
\ 5
0 5 10 I5 )) go 85 90° 95 100 105 ilo 15 120 125 - 439 13
Squares—presence of individual flavonoids in both the Gar

Fic. 5. §
of individual flavonoids in the Cornaceae only. Lower scale indicates flavonoids by numbe

nce
ryaceae and the Cornaceae. Dots—pres€ 4
ee :

LK SN ee

SYSTEMATICS OF GARRYA 49
TABLE 13. COEFFICIENTS OF :
Aucuba Aucuba_ Griselina Griselina Griselina Griselina
Garrya japonica chinensis litoralis racemosa _ ruscifolia scandens
htii 137 .074 .064 .148 .129 .142
laurifolia ssp
racemosa 129 .068 .060 37 121 132
fadyenii 129 .068 060 .137 121 132
ovata ssp
mexicana 129 .068 .060 AS7 AZ] 132
orum 148 .080 .068 .160 137 153
laurifolia ssp
laurifolia 129 .068 .060 ot Zr 132
ovata ss
lindheimeri .129 .068 .060 SZ sal oe
ovata ssp. ovata .129 .068 .060 AGe AZ) aoe
grisea 132 .071 .062 .142 Mi iss" ot
Ovata ssp
oldmanii 121 .064 .057 129 114 125
salicifolia 137 074 064 148 129 142
laurifolia ssp.
macrophylla 148 .080 .068 .160 137 .153
longifol 137 074 064 148 129 142
flavescens ssp
flavescens 129 068 .060 137 121 132
flavescens ssp
Pallida 129 068 .060 137 121 132
fremontii 148 080 068 160 137 153
veatchii 160 086 .074 13 .148 .166
buxifolia 153 083 071 166 142 160
elliptic By ie by 074 .064 .148 129 .142
glaberrima 068 000 074 064 071
*Macrocarpium officinalis is .000 for all the Garrya species above.
WRIGHTII GRISEA
VATA
as aes SALICIFOLIA
BUXIFOLIA a -
- a
FREMONT 1 VEATCHII FADYENII ee
ea
PEMESCENS........ ELLIPTICA CORVORUM LAURIFOLIA
PRE-GARRYA PRE-SADTERIA

ANCESTRAL GARRYA STOCK

Fic. 6. Probable lines of relationship within Garrya.

50 GERALD V. DAHLING

Studies of unique flavonoid constituents of the taxa of the Cornaceae
(Table 14) indicate that they, with the exception of Griselina ruscifolia
and G. scandens, have moderately high values and are well-differen-
tiated. Griselina ruscifolia and G. scandens have the lowest percentages
of unique compounds, 6.2 and 7.6% respectively. In contrast, 64.2
of the flavonoid constituents of Macrocarpium officinalis are unique.
Other taxa of the Cornaceae show intermediate unique compound
percentages ranging from 21.4 to 41.2%.

roup Affinity Indices (Table 11) of cornaceous genera range from
approximately 161 to 579. Griselina litoralis, adapted to a coastal
habitat, appears chemically differentiated from other species of the
genus. A large gap in the distribution of GAI’s separates Garrya from
the cornaceous taxa. If the isolated G. glaberrima is excluded, the
gap is about 500 points. Even when G. glaberrima is included, the
gap is substantial—approximately 300 points. Therefore, the GAI's
also indicate how chemically well-defined and different Garrya is
compared to the cornaceous genera.

Conclusions: 1. The flavonoid constituents are more variable on
an individual level than on a population level and the degree of
flavonoid differences are greater between interspecific than intraspeci-
fic populations.

2. The flavonoid content of Garrya indicates a chemical division
of the genus corresponding closely to the morphologically based
subgenera Garrya and Fadyenia.

3. Garrya species show a high group affinity and form a well-inte-
grated group.

4. Garrya glaberrima appears to be the most chemically differentiat-
ed species of the genus and shows the greatest divergence from the
Garrya mainstream.

5. Garrya species share few flavonoid constituents in common with

TABLE 14. PERCENTAGE OF UNIOUE FLAVONOIN BAS
Total Unique % Unique
Genus Species Compounds _ Compounds | Compounds_
mnie ge
Aucuba japonica 14 3 21.
chinensis 12 3 25.0
Griselina litoralis 16 5 31.2
racemos 12 5 41.6
ruscifolia 16 1 62
en 13 1 iA
Macrocarpium officinalis 14 9 64.2
Garrya laurifolia 17 15 88.3
all species 82 80 97.5

SYSTEMATICS OF GARRYA 51

Aucuba, Griselina, and Macrocarpium of the Cornaceae.

6. As indicated by the Paired and Group Affinity Indices, Aucuba,
Griselina, and Macrocarpium are chemically more closely related to
each other than to Garrya.

EcoLocy

Habitat. Garrya is a highland genus, found in both the chaparral
and coniferous forests above the lowland deserts. It generally grows
in semiarid regions, although some species have adapted to coastal
or near-coastal conditions. Garrya ellpitica, occurring from central
Oregon to Santa Cruz Island, California, is one such species and
is frequently found within a few meters of the sea. In the southern
part of the range, populations also occur inland and at higher elevations.
This is especially true near San Francisco where several collections
have been made on Mt. Tamaulpais and the hills surrounding Berkeley
and other nearby places. In the northern part of the range, G. elliptica
is almost always confined to coastal positions. Garrya fremonitii,
ranging from southern Washington to central California, is charac-
teristic of the Sierra Nevada and Cascade ranges where moisture is
plentiful, but it also extends (although less commonly) into the arid
transition zone at lower elevations. Garrya veatchii is a typical member
of the chaparral community and is quite tolerant of arid and semiarid
Conditions. It is frequently found in extremely sandy soil in the
transition zone on the lower mountain slopes along rivers, dry runs,
and canyons of southern California and Baja California. A population
also exists on Cedros Island, located about twenty miles west re)
mainland Baja California. From the coastal ranges and the Sierra
Nevada of central California, G. flavescens is distributed southward
to Baja California and eastward to Nevada, Utah, and Arizona. Garrya
flavescens ssp. flavescens is found both in semimesic canyons and
on mountain slopes. Although sometimes extending into the transition
zone, it is less characteristic of the more arid canyons and slopes
than G. flavescens ssp. pallida. Garrya flavescens ssp. congdonii is
also Upper Sonoran but shows less tolerance for aridity than does
G. flavescens ssp. pallida which is found predominantly in the arid
transition zone, Garrya buxifolia is a chaparral component of the
transition zone but is also found in disturbed areas and at higher
Moist locations in association with Pinus. Ithas a restricted distribution
and is confined to a six county area of northern California and southern
Oregon. Garrya corvorum is a limestone endemic of the high Cordil-
eran mountain forests of central Guatemala and is isolated from the
other species of subgenus Garrya by over two thousand age

Species within subgenus Fadyenia show a southern distribution

52 GERALD V. DAHLING

and are adapted to a variety of highland soils and conditions. Garrya
wrightii ranges from Arizona to west Texas and south throughout
northern Mexico, growing on arid, open, and rocky slopes of the
Upper Sonoran and transition zones. It is extremely tolerant of arid
conditions and prospers in environments where most other plants
cannot. Moreover, it shows a greater tolerance for aridity than the
other Garrya taxa, often growing between rocky crevices and boulders
of the lower mountains. Garrya ovata is adapted to limestone outcrop-
pings and is common in the mountains above the lowland deserts
in Arizona, New Mexico, Texas, and northern Mexico, with the
distributional center in the Texas-Chihuahua-Nuevo Leon area. Garrya
glaberrima, an isolated species with a limited distribution, is found
in a few scattered mountain localities of southern Coahuila, Nuevo
Leon, and Tamaulipas, growing between the lowland deserts and
the highland coniferous forests. It is found at the edge of the shrub
zone where conditions are not wholly arid. Garrya laurifolia is known
from the Sierra Madre Oriental, Sierra Madre Occidental, the volcanic
slopes surrounding Mexico City, and the mountain systems of southern
Mexico, Guatemala, Costa Rica, and Panama. Although the range

inhabits the steep, moist banks of ditches, gullies, and ravines. Garry4
longifolia, a closely related species of southern and southwesterm
Mexico, ranges from Jalisco south to the mountains of Oaxaca.

but sometimes it
slopes above strea

us, generally adapted to high

elevations, Subgenus Garrya includes species growing at both the

corvorum, with elevations of 9900-11700 feet. Garrya fremontii 15

found at 500-9000 feet, the most common interval being 3400-5

SYSTEMATICS OF GARRYA 53

feet. Garrya veatchii also shows a wide altitudinal range (750-8550
feet) with the most frequent elevation being approximately 3500 feet.
Garrya buxifolia occurs at lower elevations, ranging from 200-7000
feet, with 2000 feet being the most common for the species. Divergence
within G. flavescens appears related to both altitude and aridity. Garrya
flavescens ssp. flavescens grows from 2600-9000 feet while G. flave-
scens ssp. pallida has adapted to the drier, lower elevations, ranging
from 1500-7753 feet.

Within subgenus Fadyenia, the highest elevation is attained by
Garrya laurifolia ssp. quichensis with a range of 4500-11700 feet.
Garrya laurifolia ssp. racemosa closely follows, growing at altitudes
of 3600-10000 feet, the most common being approximately 7500 feet.
Garrya laurifolia ssp. laurifolia and G. laurifolia ssp. macrophylla
grow at lower elevations, ranging from 2000-9000 feet and 3500-8850
feet respectively. Garrya glaberrima also grows at high elevations,
ranging from 4880-9000 feet. Garrya longifolia shows an altitudinal
range of 4200-8700 feet, with the most common interval being
6000-7000 feet. The remaining species within subgenus Fadyenia
occur at lower elevations: Garrya fadyenii at 2000-8130 feet; CG.
salicifolia at 5100-6000 feet; G. ovata ssp. lindheimeri at 2000-6600
feet; and G. ovata ssp. ovata, G. ovata ssp. goldmanii, and G. ovata
Ssp. mexicana ranging from 4800-8000 feet, 4550-7800 feet, and
4000-8400 feet respectively. Garrya wrightii, which grows in arid
areas, is found at elevations of only 3000 feet. However, the maximum
elevation is about 7000 feet, with 5000-6000 feet being the most
common interval. The closely related G. grisea requires more moisture
and occurs at higher altitudes, ranging from 4500-7950 feet.

ETHNOBOTANY

Garrya elliptica, G. fremontii, and G. veatchii are used as ornamental
Shrubs in many areas throughout the world (Howes, 1974; Usher,
1974). The graceful, elongated catkins and the stately contour of the
Shrubs give it an aesthetic quality. ~

Garrya laurifolia is also used medicinally as an antidiarrhetic
throughout rural Mexico (Martinez, 1959; Armendariz, 1895; Garcia,
1886). An extract is prepared from the stem, diluted with alcohol,
and ingested. The treatment is said to yield quick and positive results,
but since experiments have shown that Garrya extracts are extremely
toxic, the antidiarrheti ties should be held suspect. Stem extracts
of.G. laurifolia were prepared and injected into the peritoneium or
veins of dogs in an effort to determine pharmacological properties
(Martinez, 1959; Olguin Hermida, 1932). At first the dogs became
highly stimulated, suffering convulsions and general muscular con-

54 GERALD V. DAHLING

tractions. Depending upon the dosage, a gradual paralysis of the central
nervous system develops, resulting in complete respiratory collapse
and death. The same physiological aberrations were induced when
pure alkaloid extracts were injected. Additional pharmacological stu-
dies of Garrya are also reported by Langman (1964).

According to legend, bark extracts of Garrya species were used
by western American Indians to treat fever. It was this legend that
led Oneto (1946) to begin characterizing Garrya alkaloids chemically
for preliminary tests as possible antimalarials.

Informal experiments in Mexico have investigated the effects of
smoking Garrya. The stems, leaves, and fruit material of Garrya
laurifolia and G. longifolia were ground into a coarse powder. Using
ordinary pipes, five 18-22 year old male volunteers slowly smoked
small quantities of the plant powder over a period of five hours.
The volunteers recorded their observations. These, along with their
feelings, showed a general similarity and were discussed after the
experiment was completed.

the result of Garrya or autosuggestion remains questionable. All the
volunteers became very relaxed and tranquil as Garrya was smoked.

that gradually gave way to a dull pain in the back of the head and
neck. In addition, muscle twitches and minor pain in the hands, feet,
and stomach were reported. The pain, resulting from muscle contrac:
tions which were mild and of short duration subsided and disappeare

*

and eventually
The volunteers stated that smoking Garrya was a pleasant experience,
€ noted that two of them experienced miscellaneous
porary loss of sexual appetite following the experiment
No hallucinations were reported. While acknowledging the many side

effects, Garrya still may hold potential for the future as a tranquilizing
agent in medicine,

PALEOBOTANY-PHYLOGENY

Paleobotanical information
genetic relationships and prop

about Garrya is scarce, making phylo- :
osed historical reconstructions difficult. at

isch lamanseceetenetiaeeniagenabcammna ovata salamat nial sipaia aaa pent eee sepegi

;
t
'

SYSTEMATICS OF GARRYA 55

The Old World fossils, which have recently been discovered and
assigned to Garrya, are of interest. Garrya tsushimense, from the lower
Tertiary, was discovered near Kushi and Saraura of the Tsushima
Islands of Japan (Matsuo, 1971). However, only leaf material was
discovered, so no firm conclusions as to the authenticity of the putative
Garrya fossils can be reached. Only the reproductive structures can
provide definite answers to the question of fossil affinity and identi-
fication, and, if they are eventually discovered, the argument for an
early evolutionary relationship between Garrya and Aucuba will be
strengthened.

It seems apparent that, in the past, Garrya had a greater distribution
than it does today (Axelrod, 1944, 1950, 1958, 1964; Wolfe & Barghoorn,
1960). Garrya axelrodi (Wolfe, 1964) was discovered in Miocene
deposits of southwest Nevada while G. idahoensis (Axelrod, 1964)
of the middle Miocene, was found in southern Idaho. The latter is
well outside the range of extant Garrya but nevertheless corresponds
to modern members in vegetative features. Unfortunately a subgeneric
assignment is not possible because the reproductive structures are
lacking. A similar situation is found in G. masoni (Dorf, 1930), a
Pliocene species found in Sonoma County, California.

Fossil leaves, seeds, and an inflorescence of Garrya have been
discovered in Pleistocene to Recent deposits at Willow Creek, Carpin-
teria, and Tomales Bay, California (Chaney & Mason, 1934, 1934a;
Mason, 1934). The fossils have been identified as G. elliptica because
they are similar to it in appearance and were found within its range.
Garrya pollen has also been identified in Pleistocene to Recent deposits
in Arizona (Gray 1960, 1961). Apparently the pollen is resistant to
decay and is easily identified.

No paleoevolutionary conclusions concerning Garrya can be reached
because of the incomplete fossil record. The fossils from the early
Tertiary are dubious. The genus probably arose in North America
during the middle Tertiary and underwent rapid evolution as a result
of aridity and climatic change. However, the question of area of origin
remains unsettled unless verified fossils of Garrya are found in the
Old World. For the present, the evidence is strongest for a New World
Origin,

On the basis of morphology, and the evident trends of reduction,
itis probable that subgenus Fadyenia most closely reflects the primitive
Condition. Members of subgenus Garrya have inflorescences which
are more reduced and floral bracts which show a greater fusion than
those of subgenus Fadyenia. Also, most of the members of subgenus
Garrya are shrubs and are distinctly less tree-like than members of
Subgenus F adyenia. The presence of a rudimentary perianth in the
emale flowers of subgenus Garrya is anomalous but absent in
Subgenus Fadyenia. However, this may not be significant if allowances

56 GERALD V. DAHLING

are made for the fact that the evolution of different features proceeds
at different rates, permitting the various organs to reach different
levels of specialization. On the whole, most of the characters associated
with subgenus Garrya appear to be more specialized than the corre-
sponding ones of subgenus Fadyenia. Because the morphological and
phytochemical differences between the subgenera are pronounced,
an early evolutionary divergence within the genus is indicated (Figure
6).

ACKNOWLEDGMENTS

I wish to express my appreciation for the time and continuous efforts of my sponsor,
Professor Reed C. Rollins, Asa Gray Professor of Systematic Botany and Director
of the Gray Herbarium, Harvard University. He has always been available for informal,

ort etings and has been helpful in other ways too numerous to mention. The
following faculty, graduate students, and staff of Harvard University have also been

Barghoorn, Otto T. Solbrig, Carroll E. Wood, Jr., and Rolla M. Tryon; Drs. Umesh

Banerjee, Ihsan Al-Shehbaz, Elizabeth Shaw, Bernice Schubert, Alice Tryon, William

Gillis, Lily Perry, Lorin Nevling, James Rodman and the late Thomas Lockwood;

Ms. Kathryn Roby, Mr. Michael Canoso, Ms. Beverly Bigwood and my helpful friends

Srs. Jesus Blase and Jose Luis of Monterrey, Mexico. In addition, my thanks to Professors

Arturo Gomez-Pompa and Ramon Riba from the University of Mexico, who helped
n

Dr. Lorin Nevling, Principal Investigator, NSF grant GB20267X; and the Society of
Sigma Xi.

TAXONOMIC TREATMENT
GARRYACEAE

Dioecious evergreen trees and shrubs, bushy, 2-40 feet high. Crown sprouts numerous,
often well-developed and indistinguishable from the main stem axis. Tap root long,
with many secon ae Young stems with ridges and grooves, becoming rounded with
age, glabrous and densely pideag? cabg aust jsatiarinag! sometimes hispid trichomes,

n, re
Leaves simple, petiolate, Pabeee "“dipbak ale “Ach eolate, sometim b
rarely oblong, stiff, coriaceous. Petioles 2-30 mm rae) opposite petioles eonnste
at the base and adnate to the stem. Leaf blades 1. 0- 20.0 cm. long and 1.0-10.2 cm.
wide; apices acute, acuminate, mucronate, or obtuse; bases frequently attenuate,

counterclockwise orientation, Staminate inflorescences catkin- like oF racemose, one

ing or unbranching, metime

Seitlary and solitary, lax or compact, densely flowered with short or ous ‘acne
taminate floral bracts connate at the base and laterally forming cup-like structures,

apices straight or recurved, bases typically cuneate, obtuse i runcat
Staminate flowers solita ary or ternate, in the axils of the partially enveloping bracts
Pedicels abbreviated or elongated. Perianth segments four, oblong-ovate, connate at
the apices form a semi-enclosed chamber with four openings, variably pubescent
on the abaxial petra typically glabrous on the — surface, white, yellow-green,
or red-brown. Stamens our, free, short, opposite t amber openings. hers oval
'o oblong, d, introrse, opening by lon aiiudinal slits. Filaments abbreviated.
issemi-

f ax mpact

ete with elongat ted conspicuous aia Pistillate floral bracts
€ piae. and laterally bg cup- -like structaron. se ag only at t the ase, tatty
vate ola foliac

lance

in the axils of bracts. Pedicels
gore unilocular, subovoid
g var a

t

= » fleshy and variously divergent , rarely

parietal placentas, anatropous it a modified berry, se to ovoid, Gibieke

°F variously pubescent, compact or lax on ash inratsences, green and fles i at
irr: ar,

aturity becoming dark blue, black or whitish-g

gu
bi si Eoiwo, rarely three, globose to ov ocellptial “with a silver = —
ically dar

J k blue-black. Embryo min
dedi in massive endosperm. Seeds paste by weathering and fragmentation of sauna

The Garryaceae is a New World family of highland areas and consists

57

58 GERALD V. DAHLING

of a single genus with two subgenera. Both the genus, which was
discovered and named by David Douglas in 1826, and the family
were described by John Lindley in 1834. Subgenus Garrya consists
of six species and is generally confined to the western United States
and northern Baja California with the exception of G. corvorum, a

central Guatemalan endemic. Subgenus Fadyenia contains eight spe-
cies and is found in the southwestern United States, throughout Mexico
and Central America, and in the Greater Antilles.

GARRYA DOUGLAS EX LINDLEY
pence Dougl. ex Lindl., Bot. Reg. 20:1686. 1834. type species: Garrya elliptica Dougl.
x Lindl.

Fadguaa Endl., Gen. Suppl. IV:38. 1847. TYPE SPECIES: Fadyenia Hookeri Endl., based
on Garrya fadyenii Hoo
Characters of the family.

KEY TO THE SPECIES

. Inflorescences catkin-like and not branching. Floral bracts connate at the bas and
beers forming cup-like structures which bear ternate flowers in the axil. Pistillatt
inflorescences compact, bracts not leaf-like, often imbricate Seen: the internodes
when in fruit: Mostly shrubs 22079) 6S rns oe Pe ubgenus Garry4.
B. Lower leaf a Riana! a with strongly curled, woolly, or villous

gra Ova ensely

. Leaves ovate to cre aly sclliptieal, pogo from 3.1-12.4 cm. long and 1.5-6.

cm. wide, coriaceous and dark green in color. Margin strongly undulate, rarely
plane. Leaf peosenetog densely cove eee ith a tomentum of curly, intertw ined
trichomes, the uppersides naan and lustrous. Masts pie shrubs but
somtimes arboreal mland 2). 200) 8) BUN eS G. ne

C. Leaves ovate to ovate-elliptical Lege from 2.3-8.3 cm. long an 7 1.0-3.2 cm
wide, strongly coriaceous, light green. Margins plane. Leaf unde — densell
covered by a mat of short, sti: interwoven bright, white trichom pper
curiuees covered ia — curly trichomes. Mostly shrubs or rarely ” gnarled
trées in sandy arid vepipus; 7 oo oe eG G. veatchit.

B. Lower leaf an Ae dates pee or densely — 3 sii or somewha

Piet at the apex, typically aetna at maturity. Female inflorescences semi
y shrubs in highland aveas. 2. oS.

D. asec leaf surfaces more or al covered with straight, wavy, or silky vale
appressed trichomes. Rare individuals may show both glabrous leaf ’ ndersilll
and silky pubescent ovari
E. Lower leaf surfaces den sely covered with long, silky, pnt a

grape trichomes. Leaves oblong to oblong-elliptical, 2.7-4.6 ong an
1.5 em. wide. Trees confined to limestone areas in central Guatetala as
ait. —-

E. Lower leaf surfaces more or less covered with straight or w wavy u
ed aon trichomes. In glabrous leaf forms, ovaries are silky or ‘moderated
pu

F. Ovaries glabrous. Lower leaf surfaces densely covered with silky apPt

silver-gray trichomes. Leaves thick, coriaceous, usually ghee? ad 4

2.3-6.5 cm. long and 1.0-3.0 cm. wide. Mostly low shrubs . . -

SYSTEMATICS OF GARRYA 59

aye a bc Mak sly: azn cass eb acne herd Mpa as a OR a eater ee Cee 5. G. buxifolia.
F. Ovaries silky pubescent. Lower leaf Sagi glabrous or covered with

straight or wavy upwardly appressed trich s. Leaves coriaceous, Laer

above, white and shiny below, oblong to siliptical 3.3-9.5 cm. long, 1.3-4.4

m. le. Shrubs or:small trees’) os) 0, ee ee a G. Aacescens

Aa Inflorescences Facemose, branching at the base. F oral ‘Breet connate at the base,
PI bearing solitary

flowers in, the axils. Pistillate inflorescences loose and spreading, bracts distinctly

leaf-like, internodes prominent during fruiting. Trees and shrubs...........
Bea oss Dies Saheees Fadyenia.

G. Plants entirely glabrous or oe with a sparse row of trichomes along the edge
of the pistillate floral bracts. Leaves 6.0-9.8 cm. long, 2.2-4.0 cm. Pelee: Shrubs
and small trees at high clevat tions. . 2.22 ee eee fe

W

glaberrima.
its ung stems, leaves, and par str Lower
ag se saci tie densely covered with t richomes when youn
H. Leaves elliptical to ovate-elliptical, mostly less than twice as ; long as broad.
gl

covered with short upwardly appressed or curly, w phi or matted trichomes.

Yellow-green, green, or gray-green shrubs and small tre

I. xn ya tt _ curly, woolly, and Sed paiieases often forming

ish-gra t. Leaves plane or markedly undulate. Ovaries glabrous.
Pehla floral roves REIS small, delaucotecdea Small shrubs and trees,
frequently ‘an. limestone. 2: Ps a re G.

. Leaves glabrous or densely covered with short ciphbad pape trichom
Leaves plane. Ovary sparsely sericeous becoming glabrous with age. Pi sitions
floral batts wee small and linear to abl ap-ottlonl fellow: -green to
gray- abit n shru ostly in miei regions.

ure leaves escnitially aepreab or oan the lower surface covered with

et upwardly appressed trichomes. Leaf surfaces yellow, green, or rare ely
light gray on the underside. Leman pistillate floral bracts pace
foliaceous. Shrubs to small trees in semiarid regions. .. . . 9. ightii.
. Mature leaves covered by short upwardly appressed cakes ly th
surfaces. Leaves uniformly gray-green or rarely bright green. cnibate floral

—

c

transition of northern Baja California ....--.---->- . grisea.

H. Leaves meeps ey Celine pobbna: oblanceolate lanceolate, a 9-4 times

as long as broad. Leaf under surfaces covered with oe trichomes ie n young

but frequently glabrous with age. Leaves dark somes o yellow with lustrous,
ua

K. Bracts of the infructescences linear-oblong ic hea tancaotie “a re
than the foliage leaves. Upper leaf surfaces ieee glabrous. aa leaf surfaces
oi

appre
lif
Leaves lanceolate to bln Wanceclats. Mostly shrubs from ¥ i

io tg eee Sw OR Re eee ee
Res ee ee © oe Pee em ee ek oe ce ae ae

bove and

found in limestone areas of the Greater Antilles... - - : ste arven
M. Leaves oblong to oblong-elliptical, pilose to villous or spars: ely so secre

frequently sates undulate, yellow-green, green, and reticulate. 7 a-

minate inflor nces lax with few flowers a nd promin a internodes.

Staminate pits a. often shorter than the flowers. Most al aa on mm
K. Bracts of the infructescences expanded and often as large as ig ——

Lower leaf surface covered with curly, woolly trichomes —— slabs

with age. Leaves elliptical, oe or oblanceolate. Upper — — a

and lustrous. Shrubs to large trees. .----* °°" :

Poke eee a ee hn
oo ea ee es He a ee Pe eS Be Re

60 GERALD V. DAHLING

SUBGENUS GARRYA

Inflorescences catkin-like and unbranched. Floral bracts connate at the base and
laterally forming cup-like structures subtending the flowers. Female inflorescences
compact, bracts often imbricat d not foli internodes short and inconspicuous.
Ovaries bear minute partially adnate perianth appendages alternating with and near
the base of the two styles. Styles elongate, slender, terete, often divergent and reflexed,
sometimes erect. Mostly shrubs.

TYPE SPECIES. Garrya elliptica Doug]. ex Lindl., Bot. Reg. 20:1686. 1834.

1. Garrya elliptica Douglas ex Lindley

Garrya elliptica Doug]. ex Lindl., Bot. Reg. 20:1686. 1834. isoryre: plentiful in rocky
situations on the seashore near the confluence on the Columbia, on the south bow
and in California more abundant, Douglas 1838 (cu).

Low shrubs or small trees, 3-25 feet high. Crown sprouts numerous. Stems short,
villous to pilose, becoming glabrate with age; reddish-brown, black, or greenish-
white. Decussate branching. Stipules absent. Leaves simple, petiolate, decussate, ellip-
tical oval, or oval-lanceolate, thick, coriaceous, persistent. Petioles (0.2-)0.4-1.0(-1.3)

. long; opposite petioles connate at the base, adnate to the stem. Leaf blades
(2.8-)4.0-9.5(-11.1) cm. long, (1.5-)2.0-4.0(-6.7) cm. wide; apices acuminate to
r

uc ate to broadly obtuse; margi ntire, strongly undulate oF
revolute; glabrous, lustrous above, lanate below with curly trichomes; papillae devel
ped; ri major veins visible above, prominent, in relief below. Staminate

minal, parietal placentas, dark blue to black

Flowering from early December to late F ebruary at elevations of
10-2750 feet, mainly along the coast but sometimes inland from cent
Oregon to near Ventura County, California (Map 1).

REPRESENTATIVE SPECIMENS. California. Alam Co.: Temesial Canyon, Biol

etti : Me
d
2/18/92 (ps); Alameda, Skolton 3/15/03 (cu). Contra Costa Co.: Oakland and Berkeley

pe ted 1420 (cu,uc); Mt. Diablo, NW corner of basin of the NE fork of sya
reek, Bowerman 1082 (uc); head of San Leandro Creek, Oakland hills, Duran Ft

28 (ps)
(cH); 3 mi. S of Crese nt Cit
at mouth of Matole River, Benson 1657 (POM) i f Trini

: , ; along Hwy. 1, about 10 mi. N 0
Dahling 1229 (cu); mouth of Luffenholtz Creek near Trinidad, Kildare 2550

ity, Wiggins 5883 (ps,pom,uc). Humboldt Co.: sea fas :

SYSTEMATICS OF GARRYA 61

Trinidad, on cliffs near the ocean, Parks 4287 (cu,uc); Patrick’s Point State Park,
Thorne 19021 (rsa); Blue Slide on Van Duzen River, Tracy 16203 (cu,uc); Big Lagoon,
Tracy 17441 (ny,uc,wrc). Marin Co.: San Francisco hills, Bolander 1863 (ny); Tomales

(rss). Mendocino Co.: along Hwy. 1 about 1 mi. S of Little River State Beach, Dahling
050 (cH); sand dunes, 3 mi. N of Fort Bragg, Mason 5403 (uc); 1-1/2 to 2 mi. W
of Haven’s Neck, Thorne & Everett 34235 (cH); 1.2 mi. S of Garcia River Bridge
on Coast Rd., Wolf & Johnson 6177 (uc,wtc). Monterey Co.: Pico Blanco, near summit,
Davy 7333 (uc); Seaside, Elmer 4933 (onr,uc); NE of Bouchers Gap Camp, Santa
Lucia Mts., Haasis 47.59 (cas); N of Pine Canyon, W side of Salinas Valley near
Spreckels, Hoover 1183 (cas); 5 mi. E of Monterey, Rose 44301 (cu,rsa); SW slope
of ridge, W of Big Sur Camp, Steward 198 (uc). Napa Co.: near Four Corner, Adam
& Eve Ridge, Howell Mt., Jepson 399 (uc). San Luis Obispo Co.: 1-1/2 mi. SE of
the mouth of Little Falls Canyon, Bolt 565 (uc); summit rd. along S fork of Las
Tablas Creek on Klau-Cambria Rd., Santa Lucia Mts., Ferris 9789 (ps,cH); Price Canyon,
Hoover 7268 (cas); 2.3 mi. NE of Slide Hill, Arroyo Grande, Lee 386 (uc). San Mateo

0.: NE side of Coal Mine Ridge, Barry 222 (ps); near La Honda, Benson 833 (pom,wrc);
Los Tancos Woods, Demaree 7312 (GH,Ny,wrc); San Mateo Ravine by the road to Santa
Cruz Peninsula, Dudley 005 (cu); Portola, Elmer 4469 (Ny,orr,UC); up hill from Trillium
Gulch, San Bruno Mt., McClintock 519683 (cas); 3-1/2 mi. N of La Honda, Santa
Cruz, Sindel 336 (uc). Santa Barbara Co.: N side of Devil's Backbone Peak, Cueva
Valdaze, Santa Cruz Island, Beeks 3 (nsa); main N ridge E of Picacho Diablo, Santa
Cruz Island, William 55 (uc). Santa Clara Co.: Saratoga, Davy 388 (uc); Black Mt.,
Santa Cruz Mts., Dudley 34 (GH,NY,ORE,WTC); canyon of Permanente Creek, Santa Cruz
Peninsula, Dudley 116708 (rsa); foothills W of Los Gatos, Heller 7220 (cu,vc). Santa

uz C i

t i : ‘
6/12/84 (NY,oRE,wrc); Huntas Creek near Pistal River, Kildare 6/65 (ps). Douglas Co.:
Douglas, Jones 213 (cu); S of Florence on Hwy. 101, Steward (DS,GH
Co.: Tahkinitch Creek, Detling 2406 (ork,wTc); Siltcoos Outlet, Eastwood 1524 ve
1525 (cu); 1 mi. N of Cape Perpetua, Peck 10626 (cx); along Hwy. 101, 1 mi.
of Sea Lion, Steward 7282 (ps,Ny).

Garrya elliptica is frequently found growing on the sandy, shrub-
covered hills and ridges along the seacoast from central Oregon .
Ventura County, California. Although it is best adapted to the humi
Coastal conditions, it has the capacity for growth at higher elevations
inland. There its habit is more arboreal, contrasting with the shrubby
habit found when it grows on the coast. nie

Undulate and iclbu leaf margins are typical of Garrya elliptica,
but the leaves of some individuals or populations are more or less
Plane. The degree of marginal leaf undulation is variable a
Garrya and is not particularly significant systematically. Arborea
Specimens from Tomales State Park, California, show nearly plane
Margins as do specimens from Santa Cruz Island. This copinantte
is sporadic but not infrequent, and should not be given systematic
Tecognition.

62 GERALD V. DAHLING |

Garrya elliptica sometimes resembles other Garrya species. For
example, in the southern and more arid parts of its range, G. elliptica
has rather small leaves and resembles G. veatchii. Populations of
G. flavescens ssp. congdonii superficially resembling G. elliptica have
been discovered at higher elevations in Glenn, Lake, and Tehama
Counties. Members of these populations approach G. elliptica in
appearance because their leaves are weakly or sometimes strongly
undulate with a covering of woolly and somewhat wavy trichomes —
on the underside. Although it is tempting to suggest that these —
populations of G. flavescens ssp. congdonii are inland forms of
subspecies of G. elliptica, this is not well-supported when the entire
distribution and variation patterns of both G. elliptica and G. flavescens
are examined. Therefore these inland populations are best excluded
from G. elliptica and referred to G. flavescens.

7

€

2. Garrya veatchii Kellogg

/
|
|
|
Garrya veatchii Kell., Proc. Calif. Acad. V:40. 1873.“HoLoryeE: Cedros Island, Veatch |
369 (CAS). ISOTYPE: GH. |
2) Garrya veatchii var. undulata Eastw., Bot. Gaz. 36:458. 1903.“ HoLoryPE: Mt. Lové, |
u cho Mt., Los Angeles Co., California, Grant 568 (CAS). it
ee) Garrya flavescens var. palmeri Wats., Bot. Calif. 1:276. 1876.”HoLoTYPE: south pa |
of San Diego County, California, 1875, Palmer 117 (GH). |
|

ort, curly, white trichomes, ee |
; lower surface with semiappressed, short, interwoven, bright graye |
trichomes; papillae well-developed; midrib and major leaf veins visible above, hide . |
the midrib. Staminate inflorescences catkin-like, unbranched, a
at tips of branchlets, or axillary and solitary, densely flowered, Pe |

nate rve
opposite bracts connate over half their lengths. Flow
Pedicels short, not exceedi
m.

. > .

rect, co inate
. Pisti floral bracts 5 mm. long, truncate, apex ee half
t to floccose; opposite bracts connate for ove’

Peennten MnretsneseOl Care hee ee a

en Lf, Distribution of Garrya: 1, G. elliptica, 2, G. flavescens ssp. congdonii (dots); G. buxifolia ar (block
* me veatchii. 4, G. fremontii (dots); G. flavescens ssp. flavescens (white stars); G. flavescens ssp. pall
stars).

SYSTEMATICS OF GARRYA

64 GERALD V. DAHLING

their lengths. Flowers ternate in axils of bracts. Pedicels minute. Ovaries bicarpellate,
rarely tricarpellate, unilocular, subglobose, subsessile, inferior, sometimes with adnate
paired perianth remnants alternating with and near the ba se of the sshd with long,

maturity brittle. Dehiscence irregular. Seeds two, or rarely three, globose to oval,
subterminal, parietal placentas, dark ale to black.

Flowering from January to early May at elevations of 750-8550
feet, mainly in southern California from San Luis Obispo to San
Bernardino county and southward into Baja California (Map 3).

REPRESENTATIVE SPECIMENS. Mexico. ‘Baja California: San Pedro sg Brandegee
173278 a bes slope of vee el erage Moran 10248 (ps,uc); NW s of Cedros
Mt., Cedro at Moran 10649 (rsa,sp); Cerro Blanco, Sierra San gad Martir,
Moam 10945 oe UC); Linke Agua sae ie N of Cerro Chato, Sierra San Pedro Martir,
Moran 11098 (ps,sp); Canada el Duranzo, Moran 11383 (ps,sp,uc); W peak of Cerro
Santa Marta, Sierra San Borja, Moran 13104 (sp,uc); 3 mi. S = Heche Sierra
Juarez, Moran 14916 (TEx,uc); summit of Cerro Pinon, 3 mi. El Alamo, Moran
17649 (sp); E slope of Cerro Bola, Moran 17806 (sp); uppermost ae fork of Guadalupe
Canyon, about 8 m mi. E of Lagun arte Olmsted 4860 (rsa); hillside on bi side
of summit, 6-8 W of Ojos Negros Rancho, Wiggins & Gillespie 4
(DS,GH,MICH,MO,US): California. Inyo Co.: ridge po junction of the two forks rot 1
Creek, 8-1/2 mi. WSW of Lone Pine, Olmsted 609 (rsa). Los Angeles Co.

Pine Canyon, San Gabriel Mts., Abrams & McGregor 671 (ps,cH); San Antonio Casal
near Claremont, Baker 3651 (cu); foothills of San Gabriel Mts. above Monrovia, Beach
21874 (cas); trail between Haines Canyon and Mt. Lukens, San Gabriel Mts., Howell
3342 (CAS,GH); Sg Antonio Mts., Parish 11975 (uc); Calabasas, Saddle Peak, Santa
Monica Mts., St. John 79 (uc); Clear Creek branch of Big Tujunga Canyon, Thorne

40612 (cas, ity Monterey Co.: ail to Cone Peak, Santa Lucia Mts., Munz 20932
(RSA). Riverside Co.: Hemet to cance San Sactaito Arai Balls 20972 (RSA); across
the Bie from Santa Rosa Springs, Santa Rosa Mts., DeBuhr 529 (uc); P dereh to Pines
hwy., San Jacinto Mts., McKelvey 5049 (cu,uc); Santa Rosa Plateau, Thorn 39331
sexy, ere Lodge, mouth of San Jacinto Canyon, Wolf 1490 (rsa). San Bern rnardin0
Co.: Creek Canyon, Abrams 2805 (cu,wrc); S of Cajon Pass, Bell 1195 (uc);
Lytle ithe area, Dahling 354 (cu); near Cajon summit and nearby slopes, Dahling

920 (cu); Lone Pine Canyon, about 2-3 mi. W of Hwy. 138, Dahling 1300 (cH); San
Beract dino, Howell 2488 (cu); between San Bernardino and summit to Victorville,
McMinn 3141 (uc); San Bernardino Mts., Parish 10793 (cu); San Diego Co.: Warners

SE of De : _ wood 9435 (G
scanso spores Gander 587 (sp,uc); Campo, Hall 171022 (uc); La Posta
beyond Beichma Springs, Harvey 13014 (sp); Cuyamaca, McGinty Mt., Jensen 100
(uc); Mt. Springs, Parish 10824 4 (cH H); Bankhead Springs, 1 mi. E of Mistletoe ae
i due

Ra, Dohling 944 (cu); 4 mi. W of Paso Robles on Peachy Canyon Rd., Ferris 9798

pein ate ridge, NW of Cuesta Pass, Thorne 31589 (cu,rsa); San Luis Obisp®
orne verett 32140 (cu). Santa Barbar. Lomp0%

Axelrod 158 (uc); El Camino Cielo R . shoe oe — Se cars Lise um

7 (uc); Nojoqui Park, Gaviota Pies Region, McMinn 4349 (uc). Ventur

tw)
Pipe and Sespe Creeks, just S of Oat Mt., Ferris 10641 (cH,RsA); Matilija Canyo™

Pollard 75074 ( (sp,uc); Ventura, Simontacchi 10856 | (GH

Garrya veatchii is common to the es of southern California |

SYSTEMATICS OF GARRYA 65

and Baja California. For the most part it is a natural and easily
recognized taxon, distinct from the other Garrya species. Minor
intergrading with G. flavescens and G. elliptica does occur, but not
to a sufficient degree to support a change in status.

The holotype of Garrya veatchii was collected by Veatch on Cedros
Island, several miles off Baja California, and is somewhat atypical
of the mainland and California material in terms of leaf and pubescence
characteristics. This variation probably resulted from differences in
the gene frequencies which would be expected to occur in small
and relatively isolated island populations.

Eastwood (1903) recognized Garrya veatchii var. palmeri and G.
veatchii var. undulata, basing her conclusions on the differences in
leaf undulation, leaf size, pedicel length, and the internode length
of the infructescences. Her treatment is not followed here since
differences in the degree of marginal leaf undulation are common
in Garrya with only a few taxa uniformly showing plane or undulate
leaves. Moreover the variation in leaf size is dependent on fluctuations
in soil moisture and other environmental influences which may cause
considerable infraspecific variation. Also, the pedicel length, as with
undulate leaf margins, is a highly plastic character showing extensive
variation throughout Garrya. In fact, the pedicel length differs not
only between populations but also on the same inflorescences or
infructescences. Since no meaningful and stable pattern to this type
of variation has been found, no infraspecific taxa within G. vgatchii

ave been recognized in the present treatment.

3. Garrya fremontii Torrey

Garrya fremontii Torr., Pac. R. Rep. 4:136. 1857.“HOLoTYPE: upper Sacramento above
the canyon, California, 1846, Fremont 369 (NY). ISOTYPES: GH, CAS. : k
Garrya fremontii var. laxa Eastw., Bot. Gaz. 36:461. 1903. HoLoTYPE: Twin Lakes,

Canyon Cr., Trinity Co., California, 1901, Eastwood 555 (cas). : :
Garrya rigida Eastw., Bot. Gaz. 36:461. 1903. HOLOTYPE: Mt. Tamalpais, Marin County,
California, 1856, Eastwood 562 (cas). ISOTYPE: CAS.

Low shrubs or bush-like trees, 3-18 feet high. Crown sprouts well-developed. neta
Stems glabrate or with whitish-gray, silky, appressed pubescence, becoming glabrate
With age, green, gray, or reddish-brown. Branc ; de
‘imple, petiolate, decussate, oblong to ovate, elliptical, coriaceous, d
Petioles (0.5-)0.8-1.3(-1.6) cm. in length; opposite petioles connate at the base an
"dnate to the stem. Leaf blades (2.3-)4.0-8.0(-9.8) em.

°m. wide; apices m , acuminate or acute;

ent with recurved apical
r ciliate; opposite bracts

66 GERALD V. DAHLING

hort,
cues for half their lengths. Flowers ternate, borne in the bract axils. Pedicels s

ose
it
n diameter, glabrous, smooth, green or red, dark blue and brittle a ae
Deaes irregular. Seeds two or rarely three, ‘omgues to oval, subterminal, p
placentas, dark blue or black.

Flowering from late January to April at elevations ranging “
approximately 3000-8000 feet from southern Washington to centr
California (Map 4).

bree
REPRESENTATIVE SPECIMENS. “California. Alpine Co.: Silver Mt., Markieey saa
207 (uc); mt. slope above Douglas Station, Hoover 4162 (ny,uc); S for Conall
River, 10 mi. S = oF Markleeville, McKinn ieee sisal Butte Co.: Jonesville, Lig

667 (cH POM oR

s ee
Co.: 1-1/2 mi. SSW of ‘Onion Valley, sonatas 129 fad Del Nort wrest along
near the Smith River, ca. 15-20 mi. NE

c
Snow Creek Trail, Dahling 679 (cu); bet
Rodin ere sa docino Co.: Bell S ; 1 (GHUO:
Station, E of Goose Lake, Balls 14652 (CAS,RSA,WTC); Ukiah, Eastwood 3351 (GH,

: : Vv

re eee 71 (veh

* 1/2 mi. W of Nevada City, Smartsville, yen (cas)
American Ranc Hill, 26 . SW of Grass Valley on McCour rtney Rd., True sip
0.: oe Gap, i. 1305 (uc); eee Gap, Jones 3301 ae HORE

S of Jackson Creek, Public Camp E
(orE); 5 mi. “Sw of Blairsden, Canb

i t. Jackson, Downieville, Rose y c). Shi Co.: near -
River Ferry wn 277 (ny); 6 mi latina, Cuff 107 (uc); Kennet, Eas River
1485 (cu); Trail Gulch, Weave ille, Lewis 28 (uc). Sierra Co.: N f Yub: pe

tween Downieville and Sierra City, Bacigalupi 1614 (POM); ae ieville, 3 wes
6759 (GH,uC). Siskiyou Co.: mi. NW of Callahan, Alexander 138 (uc); hil ” 1635
Yreka, Butler 1125 (PoM,Us); ridges an me Ss near M rble Mt., Char .
(uc); near confluence

3 ee, Highway
Wheeler 2764 (cu); 4 mi. a be junction of McCloud Rd. with the Pacific ae
Wolf 2308 ( (uc). Sonoma Co.: Hood Mt., Baker 3 (uc); Kenwood region, Hope

SYSTEMATICS OF GARRYA 67

Chamb:
el -o-cenbioaey ss 8 11263 (uc). Tehama Co.: Summerville Place, Deer Creek
yey i abil fi 40 (ny); 5 mi. E of Paynes Creek on Plum Creek Rd., Manning
e Co.: Sono aa so Bell 1189 (uc,us); near Long Barn and old ski
en

Ra
ee i Dahling 11600 (cH); Dardanelles, 1 mi. E of Baker Station, Peterson
fee 0) between C Co. Chloride Mine Trail, Dedrick, Alexander & Kellogg 5563
een Coffee and Eagle Creeks, Eastwood & Howell 4957 (cu,Ny,uc); Trinity
C

River near Don Juan P

¢ oint, Tracy 7260 (uc,wrc); N Coast Range, S fo

cia Wolf 9192 (cas,NY,WTC). Un ick n Co.: < nt’s pence to
a urber 369 (GH) )*Revada Washoe Co.: 3- "1/ 2 mi. W of Mt. Rose, Washoe

8
pa whoa see ime Hood River, Henderson 348 (ORE). bap Co.: Siski iyou summit,
Ger Cilile A ik Carberry Creek, lower end of Collings Mt. Trail, Detling 7928
Gok Leach eA egate River, Detling 3988 (ork); Rogue oh ver, 5 mi. below
ee 5 (ORE); Table Rock near Medford, Thompson 10330 (wrc). Josephine
etling 8097 (ORE);

al
ong Savage Creek, 8 mi. from Grants Pass, Henderson 5718 (ORE);

5 LAO:
Us); .

. sd eb Canyon, Crater Lake National Park, Baker 6983 (Ny).
Black ed o emia Mt., Andereus 6/27/35 (ore); Gold Hill, Detling 5674 |
6010 (cu ite n, enderson & King 15958 (ore). Linn Co.: NW of Jum

. eames Multnomah Co.: growing in Government Locks G ;
he HP of the Columbia River, 1.5 mi. up the trail, English 1931 (wrc).
of tae, “a vei Co.: several km. NW of Bingen, Suksdorf 6462 Sal near mouth
and Laurel ns (Big White Salmon), Suksdorf 6426 (GH,NY,UC /WTC); ween Gilmer
, Suksdorf 7434 (cu,uc,wtc). Skamania Co.: base of W ind M os Columbia
, Columbia taieoal Forest, Weber 2673

(NY,wr H
c); Carson, Hemlock Ranger Station, Weber 747027 (uc).

3

ORE);

G
arrya fremontii is mostly distinct from other Garrya taxa and

usual]
y shows a relatively narrow range of variation. However, in
and Sonoma Counties in

and
* Srooves, show extensive structural twisting (
trichome I ieotation

Hybridization and introgression probably

Po ii
Pulations which are intermediate between Garrya fremontii and

68 GERALD V. DAHLING

G. flavescens ssp. pallida. These intermediate populations have bright
and glabrous leaves and fruits which are less pubescent than is typical
for G. flavescens. Since morphologically they approach and are
sometimes difficult to distinguish from G. fremontii it seems likely
that they are of hybrid origin.

Although Eastwood (1903) described Garrya rigida as a species
closely related to G. fremontii, several later authors (Bacigalupi, 1924;
Wangerin, 1910), on the basis of morphology, have considered the
two as synonymous. The latter view is also supported by a phyto-
chemical analysis of flavonoids and is followed here. Populations
of G. fremontii from northern California were compared chemically
with G. rigida from the type location on Mt. Tamalpais. The results
show that both taxa are similar in terms of their flavonoids. Therefore,
the placing of G. rigida as a synonym of G. fremontii is well supported

y both morphological and chemical lines of evidence.

Garrya fremontii var. laxa was described in the same publication
(Eastwood, 1903) and was characterized as having lax or loose inflores-
cences. Because the variation in the internode length of flowering
and fruiting aments is both variable and nearly continuous throughout
G. fremontii, an infraspecific taxon is unjustified.

“4. Garrya corvorum Standley & Steyermark

Garrya corvorum Standl. & Steyerm., Publ. Field Museum Bot. 23:16, 1943. HOLOTYPE:
Guatemala, Dept. Huehuetenango, dense rocky (limestone) Juniperus forest, along
the road in the region of Chemal, Sierra de los Cuchumatanes, at km. 36, 194
Standley 81651 (F). IsoTYPE: GH.

the stem. Leaf blades (2.5-)2.9-3.7(-4.1) cm. long and (0.9-)1.0-1.3(-1.5) cm. wide
apices obtuse, acute, and apiculate; bases tapering, rounded; margins entire, plane;
labrate. lustrous. glossv. dark green i

with dense sericeous upwardly appressed trichomes, green or white; papillae well-
of ; ‘ ; ur

il
tie ier Flowers ternate within bract axils. Pedicels short, not exceeding the bracts
erianth segments 2% oblong, minute, connate at apices, with ascendi

pubescence, gray-white. Stamens free, short, alternating with the perianth se
Anthers oval-oblong, ca, 2 mm , basifixed, introrse, opening by longitudinal i

laments minute. Pollen tricolporate, reticulate, yell reen or tan. Pistillate inflore>
cences catkin-like, unbranched, y, pendulous at ends of branchlets, de ly

rey — sent cm. long, compact, internodes obscured. Pistillate floral bracts ont
mm. long, with subappressed, ascending whitish-gray silky trichomes; opposit®

SYSTEMATICS OF GARRYA 69

bracts connate for over half their lengths. Flowers ternate in the bract axils. Pedicels

minute. Ovaries bicarpellate, rarely tricarpellate, unilocular, subglobose, subsessile,

with gray or white ascending subappressed pubescence. Styles two, rarely three,
as as OV when

subglobose, 6-7 mm. in diameter, pubescent, becoming glabrate and brittle at maturity.
Dehiscence irregular. Seeds two or rarely three, oval, subterminal parietal placentas,
dark blue-black.

Flowering in December and January at elevations of 9000-11700
feet in the Sierra Cuchumatanes of Guatemala (Map 10).

REPRESENTATIVE SPECIMENS. Guatemala. Huehuetenango: between Tojiah and Chemal
at Km. 8 on Ruta Nacional 9N, Sierra de los Cuchumatanes, Beaman 3827
(GH,Msc,TEX,UC); Cuchumatanes, Hunnewell 17196 (cu); between Paquix and Llanos
San Miguel on rd. to San Juan Ixcoy, Sierra Cuchumatanes, Molina 21226 (F,Ny);
Cuchumatanes, Molina, Berber & Wallenta 16488 (F,GH); Sierra Cuchumatanes, Steyer-
mark 50134 (F); rd. to San Juan Ixcoy, Sierra Cuchumatanes, Molina, Berber & Wallenta
16551 (ny); near Chemal, Williams 22226 (F,ny); 3-15 km. N of Chemal, Williams
22200 (F,NyY).

Geographically and anatomically, Garrya corvorum, an isolated
central Guatemalan endemic, is anomalous within subgenus Garrya
and occurs over two thousand miles from its distributional center.
Garrya corvorum is geographically well-isolated and does not inter-
grade or hybridize with other Garrya species. It is commonly found
in limestone soils and is arboreal, frequently reaching heights of up
to 40 feet. While sharing the basic floral and inflorescence features
of subgenus Garrya, G. corvorum is distinct in many other charac-
teristics including leaf size, stomatal apparatus, pubescence, trichome
fine structure, and phytochemistry. Moreover, with regard to the latter
three features, G. corvorum strongly resembles members of subgenus
Fadyenia. This resemblance is definite and well-documented by both
chemical and morphological data. For example, the stomatal apparatus
of G. corvorum is small and lacks the customary wavy and undulate
subsidiary cells of subgenus Garrya. Also, the trichomes lack the
protuberances which are found within subgenus Garrya and appear
similar to those of the southern Garryas of subgenus F adyenia.

With Garrya corvorum as an exception, Garrya species form two
chemical subgroups corresponding quite well to the morphologically
based subgenera. Garrya corvorum shows an abundance of flavonoids
common to both subgenera but lacks a strong chemical affinity with
either as is reflected by the low and high values for the paired and
group chemical affinity indices respectively (Fig. 4 and Table 1).
This is unique since the other Garrya taxa are chemically well-defined
and can be easily placed in one subgroup or the other. Therefore,
Garrya corvorum appears both chemically and morphologically inter-
mediate between subgenera Garrya and Fadyenia. This may indicate
convergence or a relict from an early evolutionary stage of divergence

=

70 GERALD V. DAHLING

from ancestral Garrya representatives. The latter is the most probable

because of the isolated geographical distribution, mixture of charac-

teristics, and the general affinity of G. corvorum for members of
th subgenera.

oe. Garrya buxifolia Gray

Garrya buxifolia Gray, Proc. Amer. Acad. 7:3 1868. HoLoTyPE: Geological Survey
of California, 1867, Bolander 6579 (GH). IsOTYPES: CAS, UC
Garrya flavescens var. buxifolia Jepson, Man. F1. PI. Calif. 732. 1925.

Low bushy shrubs, 1-8 feet tall. Crown sprouts developed. Young stems with long,
appressed ‘pubescence, becoming ee, yellow-green to reddish-brown with age.
imple shortly petiola te, decussate oblong

or broadly elliptical, stiff, coriaceous, persistent. Petioles shit: (0.3-)0.5-0.7(-0.9) cm.
ong; opposite petioles connate at the base and adnate to the s stem. Leaf blades

inconspicuous. Staminate floral bracts 3.5-8.0 mm. long, apex strongly acuminate,
— 2, ces with soft, shaggy trichomes, margins ciliate; opposite bracts strongly
connate. Flowers ternate in the bract axils. Pedicels short, not exceeding bracts. Periam! anth
ae — oblong-lanceolate, 2-4 mm. long, mid-vein prominent, apices connate,

e wi ens

catkin- eo oa nche d, fasciculate at ends of ahaa poke or
lax, 4-10 cm. long, internodes prominent when young but becoming comp
age. Pistillate floral bak 3-5 mm. long, truncate, apices acute, covered w
seh le trichomes; opposite bracts connate. Flowers ternate in the bract axils. vedic
inute. Ovaries bicarpellate, rarely tricarpellate, unilocular, subglobose, pens! «
toe sometimes with perianth remnants alternating with and near the bi se of th

Fruit a berry, subglobose, 5-6 mm. in diameter, glabrous, compact on infructescences

rabsiad becoming dark blue 5 brittle at maturity. Dehiscence irregular
wo or rarely three, globose to oval, subterminal, parietal placentas, dark blue to » black

: Flowering from February to April at elevations ranging from 200-
000 feet in northern California and southern Oregon (Map 2).

"a
gs prema SPECIMENS. California. Del Norte Co.: Crescent City to Grants Pass
rams 8588 (bs,Ppom); Gasquet to Patricks, poten 8530 (ps, ais ee

pa 3, eres Dahling 1290 (cu); along Special Land Use Area Rd., Gas vet, Dah
te H); near Telephone Point, Eastwood 139 (CAS,GH,UC,WTC); a
- 90 (cas,cH); Gordo t., Kildale 9901 (ps,msc,rpom); Smith River . at

m

trail from Knapp Lod
ge toward Rattlesnake Mt., McMinn 5263 (uc); Monu
Parks & Parks 24044 (DS,GH,Msc,PoM); Smith River at 18 mi. Creek, Old Gasquet Toll

A A LY A SR SRR! RE SEIU ot ~ on = ui

SYSTEMATICS OF GARRYA TA

Rd., Parks & Parks 5650 (DS,GH,MSC,NY,ORE,UC,WTC); 18 mi. Creek, Parks & Parks 5266
(GH,wrc); Del Norte, Tracy 10888 (cu); State Line N of Monumental at head of Shelley
Creek, Tracy 16784 (uc). Humboldt Co.: Brennan Mt., near Willow Creek, Abrams
109 (ps); Grouse Mt., Kildale 15361 (rsa); Humboldt, Tracy 17841 (uc); 9 mi. from
Willow Creek, 2 mi. from summit, Wolf 1169 (ps). Siskiyou Co.: Siskiyou National
Forest, Baker 304 (cu); Siskiyou Peak, Kildale 8836 (ps). Mendocino Co.: Red Mts,,
Eastwood 8/4/03 (ny), 5/21-28/02 (cu,orE), Gankin 2698 (cas), McMurphy 594 (ps);
summit of Red Mt. on barren rocky point, Tracy 10324 (uc).*Oregon. Curry Co.:
Iron Mt., S slope, Baker 3541 (cas,uc,wrc); near Horse Sign Butte, Leach 3179, 3343
(ORE); near Craggy Trail, Leach 3491 (ork). Lane Co.: above Waldo, Henderson 5716
(ORE); Eugene, Cusick 7/02 (ork); Waldo, Howell 1506 (uc,wrc); hills near Waldo,
Howell 4/93 (ore). Josephine Co.: lower portion of grade on Oregon Mt. Rd.
1-1/2 mi. above and S of Whiskey Creek Crossing which is at the foot of the grade,
Bacigalupi 6912 (uc); rd. to Chetco Pass, about 10 mi. (airline) W of Selma, Steward,
Dennis & Haines 7428 (CAs,DS,GH,RSA,WTC); Oregon Mt., Sweaton 5/27/23 (ore); along
old rd. 10 mi. SW of Waldo, Thompson 2243 (wtc); Siskiy s. near O’Brien, Thompson
10268 (ps,GH,NY,POM,wTc); Rough and Ready Creek, Tracy 229a (uc).

a
a
5

&

Garrya buxifolia is distinct and confined to a small area in extreme
northern California and adjacent Oregon. Morphological variation in
the taxon is not extensive although individuals with less than the
characteristic dense pubescence can occasionally be found.

An examination of several lines of evidence reveals that a lowering
of taxonomic rank, as had been advocated by Jepson (1925) when
he combined Garrya buxifolia with G. flavescens as a variety, is not
Supported. For example, while leaf undersides in G. buxifolia are
densely pubescent with upwardly appressed trichomes as in
flavescens, the ovaries and mature fruits are quite different. In G.
buxifolia the ovaries are small, glabrous, and dark blue while in G.
flavescens they are densely pubescent, appearing whitish-gray at
maturity. Moreover, analysis with the SEM has shown that their
respective trichome structures are also quite different. Trichomes of

are pronounced.

The flavonoid constituents of Garrya buxifolia and G. flavescens
are substantially different. Garrya buxifolia shows only a low chemical
affinity for this as well as other species of Garrya (Fig. 4). The high

egree of chemical distinctness of G. buxifolia is maintained despite

: fo
both morphologically and chemically different, the close relationship
with G. flavescens advocated by Jepson must be rejected.

72 GERALD V. DAHLING

°° 6. Garrya flavescens Watson

w shrubs or small trees, 5-15 feet high. Crown sprouts developed. Young ee
with whitish-gray, silky, appressed pubescence, becoming glabrate with age; cinereous,

e
yellow-gray, cinereous, or sometimes dull green; lower surface glabrate or with a
or wavy upwardly appressed trichomes, yellow-green, gray, or glaucous with age; ee
developed; midrib and major veins visible above, obscured below except for at 2
in high relief. Staminate inflorescences catkin-like, unbranched, fasciculate, pen epi
owere te

and nating with the perianth segments. Anthers ov to oblong, about ee
long, basifi rorse, open by longitudinal slits. Fil ments minute. Pollen
tricolporate, reticulate, yellow-brown or tan. Pistillate inflorescences catkin-like, un-
branched, fa te, pen lets, densely flowered, c ie

extremely compact, imbricate. Pistillate floral b 3-8 mm. long, ovat lanceolate,
concave, cate, i or acute, typically cove with appressed or sh

try, oval, a mm. long and 4.8-6.5 mm. wide, a
bescent, gray to whitish-brown, dry at maturity. Dehiscence irregular. Seeds a
or rarely three, globose to oval, subterminal on parietal placentas, dark blue or black.

KEY TO THE SUBSPECIES
A. Lower leaf surfaces glabrate or s

Pparsely to moderately covered with coarse, wavy:
more or less appressed and ascending trichomes. Shrubs and small trees. . . . - 4
be eh ee . G, flavescens ssp. palli 4
A. Lower leaf surfaces densely covered with fine, straight, long, upwardly appresse

B. Lower leaf surfaces densely covered with fine straight, long, appressed, ant
strongly ascending trichom

Le ce, a. G. flavescens ssp. fo
B. Lower leaf surface densely covered with gently wavy, slightly intertwining 2
ding trichomes. Shrubs

6

Ca Garrya flavescens Watson ssp. flavescens

per oe flavescens Wats., Am. Nat. 7:301. 1873, HOLOTYPE: Kanab, South Utah, Watson

GH).

Garrya veatchii var. flavescens Coult. & Evans, Bot. Gaz. 15:96. 1890. k

Garrya mollis Greene, Leafl. Bot. Obs. and Crit. 2:86. 1910. tocorype: Oak Creek;
Arizona, 1901, Pearson 399 (us)

SYSTEMATICS OF GARRYA 73

appressed more or less aati a yellow-gray, gray-green, or yellow-green.
Abaxial leaf surfaces with dense, fine, silky, long, more or less straight upwardly
appressed trichomes; light gray-green or silvery. Fruit with dense silky appressed
trichomes, becoming less dense with age, gray or tan-brown.

Flowering from the middle of March to early May at altitudes of
2600-9000 feet in western Arizona and southern California (Map 4).

REPRESENTATIVE SPECIMENS, Arizona. Coconino Co.: — Angel Trail, 1 mi. above
Indian Garden, Bailey 1058 (uc); Mt. Trumbull, 6 mi. S of Nixon Spring, Cottam
8703 (ariz); Virgin Mts » NW Arizona, Crew 36 (ariz); Willains: Sand Gap, Demaree
42562 (ariz); mene trail to Roaring Spring, Grand Canyon, Eastwood 1038 (GH,POM);
Apache Trail, way to Mercury Mine, Eastwood 17259 (cas); Timp Point along
rim of Kaibab sath gry Osi 503-48 (ariz); Reconnisance Range, Long’s Canyon,
Hill 402 (aRrIz). Gila Co.: Matzatzal Mts., Collom 279 (cu); Sierra Ancha, Eastwood
17284 (cas); Three Bar Gans Management Unit, Tonto National Forest, Pase 885
(ARIZ), Maricopa Co.: 4 mi. N o Sunflower, Lehto 3782 (rsa). Mohave Co.: Conner
Canyon, 1 mi. NW of Parker pinay Armer 24 (uc); Secret Pass W of Kingman, Braem

um

4/15/03 (NY,POM); E slope of Providence Mts., Mojave Desert, Munz 4062 gs

Yavapai Co.: Oak Creek Canyon, ytoiapront 2718 (ariz); Oak Creek Canyon, near Sedona,
Dahling 870 (cu); Tonto Creek, ne ne, McDougal 720 (ny); rd. to Sunflower Mine
in Mazatzal Range, McKelvey 843 north 1135 — between Payson and Pine, Whitehead
2078 (ARIZ), Yuma ~ 0.: Palm Canyon, Kofa Mts., Mason 2493 (ariz), Niles 485 (ARIZ),

Los Angeles Gai’ Rock Creek Desert dunes of the San Gabriel Mts., A

(Ds); Angeles Crest Hwy., near Charlton Flats, San Gabriel Mts., Beeks 193 (rsa);
Mt. Baldy Camp, San Antonio Canyon, MacFadden 14741 (cas); ‘Strawberry Ridge,

San Gabriel Mts., Richter 108117 (ps). Riverside Co.: trail to Tahquitz Peak, San Jacinto
Mts., Ho offmann 168161 (cas); ridge E of Toro Peak, Santa Rosa Mts., Munz 15372
(Pom, wre);

318 (GH); a

3
—
fe)
=]
ga
ey
Qu
)
°
pe
N
)
=]
or
~
as)
©
=
o
-]
is)
€
=
x
8
Se!
ne
TD nwt
Go
i)
>
ae
2)
)
=]
—
©
2
=)
or
°
=
oo
<2)
3
Ss
3

eterson 319 (RSA); acosectin: Tucker 325 (uc). Tulare : Salt Check: Eastwood

32
172612 (uc). Ventura Co.: El Vallecito de los Pinos, Moseley 103 (rsa), 106 (cH);

Ortega Hill Trail, between Cherry Canyon nieces and she Ridge, apa Moe

(CAS,cu); N side of M wder 498 (RSA,UC) c
utau Flat and Mt. Pinos, S er 4

Co.: U, S. Atomic Energy Commission ide Site, Retna 10738 (RSA); Charleston _

Clokey 5550 (cu,wrc); canyon bed 1/2 mi. up canyon above Wilson’s Ranch, C oney

9 (4RIZ,CH,MEXU,UT,Wr ); :

“a, Dahling 1318 (cu); Virgin River Canyon, Bunkerville, Goodding 735 (cu ante
Ww York Mts. near California line, Shields 10/ ; :

Slope of Charleston Mts., Train 2058 (aniz,GH).Utah, Washington Co.: Pine Valley,

c

74 GERALD V. DAHLING

Cottam 6780 (ut); between Leeds and Oak Grove, Eastwood 9164 (cH,Ny,PoM); Zion
Canyon, Hunnewell 10 (cu); in canyon N of Leeds, Maguire & Richards 5201
(cH,uc); near Toquerville, McNulty 12/2/70 (vt); Beaverdam Mts., Tidestrom 9336
(GH).

Garrya flavescens ssp. flavescens is characterized by fine and
upwardly appressed trichomes on the leaves. The abaxial surfaces
are covered with a whitish-gray pubescence and frequently appear
shiny white in sunlight. While G. flavescens ssp. flavescens intergrades
with G. flavescens ssp. pallida and G. flavescens ssp. congdonii, they
are all sufficiently distinct to allow for their recognition.

A large-leaf form of Garrya flavescens found growing in semimesic
canyons of Arizona was described as G. mollis by Greene in 1910.
While the leaves of G. mollis are somewhat larger than those of typical
G. flavescens ssp. flavescens, there are similarities between the two
in pubescence, floral, and fruit characteristics. In addition, phyto-
chemical investigation of the flavonoid constituents of G. mollis and
G. flavescens showed a marked similarity and there is a general
correspondence in flavonoid content. Furthermore, field observations
repeatedly showed that environmental factors influence leaf size.
Therefore, the larger leaves of G. mollis probably reflect the more
mesic growth conditions of deep canyons rather than distinguishing
genetic differences.

Garrya flavescens was combined with G. veatchii as a variety by
Coulter and Evans (1890) but the many differences between the two
taxa make such a combination untenable. In fact, G. flavescens and
G. veatchii are so distinctive in leaf, trichome, floral, and fruit
characteristics that they are rarely confused or misidentified.

6b. Garrya flavescens ssp. pallida (Eastwood) Dahling, comb. nov.

a pallida Eastw., Proc. Cal. Acad. III. Bot. 2:287. 1902. fiocoryPe: Kings Rive!
peso fas South Fork of Kings River, Fresno County, California, 1899, Eastwood

— flavescens var. pallida Bacigalupi ex Ewan, Bull. Torrey Bot. Club 64:519.

se to small trees, 5-15 feet tall. Adaxial leaf surfaces glabrous or rarely with
avy =

: = 3; Bray r yellowish. Abaxial leaf surfaces glabrous

or with coarse, wavy appressed more or less ascending trichomes. Leaves dull gree™

yellow-green and glaucous. Frui i
: its with itch
to gray-brown or tan-brown with age oe oy ae a

green
&BiCch 0

Flowering from late February to April at elevations ranging from

1500-775 to Ap z fro
(Map 4). 5 feet from central California to northern Baja Californ!

y
cies ake gos California. Alameda Co.: Cedar Mt., Abrams 108132 (ps);
+ Mt. Ridge, Bacigalupi 1306 (ps,pom). Contra Costa Co.: Meridia?

SYSTEMATICS OF GARRYA 75

Peak summit, Bowerman 779 (uc); Mt. Diable, Mason 2711 (uc), McMinn 88 (ps),
Rose 34041 (wrc). Fresno Co.: San Juaquin River, Sierra Nevada, Blether 216835 (ps);
Kings Canyon National Park, Dahling 930 (cu); Copper eee Canyon Trail, Howell
34206 (cas); Boyden Cave, S fork of Kings River, Munz al eh Kern Co.:
Fort Tejon, Abrams 283 (ps); San Emigdio Canyon, Davy ae (uc); Mt. Breckinridge,
Grinnell 146 (us); SW of Black Mt. Saddle, Howell 38528 (CAs); Ps. a E of

Bald Eagle Peak, ee ts sat 5047 (cas); Toogate Ridge Mts., W end, selmann
12270 (cas). Lake Co.: 1-3 mi. up W side of oe Grade, Abrams 12392 fe between

8.9 mi. S of Kelseyville, Baker 11299 (uc); N ee wer Lake along the main s caisiay,
Dahling 842 (cu); N of Lower Lake along H “99, Dahling 884 (cH); og ‘s at
of Upper Lake on the road to Ukiah, Dahting 977 (cH); about ie mi. SW

29 on the Lower Lake-Knoxville Rd., Dahling 1024 (cu); about 10 mi. E of Hop eas
Dahling 1279 (cu); 7 mi. NE of jet. of wags 20 and the road “ak Bartlett ee
Thorne 31649 (us). Los Angeles Co.: Ls . N of Warm Springs, Biases Gifford 459
(uc); Gold Creek, Lewis 296 (uc). Marin near summit of Mt. Tamalpais, Dahling
1230 (GH). Mendocino Co.: Ukiah, epacopt nei 518835 = Monsees Co.: Pine Ridge,

nents Lucia Mts , Abrams pss ve i Cone Peak, Santa Lucia Mts., rigpet GATT

near Palisades, Napa River Besin; Jepson 21610 (uc). Orange Co.: Santiago Peak Trail,
Abrams AG age 1. 1 mi. : of S Acie Balls 22847 (uc); Bedford Peak, Corona,
wee

Ana Mts., Pequegnat 261681 (pom). ui Benito Co.: shoulder to San Carlos Peak above
Aurora Mine. Abrams 13842 (ps,GH); Hernancee. Eons 212587 (ps). San Bernardino
-: Cold Creek, San Bernardino Mts sage 4 (ps); S slope of Clark Mt., Alexander
495 (uc); Fountain Canyon, Pro viden ee Beal 668 (wrc); Home Creek, San
tardino Mts., Everett 19356 (rsa); cuddle fork of Lytle Creek, San Antonio Mts.,
Johnston 1579 (Pom, UC); sa Mt., Munz 12890 (cH,pom); Clark G Grade, Santa Ana
River Valley, San Bernardino Mts., Wheeler 2025 (uc); 1. 5 m. above Pachalka Spring,
Clark Mt., Wo vad 7016 hee RSA wre): San Diego Co.: Cuyamaca Mts., A Abrams 3872
(CH,POM,uc); upper SW Indian Canyon, Bastil 44111 (sp); Warners Springs, Eastwood
43346 (ps); es Isidro, Epling 224326 (ps); Monument Peak, Laguna Mts., beige 3
(Pom); San Felipe Valley, Wolf 869 (rsa). San Luis Obispo pats 4-1/2 mi. SSE of
Pozo, ebvoapei 9 (uc); Caliente Mt. along summit ri idge, Hoover 8332 (uc). Santa cua
nta Barbara Canyons, Eastwood 713 (uc); eaddlle between Junipera and Piny
Peaks, Howe 30187a (ps); S = Potrero Seco along the Matiliza-Agua- ae sail
Pollard 4 453356 (ps). Santa Clara Co.: on road to is a an e summit of Mt.

—_

-oini 3967 (Ds).
wre); Santa Clara, Heller 7507 (cx); vicinity of Mt. Umunhum, Thomas
Santa Cruz Co.: Loma Prieta Peak, Santa Cruz Mts., Mexia 634640 (uc). Solano Co.:

P
River, 3 mi. below Camp Nelson, Wolf 4589 (GH,UC,WT wrc). Ventura Co.: Topatopa Mts.,
Abrams 104 (ps); Maricopa Rd. above Bains Gorge, Pollard 2267 (cas).

Garrya flavescens ssp. pallida is characterized by wavy, coarse,
and more or less upwardly appressed trichomes on the lower leaf

76 GERALD V. DAHLING

surfaces. As the name implies, the leaf color is pale and often tinged
with gray or blue.

In describing Garrya pallida in 1902, Eastwood suggested a rela-
tionship with G. fremontii based on the similarity of their leaves.
Most G. pallida populations have leaves with the typical coarse wavy
trichomes but some populations are nearly glabrous and therefore
closely approach G. fremontii. Bacigalupi (1924) rejected Eastwood's
argument and combined G. pallida with G. flavescens as a variety
on the basis of the ovary pubescence. This is well-justified since
both taxa intergrade morphologically and are difficult to distinguish
in overlapping areas.

Chemical investigation supports a relationship between Garrya
pallida and G. flavescens since both share similar flavonoid constitu-
ents (Figs. 3 and 4). Although they bear some relationship to G.
fremontii, G. pallida and G. flavescens appear most closely allied
with each other and as such, represent infraspecific taxa. This position
is also well-supported by morphological evidence as outlined by
Bacigalupi (1924). With regard to the populations of G. flavescens
ssp. pallida which intergrade with G. fremontii, it is the author's
view that these are probably the result of hybridization and subsequent
introgression.

~’ 6c. Garrya flavescens ssp. congdonii (Eastwood) Dahling,
comb. nov.

- Garrya congdonii Eastwood, Bot. Gaz. 36:459. 1903. HoLoryPE: New Coulterville Road,
Mariposa Co., California, 1898, Congdon 556 (cas). IsOTYPE: GH.
Garrya flavescens var. venosa Jepson, Man. FI. Pl. Calif. 732. 1925.

Shrubs, 5-10 feet high. Adaxial leaf surf Ja} 4] , crinkled trichomes,
lustrous glossy, and yellow-green. Abaxial leaf surfaces with long, gently wavy, slightly
appressed and intertwining, ascending trichomes, whitish-gray or dull yellow-brown.

Fruits with shaggy gray-white ascending pubescence, with age becoming glabrate toward
the base, yellow-brown with purple or dark blue.

<loweHne February to May, mostly at elevations of 2000-6000 feet
in California from Tehama to San Benito and Mariposa counties (Map
2).

REPRESENTATIVE SPECIMENS. Californie. Bs Fas 20. just E of the
Lake-Colusa border, Baker 11645 : along Hwy. 20, ju

Stonyford, Ferris 6456a (Ds);

h Pt : 3); e Co.: Mt. St. Helena, 4 mi. below
11655 a ~ igalupi = (ps); between Burus Valley and Borax Lake, Baket
14140 a 2.5 mi. NE of Middletown, Gould 1026 (cH,uc); W of Leesville, Heller

(cas,cu); Lower Lake to Knoxville Rd., Howell 14652 (cH); Knoxville grade

Ase - — . “ i ‘= “a

SYSTEMATICS OF GARRYA ig

to Lower Lake, Jepson 21598 (uc); 13 mi. N of Calistoga, Wiggins 5772 (ps,cH);
ria E of Hough’s Mineral Springs on rd. to Williams, Wolf 1054 (os) sicstoeiaap

mi. S of Coulterville toward Bagby, Belshaw 21128 (ps); Mariposa, Congdon 57896
si Coulterville Rd., Congdon 1/98 (cu). Mendocino Co.: S fork of Eel River,
383 as Napa Co.: Pope Creek, 7 mi. from Monticello, Baker 9105 (cas);
ie "od am, Raven 2827 (cas); 5 mi. S of Reiff, Schreiber 2353 (uc). San Benito
Sa a enito County area, Hall 9940 (cu). Santa Barbara Co.: Zaca Lake Forest
2 OL “ape 713 (cH); Santa Ynez Mts., Moseley 78 (cu), 93 (GH). Stanislaus
te) | Creek, Red Mts., Mt. Hamilton Range, Sharsmith 3532 (ps). Tehama Co.:
i ga rade, along rd. from Paskenta to Covelo, Bacigalupi 2411 (ps); S fork of

attle Canyon on Ponderosa Way, near the head of Darling Ravine, Griffin 1165 (uc);
grade between Mud Flat and Bennett Spring, Heller 13000 (cas,cu); 10 mi.
Paskenta below logging rd., Wagon 2611 (cas).

Garrya congdonii is not a distinct species and should not be given
that rank because of the variation and instability in the pubescence
and leaf characteristics. The present combination with G. flavescens
appears justified because they commonly intergrade and appear closely
related morphologically. Garrya flavescens ssp. congdonii is recogniz-
able because of its thick, gently wavy trichomes on the abaxial leaf
surfaces. The trichomes are long and have the compound counter-
clockwise orientation that typifies the other G. flavescens subspecies.

7 YE

SUBGENUS FADYENIA (ENDL.) DAHLING, COMB, NOV.
Based on Fadyenia Endl., Gen. Supp. IV. 1847:38.

mmon or sometimes restricted to the base.

nflorescences racemose with branching co
aceous, sometimes as large as the foliage

a fruiting. Ovaries someti bearing minute variably adnate bracts opposite the
e es. Styles typically short, thick, fleshy, often divergent and reflexed, usually not
rect. Shrubs and tall trees.

?

2%) 7. Garrya glaberrima Wangerin

Garrya glaberrima Wang., Das Pflanzenrei

E ch IV, 56a:12. 1910. ‘HOLOTYPE: Mexico,
nearnacion, Ehrenberg 1097 (Bb), not seen.

umerous. Young stems glabrous,
-brown. Decussate branch-
elliptic to oblong, thick,
1(-1.2) cm. long; opposite petioles connate
2-)6.0-7.5(-8.6) cm. long, (2.2-)2.5-3.5(-4.0)

Ne hrubs to small trees, 4-16 feet tall. Crown sprouts n
Seed, quadrangular, green, with age becoming dark reddish

aa persistent. Petioles (0.8-)0.9-1.
c base, adnate to the stem. Leaf blades (5.

: f > onate, attenuat subcunnate; margins plane; upper
urface glabrous, bright green; lowe face glabrous, dull green papillae developed
and primary lateral veins

7 eclies 2-5 cm. long, compact, densely owered,
boas inate floral bracts ovate to elliptical, acuminate, shorter than the flowers, green,
P, vad ciliate; opposite bracts connate at the base. Flowers solitary in the bract axils,

edicels short. Perianth segments four, elliptical, connate at apices, glabrous. Stamens
h segments. Anthers oval to oblong, ca.2 mm.
udinal slits. Filaments minute. Pollen

78 GERALD V. DAHLING

tricolporate, reticulate, green. Pistillate inflorescences racemose, branched, pendulous,
fasciculate on branchlets, lax, 3-6 cm. long, glabrous. Internodes prominent. Pistillate
floral bracts foliaceous below, long elliptical above; opposite bracts connate at base.
Fl litary in the bract axils. Pedicels short. Ovaries bicarpellate, rarely tricarpellate,
unilocular, elliptical, rarely with paired bracts opposite the styles, ro
two, rarely three, persistent, short, thick, fleshy, divergent. Fruit a berry, elliptical,
-8 mm. in diameter, dark blue to black, sometimes on prominent pedicel, becoming
ry at maturity. Dehiscence irregular. Seeds two or rarely three, oval, subterminal
on parietal placentas, dark blue or black.

Flowering during the winter in northern Mexico at elevations ranging
from 4880-9000 feet, depending on the rainfall (Map 5).

REPRESENTATIVE SPECIMENS. Mieciat: Coahuila: Canon de San Lorenzo, 5-6 mi. SE
of Saltillo, McVaugh 12346 (mexu,micu). Nuevo Leon: Cerro Potosi, Chiang, Wendt
& Johnston 8063 (rex); Cerro Potosi, Gilbert 86 (rex); in the shrub zone of Potosi
Mt., Dahling 108 (cH); canyon below Las Canoas, Municipio Galeana, Mueller 2220
(GH,MICH,MO,TEX); between San Francisco Canon and Pablillo, 15 mi. SW of Galeana,
Sierra Madre Oriental, Mueller & Mueller 376 (GH,MICH,TEX,US). t: ulip 5 km.
SE of La Joya de Salas, top of Sierra Madre Oriental on trail to Gomez Farias, Martin
125 (micu); Dulces Nombres, just E of Nuevo Leon into Tamaulipas, Meyer & Rogers
2829 (cH,Mo); Sierra de Guatemala, Webster & Websier 11 (TEX).

Garrya glaberrima, as the name implies, is completely glabrous
and is the only member of Garrya lacking trichomes. It is easily
distinguished from the other taxa, and shows nearly complete repro
ductive isolation. Frequently G. glaberrima is sympatric with G.
laurifolia and G. ovata, but no intermediates or hybrids are produced.
This is unusual in Garrya where reproductive flexibility is common
and hybridization frequent.

ma are elongated and contrast with the
ristic of most other taxa.
Garrya glaberrima occurs at high altitudes in scattered mountai?

localities where collecting is diff; .
é ng is difficult, whi ins why so little
herbarium material is available. ion enihins wr

SYSTEMATICS OF GARRYA 79

ta ssp. mexicana (half circles);

vata ssp. ovata (triangles); G. ova

hwestern U.S. and Mexico. 6, G. ovata ssp. lindheimeri (dots); G.

U.S. and Mexico. 7, G. longifolia—Mexico. 8, G. laurifolia ssp.
les) M ;

(tri exico
\ g a

— P-8. Distribution of Garrya: 5, G. 0
* wrightii (dots); G. glaberrima (stars)—sout
laurif ep: goldmanii (circles)—southwestern :

olia (stars); G. laurifolia ssp. macrophylla (dots); G. laurifolia ssp. racemo

80 GERALD V. DAHLING

8. Garrya ovata Bentham

Lo es, 2-16 feet high. Crown sprouts numerous. Young stems tomentose,
’ with age zk ‘dak gray or reddish-brown. Decussate branching. Stipules absent.

ti y
long, (1.0- net od Ph cm. — ca. twice as long as aes apices mucronate, acute

below in relief above the pubescence, secondary veins taminate inflorescences
racemose, , pendulous, fasciculate on branchlets, rarely solitary, 1- ong,
clustered, internodes inconspicuous. Staminate floral bracts ovate to lanceolate, truncate
to obtuse, small, shorter than the flower, green red- ; baxial

hort. Perianth segments four, connate at sree, ‘pubescent near tips on aber ial pi
glabrous on adaxial side. Stamens ae shor ith the segmem l
Anthers oval to oblong, 2-3 mm. lon i ifixed, a open ing by ve gitudina

; : es

k hy,
a berry, subglobose, 6-7 mm. in diameter, glabrous, dark blu e, becoming dry at maturity.
Dehiscence irregular. Seeds two or rarely three, globose, pibeenadnal, dette placentas,
dark blue or black.

KEY TO THE SUBSPECIES

A. Leaves strongly undulate, mature leaves with dense woolly, non-coiled pubescence
ee a ee 8c. G. ovata ssp. goldmanii.
Leaves mostly plane or slightly undulate, mature leaves sparsely villous, densely
tomentose, or with s strongly coiled trichomes forming a mat below and a slight
covering above,
Mature ‘seat glabrate, sparingly tomentose or sometimes with villous bie s

Leaves bright Seiten... 8d. re ov vata ssp. m
M

- Mature leaves without coiled trichomes but densel tomentose on the lower
parsine es. Upper surfaces glabrate with prominent lateral veins .....-- °° *
8a

8a. Garrya ovata Bentham ssp. ovata

Garva ovata Benth, » Pl. Hartw: 14. 1839. HoLotype: CG; don the Bufa Guanajuato
1839, Hartweg 80 (K). ISOTYPE:G

enia ovata Endl., Gen. Suppl. IV: 38, 1847. Based on G. ovata Benth.

SYSTEMATICS OF GARRYA 81

Shrubs, 3-10 feet high. Adaxial leaf surfaces glabrous, rarely with sparse, separated
trichomes, lustrous, major latera Adaxial leaf surfaces
tomentose, gray becoming brown with age. Leaves plane or weakly undulate. Ovary
glabrous. Fruits dark blue-black at maturity

Depending on the rainfall, flowering during the winter at elevations
of 4800-8000 feet throughout central Mexico (Map 5).
REPRESE NTATIVE SPECIMENS. Mexico. “Chihuahua: SW Chihuahua, Palmer 319 (GH, US);

just S of Esm ralda above Sociedad Cooperative Mine, _L iang, Wendt & Johnston
90867 (Rsa, re): Carnes Pass, Pringle 2806 (GH,MEXU,US). ). Jalisco: mts. W of Hacienda

Chinampas ca. 15 mi. WSW of Ojuelos, McVaugh 17000 (micu); Sierra de Laurel,
near the Aina lioditaa pciir's r, ca. 10 mi. SW of eal McVaugh 18350 (micu).
uevo Leon: Rio Sa sa, Quintero 3897 (micH); road from Encinal to Pabillo

nta R

ca. 15 mi. SW of Galeana, Sisers Madre Oriental, Mueller & Mueller 461 (GH,MICH,TEX,US);
El Fraile Peak, Smith M654 (rsa,tex).“San Luis Potosi: region of San Luis Potosi,
Parry 3 Palmer wae (MO,US); wee Municipio Villa de Reyes, Passini 862 (rsa);
5k of the Station wlan , Municipio Catorce, Rzedowski 8155 (TEx); San Luis
Potosi, "Schaffner 289 (us " Schaffner 290 (micH,us), Schaffner 293 (us); ex convalli
San Luis Potosi, earth sig (GH,MEXxU); San Rafael Mts., ipolece alae (cx) Zacatecas:
Puerto de la Paja, 20 km. al WSW de Valparaiso sobre el camino a Hue juquilla, Rzedowski
17545 (micn, TEX). Vekvown locality: Schumann 804 (us

Because it grows in the high isolated mountains of central and
northern Mexico, Garrya ovata ssp. ovata has been collected rarely
and is poorly represented in herbaria. Consequently, little is known
of its range of variability.

It appears morphologically closely related to Garrya ovata ssp.
lindheimeri. Both are associated with limestone substrates and when
in contact they frequently intergrade. Phytochemical data also support
the idea of a close relationship. In fact, the flavonoid content of
all G. ovata subspecies shows a general correspondence which is
indicative of a tightly knit group.

8b. Garrya ovata ssp. lindheimeri (Torrey) Dahling, comb. nov.

Garrya lindheimeri Torr., Pac. Rail. Rep. IV: 136. 1857. isotyPE: Western Texas to
(GH).

El Paso, New Mexico, 1849, Wright 633 (cH

12 feet high. Crown sprouts developed. Leaf margins

Low shrubs or small trees, 3- tomentose or

Plane, sometimes undulate. Upper leaf surface green, sparingly
trichomes, whitish- -gray or gray-green. Ovaries
the styles, conspicuous. Fruits dark blue at rei ity.

Flowering from March through April at elevations of 200-6600 feet
throughout southwestern Texas and northern Mexico (Map 6).

© yamablokis SPECIMENS. “Mexico. Coahuila: Canon del Mulato, Serranias del agin
Ca. 65 mi. NW of Sabinas, Gould 10609 (micH, TEx); 18 mi. S of Sa Itillo on A f
Hess & Hall 520 (micu); Canon de Tinaja Blanca, Sierra de ie Cruces, We of Santa

82 GERALD V. DAHLING

Elena Mines, Johnston & Muller 258 (cu,TEx); Sierra de Hechiceros, Canon del Indio
Felipe, Johnston & Muller 1332 (GH,TEX); Sierra de la Rata, Johnston, Wendt & Chiang
101611 (TEx); Sierra de la Madera, S side of the lower at of Canon de la Hacienda,

( nd

& ene 11709a (TEx); Muzquiz, Marsh He ges Santa Rosa Mt., Marsh
1430 (rex); Canon Espantosa, W slope of Sierra de San Vicente, central part of the
state, Schroeder 98a (Gu); Canon de Milagro, sonity de a pe uajes, 12 km. W of Hacienda
de la Encantada, Stewart 1517 (cu,TEXx); Canon de Ybarra, Sierra del Pino, Stewart
1823 (TEx); 8 km. W of Santa Elena, Stewart 2950 (TEX); vicinity of Santa Hew Mines,
E foothills of Sierra de las Cruces, Stewart 2272 (Gu,rEx); Muzquiz, near Puerto Santa
Ana, Wynd & Mueller 276 (us); Rancho Agua,Dulce, Sierra de San Manuel, Mucha
de Muzquiz, Wynd & a, age MICH,MSC) coco I Leon: Rancho Resendez, Lampazos,
Edwards ~ (MO,TEX ne ndera Co.: along Seco Creek, W of Tarpley, Correll
29025 (TEX); 6-1/2 mi. N of oh eae Cory 34706 (cu); ravine, 9 mi. W of Boerne,
ohnson *y Webster — (TEx,us); canyon of Can Creek, Lawless Ranch, 7 mi. N of
Vanderpool, McVaugh 7056 (micH,TEX). Bexar Co.: re mi. N of San pokes Burr
532 (TEx); N of Helotes, Parks 29441 (cu); vicinity of San Antonio, von Schrenk 4/19
(Mo). Blanco Co.: oo 1549 (mo); Rio Blanco, Sargent 5/29/85 (GH); a —_—
Sargent 5/30/85 (GH). Brewster Co.: NE side of Casa Grande, Chisos Mts., pat
National Correll 20674 (TEx); Juniper Canyon ea Upper Juniper pear:
Spring, Ferris & Duncan 2776 (mo); S rim of Chisos Mts., Big Bend National hae
Nelson & pe 5103 (cH); Chisos Mts., Pulliam Canyon, Sperry 471 (us); Chisos
Mts., Warnock 926 (us), Comal Co.: along Ranch gin a ca. 5 mi. E of Fischer in
the Devil’s Backbone Area, Dahling 804° (cH); 12 mi. W of San Marcos, Ethridge
3/16/64 bee pamene Spring, New aoe Sy Lindheimer 842, 843, 844
(GH,MO,TEX,Us). Edwards Co.: Polecat Creek, Cory 41160 (cH,TEx), Cory 41162, 41163
(GH); Frio Water she Hill 59 (us). Gillespie Co.: Jermy 128 (us). Hays Co.: Wimberley,
Fisher 43056 (us us); San Marcos, Kiddler 85 (cu). Jeff Davis Co.: along Limpia C Creek

Real C Co.: E % rk of ies River, Correll 13434 (1x); 16 mi. N of Leaky, Cory

ee)
Log
@
g
~
<°
R
Q
*S
3
®
~
8
=
i ~¥)
Fy
a

os
fe
GS
>
c

+
5

&
=

Q

'S

28:
a
~
zs

(cake: near Austin, Coville 1831 “es Ht L ngenbie
ake Austin, near dam, Crutchfield 2426 (1#X);
pee a ey Burnell, Hall 1070021 (us); Austin, University of Texas, Heimsch & Tha
Pai — Pi nnell, NW of Austin, Warnock 46030 (cH,rrx). Uvalde Co.: Montell
Coome eee 23785 (cu); 6 mi. N of Montell, Montell Creek, Cory 42910 (TEX);
10 River, Palmer 10184 (mo,us). Val Verde Co.: Comstock, Blair 209 (US)-

Garrya ovata ssp. lindheimeri is distributed throughout southwesterm
exas and adjacent Mexico where it typically inhabits limestone
H eINS Se and also the edges and faces of cliffs. For example.
shies fo southwestern Teaas, G. ovata ssp. lindheimeri is a commo?
at of the limestone cliffs overlooking rivers and streams. While
ne ste ek ees on cliffs and ledges, it becomes arboreal in
cted areas such as the moist, rock steep:
semiarid slopes. y ravines and gullies of steeP
. bebe Ae ssp. lindheimeri is distinct in Texas, but it tends '
ge with both G. ovata ssp. ovata and G. ovata ssp. goldmanii

El ES RSS 8 er PRR RSS Se AER or ln ee) “Sl ae ASO A a Y= NS SEN

SL cr ee sities

eat
—— ae ame anh

SYSTEMATICS OF GARRYA 83

in northern Mexico. Apparently gene flow is frequent between these
subspecies when they are sympatric.

a v
Te. Garrya ovata ssp. goldmanii (Wooton & Standley) Dahling,
comb. nov.

St
~ Garrya goldmanii Woot. & Standl., Contrib. U.S. Nat. Herb. 16:157. 1913.HoLoryPE:
limestone ledges, Queen, Eddy County, New Mexico, 1909, Wooton 562308 (us).

Small shrubs, 1-6 feet high. Crown sprouts developed. Leaves strongly undulate.
Upper leaf surfac paraiso or with sparse variably ieee pubescence, lustrous
major lateral aoe inconspicuous. Lower leaf surface lanate, sometimes tomentose,
whitish- “gray or ye ellow- pata. "Ovary glabrous, bracts opposite the styles variably adnate
and conspicuous. Fruits glabrous, dark blue at maturity.

Flowering from March to May at elevations ranging from 4550
to 7880 feet throughout southwestern Texas, southeastern New Mexico,
and northern Mexico (Map 6).

REPRESENTATIVE SPECIMENS. Mexico.*Chihuahua: Sierra de Chupaderes, extreme S end
of Rancho Las Pampas, ENE of Jimenez, Chiang, Wendt & Johnston 8900 (rrx); canon
in N face of Sierra Rica, S of Rancho La Consolacion, Johnston, Wendt & Chiang
10756 (TEx); cafon in N face of Sierra aes S of Rancho La Consolacion, Johnston,
Wendt & Chiang 10769 (1rx); Sierra G ca. 3 km. E of Rancho El Muscielago,
Johnston, Wendt & Chiang 11288A en ee de los Ps LeSeur 1524 (mo, TEX);
Santa Eulalia Mts., Pringle 131 (GH,MicH,kSA,US). Coahuila: 4 mi. W an nd 10 mi. S
of Ocampo, Graber 173 (rEx); Sierra de la Madera, pa eg Pajarito, Huller 3190
(MICH); ae de la Madera, vicinity of La Cueva, Johnston 8902 (rEx); Sierra del
Pino, vicinity of La Noria, Johnston & Muller 524 Sean o. del Pino, W of cam
at La Noria, Johnston & Muller 599 (cu,TEX); Del Carmen Mts., Marsh, Jr. 802 (GH,TEX);
Puerto de San arash Sierra de San Lazaro, Manicinio de regen aia 308

ics. Glass Mts. p hiabeole k W322 (GH H,TEX). gars Co. ai oie ‘ts woe Mts., Bailey

452 (us); McKittrick Canyon, Guadalupe Mts., C rrell & Johaston 18493 (TEX); Pine

Spring, Guadalupe Mts., Whitehouse 8686 (TEx). jeff Davis on : Mt. Livermore, Hinckley

10/13/34 (TEx); Fern Canyon, NW of Mitre Peak, 14 mi. NW of Alpine, Muller 8183

(GH,MICH, TEX); Davis Mts., Palmer 34340 (mo); Rose past N of Alpine, Warnock
I - H,TEX,UC); Davis Mts., Sieh 8/13/14 (1x). Pecos Co.: Gap Tank, Glass

Mts., 25 mi. N of Marathon n, Warnock 5023 (rx). Presidio Co.: near Vieja, Tierra

s).

Vieja Nes, Hinckley 20951 (us). Real € Harris 14 (us

Garrya ovata ssp. goldmanii is distinctive because of its small stature
and short, narrow, very undulate leaf margins. When inhabiting the
upper Sonoran zone, it commonly grows in the shade as a small
shrub, However, sometimes it extends into the semiarid transition
zone where, because of the exposure to the sun and general aridity,

84 GERALD V. DAHLING

it becomes nearly prostrate and forms a low-growing mat.

Garrya ovata ssp. goldmanii has very weak reproductive barriers
and intergrades morphologically when in contact with other G. ovata
subspecies. This is especially true in northern Mexico where the ranges
of all G. ovata subspecies overlap. Also, like the other G. ovata
subspecies, G. ovata ssp. goldmanii is often found in, but not restricted
to, limestone areas.

v’ 8d. Garrya ovata Bentham ssp. mexicana Dahling, ssp. nov.

Arbores 3-4 m. altis. Folia late ellipticis vel ovatis, acuminatis vel acutis, obtusis,

petioli (1.1-)1.2-1.8(-2.1) em. longis; folia (5.4-)7.5-11.5(-13.1) em. longis, (2.1-)3.0-
.O(-5.7) cm. latis, glabratis vel parce villosis.

“HOLOTYPE: in the Gray Herbarium, collected on the mountain top above El Cercado,

Nuevo Leon, Mexico, February, 1972, Dahling 1180. isoryrrs: to be distributed.

Trees 3-5 m. tall. Leaves broadly elliptical or ovate, acuminate, acute or obtuse;

petioles (1.1-)1.2-1.8(-2.1) cm. long; leaves (5.4-)7.5-11.5(-13.1) em. long and (2.1-)3.0-
4.0(-5.7) cm. wide, glabrate or sparsely villous.

Flowering from February to March at elevations of 4000-8400 feet
throughout southern Nuevo Leon (Map 5).

REPRESENTATIVE SPECIMENS. Mexico. Nuevo Leon: Chipinque, Calzado 2931 (MEMO);
near the tops of the mts. surrounding Monterrey, about 1/2 days climb, Dahling |
GH); Mesa de Chipinque, Landaw 0935 (MEMO); Villa de Santiago, Leavenworth 157
(mo); Canon Guajuco, Rancho Vista Hermosa Municipio Villa Santiago, Mueller 2031
(GH,MICH,MO,TEX); near Monterrey, Pringle 2104 (MExu); Sierra Madre near Monterrey,
Pringle 2395 (mict,Mo,Msu,us), 11816 (GH,MICH,MO,MSC,TEX), 11817 (MICH,MO,MSC,TEX,US);
at the point farthest E on the Chipinque Rd. in the thorn-oak ecotonal area, Monterrey,
Smith 450 (Gu,rEx); Cerro de la Silla near Monterrey, White 1484 (GH,MICH).

Garrya ovata ssp. mexicana is arboreal or shrubby and lacks the
dense covering of trichomes common to other G. ovata subspecies:
It appears most similar to G. ovata ssp. lindheimeri in leaf shape
dimension, and pollen exine structure. In addition, the fine structure
of the trichomes is similar in its system of ridges, furrows, a?
prominent protuberances.

Garrya ovata ssp. mexicana inhabits the semiarid mountain summits
near Monterrey, Nuevo Leon, Mexico. It is also found in the shrub
zone on steep mountain slopes and on limestone near cliffs. It i
easily distinguished from other members within G. ovata by the spats¢
pubescence and the more arboreal habit.

9. Garrya wrightii Torrey

Garrya wrightii Torr.,

Francisco Mt., Ne rae Be Rep. 4:136. 1857. LecroryPE: on rocks at base of oe

xico;
NY). ISOTYPE: GH

1851-52, Wright 1789 (wv) conto at Copper Mines (Santa Rita), New Meni

SYSTEMATICS OF GARRYA 85

Shrubs, 2-10 feet tall. Crown sprouts developed. Young stems with appressed, silky,
long, whitish- -gray pubescence; with age ve gover ol aragurd tines brown, or dar
gray. Decussate branching. S ate, decussate, pees
or narrowly elliptical to obovate, thick, coriaceous, ion chen Peto (0.3-)0.4-0.5(-1.0)

or sparsely pilose i pa young, bright yellow or gray- sab een Bape ta midrib
wig major leaf veins ee and yellow above, in reli ate inflores-
cences racemose, as ched, pendulous, fasciculate on slay Heiter "2-5 cm. long,
internodes short. Staminate floral bracts linear-lanceolate or oblong, apices acuminate,
recurved, ca. 5 mm. long, green-red-brown, densely pilose with appressed trichomes
on abaxial surfaces; opposite bracts connate only at the base. Flowers nenperdy partially
enveloped and in bract axils. Pedicels 1-2 mm. long. Perianth segments oblong-

minute. Pollen tricolpornte, reticulate, yellow-green or dark orange-brown. Pistillate
inflorescences racemose, branched, pendulous, solitary or fasciculate on branchlets,
isti ct mm. |

|
3
fant
°
=]
qa
~
=]
-~
oO
at
~
©
Qu.
oO
“a
°
°
i]
”
.
i?)
—
°
i=
ag
a
n
te
con
eo
=
o
se
°
at
iS]
joi
a
bel
7)
r
ve)
=]
ga
Len a]

eshy, variously divergent. _ a berry, sc .0-6.2 mm. in diameter, —
dark blue-black or gray with age, beco pg x“ maturity. Dehisc

See rd two or rarely three, globose to path ncitkal. parietal placentas, dark blue
or black.

Flowering March to August at elevations of 3000-7000 feet, ranging
from western Texas, northern Mexico to central and southern Arizona

(Map 5)

Muller 3702 (GH,MICH,TEX,UC); near Lake Santa Maria,
de Las Varas, Santa Clara Mts., i. e 7944 (ariz,us); 15 rd.
Juarez in “The Tinaja,” Wilson Pape sat 8453 (1Ex). Coahuila: Sierra de la 7 aha
km. E of Picacho de zaya, Johnston 9033 ie a del Pino ) W ofc
n

Angos Dod,
573 Reka) “El eat ta de la Nacha, 25 mi. W of La Angos
(cH,Micu); El Rancho de Roble, NE of El Tigre, White 43
del Cumarito, Sierra de la Cabellera, White 4640 (aRIz,GH,MICH,TEX
+: Saddle ek, Chiri

Jones 4261 (aniz,cu,Msc); Ramsey Canyon, Huachuca

: 66 (
Bixbee, McKelvey 621, 626 pad Dragon Mts., near road into Cochise Stronghold

86 GERALD V. DAHLING

Area, S13, T17S, R23E, Newsome eee (ariz); S fork of Cave Creek, —
in s., Ordway 185 (ariz); near Fort Huachuca, Wilcox 9/94 (Ny ,us). Coconino Co.:
n Francisco Mts., Coville & ce 8 (ny,us); about 3 mi. W o ms pe aa
~ 89A, " Dahting 878 (cu); along highway 89A before the entrance to Oak Creek
Canyon, Dahling 1076 (cu); Grand View Trail, Grand Canyon , Thornber 8432 (aRiz).
Gila Co.: San Carlos Indian raat San Carlos to Cazador Spring, Coville 1924
(NY,US); — Rt. Sik ca. si 1/2 mi. W of Miami, Dahling 843 (GH); Superior to Miami
Hwy. near the Ec y line, Gillespie 8633 (cH,us); Calbecue Ridge Fort, pete
Indian Reservation, pte 67-81b ies Graham Co.: W slope of Gila Mts., Goldma
2355 (us); Mt. Graham, Lemmon 8/80 (cu,Mo,us); Mt. Graham, Peebles, Harrison k
Kearney 4487 (anIz,uUs). Greenlee _ Blue aio gaan 2410 (ny,us); Santa Rita
Mts., — 1470 (cu, peli Whitehouse Cany the Santa Rita Mts., McKelvey 506
(GH). Maricopa Co.: Four Peaks, Fe ell 1028 ne saat Co.: 9 mi. W of Cibeque,
Bohrer 1609 (ARIZ); cee Reservation, Fish 29 (uc). Pima Co.: Box Canyon, Santa
Rita Mts., Barr 67-371 (ariz); a aaa Lemmon Mts., road E of Tucson, 2.5 mi. belo
the Prison Camp, Bell 1202 (ariz); Sabino Canyon, Santa Catalina, Benson 6150 (ARIZ);
Rincon Mts., Goodding 23 (ariz); Baboquivari Mts., Goodding 4578 (Ariz); El Hambra
region, Papago Indian Reservation, Goodding 418-45 (ariz); Quinlin Mts., Goodding
sher ;

o

& Lusher 174-45 (ariz). Pinal Co.: near the county line Superior to Miami
Hwy., Bell 1207 (us); Belle Ridge, 0.75 of er distance between the gate and the
top of the ridge, Caldwell 171 (ariz). Santa Cc Nogales, Coville 1620

(us), Purpus 106046 (uc). Yavapai Co.: S of Mt. ara Prescott, Benham 7162 (ARIZ);
ca. 1.2 mi. before the jct. of Hwy. 89 and 89A near Prescott, Dahling 1009 (cH);
along Hwy. 89 ca. 2 mi. W of Sedona, eta: 1020 (cu); ca. 4 mi. S y Prescott,
Dahling 1171 (cu); N of Prescott, along Hwy. 89A, Dahling 1231 (GH); 14 mi. S of
Prescott, Gillespie 8510 (cu, ee “French, Gulch Senator ee paeeant bree (cH);
9.3 mi. NE of Wihoit, Vasek 328 (rsa). New Mexico. Catr : Rio San Francisco

- of Heuvas Rocks, 14 mi. E of Las Cruces, Dunn 7949 (rsa); Finley Canyon, Organ
ts., Everett 10/21/74 me com ape Fort Bayard Watershed, Blumer 152A (us);
152b (cn), ego (us); ca. 1/2 mi. W of Cental along Hwy. 80, Dahling 822 (GH);
ca. 3-1 E of Central on gris, 90, Dahling 1159 (cH); Silver City, Demaree

of ‘Anhiea a Animas Mts., Goldman
, 1347 (ny,us). Lincoln Co.: Godfrey Mts.
e oe 6044 ay Oke Lana G: Florita Mts., Goldman 1487 (ny, ian Boe 9/7/ 03
+ Fresnal Canyon, oo Mts., 6 mi. NE of Alamor
mi.

Horn, Bailey 483 (ny,us); Fra A
Lee & Tharp 46231 (cx), Hudspet Mts., Barlow 876039 (uc), Hueco Tanks, Ber

along Hwy. 80, about 6-10 mi

pe NY,TEX,US).

BR oe is singular in its adaptation to harsh arid environ
it is f d is particularly evident throughout western Texas wher
ound amid rocks and boulders at the base and on the sides

seni cabana teemaneneineliney eliiiblies wi ddeiiant> <aaueniieniiin> aang. — ce, ca eS

—— a

eee

UN A

SYSTEMATICS OF GARRYA 87

of small mountains. Although growth conditions are harsh, the shrubby
G. wrightii commonly thrives there, producing fruit and seed.

Leaves of Garrya wrightii are usually more or less glabrous although
a few rather atypical populations in northern Mexico have leaves
with upwardly appressed pubescence.

Within subgenus Fadyenia, only Garrya wrightii, G. ovata, and
G. grisea have trichomes bearing protuberances. These are large,
well-developed, and similar in G. ovata and G. wrightii, but appear
as small swellings associated with the ridges in G. grisea. Whether
these swellings represent reduced or incipient protuberances remains
unsettled.

Phytochemical investigation of the flavonoids suggests a close
chemical relationship between Garrya wrightii and G. grisea. While
populations of G. wrightii and G. grisea show a high paired affinity
index, populations of G. wrightii and G. ovata generally do not.

The variation in Garrya wrightii is within narrow limits, making
the taxon distinct in most parts of its range. However, this is less
the rule in Mexican populations toward the southern part of its range
where larger leafed individuals, superficially resembling G. laurifolia
Ssp. macrophylla, can be found. Moreover, some of these individuals
also have larger and somewhat foliaceous floral bracts which are more
characteristic of G. laurifolia than of G. wrightii. It appears that the
two taxa hybridize in their areas of contact, which would account
or this convergence. Although G. laurifolia ssp. macrophylla occa-
sionally shows glabrous leaves, it can be distinguished from the
large-leafed G. wrightii forms or hybrids by its more arboreal habit.

2°10. Garrya grisea Wiggins

Garrya grisea Wiggins, Contrib. Dudley Herb. of Stan. Univ. 1, no. 5: 172. 1933.
“HOLOTYPE: collected in a small canyon at the upper end of the meadow at La Encantada,
Sierra San Pedro Martir, 1930, Wiggins & Demaree 4988 (DS). ISOTYPES: GH, CAS.

lane €, acuminate with recu or ght apices, gray-green, appressed ne
on abaxial surfaces; opposite bracts connate only at th lowers solitary, partially
enveloped an e in the b ils. Pedicels 1 mm. long. Perianth segments four,

88 GERALD V. DAHLING

nthers oval, minute, basi-fixed, introrse, opening by longitudinal slits. Filaments
abbreviated. Pollen tricolporate, reticulate, yellow or orange-brown. Pistillate inflores-

paired bracts opposite the styles, sparsely pilose, soon becoming glabrous. Styles two,
rarely three, short, thick, fleshy, persistent. Fruit a subglobose berry, 5-6 mm. in diameter,

i us or occasionally with appressed pubescence near the base, dark blue becoming
dry at maturity. Dehiscence irregular. Seeds two, rarely three, oval to globose, subter-
minal, parietal placentas, dark blue or black.

Flowering February-April at elevations ranging from 4900-7950
feet in northern and central Baja California, Mexico (not mapped).

REPRESENTATIVE SPECIMENS. Mexico. ‘Baja California: San Pedro Martir, Brandegee
106054 (uc); Sierra San Pedro Martir, near Vallecitos, Municipio Ensenada, Breedlove
595 (ps,uc); Cerro Chato Summit, Moran 11107 (ps,sp,uc); Cerro la Sandia, Moran
11531 (ps,Rsa,sp); Cerro Azufre, Moran 11639 (Ds,MEXxU,RSA,UC); N slope of Cerro Azufre,
Moran 11644 (ariz); Volcan Las Tres Virgenes, Moran 11665 (ps,rsa,sp,uc); Calbozo,

mi. SE of Rancho Calbozo, Moran 13488-1/2 (rsa); head of Arroyo Copal, Moran
15464 (sp); S of Cerro Venado Blanco, Moran 15681 (sp); Cerro Prieto, Moran 18122
(sp); Volcan Las Tres Virgenes, Moran 20431 (sp); Cerro Matomi, Moran 20781 (sb):

_ Herbarium material of Garrya grisea is not plentiful because it
is endemic and confined to rugged northern and central Baja California.
Conclusions are based on admittedly limited material and may require
future modification when more material is available for study.
Wiggins (1933), in a note following his original description of the
species, suggested a possible relationship with Garrya wrightii. Leaves
of G. grisea are densely pubescent with upwardly appressed trichomes
on both the upper and lower surfaces. This contrasts with G. wrightii
in which the leaves are typically glabrate or with sparsely appressed
pubescence when young. Some Mexican populations of G. wrig tii
are densely pubescent and are more or less covered on the lowe!

how a system of ridges and furrows
berances. Swelling can be seen periodi-
es but the question of their representing

|
|
|
|
|
|
|
|
|
|
|
|
|
|

SYSTEMATICS OF GARRYA 89

incipient or reduced protuberances remains open. However, in either
case the trichome structures of G. wrightii and G. grisea support the
idea of a relationship between the taxa.

Although the fine structure of the exine is similar to Garrya lauri 'folia,
pollen. size-in .G. grisea-is similar to that.of G. .wrightii.and G. ovata.
Phytochemical data also support the idea of a relationship with G.
wrightii since the chemical paired affinity index is high. While an
affinity, as suggested by Wiggins, is provisionally accepted, sufficient
differences exist to warrant the recognition of G. grisea and G. wrightii
as separate species. The inflorescences of G. grisea show reduced
branching with typically smaller flowers than G. wrightii. In addition,
the latter is adapted to extreme arid conditions and is generally found
at lower elevations than G. grisea. A close relationship with other
Garrya taxa is not indicated.

11. Garrya salicifolia Eastwood

Garrya salicifolia Eastw., Bot. Gaz. XXXVI:463. 1903."HoLoryr: Sierra de la" Laguna,
aja California, Mexico, 1890, Brandegee 259 (cas).

Shrubs to small trees, 6-15 feet tall. Crown sprouts developed. Young stems slender,
slightly pubescent or glabrate, rough, lenticels prominent, becoming reddish-brown
simple, petiolate, decussate,

>
fo]
is)
a
=
5
ga
Qu
2}
fe)
S
n
nn
~
Comal
®
”
cS.
uo)
i
®
~“
fo
o-
w
®
3;
®
fe
<
o

e
lustrous; lower surfaces glabrate or with appressed pubescence near the petiole,
ts Wk J pak, poet Pee i t. Staminate

inflorescences racemose, branched, pendulous, fasciculate or solitary on branchlets,
slender, lax, internodes conspicuous. Staminate floral bracts small; opposite bracts
connate at the base. Flowers solitary in the bract axils. Perianth segments four, pubescent

a
the perianth segments ute, oblong, basifixed, introrse, opening by : itu-
dinal slits Filaments abbreviated, slightly longer than the anthers. Pollen tricolpor ~
reticulate yellow, green, or or. brown. Pistillate inflorescences r oO ed,

; » Ca. "I iameter, gree : i ol ntas
"Tegular. Infructescences suberect. Seeds two, globose, subterminal, parietal placentas,
blue or black.

Flowering August to December at elevations of 5100-6000 feet in
Southern Baja California, Mexico (not mapped).

a i a : i de la Laguna,
REPRESENTATIVE SPECIMENS, Mexico. ‘Baja California: La Laguna, Sierra

E of Todos Sanctos, Alexander, Kellogg & Carter 2305a (ps,GH,TEX,UC); La EN

Brandegee 259 (cu,uc); Sierra de la Laguna, Brandegee 1908 (ow; 1s 9/22/30

Brandegee 106049 (uc); La Laguna, Gentry 4423 (ps), Jones 522101 (ps,cH,POM), 9/

90 GERALD V. DAHLING

(GH,DS,Mo,POM), Nelson & Goldman 7458 (cu); Laguna, Meadow Laguna, Peters 249
(uc),

Garrya salicifolia is difficult to.characterize because only limited
collections are available. Because it is confined to the lower and until
recently generally inaccessible regions of Baja California, few speci-
mens have been collected and herbarium material consists of sterile
or poorly preserved flowering specimens. Consequently, only tentative
conclusions can be drawn about the diversity and variation within
the taxon.

The trichome structure of Garrya salicifolia is interesting because
both a counterclockwise orientation of ridges and furrows and a
counterclockwise structural twisting are present. The occurrence of
this compound trichome orientation is rare within Garrya and is found
elsewhere only in G. fremontii and G. flavescens of subgenus Garrya.
The trichomes of G. salicifolia lack protuberances and are therefore
similar to G. laurifolia and other southern Garryas of subgenus
Fadyenia. An SEM examination of pollen exines also shows a gross
similarity with G. laurifolia. However, the leaves of G. salicifolia
are not as coriaceous as are those of G. laurifolia and more closely
resemble those of G. longifolia. This is also the case with the
inflorescences which are conspicuously lax and have elongated inter-
nodes. While the flavonoid content of G. salicifolia suggests a relation-
ship with G. laurifolia, the correspondence of flavonoids is not close
enough to support a change in taxonomic rank.

Because of its complete geographical isolation, Garrya salicifolia
does not hybridize with other members of the genus. However, the
general morphological and chemical similarity suggests that G. sali-
cifolia was evolutionarily derived from CG. laurifolia-like ancestors.

12. Garrya fadyenii Hooker

rari: fadyenii Hook., Icon. Pl. t. 333. 1840. 'aocorypr: farnaica; Fadyen H1859/73-56

Fadyenia Hookeri Endl. Gen. Suppl. IV:38. 1847. Based on Garrya fadyenit Hook.
Shrubs or trees, 12-25 feet hi i

eae gh. Crown sprouts present. Young stems pilose

densely tomentose, with age becoming glabrate and dark reddish-brown oF black.

developed; midri : om a
; rib and m : . . b
nubesceiies bisisae Sica ivi lateral veins conspicuous above, veins obscured >

ne tainty
i ulate
or solitary Gv becicat ee eronne , branched, pendulous, fasci¢
OF Botany pyres about 4 cm. long, compact, densely pubescent; inte odes
preuous. Staminate floral bracts ovate to | Sek aot e sometimes

ee

SYSTEMATICS OF GARRYA 91

concave, 3-6 mm. long, opposite bracts connate at the base. Flowers solitary in bract
t

Town or orange. Pistillate inflorescences racemose, bra
or solitary on branchlets, ca. 6 cm. long, lax, gage or internodes prominent. Pistillate
o2Z2ec

Opposite bracts connat €,

varies bicarpellate, rarely tricarpellate, unilocular, oval, subsessile, imes
paired bracts opposit s, tomentose. Styles two ean three, persistent, short,
thick, fleshy, nearly erect. Fruit a subglobose berry, blue or brown, becoming

brittle at maturity. oe ean Seeds two or ie three, globose, subterminal,
parietal placentas, dark b

Flowering during the winter at elevations ranging from 2000-8130
feet throughout the Greater Antilles (Map 9).

REPRESENTATIVE SPECIMENS. Cuba. Oriente: Guantanamo, Ekman 10305 (us); Sierra
eS estra, on top of Palmamocha, Ekman 14316 (ny); Pico Turquino, Sierra estra,
ct 10710 (cu,ny); Arroyo Frio, Sierra Jurias, S. Baracoa region, Leon 72197 (Ny);
“Sa Peladero Arriba, Sierra Maestra, Lopez-F. 2174 (us); Farallon de la Perla peal
P afer 8755 (cu), 8782 i) Oriente, Wright 492 (cu,Mo). Las Villas: summit of Pico

otrerillo, Trinidad Mts., Br. Alain 6339 (cu,us); mts. of Trinidad, Pico Pouediie,
Ekman 14019 (micu); mts. of the ‘ cewiita Trinidad group, valley of Rio Hanabanilla,
Ekman 18486 (us); San Blas-Buenos Aires, Trinidad Mts., Gonzales 637 (mic); Las

Jack 5959 (F,GH,us); Santa pase Jack 6816 (GH);

as Lagunas, Buenos Aires, Santa Clara, Jack 6826 (us
= orton 10373 (us); Santa Clara near reservoir
ressler, Jones, Schubert & Wilson 212 (cu). Dominican Republic. B

la Selle, Port au Prince, near Dufresnay, Ekma

a la Selle, Proctor 10759 (vs); Morne de Ceneay

. — de, se at no. 4, = sh & Taylor 1255 (ny,us)" Jamaica
ap, Crosby, Hespenheide & Anderson a (DUKE, MICH, TEX); Gordon

(us); “deeseg Gap, eras 7/11/32 (Ny); New Haven

sie vicinity of St. Helen’s Gap, Maxon & Killip 1334 (GH,US).

3 ams 8778 (pukk); Craig Hill, erren 11198 (puKr,Mo); Cinchona, And:

Aceh (GH,MicH); ridge from Morces Gap to John Crow Peak, near the Portland border,

Sud son & Sternberg 3460 (DUKE,GH,MICH); near Belli o

eeward slopes, Harris & Laurence C1518] (us); near Cuichona,

y below Cinchona, Maxon & Killip

era Cinchona, Phillipson 908 (Mo);

s., Proctor 23561 (MICH,NY,TEX);

s a

Garr rya fadyeniti is restricted to the Greater Antilles andi
floral bracts

isolated from other Garrya taxa. Except for the smaller

92 GERALD V. DAHLING

Maps 9-10. Distri i
Re Shares — of Garrya: 9, G. fadyenii—West Indies. 10, G. corvorum (dots);
7 g Central America. ee

G. laurifolia s5P

SYSTEMATICS OF GARRYA 93

and the tomentose leaf pubescence, G. fadyenii morphologically
resembles G. laurifolia. The trichomes of G. fadyenii appear similar
to those of G. longifolia, G. laurifolia, and G. salicifolia, showing
a counterclockwise orientation of very coarse and widely spaced ridges
and furrows. Protuberances, as with the other southern Garryas, are
absent. The size, shape, and fine structure of G. fadyenii pollen is
quite similar to that of G. laurifolia. An examination of the flavonoid
constituents within G. fadyenii reveals a high number of unique
compounds. These probably developed in response to environmental
influences and geographical isolation which allowed for their
independent development unimpaired by the effects of hybridization
and gene flow.

Garrya fadyenii is a tree species commonly associated with limestone
soils at high elevations. While approaching G. laurifolia in some
respects, it is sufficiently distinct to warrant specific status.

13. Garrya longifolia Rose

we lashes see Rose, Contrib. U.S. Nat. Herb. VIII: 55. 1903-0

Lg TYPE: “Mexico. ‘Morelos, Sierra de Tepoxtlan, 1899, eit 6998 (US). ISOTYPE:
I. GH:

‘Garrya gracilis Wang., Das Pflanzenreich IV, 56a: 16. 1910. LecToryPE: Mexies Morelos,

‘i ierra de Tepoxtlan, 2500 meters, Pringle 8363 (B), not seen. ISOTYPES: F, GH, MI
, US

Shrubs to trees, 10-25 feet high. Crown sprouts well-developed. Young stems
pubescent, glabrate with age, silver-gray, green, reddish- brown or dark gray. Branching
decussate and ascendin ng. Stipules absent. Leaves simple, petiolate, decussate, ve ng-
elliptical or lanceolate, delicate, subcoriaceous, persistent. Seton (0.5-)0.8-1.4(-1.6)
cm. long; opposite petioles connate at the base, adnate to the stem. pe ‘blades
(4.5-)7.0-16.0(-18.0) cm. long, (1.5-)2.5-5.5(-6.5) cm. w wide; apices acute, apic culate or

rang :
aan Oi aa pendu ee fasciculate or ony. on branchlets, lax, sparsely
n

fonnate at the base. Flowers solitary, often gna in the bract axils. Pedicels minute.
. nt biel young. Styles
two, rarely three, persistent, short, thick, fleshy, sem iinet Fruit a berry angele
th dark blue and brittle at maturity. Dehis cence ieee eeds
two, rarely three, globose, subterminal on parietal placentas, dark blue or black.

94 GERALD V. DAHLING

Flowering from January through March at elevations ranging from
4200-8700 feet throughout central Mexico depending on the rainfall
(Map 7).

REPRESENTATIVE SPECIMENS. “Mexico. Durango: ca. 20 mi. SE of El Salto along Hwy.
40, Dahling 372 (cu); Cerro Huehueto, S of Huachicheles, ca. 75 mi. W of C. Durango,

tatlan on road to Teapa at km. 123-124, Anderson & Anderson
M

e el camino de Xochipala al aserradero Agua Frio, approximately 43 km.
Chilpancingo, Rzedowski & McYaugh 304 (MICH,MSC,TEX); shelf of bluff W of Chilpan-
cingo, Sharp 441404 ( ) ali Si de San Sebastian, 15-30 km. N of Mascota
on the road to San Sebastian, Anderson & Anderson 5966 (micH); 7-8 mi. NW 0
Los Volcanos along road between Ayutla and Mascota, McVaugh 12199 (MEXU,MICH);
NE slopes of Nevado de Colima below Canoa de Leoncito, McVaugh 1347
(MEXU,MICH,TEX); Sierra de Cuale, SW of Talpa de Allende, SW of Pidra Rajada, McVaugh
14367 (MEXU,MICH,TEX), 14367A (micH); N of La Cuesta below pass to Talpa de Allende,

741 (cu); Cajones, Temascaltepec, Hinton 2385 (mo,us), 3201 (F,mMo,us), 3666 (vs).
es NW of Quiroga in view of the

e
main highway, Dahling 670 (cx); Tancitaro, pedregal, Hinton 15641 (us), 15698
(GH,MEXU,MICH,US); between Jerdan and Uruapan, Ikhangman 3255 (Mexu); E of San
Jua evo about 8 km. S of Uruapan, King & Soderstrom 4755 (MEXU,MICH,TEX,US);

de Salto and La Polvilla about 18 mi. E of Morelia, King & Soderstrom
5102 (MEXU,MICH,TEX,US); 2 mi

W of Aguililla, about 6-7 km. S of Aserradero Dos
; Sierra de San Joaquin, 11 km. al S de Tlalpujahua,
90 (GH); wet ravines near Patzcuaro, Pringle

s 612 (

pe

Cuernavaca, Lyonnet 730 (MO,uU

peeves — de Tepoxtlan, Pringle 6988 (F,Micu,Mo,us); Sierra de Tepoxtlan, P ringle
»MICH,MO,US); Sierra de Tepoxtlan, Pringle 9819 (F,Mo,us); 10 km. NW de Tepoxtlan

sobre la autopista Mexico-Cuernav

Fe aca, Rzed. + Oaxaca,
Bruff 1276 (mexu); Cerro San Blas Huan owski 26931 (MICH, MSC). axaca: O: per
wy.

of _ 190
, Dahling 260 (cu); about 27-1/2 mi. S of Oaxaca
on Hwy. 190 near top of mt. chain in view of the Microwave cue Dahling 999 (cH).

Garrya longifolia is distin

its long, branching, guished from other Garrya species

lax inflorescences. Both the male and female

=

SYSTEMATICS OF GARRYA 95

inflorescences show elongated internodes and reduced floral bracts
(Plate 8A). Although plants of this species are typically trees, shrubs
are sometimes found in disturbed areas near roadways. The larger
individuals usually are located in forested areas at elevations up to
9000 feet.

Over the years, the taxonomy of Garrya longifolia has become
somewhat confused. The description of G. longifolia in Wangerin’s
monograph (1910) states “Flores masculi ignoti.” They were unknown
to Wangerin because he placed the male-flowered specimens in his
new species, G. gracilis, which was described in the same work.
This was curious, since the female elements of G. gracilis, although
collected outside the G. longifolia type locality, were themselves typical
members of G. longifolia as described by Rose (1903-05).

Garrya longifolia appears closely related to, and often intergrades
morphologically with, G. laurifolia when they are in contact. SEM
investigation shows that the trichomes of G. longifolia, G. laurifolia,
G. salicifolia, and G. fadyenii are quite similar. The leaves of G.
longifolia are subcoriaceous and therefore resemble leaves of G.
salicifolia and G. laurifolia ssp. racemosa. The pistillate inflorescences
are also quite similar to the lax inflorescences of G. salicifolia. Although
the size of the G. longifolia pollen is similar to that of G. laurifolia,
the fine structure of the exine is quite different and can be readily
distinguished from the pollen of other Garrya taxa.

While complete reproductive isolation has not been achieved, and
although hybridization with Garrya laurifolia does occur, G. longifolia
is sufficiently distinct in most areas to justify its retention at specific
rank. Moreover, this is also suggested by phytochemical data showing
that the flavonoids of G. longifolia are rather different from those
of other members of Garrya.

14. Garrya laurifolia Bentham

Shrubs to tall trees, 10-35 feet tall. Crown sprouts numerous. Young stems tomentose,
with age reddish-brown and glabrate. Decussate, sometimes weakly ascending branch-
ing. Stipules a ea imple, petiolate, decussate, lanceolate to elliptical or

f
inconspicuous. Staminate floral bracts ovate to lanceolate, 3-6 or rarely 8 mm. long,
apices acut Focus ee eke te at the base. Flowers solitary, partially
enveloped and in the bract axils. Pedicels 2-3 mm. long or lacking. Perianth segments
four, oblong to elliptical, connate at the apices, abaxial side variably pubescent, adaxial

96 GERALD V. DAHLING

side gar Stamens four, free, short, alternating with the perianth segments. Anthers
oblong, 2-4 mm., basifixed, introrse, opening by longitudinal slits. Filaments about

inflorescences racemose, bran , pendulous, fasciculate or solitary on branchlets,
lax, tomentose, 5-12 cm. long, solartiodes conspicuous. Pistillate floral bracts foliaceous,

r

with age; opposite bracts connate at the base. Flowers solitary in the bract axils.
Pedicels minute, rarely 5 mm. or larger. Ovaries bicarpellate, rarely hago
unilocular, subglobose, sometime , pilose, glabra
with age. Styles two, (aera dares: persistent, short, thick, fleshy, * Aieleeak ruit a
berry, glabrate, green, dark blue, rarely whitish-gray, brittle at maturi rity. Dehiscence
irregular. Seeds two or rarely three, oval, subterminal, parietal placentas, dark blue

k.

KEY TO THE SUBSPECIES

A. Leaves broadly elliptical, 10-18 cm. long and 4. os © om: wide. 22-2. sia
OHI. Scie oes ONE ee lanl arabic 14b. G. laurifolia ssp. macrophylla.
A. os long elliptical sometimes oblanceolate, 6.0-19.0 cm. long and 2.0- 5.5 cm.

B. vane re leaves tomentose below
B. Mature leaves glabrate or slightly pubescent ees
C. Leaves oblanceolate-long elliptical Some pistillate floral bracts distinctly leaf-

Pee ind 4a. G. laurifolia ssp. laurifolia.

like and as long as the foliage leaves... .. . 14c. G. laurifolia ssp. racemos
C. Leaves elliptical or long elliptical. Pistillate floral bracts not expanded to the
Size of foliave leaves 3 2. c vice cscs ica 14d. G. laurifolia ssp. quichensis.

“14a. Garrya laurifolia Bentham ssp. laurifolia

Garrya laurifolia Benth., Pl. cage 14. — HOLOTYPE: Mexico, near Guanajuato,

1839, Hartweg 81 (xk). IsOTYPEs: xk,
i parece laurifolia Endl., Gen. Suppl. IV: 37. 7 Based on Garrya laurifolia Bent th.

Garrya _ — Pl. Hartweg. 51. 1840. "HoLoryPE: Mexico, hills near Regla,
1839, sores weg 385

Fadyenia oblonga Endl, Gen. oe IV:38. 1847. Based on Garrya oblonga Ben th.

— laurifolia var. genuina Wang., Das Pflanzenreich IV, 56a: 14. 1910. Based

n Garrya laurifolia Benth.

Le rubs to trees, 6-20 feet high. Crown sprouts numerous. Young stems pubesce

eaves elliptical-long elliptical plane; upper surfaces glabrous, — bre
e; lower surfaces tomentose, whitish- -gray or brown with age. Ovary pu scent,

Frequent with prominent adnate bracts opposite the styles. Fruit glabrate, dark blue

Flowering February through March at elevations of 2000-9000 feet
throughout central Mexico (Map 8).

REPRESENTATIVE SPECIMENS, Mexi ico. Aguascali entes: Si 7
ierra de Laurel, ca. mi.

e > fa aires & Koelz oe (MicH). es Arroyo Hondo, Sierra Charu,
(MEXU,MICH,US); Guayanopa Canon, Sierra Madre Mts., Jones 1905

Jones 77759 (pom); Mojarachic, Knoblock beet
no
7053 (msc,us); below Basaseachi sang , SC), Knoblock 5788 (msc,vC), F r 825

; SC,TEX,US).
Breedlove 18839 (cas, RSA); i. N of Estacion Coyotes, N of La Dieneguit®

1
Pete esi Fisher 44274 (cu,mo); 10-12 mi. W of La Civ udad, a

24?
Garrya macrophylla Benth., Pl. Hartweg. 50. 1840. Ho
M4 Encarnacion, 1839, Hartwe.

SYSTEMATICS OF GARRYA 97

38-40 mi. W of El Salto along the rd. to Mazatlan from C. Durango, Gentry & Gilly
10613. (mEXU,MiICH,TEX); San Luis, 51 rd. mi. NW of Coyotes, Maysilles 7215
(mExU,MICH,US), Maysilles 7951 (MICH,TEX); Quebrada de San Juan, 26 mi. N of railroad
at Coyotes on rd. to San Luis, Maysilles 8299 (MExU,MICH); 34 mi. N of railroad at
Coyotes, Laguna de Progresso, Maysilles 8384 (micu); Llano Grande, 42 mi. WSW

°
baa)

2049 (us). Guanajuato: along Hwy. 110 between Santa Rosa and Guanajuato near
the 88 km. post, Dahling 294 (cu); 14.5 mi. from Guanajuato on rd. to Dolores Hidalgo,
Johnston 2643 (MEXU,MICH,TEX); mts. ESE of San Jose Iturbide and ca. 5 mi. W of
Cerro Zamorano near Mesa de Gato, McVaugh 10387 (mMExU,MICH). Hidalgo: Cerro
Juarez, 8 km. N de Tasquillo, Qunitero 2932 (MICH,TEX); Penas Largas, cerca de Texoantla,
municipio de Real del Monte, Rzedowski 22194 (MICH,MSC,TEX) Mexico: Valley of Mexico,
Rose & Hay 5366 (us). ‘Michoacan: Quincho Cascade, vicinity of Morelia, Br. Arsene
5409 (cu,Mo,us). ‘Nayarit: Volcan Caboruco, Paray 3407 (mick). ‘Puebla: Chipantla,

3.5 km. al S de Magdalen Jicotlan, Cisneros 2257 (micu,Rsa); near Reyes, Nelson 1730
(ny,Us).“San Luis Potosi: San Luis Potosi, Schaffner 2gi (us). "Veracruz: near the Puebla

order, Martinez 81 (us)."Zacat 38 km. al Wde Jalpa sobre la carretera a Tlaltenango,
30 km. del entranque con la carreterra Jalpa-Juchipila, Rzedowski & McVaugh 1032

(MICH)

Garrya laurifolia ssp. laurifolia is a common constituent of the
shrub zone on high mountain slopes. Although it usually inhabits
the transition region between the arid lowland and the moist highland
forests, G. laurifolia ssp. laurifolia is not restricted to this zone but
occasionally extends into the forests. Individuals at these higher
elevations commonly have narrower and less coriaceous leaves than
is typical for other members of the taxon. Of the other subspecies
within G. laurifolia, ssp. macrophylla appears to be closest to ssp.
laurifolia as shown by similarities in pollen, pubescence, and leaf
shape. Moreover, the flavonoid content of all the subspecies is fairly
uniform suggesting close infraspecific relationships.

4b. Garrya laurifolia ssp. macrophylla (Bentham) Dahling,
comb. nov.

LOTYPE: Mexico, Barranca del

50 (k).
“aay macrophylla Endl., Gen. Suppl. IV:38. 1847. Based on Garrya macrophylla
enth.

Shrubs to trees, 8-18 feet high. Crown sprouts developed. Young stems pubescent.
Leaves large, somewhat revolute, broadly elliptical, up to 8 cm. long a m. wide;
leaf upper surfaces glabrous, lustrous, reticulate; leaf undersides tomentose, gray Or
Fruit glabrate, dark blue, brittle at maturity.

Flowering March through April at elevations of 3500-8850 feet
from northern to southern Mexico (Map 8).
13 mi. NE of Zimapan, Anderson &

2 mi. N of Zimapan near the oF
along Hwy. 85 near the 89 km.

REPRESENTATIVE SPECIMENS: “Mexico. ‘flidalgo:
Spada 4028 (cu,micu,us); along Hwy. 85, ca.
m. post, Dahling 318 (cu); ca. 11 mi. SW of Jacala,

98 GERALD V. DAHLING

post, Dahling 590 (cu); at second zinc mine site, 12 mi. up rd. W of Mexico 85,
mi. N of Zimapan, Mears 292A (TEx); NE of Jacala at km. 281, Puerto de la Zorro,

Moore 3802 (mexu); 7 km. NE of Jacala, Quintero 1331 (micH,msc); 12 ke from

Alfajayucan, Quintero 2056 (ps); 3 km. N of Jacala, Rzedowski 27675 (ps).

E de Mexico, between Tultenango and Tulpetlac, Villada 5 (MExu).Nuevo Leon: Potosi

Mt., along the Communication Tower Rd., a few miles NW of Galeana, Dahling 100

u onterre Nombres

municipality of Zaragoza, Cerro del Viejo, Meyer & Rogers 3040 (mo); Sierra Madre
ts., Monterrey, Mueller & Mueller 210 (GH,MEXU,TEX); between San Francisco Canon
and Pablillo ca. 15 mi. SW of Galeana, Mueller & Mueller 373, 382, 392, 634 (GH,MICH,
TEX); Guajuco Canon, El Cercado, Mueller & Mueller 1313 (GH,MICH,TEX,US); Sierra
Madre Oriental above the Puertos, El Cercado, Mueller & Mueller 1341 (GH,MEXU,MICH,
TEX,US); Cerro Potosi, Rzedowski & Hinton 17281 (micu); Cerca de Los Hoyos, Km.

Tultitlanapa, near Oaxaca border, Purpus 3370 (cu,Mo,us), Purpus 3371 (No
km. a e Tehuacan cerca del Puerto Olivo, Rzedowski 18850 (MEXU,MICH,TEX). San
Luis Potosi: 20-35 mi. W of San Luis Potosi along the road to Rio Verde, Gentry,
Barclay & Arguelles 20166 (us); Km. 55 carretera San Luis-Rioverde, Rzedowski 4471

tt
erro Carrizo, Bartlett 10516 (F,micu,us); Gomez Farias re ierra
above Rancho del Cielo Biological Station, Johnston 7407 (TEx); Lagua Zarca, SE 0
La Joya de Salas, Martin H12 (micu); La Joya Prieta, 4 km. SE of Carabanchel on
trail to Montechristo, Martin H13 (micu); Villa Mainero, Arroyo La Oveja, 10 km.
SW of Pueblo, Martinez 2920 (mexu); 8 mi. E of Dulces Nombres, Meyer & Rogers

1220 (GH,MEXU,MICH,MO,TEX); Coscomatepec, Matuda & Standley 1325 (GH,MEXU,
MICH,MO,TEX); Cerro San Cristobal, Orizaba, Miranda 4861 (mExu); El Esquilon,
municipio de Jilotepec, Ventura 4727 (MICH,RSA,TEX); Acejete, Ventura 5308 (MICH,RSA,TEX)-

Garrya laurifolia ssp. macrophylla can usually be distinguished
from other Garrya taxa by the large and extremely wide leaves. While
distinct in some areas, it intergrades extensively with the closely related
ssp. laurifolia. F requently, G. laurifolia ssp. macrophylla is a shru
gues dry localities but it becomes arboreal in moist sites. Also, the leaf
dimension varies and appears related to the degree of aridity since
in drier semiarid sites the leaves are smaller and less wide. AS Wi
most Garryas, G. laurifolia ssp. macrophylla is well-adapted to aridity
and can maintain a green color even during the dry season when
other plants have long turned brown.

- 14c. Garrya laurifolia Ssp. racemosa (Ramirez) Dahling, comb. noV-

: pre racemosa Ramirez, Anal. Inst. Med. Nac. Mex. 1:298. 1895. HoLoTyPE: Mexic®
—— acto, Ramirez 431366 (us
a arrya taurifolia var. lanceolata Wan erin .
) i. E 56a:16. 19
“ HOLOTYPE: Mexico, Uhde 34 (B), not to sna Herpes i

SYSTEMATICS OF GARRYA 99

Large shrubs and trees, 8-35 feet high. Crown sprouts developed. Young stems
pubescent. Leaves long-elliptical or oblanceolate, up to 4 cm. wide and 18 cm. long;
margins plane; upper leaf sapere ppg par sat Hes de Arrtetes leaf surfaces
glabrate, bright green or green-yellow with sparse trichomes. Floral bracts as
large as the foliage leaves. onan caesar Jered fe reset = maturity. Fruit
dark blue or black.

Flowering from December through March at elevations of 3600-
10000 feet throughout central Mexico (Map 8)

REPRESENTATIVE SPECIMENS. Mexico. ‘Distrito Federal: ce de los Leones, W of
Mexico City, Dahling 567 (cu); Canada de Contreras, Espinosa 505 (micu,TEx), 603
(us), Rzedowski 18276 (MICH,NY,TEX,US); Canada de Con aa junto al acueducto,
Espinosa 649 (meExU,MICH,TEX); La Marquesa, Desierto de los Leones, Hernandez 529
(TEx); Canada de Contreras, cerca del Cuarto Dinamo, Hernandez 3/21/65 (msc); Mixoac,
Lyonnet 1262, 2971 (us); Desierto de los Leones, Lyonnet 2592 (us). aca Omiltemi,
20 km. al W de Chilpancingo, Rzedowski 15964 (micu). “Hidalgo: Money Station,
Trinidad, Pringle 9/9/06 (micu); 45 km. al E de Re oe = Monte, iis 12/19/64
(MICH,Msc). Jalisco: Cerro de Tequila al S Tequila, Diaz 414 (micu); NW slopes of
Nevado de Colima above Jazmin, 2-3 km. above the peters of Fl Isote, McVaugh

along lumber rd. E of crossing called La Cumbre between El Chante and Cuzalapa,
McVaugh 23155 (micu). “Mexico: near Amecameca, Sacromote Hill, Beauchamp 1926
(Mo); San Nicolas, Bourgeau 997 (us); Km. 26.5 de la autopista Mexico- beer iehanngt
532 (msc); Valle de Bravo National Park, rd. leading to Valle de Bravo, 2

mi. from Hwy. 15, Dahling 650 (cu); Nanchititla, Hinton 3605 (mo); Nanchititla- iia:
scaltepec, Hinton 3606 (us); Cumbre Temascaltepec, Hinton 5922 (us); Cruces Tema-
scaltepec, Hinton 6852 (micu,us); Temascaltepec, Hinton 8948 (us); Ixtaccahautl, Purpus
237 (mMo,ny), 1695 (mo,us); Sarrania de Ajusco, Pringle 12363 (us); 3 km. al E de San
Rafael, municipio de Tlalmanalco, Rzedowski 19344 (micH,Msc,TEX); 4 km. al NW de
Santiago Tlazala, Rzedowski 27108 (MiCH,MSC,RSA,TEX); Sierra de ye pet 6 km.
a de Co ; :

mi. NW of Los Omos along rd. to Mocorito, — of Sinaloa and Vela, Sierra
Surotato, Breedlove & Thorne 18264 het RSA); 3 N of Los Ornos along rd. to
Ocurahui rotato, Uriedtove & Thorne 18309 (Mo,RSA);
Ocurahui, Stave Surotato, Gentry 6370 (Fr, GH,MICH,MO). Unknown: mts., near Cuapinalpa,
Gregg 686 (mo).

Garrya laurifolia ssp. racemosa is a distinctive tree, reaching heights
of over 35 feet. It is often an important constituent of the secondary
communities within high altitude coniferous forests where it grows
well in moist sites. Although occasionally G. laurifolia ssp. laurifolia
and ssp. macrophylla extend to the forest edges and intergrade with
SSp. ihdsaasie the latter is gonerayry separated from the other subspe-
Cies by its in ce forari ditions. The most distinctive character
of G. nome ssp. racemosa is its large foliaceous bracts found
uP the pistillate inflorescences. They reach lengths of nearly 20 cm.

aI

laa, Garrya laurifolia ssp. quichonsts (Smith) Dahling, stat. nov.

»

100 GERALD V. DAHLING

: Garrya laurifolia var. quichensis Smith, Bot. Gaz. 54:237. 1912. HoLotyPE: Guatemala,

Department Quiche, San Miguel, Uspantan, 1892, Smith 3175 (us).

Shrubs to small trees, 10-35 feet high. Young stems pubescent, becoming glabrous,
reddish-brown, or dark-brown with age. Leaves elliptical-long elliptical; petioles
(0.8-)1.0-1.4(-1.6) cm. long, leaf blades (6.0-)7.5-10.5(-12.5) long, (1.8-)2.0-4.0(-5.5)
cm. _ wide; margins plane; upper leaf surfaces glabrous, lustrous, reticulate; lower: leaf
]
not expanded as foliage leaves. Ovary sparsely pubescent, becoming glabrous. Fruits
glabrous, dark blue or black at maturity.

Flowering December to March at elevations of 4500-11700 feet
throughout central and eastern Guatemala (Map 10

REPRESENTATIVE SPECIMENS. Costa Rica. *Cartago: sey 5 km. N of San Rafael on Costa
Rica Rd. 8 to Volcan Irazu, Almeda 684 (pukE); Volcan Irazu, S slope, Hatheway
& Perez 1476 (us); S slopes of Volcan de Turrialba, Standley 34968 (us); N of Irazu,
Potrero on Bridle Path, Stork 1271 (micu); 1 km. from Rio Birris, —_ 9 km. NE
of Cartago, Wilbur 14297 (cu,micu,Mo); NW of Ascuncion of the Cerro de La Nuerte,

about 16 km. NW of Ascuncion, Wilbur 14585 (puKr); Volcan Turrialba, trail from
Finca Quecada, Wilbur 14640 (puxr); near San te border, about 2 mi. NW of La
Ascuncion and El Empalme on Interamerican Hwy., Wilbur 17392 (puKE); Volcan

of Valle de los Conejos along upper Rio Talari, Burger & G some pa 8299 (DUKE,MO);
Cerro de las Vueltas, Standley & Valerio 43976 ( (us). “sail ‘Chimaltenango:
Chichavac, Skutch 493 (cH,micH,us), 700 (us); Cerro de Tecpan, region of Santa Elena,
Standley 58738 (Gu); Barranca de La Sierra, SE of Patzun, Standley 61567 (Gu); between
Los Idolos and Chocoyos, Wallenta 6220 (vy). Guatemala: near Finca La Aurora, Aguilar
292 (Gu); 20 mi. E of Guatemala City, Molina 13539 (ny); 20 km. N of Ciudad Guatemala,
Molina & Molina 12394 (us); slopes of Volcan de Pecaya, between San Francisco
Sales and = base of the active cone, Standley 80720 (GH); Jalapa, — Jalapa
d Montana Miramundo, Btaperrnant: 32879 ( (c 3H). ‘Huehuetenengo: 4 mi. of San
Maina tsar a. rd. to Barilla i tg TEX);
5 mi. S of San Juan Ixcoy, efit de los Cuchumatanes, Breedlove 11512 (msc); NW
of Cuilco, up Cerro Chiquihui above carrizal, Yate 50810 — mts. SE | fe)

on a (i Rd., ontreras 4969 pieces: San Mi uel hh Heyde 31 75 (us);
S of Nebaj, Proctor 25177 (TEx); San Miguel Uspantan, ak 3175 (us).Sacatepequez:
Volcan Acatenango, Kellerman 4805; Volcano Aqua, Kellerman 7443 (NY, = eg (GH,US);
near San Luca, spo, Me 15341 (GH,NY); Volcan es between Santa Maria de Jesus

oO neja trail to Paraiso,

municipio of Tenejapa ; Beadlane 6887 (cu); along trail - io me Sone | Antonio from

Tenejapa, a das of Tenejapa, Breedlove 7020 0 (mic); along trail from a

center to Pokolum, Barrio of Ho’Ho’Ch’ en, Paraje of Shishintonil, Breedlov

So »MICH); Chabul Ch’en in the paraje of Siganil Ha’, municipio of Tenejapa, Breedlove
4 (micu); 5 mi. N of Chamula Center 2 th rd. to Chenalho, municipio of Ch hamula,

|
|
|
|
|
|
|
|
|
1

SYSTEMATICS OF GARRYA 101

Breedlove 8146 (cu); SW of Hwy. 190 near Rancho Nuevo, 9 mi. SW of San Cristobal
Las Casas, B 3H,MEXU,MICH,TEX); barrio of Chinhk Ha’, paraje of Yashanal,
municipio “ Tenejapa, ese e 9380 (TEX); 1 mi. W of Nabenchauk along Hwy.

90, municipio of Zinacantan, Psion 9527 (Msc); near the NE border of Aguacaten-

along trail to Kulak’tik, municipio of Tenejapa, Breedlove 10936 (msc); along creek
near the center of Amatenango, Breedlove 12170 (mrExu,msc); Teopisca, Goldman 975
(us); Zinacantan Center Valley floor, herent re — XU,MICH); Kampan ei a dhe
Hwy. 190, 3 mi. W of Paraje Navenchauk, pio of Zinacantan ree

(MEXU,MICH); between Teopisca and San Cris ovat lataede 2727 (MEXU); Pous ti

(DUKE); 3 mi. NW of Comitan, Webster, Miller & Miller 12945 (mExu,Mo).“Panama.

/Chiriqui: E rie of Volcan Baru, from 3000 m. to just below the summit, Mori &
MO).

Bolten 7433

Garrya laurifolia ssp. quichensis is confined to Central America
and grows as a large shrub or tree in high mountainous areas and
on volcanoes. As is common with other Garrya taxa such as G.
corvorum, G. fadyenii, and G. ovata, G. laurifolia ssp. quichensis
is frequently associated with limestone soil and outcroppings. It is
the most geographically isolated of the subspecies of G. laurifolia
since the Isthmus of Tehuantepec cuts the range of G. laurifolia
and isolates G. laurifolia ssp. quichensis to the south.

Study of flavonoid constituents shows that Garrya laurifolia ssp.
quichensis has become chemically different since its paired affinity
index is distinctly lower than that of the other subspecies. Since its
range does not overlap with the other G. laurifolia subspecies, these
distinguishing chemical differences most likely developed in response
to the greater isolation maintained by G. laurifolia ssp. quichensis
throughout its history. Nevertheless, other morphological characters
such as habit, leaf, pollen, and trichome structure support the connec-
tion with G. laurifolia and justify an infraspecific rank.

LITERATURE CITED
Atston, R. E. & B. L. Turner. 1963. Biochemical Systematics. Prentice Hall, N.J.
4 pp
Arsen, E. 1895. eo para la materia medica Mexicana. Part I. Inst. Med. Nac.,
0, pp. 183-19

Mex
AXE non, D. me 1944. oe Sonoma flora. Carnegie Inst. Publ. 553:176-206.
1950. Studies in late Tertiary paleobotany. Carnegie Inst. Publ. 590:1-22,

217-306.
Rese sai Evolution of the modern Tertiary geoflora. Bot. Rev. 24:433-509.
oo e Miocene trapper flora of southern Idaho. Univ. Calif. Publ.
Ge BRE a Sei. 51: 2131-132.
Bacicatupt, R. 1924. A adapt study of the genus Garrya. Unpublished Master’s

pg en ete London. pp. 72-83.
2. 1879. Traite du dévellopement de la fleur et du fruit. Adansonia 12:262-269.
BenrHam, G. 1839-1857. Plantae Hartwegianae. London. pp. 14, 50-51,

102 GERALD V. DAHLING

————— & J. Hooker. — Genera Plantarum. L. Reeve, London. pp. 256, 947-952.

Bourgue.ot, E. & H. Herissry. 1902. Sur un glu eee nouveau, l’aucubine, retire
des graines d’Aucuba isseais a, Bot. Centralblatt 90

Cuak, L. 1938. Standardization of terms for vessel Sudaiee and ray width. Trop.
- 55: 16-23

Cuaney, R. W. & H. L. Mason. “ae eens flora from the Santa Cruz Island,
Cali a oo. Inst. Publ. 415:1-24.

_. A Pleistocene flora ey ‘ie asphalt deposits at Carpinteria, California.
. er Publ. 415:45-49,

puree on the Standardization of Terms of Cell Size, International Association

ood Anatomists. 1937. Standard terms of length of vessel members and wood

pales Trop — ds 51:21.

CoutT-er, J. M . H. Evans. 1890. A revision of North American Cornaceae II.

oe

Dr CANDOLLE, A. 1869. Prodomus ime sama Regni Mil agente 16(1):486-488.
ora. C., C. R. Smirn, A. E. Lippman, S. K. Fic & J. Herran. 1955. The structures
the diterpenoid alkaloids dicirifsline and Seapblabhiclne: Jour. Amer. Che
ee 77:4801, 66.
Dorr, E. 1930. Lele floras of California. Carnegie Inst. Publ. 412:1-108.
Eastwoop, A. 1902. New a from the Sierra Nevada Mountains ae California.
Pro slayer Acad. 2(9):2
———_. otes on seni with descriptions of new species and key. Bot.
Gaz. 36: oo
Exuison, W. L., R. E. Avston & B. L. pkg 1962. Methods of presentation of crude
iochemical data for systematic purposes with particular reference to the genus
Bahia oy ee Amer. Jour. Bot. 4916). 594-604
Enpuicuer, S. 1836-41. Genera Plantarum Sipeaditn. Ordines Naturales Deposita.
Vin arabia) p. 288.
1847. Generum Plantarum Supplementum 4:38.
ENcuer, A. & E. Gite. 1924. Syllabus der Pflanzenfamilien. 9th ed. pp. 161-162.
Shane G. 1952 - Pollen Morphology and Plant Taxonomy. Angiosperms, Chronica
Botanica Co., Wa Itham, Mass. pp. 129-131, 182-183.
. 1969. 9. Handbook of Palynology. Hafner Publ. Co., N.Y. pp. 21-50, 213-216,

EypkE, . 1964. Inferior ovary and genetic affinities of Garrya. Amer. 1B Bot.
Sito. 1083-1092.

——. 7. The peculiar gynoecial vasculature of Cornaceae and its systematic
significance. Phytomorpholgy 17: :172-1 182.

oe A. L, 1924. I

tude

de la famille
s Cornacées. Fac. Med. Pharm. Lille, These Doct. Univ. 218 pp. > Unpublished
Fost, AS S. 1949. Practical Plant Anatomy. Van Nostrand Co., Inc., New York. pP-

Furuya, M., A. W. on & B. B. Srowr. 1962. Isolation from peas of co-factors
d inhibitors 7 hidolvisaaus acid oxidase. Nature 193:456.
Gua C. Pome Una nueva planta como subcedanea de la Cuasia. Voz de Hipocrates

1969. Organic chemistry of secondary plant metabolism. Freeman,
ooper & Co., San Francisco. pp. 238, 310-311
Gisss, - - 19 BrciSlen ney otaxonomy of Flowering Plants. Vols. I-IV. McGill-Queen’s
ess, Lon
Gna, V. ven Plant Speciation. Columbia Univ. Press, N.Y, 435 p
RAY Late Tertiary microflora f, Arizona.
Selnics tas ts, ora from the basin and range Novae.
————. 1961. Early
Science 133:38-39.
en EENE, E. L. 1910-1912. Miscellaneous specific types. Leafl. Bot. Obs. Crit. 2:86-87.
a - & L. Patscrixe. 1960. Zur Biogenese der Isoflavone III. Chem. Bet:

Pleistocene paleoclimatic record from Sonoran Desert, Arizon@

SYSTEMATICS OF GARRYA 103

Hapwicrr, L. A. 1966. The biosynthesis of pisatin. Phytochem. 5:523-525.
Hatiock, F. A. 1930. The relationship of Garrya. Ann. Bot. 44:771-812.
Harms, H. 1898. Cornaceae. In Engler & Prantl, Die Naturlichen Pflanzenfamilien
III. 8:256-257.
Hecnauer, R. 1966. Chemotaxonomie der Pflanzen. Birkhauser Verlag, Basel und
Stuttgart. pp. gale
HeErDERSHOTT, C. & D. R. Wackkr. 1959. — be a growth inhibitor from
extracts of oid peach flower buds. Scie 130:798-799.
Herissky, H. & C. Lepas. 1910. Presence de ras sees ine plusieur especies du
genre page Bot. Centralblatt 117:1
gia Se N. 1974. A —— of useful and everyday plants and their common
es. Pacikvies O . Press, London. p.
ero. Ww. L. 1925. A “deel of the Flowering Plants of California.. Univ. Calif.
Press, Berkeley. pp. 3-20, 731-732.
Kapit, R. N. & P. R. MosHanc. 1966. Studies of Garryaceae II. Reiayoes a
position of Garrya tice ito ex ene Phytomorphology 16:56
Kapp, R. O. 1969. Pollen and spores. W. C. Brown, Dubuque. pp. 3- 13, 23 33-2
Kuun, R. & I. Low. 1949. Marenoece und Chemie der Scbotess von Forsythea.
Chem. Ber. 82:474.
Lancman, I. K. 1964. A Selected Guide to the Literature on the Flowering Plants
of Mexico. Univ. of Pa. Press, Philadelphia. 1015 pp.
Lesas, C. 1911. Sur la pheanpat d’aucubine dans diverses variétiés d’Aucuba japonica
L. Bot. Centralblatt 117:176.
Linbtey, J. 1834. —o stiotion Bot. Reg. - aap
——_—_——.. 1847. Vegetable Kingdom. Bradbu & Evans, London. p. 295.
MakTINEZz, ee 1959. Las plantas seis oe Mexico. sted Botas, Mexico,
Mason, H. te 1934. Pleistocene flora of the Tomales formation. Carnegie Inst. Publ.
415:81-179.
Se H. 1971. Fossil flora of the Tsushime Islands. Bull. Natl. Sci. Mus. Tokyo
14(4):702-705.
Metcatrr, C. R. & L. Cuack. 1950. Anatomy of the Dicotyledone II. Oxford Univ.
Press, London. pp. 745-748.
eseuaphe M. F., Jn. & R. M. Breks. 1955. Studies of ye Garryaceae I. The comparative
orphology and phylogeny. by Seventeen 1 jeer
Mosiinsc E. 1961. Garrya alkaloids. — Chem. Bee. 73:
Mumror , D. H. Smitu & LE, 1961. An pci of indoleacetic acid
hone pea Hoe. Pl. Physiol. 36:7 752-7
OLcuIN she - 1932. Estudio saechiis de la Garrya laurifolia. Instituto de Biologia,
0. 74
OneETo, : F. 1946. The alkaloids of species of Garrya. Jour. Amer. Pharm. Assoc.
35:204-207.
aaa S S. & L. Kakkar. 1970. Leaf anatomy of some Garrya species. Jour. Linn.
Soc. Bot. 63:81-90.
angle S. W. 1960. The correlation of the atisine and veatchin
alkaloids. Jour. Amer. Chem. Soc. 82:23
Rao, M. 1963. Suspensor polyembryony i in Garrya veatc
32:468—469.
cae R. fie 1943. Cor oft
oot in Garrya stiotiea Amer. 1 Bot. 30:608-619. :
ae T. 1967. The Organic Constituents of Higher Plants. Burgess Press, Min-
neapolis. pp. 121-210, 256-

e series of diterpene
hii Kellogg. Curr. Sic. Bangalore

ci antl ith illary vegetative

Rose, J. 1903-1905. Studies of iiocics and Central American plants no. 3. Contrib.
8 S. Nat. Herb. Ch Soc. 39:
"heal ERY th ee Oe
962. Sua of anthocyanins. J aay inbvin cane Press,

Stespins, G. L., Jr. 1950, Variation and sat oa in Plant
N.Y. 643 p

: Pp.
———_. 1952. Aridity as a stimulus to evolution. Amer. Nat. 86:33-44.

104 GERALD V. DAHLING

Swain, T. 1963. Chemical Plant Taxonomy. Academic Press, N.Y. 543 pp.
Turner, W. 1960. Chromosome numbers. Madrono — 220.
UsHER, “ <a A nicer omaed of Plants Used by Man. Hafner Press, NY
WANGER Die os und sere der Familie pid ‘Cornaceae.
ho Tah cope
—_~-—. 1910: Carryaceae In Engler, A., Das Pflanzenreich. IV(56a):1-18. W.
ues. Verlag, Leipz
— K., S. K. Ficpor, M. Fr BaRTLETT & D. R. Henperson. 1952. Garrya alkaloids.
nad, i Chem. 30:608.
dct cins, I. L. 1933. New slants from Baja California. Contrib. Dudley Herb. 1(5):172-
173.
WOLFE, A. 1964. Miocene floras from eeieoneeny Washington, southwestern Nevada.
S. Geol. tg Prof. Paper 454N:N 36.

——_ & E. S. Barcuoorn. 1960. anhaar change in Tertiary floras in relation
to age. Amer. Jour. Sci. 258A:338-399.

from the

ibutions
GRAY
HERBARIUM

Contr

No. 210

wrssourt ornetets

980

@nncen LUBRARY

aire 26

Ss

Reed C. Roll

;
2
‘3

=

NOTICE

Research papers by persons not associated officially with the Gray
Herbarium but resulting from investigations based in full or in part
from studies of the collections of the Gray Herbarium will be consid-
ered for publication in future issues of the Contributions from the
Gray Herbarium. A page charge will be assessed, which can be waived
for those unable to secure outside funds, but not beyond eight printed
pages. Authors of monographs citing the Gray Herbarium material
are especially encouraged to submit papers for consideration to the
Contributions of the Gray Herbarium.

C5494

Contributions from the
GRAY
HERBARIUM

1980 No. 210

ANOTHER CRUCIFEROUS
Reed C. Rollins WEED ESTABLISHES ITSELF
IN NORTH AMERICA

THE GENUS PENNELLIA (CRUCIFERAE)

Reed C. Rolli
said IN NORTH AMERICA

STUDIES ON SOLANACEAE XII
Armando T. Hunziker ADDITIONS TO THE GENUS
CHAMAESARACHA

A. Li
ghee Bogle A GENERIC ATLAS OF
C. Thomas Philbrick ee

EDITED By Otto T. Solbrig
Kathryn Rollins

Missounl BOTANICAL
AnDea Hisrany
PUBLISHED BY
THE GRAY HERBARIUM OF HARVARD UNIVERSITY
ISSN: 0195-6094 IssuED JuLy 15, 1980

ANOTHER CRUCIFEROUS WEED ESTABLISHES ITSELF IN
NORTH AMERICA'

REED C. ROLLINS

The Cruciferae is one of the plant families that has contributed
a disproportionate number of introduced weedy species to the flora
of North America. Many of these are of European origin but others
came from North Africa, the Middle East and parts of Asia. In fact,
the exact origin of most weedy crucifers now established in our flora
cannot be ascertained. But the extent to which members of this family
occupy vast territories once the haven of only native species is a
cause for concern and regret. That these weeds occur in old fields
and waste places is one thing but the many square miles of open
desert in Utah and Nevada covered by Sismybrium altissimum L.
or the thousands of acres of the eastern Mohave Desert of California
to Arizona with nearly a continuous stand of Sisymbrium irio L.
are unwanted sights. I remember my first encounter with Malcolmia
africana (L.) R. Br. in western Colorado in the year 1938. At that
time, the species was scarcely known outside of the Salt Lake Valley
in Utah, but now it is everywhere on the sheep ranges of western
United States. Chorispora tenella (Pall.) DC. is another introduction
whose geographic range has burgeoned in the past twenty years.
Although widespread elsewhere in the west, it has found the western
plains a particularly hospitable region in which to proliferate. Nearly
forty years ago (Rollins, Contrib. Dudl. Herb. 3: 183, 1941), I reported
the first known North American station for Alyssum desertorum Stapf,
a species now widespread in the Intermountain West. Alyssum alys-
soides L. has been around a long time and is common in old fields
and waste places (as well as range-lands) all across the United States
and southern Canada. Relative newcomers in this genus, Alyssum
minus (L.) Rothm. and Alyssum szowitsianum Fisch. & Meyer are
presently more restricted in their distribution in North America but
they are spreading rapidly.

Although some of the annual species of the Cruciferae are difficult
to control, especially in grainfields, gardens, etc., the real noxious
members of this family are the perennials. For example, such species
as Cardaria draba (L.) Desv., C. pubescens (Meyer) Jarm., Rorippa
sylvestris (L.) Bess., and Lepidium latifolium L. are nearly impossible
to clear out of irrigated lands once they get well-established. Therefore,
it is with some alarm that I report a recently introduced perennial
member of the mustard family. The comparative newcomer is Brassica
elongata Ehrh. This species has turned up in a number of sites in
eastern Nevada, particularly along U.S. Highway 50. The danger is

"Research supported by National Science Foundation Grant DEB 78-08766.

1

2 REED C. ROLLINS

that the species will spread into the open desert where it could displace
some of the native vegetation. A list of the localities where it has
been found is given below.

Brassica elongata was brought to my attention by a specimen sent
for determination by Mr. Sherel Goodrich (no. 8387) collected along
Highway 50 about 50 miles east of Austin, Eureka County, Nevada,
July 7, 1977. A search in the Gray Herbarium turned up a specimen
misidentified as Thelypodium which was collected 13 miles west
of Pancake Summit between Eureka and Ely, White Pine County,
Sept. 2, 1968, by John Thomas Howell and Gordon H. True, their
no. 44609. This appears to be the first record for the infestation in
eastern Nevada but older specimens collected on ballast at Linnton,
near Portland, Oregon, Sept. 2 and Nov. 3, 1911, Wilhelm N. Suksdorf
1704 and 1749 (cu) shows that the species came to North America
at least once at a much earlier time. It is probable, as with many
plants introduced on ballast, that B. elongata did not persist where
Suksdorf first found it. At least, the species has not been recognized
as part of the adventive flora of northwestern United States (Hitchcock
and Cronquist, Fl. Pac. NW. 156, 1973). It is doubtful whether there
is any connection between the Oregon introduction and that of Nevada.
However, the source of the latter is not determinable at this time.

Following is a listing of more recent collections of Brassica elongata,
all from Nevada. Eureka County: 26.6 miles west of Eureka, June
12, 1979, Reed C. and Kathryn W. Rollins 79216 (cu, duplicates to
be distributed). White Pine County: 2.5 miles south of junction of
highways 50 and 93, roadcut near KOA campground, June 28, 1978,
Margaret J. Williams 78-164 (GH); same locality, June 25, 1979,
Margaret J. Williams 79-97. Laurie Birdsey and Arnold Tiehm (cu);
12.1 miles east of Eureka, June 12, 1979, Reed C. and Kathryn W.
Rollins 79215 (cu, duplicates to be distributed); 36.6 miles east of
Eureka, June 12, 1979, Reed C. and Kathryn W. Rollins 79214 (cu,
duplicates to be distributed); 30.2 road miles east of Fureka on Highway
50, near turn to Belmont Mine, June 28, 1979, Arnold Tiehm 5348,
Laurie Birdsey and Margaret J. Williams (GH).

Brassica elongata is a deep-rooted fleshy perennial with numerous
branches beginning just above the base and extending to the top
of individual plants. Usually several main stems arise from a single
root. The branches diverge at a wide angle and each is divided into
many smaller branches which ultimately bear inflorescences of many
flowers with yellow buds and bright yellow to orange-yellow petals.

ANOTHER CRUCIFEROUS WEED 3

Many patches of Brassica elongata were seen by the roadside
(sometimes extending away from the road into the open desert) at
intervals both east and west of Eureka for a distance of more than
100 miles. The species is obviously well-established and unless it
is eradicated soon, an infestation of much larger proportions could
easily take place.

phere.

a

THE GENUS PENNELLIA (CRUCIFERAE) IN NORTH
RICA’

REED C. ROLLINS

The genus Pennellia had its origin in the recognition by Robinson
of section Heterothrix of Thelypodium in his treatment for the S ynopti-
cal Flora of North America (Robinson in Gray, 1895). Two species
were included, Thelypodium longifolium (Benth.) S. Wats. and T.
micrantha (Gray) S. Wats. Subsequently Rydberg (1907) raised sec-
tion Heterothrix to generic rank. Nieuwland (1918) pointed out that
Heterothrix had been used previously (Mueller, 1860) for a genus in
the family Apocynaceae, and therefore Heterothrix of Rydberg is
illegitimate. He proposed Pennellia as a replacement for Heterothrix
and transferred H. micrantha to it. Thus Pennellia micrantha became
the type species of Pennellia. However, Schulz (1924) used Rydberg’s
illegitimate Heterothrix when treating the group for Das Pflanzenreich
and expanded it to include six species, three from North America
and three from South America. In the course of his study, he became
convinced that the species formerly known as Heterothrix longifolia
should not be associated in the same genus with H. micrantha and
described Lamprophragma as a monotypic genus to include it. Al-
though no specific points were made by Schulz to justify splitting
Heterothrix longifolia away from H. micrantha, it is clear from the
protologue of Lamprophragma that he was much impressed with the
slight zygomorphy present in the flower of H. longifolia. However,
as distinctive as this feature is in Pennellia longifolia, it breaks down
completely in P. hunnewellii where there is little or no zygomorphy
shown by the flowers; yet this species is so close to what is here
called P. longifolia that the collections of it have been included there
up to the present. Schulz (1936) later recognized Pennellia.

The first two species of Pennellia described, P. longifolia and P.
micrantha, were placed in Streptanthus (Bentham, 1839; Gray, 1849)
but most subsequent authors have agreed that they do not belong
in that genus. In fact, Pennellia is not closely related to Streptanthus.
An early different disposition of these species was that of Watson
(1871, 1882) who referred them to Thelypodium. However, Al-Shehbaz
(1973) has concluded that species presently included in Pennellia
do not belong to Thelypodium.

As Pennellia is now known from North America, including Mexico
and Central America, it consists of eight species. We have deliberately
not attempted to treat the species of South America because there
is so little material available for study. Even with the North American
Species, one (P. juncea) is known from only one collection, a second

"Research supported by National Science Foundation Grant DEB 78-08766.

5

6 REED C. ROLLINS

(P. mcvaughii) is known from but two collections, and a third species
(P. robinsonii) from only three collections. The genus is primarily
Mexican with only two species occurring northward into southwestern
United States. These are the two most widespread and best known
species, P. longifolia and P. micrantha.

One of the distinctive features of Pennellia is the very small flower.
In most species, just prior to full anthesis, the flower is cup-like
in appearance with the petals only slightly exceeding the sepals. In
some, the flower is almost closed. But as flower growth continues,
both sepals and petals not only increase in length but they are less
tightly associated than in early stages. The largest flower, with petals
nearly twice the length of the sepals when fully expanded, is found
in P. mevaughii and this is exceptional. However, the flower form
is in almost exactly the same mold as the other species with the
exception that the petals only slightly exceed the sepals in all other
species. The siliques of Pennellia are terete to very slightly flattened,
very narrow (often ca. 1 mm. wide) and elongated. The seeds are
numerous, either in a single or partially double row, and are often
very crowded so that the seeds themselves are distorted in shape.
The cotyledons are incumbent.

Pennellia is most closely related to the genus Halimolobos which
in turn has been considered very close to Sisymbrium. Many of the
species now placed in Halimolobos have been treated as Sisymbrium

Pennellia Nieuwland, Amer. Mid]. Natur. 5: 224. 1918.

page Rydberg, Bull. Torr. Bot. Club 34: 435, 1907; not Heterothrix Mueller,

Lamprophragma O. E. Schulz, Das Pflanzenreich IV. 105: 298. 1924.

F — or perennial herbs, densely pubescent with mostly branched trichomes

stems usually single from base, erect, branched above, up to 1.5 m. tall;
al leaves oblanceolate to broadly oblong, entire or sinuate dentate to shallowly

ie “iy ——

THE GENUS PENNELLIA -

lobed, petiolate, ne petiole often short, usually densely pubescent with branched
trichomes (except in P. juncea which is glabrous throughout); cauline leaves petiolate
n dy i rescen

s
y line
agar to septum, usually straight or near arly so and one-nerved at least below, sessile
with a short oe styles present: or absent; seeds small, erated 1 mm. or
ie long, wingless plum ; cotyledons

mbers are known for 0 nly two s species: PF. etree eta. Roll., =]
Asiana Roll., n = 8, 2n = 16.

KEY TO THE SPECIES

A. Siliques pendant; pedicels arched downward; oo usually sec
B. Petals 4.5-6.0 mm. long, barely exceeding sepals; paired stamens tnalintod or
nearly so; other floral parts relatively small.

C. Lower stems hirsute with coarse minidisc g simple or forked trichomes; basal
and lower cauline leaves pubescent with coarse simple or few-branched stalked
trichomes; buds and aevads glabrous; flowers slightly oe pedicel
ApICeS ASVMMEWICAN. Yr eee Cee eae a ngifolia,
Lower stems pubescent with fine appressed many-branched sed hea basal

CS.
and lower c set Yoae leaves pubescent with fine dendritically branched trichomes;
buds and Rane siustes pubescent; flowers regular; pedicel apices symmetrical
Ob MORRIS OC od en ie a ae 3 2. P. hunnewellii.

B. Petals 9-21 mm. long, exceeding sepals by nearly half their length; paired sta-
mens exserted by about one-third their lengths; other floral parts relatively large
Par green APA a rene 00) URES os oir geu ir Syn uly detes Oui cr Reyne Uo ate 3. P. mevaughii.
A. Siliques erect or ascending; pedicels erect or divaricately ascending; infrustescences
not secund.
D. Pedicels and siliques erect or nearly so, usually appressed to rachis; infructescence
narrow, usually dense
E. Buds and sepals glabrous; beta pedicels less than 8 mm. long, usually 4-7
mm.; siliques less than 4
F. Siliques agen sligh a compressed less than 2 cm. long; styles evident;
sepals deep purple; petals purplish ............... 4. P. robinsonii.
F. Siliques il a terete, more t than 2.5 cm. long; styles obsolete or nearly s
sepals usually greenish; p WHO aki 5. sale
E. rep and sepals sei Br a sabes pedicels more than 1 cm. long, siliques
A TONS a es ee ee iocalycina.
D. pea rigidly pack igi egal erect to divaricate; spb aicsieoinces broader,
loose, pedicels usually
G. Lower stems ee nell gear, fruiting pedicels 1 cm. or less re

the ey cen sade ee ae Iie, cat ved ihe ae aul et wate elie ae OS ie tye eae et ct Nm GA ge fee es ae en oa ee Sie Ge

1. Pennellia longifolia (Benth.) Rollins

P. longifolia a — —* 62: 16. 1960, based on Streptanthus longifolius
pay = 9. Type collected “In pasceris montosis, Aguae
ali "Meso. ey gaan 52. Holotype not seen, photo of type at Kew
sais ave

eo

8 REED C. ROLLINS

Thelypodium longifolium (Benth.) S. Wats., U.S. Geol. Explor. Fortieth Parallel (Bot.)
V: 25; 1871

Heterothrix longifolia (Benth.) Rydberg, Bull. Torr. Bot. Club 34: 435. 1907.

Lamprophragma longifolium (Benth.) O.E. Schulz, Das Pflanzenreich, Heft. 86. IV.
105: 299, fig. 63. 1924.

Thelypodium anisopetalum Greene, Pittonia 3: 247. 1897. Type collected in the Valley
of Mexico, Federal District, Mexico, Sept. 30, 1896, C. G. Pringle 6548. Holotype
not seen; isotypes F; GH, 2 sheets; MO; MSC; NY; PH; UC; US, 2 sheets.

Biennial; stems erect, most often single from base, branched above, usually virgate,
5-15 dm. tall, pubescent with coarse simple or forked spreading trichomes towar
base, glabrous above; basal leaves usually not persisting beyond first year, rarely present
on flowering or fruiting specimens, those seen petiolate, oblanceolate, entire to slightly
sinuate, sometimes dentate, obtuse and short-petioled, i.e., petiole shorter than blade,

Ww cm.

1.0-1.2 mm. long, filament attachment below middle; glandular tissue surrounding
base of single stamen filaments, only subtending paired filaments; pedicels slender,
qi + 5 EE | J
lab

slightly ing ard, glabrous, 8-11 mm. long, markedly expanded
and asymmetrical at apex; infructescences usually secund, 2-4 dm. long; siliques
pendant, straight or nearly so, terete, glabrous, 6-8 cm. ong, ca. 1 mm. wide, one-nerve

from base nearly to apex; styles slender, 1.0-1.2 mm. long; seeds small, crowded,
irregularly shaped, often angled, plump, marginless, variable in size, 0.7-1.3 mm. long,
0.4-0.5 mm. wide; funiculi slender, less than 1 mm. long; cotyledons incumbent,
oblique or accumbent. 2n = 16 (Rollins and Rudenberg, 1977).

DISTRIBUTION: Arizona, Colorado, New Mexico, western Texas, and Mexico south to
the Federal District, Mostly at elevations above 6,000 ft.

One of the outstanding and disti ctive features of Pennellia longifolia
when seen growing is the virgate or wand-like branches that bear
pendulous siliques in a secund raceme. The plants are often a meter
or more in height and may be abundant in favorable sites. They
are sometimes aggressive colonizers, then tending to become estab-
lished in disturbed places near or in cultivated fields, particularly
in the mountains of Mexico.

cena | ree

THE GENUS PENNELLIA 9

on the leaves, and simple or forked and spreading on the lower stems
in P. longifolia while in P. micrantha, the leaf and stem trichomes
are fine, many-branched, and extend farther up the stems than in
the former species.

There is some variation in the intensity of pigmentation of both
sepals and petals taking the full geographic range of Pennellia longifolia
into account. In some populations, the sepals are markedly dark
purple-red, other populations show less dense or incomplete coloration,
the margins becoming whitish. The petals are similarly colored except
the intensity is not as great or the coverage as complete as in the
sepals. In some populations, the petals are nearly white.

In the protologue of Pennellia longifolia, Bentham (1839) did not
give the place of collection of Hartweg no. 52, but supplied it in
a subsequent fascicle of the publication as given above. According
to McVaugh (1970), Hartweg collected in Aguascalientes between July
13 and September 17, 1837. The isotype specimen in the Gray
Herbarium gives only the year, Mexico, the name Hartweg and the
number 52 which Bentham assigned to the collection for purposes
of publication and distribution.

REPRES = SPECIMENS. Mexico. Chihuahua: Sierra de Santa Barbara, about 4 miles
southwest of Villa Matamoros, 6,300 ft., Oct. 4-5, 1959, D. S. Correll & H. S. Gentry
22795 (cu); valley of Gonogochic, east of Creel and San Ignacio, Mnpo. de Bocoyna,
7,400 ft., Aug. 24, 1973, Robert A. Bye, Jr. 4778 (cu); near Colonia Garcia in the
Sierra Madres, 7,500 ft., July 29, 1899, C.H.T. Townsend & C.M. Barber 187 (F, cu,

MC, NY, POM, RM, UC, US); Memclichi, Rio Mayo, 7,500 ft., Sept. 16, 1936,
8: Gentry 2738 (ariz, F, GH, MEXU, MO, UC us). Coahuila: near summit at Las Cumbres
Pass, 12 miles east of San Antonio, Sept. 28, 1974, Reed C. Rollins and K. W. Roby
7479 (cu); Canyon de Centinela, Sierra del Carmen, July 31, 1973, James Henrickson
11692 (cu); 26 km. northwest of Fraile, 3550 m., July 16, 1941, L. R. Stanford,
A. i, Retherford & R. D. Northcraft 439 (ariz, GH, MO, NY). Durango: Otinapa, July
25-Aug. 5, 1906, Edward Palmer 546 (cu, Ny, Us); about 48 miles west of Parral and
12.5 miles west of Ojito, Sept. 13, 1972, James L. Reveal & William J. Hess 3053
(cH, MSC); arroyo of Rio Chico, 17 miles west /southwest of C. Durango, Sept. 1,195).
James H. Maysilles 7659 (cu, micu). Federal District: cerca del Cerro Conejo, al ENE
de Ajusco, 2750 m., 23-VII-1967, Rzedowski 24118 (ariz, F). Guanajuato: summit east
of Guanajuato, 8,000 ft., July 16, 1963, Stuart K. Harris 25887 (cu). Hidalgo: Tezoantha,
Sept. 1945, Maximino Martinez s.n. (us); Sierra de Pachuca, Sept. 1, 1903, J. N. Rose
& Jos. H. Painter 6734 (us); Pachuca, July, 1905, J. N. Rose, Jos. H. Painter & J. S.
Rose 8850 (cu, us); below Guerrero on road from Real de Monte to Omitlan, Atotonilco
el Grande, 2,700 m., July 30, 1948, H. E. Moore, Jr. & C. E. Wood, Jr. 4141 (cu).
Jali 15, 1970,
ose & Jos.

a
)
3.

id
2
°
g
3
fe
&
ro

Q
e

3

z
-

5

-

6

Lek

2.

)
e
ex
=.
S
3
ie
°
bs)
Pa
¢
ge

i i i : Henry E. Seaton
: TEX, US); vines, Mt. Orizaba, Aug. 8, 1891,
250 (us); 8 km. al E de Coatlinchan, municipio de Texcoco, July 20, 1967, Rzedowski
24023 (MICH Msc). Michoacan: vicinity of Morelia, Sept. 1, 1909, G. Arséne 7262 tg
Nuevo Leon: cerro Potosi, near microwave tower, * ae vee _ soe > a
McG ; i thwest of Galeana, June 28, 2. i.
& regor et al. 323 (cu); about 15 miles sou : SS

- I. Mueller 907 (Fr, GH, MICH, TEX); Pefia Nevada, 26 miles fDi
Arroyo, July 4, 1959, John H. Beaman 2707 (msc). Puebla: Las Derrumbadas, municipio
de Buenos Aires, Aug. 24, 1972, F. Ventura A. 5934 (cu). San Luis Potosi: in montibus

10 REED C. ROLLINS

San Miguelito, 1876, J. G. Schaffer 156 in part (ca, GH, MEXU, ies Sierra de San Miguelito,
cerca de El Capulin, Sept. 5, 1954, Rzedowski 3993 (cu); San Luis Posoti, 1879, J. G.
co 556 (us). oye steep slopes of barranca, 3 hey alae of El Palmito,
1950 m., Aug. 12, 1974, D. E. Breedlove 36458 (cas, GH); 3 miles north of Los Ornos
along road to Geurahui, Sierra Surutato, Municipio de Badiraguato, 6,500 ft., Oct.
2, 1970, D. E. Breedlove & R. F. Thorne 18344 (cas); Ocurahui, oe Surutato, 6-7,000
ft., Aug. 27-30, 1941, H. S. Gentry 6203 (ariz, DS, GH, MICH, MO, NY, PH). Ta: maulipas:
on Pena Nevada, July 18, 1949, Stanford, Lauber & Taylor 2512 (Ny, US) and 2512A
(us). USA. Arizona. Apache Co.: 1.5 miles east of Sheep Spring, 24-mile Draw, Sept.

m., Aug. 24, 1920, W. W. Eggleston 17158 (F, Gu); 8 miles E of Nutrioso, Aug. 24,
1951, K. F. Parker & E. McClintock 7516 (ARIZ, cas, us). Cochise Co.: Rustler’s Park,
Chiricahua Mts., 8,000 Oct. 21, 1974, Reed C. Rollins & K. W. Roby 74197 (cu);
same locality, Nov. 22, 1974, Reed C. Rollins, Charles T. Mason, Jr. . te, B.

Syoinarelg 74202 (cu). Coconino Co.: 13 miles north of F ee San Fra o Mts
00 ft., July 20, 1946, K. F. Parker 5992 (ARIZ, CAS, RSA); 3.5 m Sie F “lagstaff
on n Schulz Pass a San Francisco Mts., 7,300 ft., Sept. 11, 1968, Sette Thomas Howell
on H. Tru (CAs, GH); San Francisco Mt., October, Sitgreaves Expedition
of 1861 ( (GH) Sais Co.: Spud Ranch, Rincon Mts., 7,400 ft., Aug. 31, f
Blumer 3316 (ariz, F, GH, MO). Pinal Co.: Santa Catalina Mts., Aug. 23, Grd
arrison & T. H. Kearney 8115 (ariz, cas, F). Yavapai Co.: hills of s near Prescott

er

on road to Ash Fork, Nov. 7, 1928, Alice Eastwood 16728 (cas). Colorado. La Plata
Co.: river bottom, Bayfield, Aug. 9, 1917, E. B. Payson 1151 (xm). New Mexico. Catron
Co.: Mogollon Mts., 18 mi. northeast of a 7,500 ft., Aug. 1, 1938, C. L.
i ‘ M, UC, UCLA, UTC, WS, WTU). Colfax Co.: vicinity of Ute
Park, Aug. 24, 1916, Paul C. Standley 13719 pie Ceant Co.: Fort Bayard ——
Sept. 3, 1905, J. C. Blumer 53 (GH, Ny). Lincoln Co.: wie Mts., 7,000 ft., Aug. 11,

iv ;
. B. Metcalfe 1417 (Fr, ny). apap = Mogollon Mts., 7,500 ft., Aug. 2,
B. Metcalfe 350 (ARIZ, MO, NMC, NY, uc). Texas. Jeff Davis Co.: Madera Ca anyon,
Mt. Livermore, Aug. 5, 1935, -L. C. ‘Hinckley 288 ‘iy, NY, TEX); Mt. Livermore, Davis
Mts., ca. 2250 m., July, 1936, L. C. Hinckley s.n. (GH).

2. Pennellia hunnewellii Rollins, sp. nov.

Perennial; stems one or few from an entree crown, 5-12 dm. tall, branched above,
pubescent below with appressed sever

al ranched
sparsely pubescent above; basal le eaves petiolate pi eran oe sinuate dentate, obtuse,
15

sepals greenish to dull purple, broadly
rplish, 4-

stene :
petals lingulate, white to pu a 8, scarious ma
Ss ed

Ce)
ong, ca. 1 mm. wi ae fruiting pedicels
y pu ihe > abies 8-12 m ong, apex
es widely pre and pendulous, inane or nearly
f llel to septum to nearly terete, 5-8 cm . long, 1-nerve
© ‘0 apex, nearly sessile or with a short gynophore less et 1 mm. long;
long; seeds in an irregular row, plump, wingless,
wide; cotyledons incumben

ong,
Herba perennis, ehullbos erectis superne ramosis sparse pubescentibus vel glabris

eee | ones

THE GENUS PENNELLIA Li

inferne pubescentibus 5-12 dm. altis, foliis basalibus petiolatis oblanceolatis sinuato-
dentatis obtusis 4-8 cm. longis, foliis caulinis linearis vel filiformibus sparse dentatis
ib i s

siliquis pendulis plus minusve teretibus rectis glabris 5-8 cm. ongis, seminibus
immarginatis noncompressis ca. 1 mm. longis s, cotyledonibus apa us.
i i llected ] n Acatenango,
8,500 ft., Dept. of Sacatepéquez, Guatemala, Feb. 16, 1937, F. WwW. ‘Meisanedl 1482.
DISTRIBUTION: Hidalgo and Jalisco, Mexico to Guatemala.

This species is named for the collector, Mr. Francis Welles Hun-
newell (1880-1964) who was Phanerogamic Curator of the New
England Botanical Club from 1913 until 1958 anda Research Associate
of the Gray Herbarium for most of his adult life. He was Comptroller
of Harvard University and later Secretary to the Corporation of the
University. In establishing the Fernald Fund in Harvard College for
field research in systematic Botany, Mr. Hunnewell left an enduring
commitment to botanical field work, which he so enthusiastically
enjoyed.

Most of the material cited below has been considered to belong
to Pennellia longifolia. Certainly the phenological aspects of the
specimens are very similar to that species and it is assumed P.
hunnewellii and P. longifolia are closely related. In general, the
geographical range of P. hunnewellii is to the south of P. longifolia
but there is a broad area of overlap in the highland region of Mexico.
The two species may be distinguished from each other by carefully
examining the trichomes, particularly on the leaf surfaces and toward
the base of the stems. The trichomes of P. hunnewellii are fine and
many-branched whereas those of P. longifolia are coarse and simple
or forked. The latter are spreading and produce a hirsute condition,
especially on the stems. On the other hand, the trichomes of P.
hunnewellii are so small that they scarcely show any divergence from
the stem or leaf surfaces.

The siliques of Pennellia }
to the septum while those of P. longifolia. are = strictly terete. The
former species is a perennial with strong tendencies to produce a
woody foot at the apex of which the lower leaves are clustered. The
latter species is biennial and the lower leaves are infrequently present
on fruiting specimens. A woody foot is not produced.

ae oe d parallel

km. on
80: Sierra d Cuch tanes, betwee Tojiah oo San Juan Ixcoy at

Ruta Nacio ck ON a oe a. 3200 m. » Aug. A, 960, John A. Beaman 3951 (msc); small
limestone rid PEt los Chuchumatanes, between

Paquix and Chena at km. 311 on Ruta as! 9N, ca. 3360 m., Aug. 2, 1959,

12 REED C. ROLLINS

John H. Beaman 2964 (cu, Msc). Sacatepéquez: me Santa Maria de Jesus, north-facing
ra of Shei aay de Agua, by trail, Nov. 8, 1958, J. G. Hawkes et al. 1911 (cn).
Mexi idalgo: Cerro Alto, 3 km. al SE de Shoe July 19, 1963, Rzedowski
16911 ( sais aitce: Nevado de Psapp a few miles south of Ciudad Gassnan (Zapotlan),
2700-28! * ; vid P. Gregory & George Eiten 307 (micu); Nevado
de Colima, northeast side oF sored ear Puerto de los Cruces, ca. 3600 m., Aug.
27, 1958, John H. Beaman 2385 (msc); northeastern oe of Nevado de Colima, below
Canoa de Leoncito, Sept. 10, 1952, Rogers McVaugh 12836 (cH, MicH). México: 19
km. NE de Texcoco, sobre la tet a ne eae ae asi 17, 1971 Pinta 28498
(micH); Crucero, Temascaltepec, 3400 m., Oct. 24 1933: Geo. B. ‘Hint n 4920 (cH);
1 km. al N de Llano Grande, C ehcten. en ‘lee faldas del Cerro her July 26, 1964,
Rzedowski 18420 (cu); near Contreras, jig pe: 1944, GC. . Goodman pape .

de San Filipe, 10,000 ft., Sept. 18, 1894, C. G. Prin gle 5622 (cH). Queretaro: near
summit of Cerro Zamorano, 2950 m . Aug. 3, bap ‘Melinda F. Denton 1958 (micw);
parte mas alta Cerro Zamorano, Colén, 3,200-3 , Nov. 13, 1971, Rzedowski
& McVaugh 419 (micu). Veracruz: faldas del Pico id pr aditde 3220 m., Oct. 16, 1971,
R. Hernandez M. 1321 (cu).

3. Pennellia mevaughii Rollins

Pennellia mevaughii Rollins, Taxon 28: 24.

oe ollected near som an Highway ge miles east of El Salto, Sierra Madre
Occidental, Durango, Mexico, Sept. 21, 1974, Reed C. Rollins & K. W. Ro by 7423
aca GH). Isotypes to "sag distributed.

zygomorphic stamen pair rcurved dette ;
sepals eer scarious toward Si 7-9 mm. ator 3-4 mm. wi uter saccate,

inner pair plain; petals purple, n narrowly lingulate, Gone paca
below except at poin ea A cant straight, 9-12 2 mm. long, 2-3 mm. wide; filaments
gradually broadened toward ie hae se of paired stamens 8-10 mm. long, stamens
excerted, anthers rect 1.5-2.0 mm. long; ovary and se slightly curved upward;

stigma entire, slightly exceeding style in diameter; infructescences usually secund;
fruiting pedicels a ascendin net then arched downward, glabro ous, 7-11 m m. long,
markedly expanded at expa t ge

&

. dei slightly tad i terete to slightl site tt labrous
ise get dines — is “4 a ened parallel to septum, glabrous,
1.0-1.5 mm. long; seeds crowded, see ager ca. 1 mm. long, less than 1 mm. wi

somewhat — pendant on funiculi c - long. n = 8, 2n = 16 (Rollins and
Rudenberg, 1977).

nown i from the Sierra Madre Occidentale west of C. Durango, Mexico.

THE GENUS PENNELLIA £3.

species was made but none were found. Notes made at the time of
this collection suggested that P. mevaughii is probably a perennial
but there was no positive evidence available. The root is sufficiently
thick and the leafscars cover a sufficient distance on the crown to
Suggest a perennial habit. However, the fact that all basal leaves have
been shed suggests a biennial which produces a rosette the first year,
then when flowering occurs, the basal leaves wither and are ultimately
shed. In any case, it is certainly not an annual species.

It is clear that Pennellia mcvaughii is most closely related to P.
longifolia. These species are very similar in habit and in general
they occupy the same types of habitats. Mostly, they occur in open
conifer or oak forest areas at relatively high elevations in the mountains.
Pennellia mcvaughii was found in the pine zone at about 8,000 ft.

SPECIMEN CITED OTHER THAN TYPE. Mexico. Durango: frequent in moist sloping meadow,
Pine-forest zone, 3 miles north of Coyotes, Sierra Madre Occidentale, ca. 2400 m.,
about 80 km. west of C. Durango, Sept. 28, 1962, Rogers McVaugh 21680 (micu,
NY).

4. Pennellia robinsonii Rollins, sp. nov.

Perennial with a multicipital caudex; stems one to several, up to 1 m. tall, branched
above, pubescent with fine dendritic trichomes below, glabrous above, old stems often
Persisting; basal leaves tufted at apex of caudex branches, oblanceolate, obtuse, sinuate
d te to entire, petiolate, with a strong central vein, densely pubescent with fine
dendritic trichomes, 5-15 cm. long, 6-20 mm. wide; lower cauline leaves linear to
linear-lanceolate, sessile and tapered toward base but scarcely petiolate, entire or nearly
SO, sparsely pubescent or glabrous; upper cauline leaves narrowly linear to filiform,
glabrous; inflorescence narrow, greatly elongating in fruit; buds globose to obovoid;
glabrous; flowers cup-shaped; sepals dark purple, broadly oblong to ovate, nonsaccate,
3.5-4.0 mm. long, ca. 2 mm. wide; petals tipped with light purple, whitish below,
oblong, not tapering toward base, 4-5 mm. long, ca. 1.5 mm. wide, barely exceeding
sepals; fruiting pedicels strictly erect, appressed to rachis, glabrous, 4-6 . long;
stamens erect, paired and single nearly the same length; filaments stocky, gradually
broadening toward base; 2.5-3.0 mm. long; anthers ca. 1 mm. long, ca. 1 mm. broad;
glandular tissue well-developed, surrounding base of single stamen, subtending base

Parallel to septum, indistinctly one-nerved below, 1.5-2.0 cm. long, obscurely —
Phorate; styles ca. 0.5 mm. long; stigma elongated over valves; seeds numerous, —
Mmarginless, plump, angled, ca. 1 mm. long, in two distinct rows in each loculus;
funiculi filiform; cotyledons incumbent.
erba perennis, caudicibus ramosis, ca an & : :
Pubescentibus, foliis basalibus petiolatis oblanceolatis sinuato-dentatis vel integris,
foliis caulinis linearis vel filiformis superne glabris, inflorescentiis angustis e eit
floribus poculiformibus, sepalis purpureis late oblongis vel ovatis nonsaccatis . ,
mm. longis, ca. 2 mm. latis, petalis oblongis 4-5 mm. longis, pedicellis fructi oS
€rectis glabris 4-6 mm. longis, siliquis erectis 1.5-2.0 cm. longis plus minusve teretibus,
Stylis ca. 0.5 m longis, seminibus noncompressis immarginatis ca. 1 mm. longis,

.S mm.
cotyledonibus incumbentibus. - ae illsi
olotype in the Gray Herbarium, collected on a steep granitic pal poner ge ids
miles south of Chihuahua City, Chihuahua, Mexico, Oct. 15, 1974, - : 0 ad
and Kathryn W. Roby 74182. Named for the late Professor B. L. Robinson, former

lh pS is glabris inferne

14 REED C. ROLLINS

curator of the Gray Herbarium who first recognized Pennellia as section Heterothrix
of Thelypodium and who annotated the Pringle specimen in the Gray Herbarium
calling attention to the elongated stigma which has the lobes over the valves.

Known only from the state of Chihuahua, Mexico.

Pennellia robinsonii is nearest related to P. micrantha and the
specimens of it have been distributed under that name. One of the
most striking features of P. robinsonii is the multicipital caudex found
in older plants. The short acute fruits are also distinctive when
compared to those of P. micrantha. The flowers of P. robinsonii are
cup-shaped and barely open during anthesis. They are fully as large
as those of P. longifolia. The sepals are dark purple instead of greenish
as in P. micrantha and the petals are purple-tipped, oblong, and
non-tapering toward the base in P. robinsonii as contrasted with white
lingulate petals that taper toward their bases in P. micrantha.

The habit illustration given by Schulz (1924, p. 296, fig. 62) is
that of Pennellia robinsonii and probably was made from Pringle 636
which he cites under Heterothrix micrantha. Unfortunately, the source
of the illustration is not given in the legend. The different petal shapes
“D-E Petala speciminum diversorum” are of two species. D illustrates

Our field notes made at the time the type of Pennellia robinsonii
was collected indicate that plants of the species were infrequent in
the area. There were granitic outcrops on a steep, otherwise grassy
hillside and the plants were growing at the base of these outcrops.

OTHER SPECIMENS STUDIED,

e
12, 1885, C. G. Pringle 636 (Ds
C. G. Pringle 294 (us).

xico. Chihuahua: rocky hills near Chihuahua, Sept.
» F, GH, NA, NY, PH, RSA, US); same locality, Oct. 1885,

5. Pennellia micrantha (Gray) Nieuwl.

Pennellia micrantha (Gray) Nieuwland, Amer, Midl. Natur. 5: 224, 1918, based on
Streptanthus micranthus Gray, Mem. Amer. Acad. Arts and Sci. IV: 7. 1849. Type
collected near Santa Fe Creek, New Mexico, A. Fendler 23, July, 1847. Holotype
GH; isotypes GH

— mniCrantin (Gray) Wats., Proc. Amer. Acad. Arts and Sci. XVII: 321.

T. longifolium (Benth.) Wats. var. catalinense M. E. Jones, Contrib. West. Bot. 12:2.
1908. pe collected in Sabino Canyon, Catalina Mts., Arizona, 3000 ft., August

20, 1903, reus E. Jones s.n. Holotype pom; isotypes MO, UC, US.

Heterothrix micrantha (Gray) Rydberg, Bull. Torrey Bot. Club 34: 435. 1907.

Perennial or biennial, stems usuall
tall, branched above, pubescent below w
glabrous above or with scattered dendritic trichomes; basal leaves oblanceolate, sinuate

trichomes, lower ca line leaves si

milar to basal, etiolate, up to 10
shorter, upper cauline leaves m A 2

cm.
uch reduced, cuneate at base, sparsely pubescent to

THE GENUS PENNELLIA 15

glabrous, usually entire; ——— narrow, elongated; buds globular to obovoid,

glabrous or with a few trichomes near apex, usually ee = broadly oblong,

scarious margined above, oe 3.0 mm. long, 1.5-2.0 mm. wide, outer pairs slightly

saccate, inner pair plain; petals white or rarely purplish, siatalate to Nunes lingulate,

gradually narrowed toward base, 3.5-4.5 mm. lon g, 1.0-1.3 mm. wide; stamens He Htoy

nearly equal, paired stamens straight, single sinh mens curved, filaments 1.5-2.5 mm.
e 1

. long; s g
or rarely et pubescent when young, (2-)2.5-3.5(-4.5) cm. long; styles Shestees
or eee so; seeds numerous, crowded, plump, angled, marginless, 1.0-1.2 mm. long,
ca. 0.5 mm. wide; cotyledons incumbent.
eacinitt UTION: xine. to southern Arizona, New Mexico, west Texas and the moun-
tains of Mexico south to the state of San Luis Potosi.

This is the most variable species of the North American members
of the genus. Silique length, pedicel length, density of the indument,
style length and the number of flowers per raceme all vary considerably.
In all but three specimens seen, the siliques are sessile or very nearly
so. In one specimen at the Gray Herbarium, collected by Wright,
presumably in the Fort Davis area of Texas, the siliques are on a
gynophore 1-2 mm. long. Similarly, specimens of Stephen S. White
3494 from northeastern Sonora, Mexico, at the University of Michigan
show a definite gynophore. The siliques of a specimen at the U.S.
National Herbarium, Standley 40609 from the Guadalupe Mts., Texas,
are not quite sessile. Here the gynophore is less than 0.5 mm. long.
Usually in the Cruciferae, a definite gynophore is a significant
distinction often correlated with other differences that set off separate
taxa. However, in this instance I have not been able to discover any
correlated distinctions from other specimens of Pennellia micrantha
and it seems best to regard the presence of a gynophore as an unusual
feature of erratic occurrence. It is in the Wright specimen collected
in 1851 where the striking gynophore suggests a closer look at other
material from the area. However, with the exceptions given above,
in all Texas material as well as that from elsewhere in the species
Tange, the siliques are sessile or very nearly so.

The holotype of Pennellia micrantha consists of two branches, one
bearing flowers, the other having more or less mature siliques
lower parts of the plants are missing. However, Fendler 22, iiecied
at the “foot of mts. on Santa Fe Creek” is from the same area as
the holotype and the specimen on one sheet (GH) is complete except
for lower and basal leaves. In reviewing the application of the name,
. micrantha, I have taken into account both Fendler 23 and Fendler
22. But the latter is a mixed collection with specimens of both P.
micrantha and P. longifolia present on one sheet at GH. :

The two more northerly collections of Pennellia micrantha in

16 REED C. ROLLINS

Colorado have unusually small flowers and the specimens differ in
other minor ways from those collected further south. However, the
material is inadequate to fully test the possibility that a distinct taxon
is represented.

REPRESENTATIVE SPECIMENS. Mexico. Chihuahua: Culebra Mts., Aug. 18, 1936, Harde
LeSueur 749 and 643 (r, Poe 7 road miles ea ary Colonia Juarez in “The Tinaja,”
1600 m., July 28, 1972, M. & E. Wilson, L. A. C. rileon 8431 (GH); Canon
de St. Diexo:; Sierra an ae oaee Sept. fe a Hartman 804 (F, Gu,
NY, UC, US); same locality, 6,600 ft., t. 16, 1903, Marcus - : yee S.n. (POM, 2 sheets).
Coahuila: Sierra de Santa Rosa, cas of Miizquiz, July 25, 1938, Ernest G. sg
1459 (F, Gu, TEx); Cahon Hundido on N side of Pico de Centinela, Sierra del Jardi
1500-2250 m., July 27, 1973, M. C. Johnston et af 11799b (teEx-LL); Hidalgo Piaaes
Blanca, Villa Acufa, 1936, no. 6647 (no collector given mexu); Canon de Centinela

cH, US);
NE Sonora, Aug. 19, 1940, Stephen S. White 3367 (ARIZ, GH, MICH, Us). USA. Arizona.
Apache County: 4.8 miles south of junction on Green’s Mountain road, July 26, 1973,
Lehto, McGill and Pinkava 11505 (NY). part wag Co.: Guadelupe Canyon, Guadelupe
Mts., 4300 ft., Oct. 4 & H. S. Haskell 4532 (ariz); Cave Creek,
Chiricahua Mts., Sept. 21, 1929, G. J. Se <7) bh ntti 6166 (cu); Barefoot
Park, Chiricahua Mts., Sept. 12, 1906, J. C. Blumer 1363 (ariz, GH); Garden Canyon,

we
Nl
8
|
Q*
v3
a=

iv t.

Sept. . Davidson 848 (cu). Pima Coe “White House Canyon Santa Rita Mts.,
5200 ft., hig 27, 1939, Lyman Benson 9722 (ARIZ, Pom); South Canyon, Baboquivari
Mts., 3600-4000 ft. Aug. 31, 1940, T. H. Kearney & R. H. Peebles 14948 (aniZ, NY,
t

rn weighs 10457 (cas). Teller Co.: rocky cliffs, Mountain View, Pikes Peak, July
re - Clements s.n. (ny). New Mexico. Santa Rita de Cobre, Aug. & Oct., 1880,

E. O. Wooton s.n. (Nuc, Ny); Organ Mt ts., Aug. 16, 1895, E. O. Wooton on: (us us). Grant
Co.: mts. near copper mines (Santa pee Aug 1851, C. Wright 844 (cu, ny, pH). Lincoln
900 te Ear

rro
Mogollon Creek, Mogollon Mts., July 18, 1903, O. B. gra aco (ARIZ, MO,
NMC, UC); Mogollon oa Aug. 1887, Henry H. Rusby 25-1/2 (F, mo, magi :
Brewster Co.: Toron o Canyon, ca. 5 miles west of Alpine, “eh 26. igs Tr.
Sperry T1179 (us); Gis Mts., C. H. Mueller 8004 (F, TEX). pied ‘Cac Pine
ant on, Guadalupe Mts., Aug. 15-17, 1924, Paul C. Standley 40609 (us); vicinity
of Frijole Post Office, Aug. 10, 1930, Carl O. Grasel 187 (micu). Hudspeth Co.: Eagle

oxana §. Ferris and Carl D. Duncan 2570 (cas); M ;
; Madera Canyon, Mt. Livermore,
oe 1936, L. C. Hinckley s.n. (ariz); Fort Davis, 1883, V. Havard 214 (cu); Limpia

THE GENUS PENNELLIA 17

Canyon, 1889, J. C. Nealley 11 (F); mountains and arroyos of the Limpia, July, 1852,
J. M. Bigelow 33 (ny). Presidio Co.: west branch of ZH Canyon, above mouth, July
16,1941 L.-C: Hinckley 1003 (aniz, GH, TEX, us); head of Pinto Canyon, northwest
of Chinati Peak on the Marfa-Ruidosa road, Sept. 10, 1961, D. S. Correll and M.
C. Johnston 24385 (Gu, TEX-LL).

6. Pennellia lasiocalycina (Schulz) Rollins, comb. nov.

Based on Heterothrix micrantha (Gray) Rydberg var. lasiocalycina O. E. Schulz, Das
Pflanzenreich IV. 105. Cruciferae-Sisymbrieae. 86: 296. 1924. Collected in the Sierra
de Parras, Coahuila, Mexico, 8-9000 ft., July, 1910, C. A. Purpus 4604 (holotype
B, not seen; isotypes F, GH, MO, UC, US).

Pennellia micrantha (Gray) Nieuwl. var. lasiocalycina (Schulz) Rollins, Contrib. Gray
Herb. no. 206: 8. 1976.

Perennial; stems one or few, branched above, 6-10 dm. tall, densely to sparsely
pubescent with fine dendritically branched trichomes sometimes glabrous above; basal
leaves petiolate, oblanceolate, sinuate dentate to entire, obtuse, densely pubescent with
stalked dendritically branched trichomes, usually the indument matted on lowermost
leaves, 2-4 cm. long, 6-10 mm. wide, lower cauline leaves similar to basal leaves
but less densely pubescent, cauline leaves gradually reduced upward, short-petioled
to cuneate at base, pubescence reduced upward on both stems and leaves; inflorescences
terminating each branch, narrow, lax; buds nearly globose, pubescent with branched
trichomes; flower pedicels slender, erect, usually pubescent; flowers cup-shaped, regular,
petals barely exceeding sepals; sepals purplish, pubescent, broadly oblong, scarious-
margined, 2.5-3.0 mm, long, 1.5-2.0 mm. wide; petals white to tinged with purple,
narrowly oblong, not differentiated into blade and claw, not tapered below, 3.0-3.5
mm. long, ca. 1 mm. wide; stamens included; fruiting pedicels strictly erect, pubescent
or glabrous, 1.0-1.5 cm. long; siliques glabrous, erect, slightly flattened parallel to
Septum to nearly terete, 4-6 cm. ong, acute at apex; styles evident but less than
1 mm. long; seeds numerous, marginless, plump, often misshapen, slightly more than

mm. long, ca. 0.6 mm. wide, in an irregular combination of one and two rows;
cotyledons obliquely incumbent.

DISTRIBUTION: Coahuila to Hidalgo and Nuevo Leon, Mexico.

Unfortunately, most of the specimens studied are in flower, some-
times with young fruit, but they do not give an adequate basis for
determining the variation undoubtedly present in this species. The
isotypes show relatively long siliques compared to those of Pennellia
micrantha and in the one specimen with mature siliques, Chiang et
al. 9435, the siliques are about 6 cm. long. The siliques and pedicels
of P. lasiocalycina are strictly erect and closely appressed to the rachis
of the infructescence. The flowers are relatively larger and broader
than those of P. micrantha and in general, the plants are taller. If
there is any confusion of identity, it would most likely be with P.

sparsely represented by collections in the herbaria consulted.

18 REED C. ROLLINS

SPECIMENS EXAMINED. Mexico. Coahuila: Sierra de la Madera SE and SSE of Ranchero
Cerro de la Madera, 1500-2900 m., Sept. 20, 1972, F. Chiang, T. Wendt and M. C.
Johnston 9435 (tex); Sierra Madre, 40 miles south of Saltillo, July, 1880, Edward
Palmer 37 (cu); ca. 35 km. W of Cuatro Cienegas, above Canon de la Hacienda,
in limestone, Sierra de la Madera, 8900 ft., August 5, 1973, James Henrickson and
T. Wendt 11947 (cu). Hidalgo: cerro alto, 3 km. al SE de Epazoyucan, 2500 m., July
19, 1963, Rzedowski 16911 (MicH); cerro de Santa Monica, N of Santa Monica, 40
km. NW of Apam on Pachuca highway, 2650 to 2850 m., July 11, 1966, Robert C.
West P-11 (micu). Nuevo Leon: wooded slope near Ojo de Agua at foot of Cerro
de Potosi, 6600 ft., July 14, 1945, A. J. Sharp 45732 (GH).

7. Pennellia patens (Schulz) Rollins

Pennellia patens (Schulz) Rollins, Rhodora 62: 15, 1960; based on Heterothrix patens
O. E. Schulz, Pflanzenr. IV. fam. 105: 296. 1924. Lectotype from District Comitan,
Chiapas, Mexico, 18 Aug. 1898, Caec. and Ed. Seler 3038 (holotype B, not seen;
isotype Gu).

Biennial; stems usually single from base, up to 1.3 m. tall, branched above, sparsely
pubescent below with fine dendritically branched trichomes or glabrous, glabrous
above; basal leaves usually not persisting beyond first year, rarely present on flowering
or fruiting specimens; sinu i .

, often apiculate, lowermost usually missing; upper cauline
y linear, sparsely pubescent to glabrous; inflorescences greatly
elongated, terminating each branch; buds ovoid, with a few tricl tap glab

flowers remote; sepals greenish to purplish, oblong, glabrous, 3-4 mm. long, ca. 1-3
mm. wide; petals white, scarcely differentiated into blade and claw, slightly narrowed
toward base, ca. 4 mm. ong, ca. 1 mm. wide; stamens erect, single only slightly
shorter than paired; fruiting pedicels slender, divaricate, glabrous, stiff, nearly straight,
(4-)5-9(-9) mm. long; Siliques divaricately ascending, terete, 3-6 cm. long, less than
S numerous in a single row to densely packed in semi-double rows,

mm. wide; se e
marginless, oblong, slightly over 1 mm. to less than 1 mm. long, ca. 0.6 mm. wide;
cotyledons incumbent.

DISTRIBUTION:

rous;

Mexico from Durango to Chiapas.

There are some inconsistencies evident in the material I have referred
0 Pennellia patens and it is possible that when a greater knowledge

of the populations is available, more than one taxon will be recognized.
The problem centers on the lack
the size of th

THE GENUS PENNELLIA 19

if ultimately correlated with other features, could provide for the
recognition of more than one taxon. But such correlations are not
possible from the material under study. There are some puzzling
specimens from Durango and Aguascalientes that I have included
under P. patens, but which have shorter pedicels and siliques than
the usual material of this species from further south. These in particular
may represent a distinct taxon.

Aside from the divaricate pedicels and longer siliques distinguishing
Pennellia patens from P. micrantha, the long pedicels and narrow
fruits are distinctive. As the siliques become mature, they tend to
take on a purplish pigmentation.

SPECIMENS STUDIED. Mexico. Aguascalientes: ca. 20 km. east of Rincon de Romos,
road to Asientos, between Cerro Altamire and Cerro de San Juan, 2200-2450 m., Sept.
4-8, 1967, Rogers McVaugh 23754 (GH, MICH). Distrito Federal: pedrigal (lava beds),
Valley of Mexico, Aug. 21, 1896, C. G. Pringle 6454 (cas, ps, F, GH,

MSC, NY, PH, POM, UC, us); Tlalpan, Aug. 16, 1910, C. R. Orcutt 3644 (F, GH, MO, Us);
same locality, July, 1905, J. N. Rose et al. 8489 (GH, NY, us); Vertiente E del Cerro
de Cc :

of Ajuelos, Aug. 16, 1957, Rogers McVaugh 17009 (cu, mic). México: Vertiente E
del Cerro del Pino, cerca de Ayotla, July 13, 1967, Rzedowski 23996 (micu, rsa); Entre
El Oro y Via Victoria, Jun. 19-20, 19.

Cc

vicinity of Morelia, north of Zapote, Aug. 4, 1910, G. Arséne 6850 (cH, Mo, us); vicinity
of Morelia, Jaripeo, July 13, 1911, G. Arséne 5598 (cu, Mo). Oaxaca: us km. southwest
‘ ; ee.

6 .
Jul. 15, 1932, F. Miranda 2814 (MExu). San Luis Potosi: alrededores de La Salitrera,
Municipio de Zaragoza, Aug. 2, 1959, Rzedowski 11408 (ENcB).

8. Pennellia juncea (Schulz) Rollins, comb. nov.

Based on Heterothrix juncea O. E. Schulz, Das Pflanzenreich IV. 105. lage aba ssaal
brieae 86: 297, 1924. Collected in San Luis, Puebla, Mexico, 1908, a “dl ge
3486a (holotype ps), vicinity of San Luis, Tultitlanapa, Puebla, near Oaxaca, July,
1908, C. 4. Purpus 3486 (uc, probable isotype).

Biennial, or Possibly perennial, glabrous throughout; stems stiffly , —
branched above, 4-6 dm. tall, purplish especially above; basal a - ae ar
lower cauline leaves petiolate, narrowly oblanceolate, entire to shallow y den ,
tuse, 3-5 cm, long including petiole, 4-8 mm. wide; upper cauline leaves na

20 REED C. ROLLINS

linear, acute, entire; infl inating each

bracts; buds globose to slightly elongated; sepals erect at anthesis, non-saccate, narrowly
oblong, 3-4 mm. long, ca. 1.2 mm. wide; petals white to faint lavender, spatulate,
not differentiated into blade and claw, 5-6 mm. long, ca. 1.5 mm. wi

rigid, remote, 12-18 mm. long; siliques terete, divaricately ascending, 3.0-3.5 cm. long,
ca. 1.2 mm. wide; valves obscurely nerved toward base, purplish; styles obsolete to
less than 0.5 mm. long; stigma entire; seeds oblong, in a single row, not crowded,
somewhat embedded in tissue of septum, marginless, 1.0-1.2 mm. long, less than
mm. wide; radicle exceeding cotyledons; cotyledons incumbent.
Known only from the type collection from near Oaxaca, Mexico,

Pennellia juncea is most closely related to Pennellia patens with
which it shares the distinctive feature of having smooth oblong seeds
that occur in a single row in the silique. This feature apparently
characterizes the more southerly populations of P. patens but in the
northerly populations, the seeds are crowded and misshapen as in
most species of Pennellia. From the specimens available, the impres-
sion of rigidity of stems, pedicels and siliques is a striking feature
of P. juncea. This is the only North American species we have seen
that is completely glabrous.

are immature.

LITERATURE CITED

AL-SHEHBAZ, IsHAN A, 1973. Th
Contrib. Gray Herb, 204: ]_ :
Bentuam, Grorce. 1839-1857. Plantae Hartwegianae. London. 1-393.
Gray, A. 1849. Plantae Fendlerianae. Mem. Amer. Acad. Arts & Sci. IV (1): 1-16.
McVaucu, Rocers. 1970. Introduction to the Facsimile Reprint of George Bentham’s
102.

e biosystematics of the genus Thelypodium (Cruciferae).

MUELLER, Arc. 1860, Apocynaceae in Martius, Fl. Bras. VI(1): 1-195. (p: 138, t 40).
NieuwLanp, J. A. 1918. Heterothrix (B. L. Robins.) Rydb. A synonym and other notes.
25

R : Synop. Flora N. Amer. 1(1): 98-180.
peggtacs R. C. and L. Riipenserc. 1 7. Chromosome numbers in the Cruciferae III.
Contrib. Gray Herb. no. 207: 101-116.

THE GENUS PENNELLIA pat

RypgBerc, P. A. 1907. Studies on the Rocky Mountain flora—XVIII. Bull. Torr. Bot.
Club 34: 417-437.

Scuutz, O. E. 1924. Cruciferae- Sisymbrieae. Das Pflanzenreich, heft 86, Iv (105):
294-300

. 1936. A. Engler and K. Prantl, Die Natur. Pflanzenf. ed. 2, 17b: 643-644.

STUDIES ON SOLANACEAE. XII. ADDITIONS TO THE
ENUS CHAMAESARACHA

ARMANDO T. HuNZIKER!

SUMMARY

aracha rzedowskiana A. T. Hunz. sp. nov. is Geacnied and .
is Potos

rooting at stem nodes, while the other two are erect plants up to 1 m high; probably
the three are annuals) and their southern range where they occur in different types
of forest vegetation. An Shontiticnea key is presented.

RESUMEN

e describe e ilustra Chamaesaracha rzedowskiana A. T. Hunz. sp. nov. (México:
a Luis Potosi), Sf con Ch. potosina Rob. et Greenm. (México: San Luis Potosi)
cernua (Donn. Smith) A comb. no éxic inaloa y Guerrero

n a s
del género; al efecto, se trae a colacién una referencia bibliografica pertinente. Se
incluye por fin una clave diferencial, para -facilitar ‘al reconocimiento de las especies

atadas.

Chamaesaracha rzedowskiana A. T. Hunziker, sp. nov.

(FIGURE 1)
Herbae depressae (annuae?) ubique pubescens caulibus repentibus. Laminae conco
lores parvae, late ov atae, leviter cordatae, — acutae, (0.48-)0.9-2.7(-4.6) cm longae

et (0.45- 08 82- 2.2(-3.0) cm latae; petioli 0.6-2.3 cm longi. Flores ere solitarii;
Pedicelli (14— ~)18-23(-35) mm longi. Calyx campanulatus 5-lobatus, 1.5-2.0 mm longus
extus piliferus (trichomata magi apie vel glandulosa antag unicelulan
lobi t triangulares ca. m longi et (basi) 0.9-1.1 mm lati, nata
aliquantum breviori. Corolla + rotata 9-1 1 mm diam; lobuli 5 late triangulares acuti,

fai Ca. 00 longus" intus in parte inferiore ‘glaber, sed parte inferiore bea
Corollae cum trichomatibus simplicibus et plerumque ramosis sparse obsita. Fi wen
haga interdum ene inaequalia; antherae oblongae filamentis eating ca. 1.2-1
m latae Stylus leviter obsubulatus; stigma c um, depressum

Pi edicelli fructiferi baie elongati usque ad 40 mm longi; clatter alee accrescens
3.1-3.3 mm longus) fere omnino baccae arcte ie ean aa pericarpium tenue, in co
translucens; granula sclerotica desunt. Semina pauca (14-25) Ske tuberculato,
ve 0S longa, 0.9 mm lata et 0.45 mm crassa.

TYPE: México eo peels Potosi: Municipio de Xilitla, Las Crucitas. 600 m ;
Rzedowski 10103, 1 Mar. 1959. “Ladera caliza con yegetacion de bosque cous
Secundario,” (Holotype: enc; isotype: CORD) ;

‘Me ember of the ‘ “Carrera del ruins (CONICET, Argentina). Address: Museo Botanico,
ond Nacional de Cérdoba, Casilla de Correo 495, Cordoba, Argentina.

as

Fic. 3 Chamaesaracha Tze

ventral views)

view) with two attached stame

>

ARMANDO T. HUNZIKER

bts2 f—

WES
sa
a
Y

dowskiana: A, D, anther (dorsal and
x 16.5; B, flower, x 6.5; C, sector of corolla (internal
ns, X 6.5; E, K, L, seed (transverse

}
Ly
ete
$e
ae) a
oF me
A
v

and longitudinal sections and lateral view), x 25; F, fruit, x 6.5; G,

Synoecium, X 8.5; H, I, calyx trichomes,

x 83; J, branch, x 0.83.

STUDIES ON SOLANACEAE pase

Herbaceous pubescent plants (annual?) with prostrate stems rooting
at the nodes. Leaves small, thin; blades concolorous (0.48-)0.9-2.7(-4.6)
cm long and (0.45-)0.82-2.20(-3.0) cm wide; petioles 0.6-2.3 cm long.
Flowers solitary; pedicels (14-)18-23(-35) mm long. Calyx campanu-
late, five-lobed almost to the middle, 1.5-2.0 mm long; external surface
pubescent with few-celled simple and glandular trichomes (the latter
with unicellular heads); lobes deltoid, somewhat shorter than the tube
plus limb, ca. 0.85 mm long and 0.9-1.1 mm wide at the base. Corolla
rotate 9-11 mm diam, with five small broadly triangular acute lobes,
pubescent externally (particularly at the apex and margins) with simple
and branched trichomes; tube ca. 0.95 mm long, interior glabrous;
lower part of the limb with simple and branched trichomes. Filaments
glabrous, sometimes slightly unequal in length; anthers oblong, shorter
than the filaments, ca. 1.2-1.4 mm long and 0.75 mm wide. Style
somewhat obsubulate; stigma capitate, depressed. Fruiting pedicels
up to 40 mm long; the accrescent calyx (3.1-3.3 mm long) tightly
enclosing the berry; pericarp thin, translucent when dry, lacking
sclerotic nodules. Seeds few (14-25), tuberculate, ca. 0.95 mm long,
0.9 mm wide and 0.45 mm thick.

During a visit to the herbarium of the Escuela Nacional de Ciencias
Biologicas (ENCB), I found Chamaesaracha rzedowskiana in the
collections of the Solanaceae there. Although the materials do not
allow definite conclusions about the underground organs and the
Persistence of individual plants, most probably they are annuals,
lacking a woody tap root and secondary gemmiferous roots typical
of the more northerly xerophytic species of the same genus.

Chamaesaracha rzedowskiana differs from the other species of the
8enus because of its thin leaves, with smaller and broader blades.
The prostrate stems rooting at the nodes are another distinctive
vegetative trait. The most closely related species are Ch. potosina
Rob. and Greenm. and Ch. cernua (Donn. Smith) A. T. Hunz., which
are considered below.

Chamaesaracha cernua (Donnell Smith) A. T. Hunziker, comb. nov.

Athenaea cernua Donnell Smith, Bot. Gaz. 48: 297. 1909. TyPE COLLECTION: parece
Dept. Alta Verapaz, Sasia, 900 m alt. H. von Tuerckheim 1.2245, May 1908. HOLOTYPE:
US; ISOTYPEs: CORD, W. z > : exi

Physalis capsicoides Bitter, Repert. Sp. Nov. 20: 371. 1924. tyr Seecamnenanan
Veracruz, Papantla, Schiede 1191, Jan. 1829. I nie -— in eve
of this collection, but the long and precise description by Bitter agrees i ry
detail with the plant under discussion. ,

Physalis heb (Robins.) Bitter var. cernua (Donnell Smith) Waterfall, Rhodora
69: 99. 1967.

. i . Calyx 3.5-4
Herbaceous pl ti high. Flowers solitary, rarely geminate : :
long, deeply ik aa ickny oblong lobes and with three types of trichomes (simple

26 ARMANDO T. HUNZIKER

multicellular, long-stalked with unicellular glandular heads, and short-stalked with
multicellular glandular heads). Corolla rotate, 6-7 mm long. Anthers 1.3-1.8 mm long,
on filaments exceeding them in length. Fruits orange-red with a thin and t anslucent
pericarp, lacking sclerotic nodules; accrescent calyx 9-10 mm in length, tightly enclosing
most of the pericarp. Seeds tuberculate, 1.2-1.4 mm long.

few and incomplete.
Athenaea Sendtner is a small Brazilian genus of about ten woody

unacceptable for this ‘very distinctive element in the Mexican and
Central American floras;” its floral and fruit characters strongly advise
against such an action. It should be added that for the same reason

A. cernua. The genus where this problematic species fits adequately
is Chamaesaracha (A. Gray) A. Gray. This is especially evident since
the discovery of Ch, rzedowskiana described above.

ADDITIONAL SPECIMENS EXAMINED. MEXICO. GUERRERO: Distr. Galeana, Carrizo-El Rio,
800 m alt., Hinton 14689, 20 Oct. 1939, “by stream in mixed forest; 1 m high, fl.

; H, NY, US). :
de Tarahumare, Sierra Surotato, 3000-4000 ft, H. Scott Gentry 7306, 17 /24 Mar. 1945,
Deep wooded canyon with mixed subtropical vegetation; shade. Wide spreading bush

TABLE 1. SOME REPRODUCTIVE STRUCTURES OF ATHENAEA AND CHAMAESARACHA CERNUA

Athenaea Ch ha cernua

Calyx cleft almost to its base, the lobes
tube.

Calyx with five lobes shorter than the
much longer than the tube

Corolla aestivation plicate; lobes
broader than long, shorter than the

aa 2 tube and the limb; limb with a basal
Ss
Fi ng of sc.

ilaments short (usually shorter to i eres long trichomes

: Filaments longer than the anthers,

lacking lateral indentations toward
part adnate to corolla tube, two lateral base. °
ase

indentations present toward b

Pericarp thick, coriaceous and Opaque. Pericarp thin, membranaceous and
1

translucent.

STUDIES ON SOLANACEAE 27

1 m high” (us). GUATEMALA. DEPT. PETEN: Parque Nacional de Tikal, en camino del
Templo no. 4, Tun Ortiz 998, 23 Apr. 1970, “Hierba en foresta alta; fruto colorado”
(US, Ny). BELIZE: El Cayo District, Arenal-Valentin road, Lundell 6181, Jun.-Aug. 1936,
“Along roadside; herb; height 1 m”, (cu, us, NY).

According to studies of Averett (1973), the genus Chamaesaracha
comprises seven species, “largely restricted to the arid regions of
southwestern United States and northern Mexico.” The addition of
Ch. potosina,’Ch. cernua and Ch. rzedowskiana increases the genus
to ten species; also the geographical distribution is enlarged, with
the inclusion of Belize, Guatemala, and regions of southern and western
Mexico. The distinct geographical distribution of this meridional and
not strictly desert inhabiting group of three species, coupled with
some morphological characters (for example, the quality of the indu-
ment), may in the future permit the recognition of two sections in
the genus. In case this position is adopted, the section Capsicophysalis
Bitter (Repert, Sp. Nov. 20: 370, 1924) proposed to accomodate Ch.
cernua (under the name Physalis capsicoides (Bitter) should be used.
Chemical data, such as those of Averett (1973) for the xerophytic
species, might be useful in elucidating this problem. For instance,
it is not known whether these plants are perennials or annuals. The
chances are in favor of them being annuals; the type collection of
Physalis capsicoides (Schiede 1191) was considered annual by Schiede
(Schlechtendal et Chamisso, 1831: 378). This is an important feature

€cause it is a well-established fact that the seven northern xerophytic
Species are perennials, frequently with gemmiferous roots.

The following key is provided to facilitate the identification of
the three species treated in this paper.

KEY TO THE SPECIES

l. Stems Prostrate, rooting at the nodes. Corollas a mm pee aie er ~~
I f the corolla. Anthers oblong, 1.2-1.
ioe ee eS ae . : a Se 1. Ch. rzedowskiana.
1’. Stems erect, up mim high.
2. Corolla 40-48 mm long, with lobes slightly shorter than the tube and ~ limb.
Anthers ca. 0.6 mm in length (slightly less in breadth) ....... ; . e oe
2’. Corolla ca. 7 mm long, with lobes about 1/2 as long as the tube “ a hea
Anthers oblong, 13-18 mm lone 666 i 6 ee se ee | CR: ;

ACKNOWLEDGEMENTS

F i tudy at
Thanks to the generosity of Dr. J. Rzedowski, I had the opportunity to stud
the Herbarium peta of ee Escuela Nacional de Ciencias Biologicas (Instituto

Sen rnacearacha potosine Robinson & Greenman, Amer. J Sci St 16 155, wouors: México,
an Luis Potosi, Tamasopo Canon, Pringle , 25 Nov. dais centre firey

only from the type colleen: is closely related to the two other species ssa Bes — ct

The evidence makes wholly unacceptable its transference to Saracha (Averett, . Mo. :

37(3): 380, 1970) which is quite a different genus indeed.

28 ARMANDO T. HUNZIKER

Politecnico Nacional, cri o, D. F.) ] teful to the staff of the G I
and the Arnold Arboretum of Harv os ae for the facilities ‘mbes available
including access to net Harvard Ghivecs sity library and herbarium. Thanks are due

as well to Prof. Reed C. Rollins for reviewing ‘ie manuscript and oo calustdi
suggestions. The illustration was skillfully done by Mrs. N. M. de

LITERATURE CITED

AVERETT, Has 1970. New ec ueepinen in peed neon (Solanaceae) and comments
lis. Ann. Mo. Bot. Gard, 57(3): 380-391.
—————.. 1973. Biosystematic wai — Ebay ects (Solanaceae). Rhodora
75(803): sis hol. f. 1-13.
Bitter, G, 1924. Zur Gattung Physalis. II. Repert. Sp. Nov. 20(577- 580): 369-372.
saeg J. L. 1973. Studies in Mexican and Central American Solanaceae. Phytologia
26(4): 265-278, f. 1-2
——_——. . STANDLEY. 1974, Solanaceae [Flora of Guatemala]. Fieldiana, Bot.
24(10)1-2: Ee 151, ep 1-20.
Hunziker, A. T. 1979. South American Solanaceae: A s pti In . Hawkes,
. N. Lester & A. : Skelding (eds.), The TR pe and Taxonomy of the Solanaceae.
-11.

Linnean Society obi aetna share rt ae:
Rosinson, B. L. & J. M. Green _ 1895. ew eee noteworthy — yeseTe from
axaca, collected by Wes: C. G. engu: L. C. Smith and E. W. Nelson. Amer.

J. Sci. 50: 150-168
SCHLECHTENDAL, D. pe 2 & A AD. DE Cuamisso. 1831. Plantarum mexicanarum a cel. viris
Schiede et Deppe pe ae Sa Recensio brevis. Linnaea 6: 352-384.

A GENERIC ATLAS OF HAMAMELIDACEQUS POLLENS

A. Linn Boc.e?”
C. THoMas PHILBRICK”

The Hamamelidaceae, when considered in the broad sense, be
including the subfamily Liquidambaroideae, which is often split off
as a segregate family Altingiaceae, presently consists of 28 genera,
or 29 if the poorly known segregate genus Semiliquidambar H. T.
Chang (1962) is admitted.

Among angiosperms the Hamamelidaceae appear to be a relatively
ancient family, with a very high proportion of small or monotypic
genera having highly restricted or narrowly endemic distributions,
and a high concentration of both genera and relatively primitive floral
types in Southeast Asia. Concomitant with this phytogeographical
evidence of antiquity is a fossil record which, in the form of the
readily identifiable periporate pollen grains of Liquidambar, extends
back at least as far as the Paleocene (Muller, 1970) or possibly even
the Cretaceous on the basis of macrofossils of Liguidambar (Brown,
1933a, b). But the pollen of Liquidambar is relatively specialized
within the family. If reliable identification of the more generalized
and primitive tricolpate pollens which characterize the large majority
of hamamelidaceous genera becomes possible, the microfossil record
of the family may be found to extend considerably further back into
the Cretaceous.

The pollen morphology of the family has been the subject of several
investigations in recent years, but none of these has covered the entire
family, and most have considered only a single genus or a few of
the more common genera. Several of the more significant studies
include the following:

Simpson (1936) categorized the pollen grains of 18 extant hamame-
lidaceous genera in six morphological groups which he considered
“onvenient, and described fossil pollens of six genera which he
attributes to the family from Tertiary lignitic coals of Argyllshire,
western Scotland. ee

Erdtman (1943) provided descriptions of Hamamelis and Liquidam-
bar as examples of Hamamelidaceae. He later (1946, 1952) provided
brief descriptions of the pollen of 18 species from 15 genera, including
Distylium guatemalense (= Molinadendron guatemalense (Radlk. ex
Harms) Endress) as representative of the genus Distylium, and S$ nat
dunnii (= Distyliopsis dunnii (Hemsl.) Endress) as representative o
the genus Sycopsis. In this respect a currently recognized os aaen e
of Distylium is not included in his descriptions, but one of Sycopsis

Me : ; iment Station.
aocientific Contribution Number 1D32 from the New Hampshire ae ne i iat ar aa
* a address: University of New Hampshire, Botany and Plant Pathology Dep :

“A. 03824,

29

30 BOGLE AND PHILBRICK

is by virtue of the fact that he added a description of Sycopsis
griffithiana Oliv. in 1952. The 1946 descriptions apparently provided
the basis for his general description of the pollen morphology of
the family (1952).

Ikuse (1956) also provided brief descriptions of the pollen of ten
species of six genera native or cultivated in Japan (Liquidambar,
Disanthus, Loropetalum, Hamamelis, Corylopsis, Distylium).

Lee (1969), in a master’s thesis (unpublished) written at the Univer-
sity of Pennsylvania, described the pollen of 22 species of ten genera,
and recognized four pollen types: tricolpate, tricolporoidate, tri-
colporate, periporate.

Several authors have published detailed analyses of the pollen of
Liquidambar and Altingia in relation t iderat of the taxonomic
position and evolutionary history of these genera (e.g., C. T. Chang,
1958, 1959; Kuprianova, 1960; Makarova, 1957; Samorodova-Bianki,
1957; Sears, 1930).

The most extensive study to date has been that of C. T. Chang
(1964), who examined and described (in Russian) 57 species of 21
genera of the family, including seven species of Altingia and Liqui-
dambar which he separates in a family Altingiaceae, and 50 species
of 19 genera of Hamamelidaceae. His descriptions are accompanied
by 21 plates of photomicrographs. Several very rare genera were not
included in this study for lack of material, and several new genera
have since been recognized in the family.

Hesse (1978) describes the ultrastructural characteristics of the exine
and “pollenkitt” of Hamamelis vernalis, H. virginiana, Corylopsis
platypetala, and Parrotia persica in relation to their entomophilous
(sticky pollen) or anemophilous (dry pollen) pollination ecology.

Most of these studies preceded the development and widespread
use of the scanning electron microscope as a research tool in morpho-
logical studies, and were therefore conducted at the level of the light
microscope, which is still the basic tool in pollen analysis. However,
scanning electron micrographs of pollen grains, because of their
three-dimensional appearance and great magnification, are much more
useful for illustrating the surface details of grains than are light
micrographs, and are increasingly being used in conjunction with
light microscope analyses in palynological studies.

Scanning electron micrographs of hamamelidaceous pollen grains
are infrequent in the literature, the largest assemblage that we know
of being a group of eight photographs appearing in a recent review
paper on evolutionary trends in the Hamamelidales-Fagales group
(Endress, 1977). The SEM photographs presented here have resulted
from a continuing effort to develop descriptions and illustrations of
all 28 genera, in the hope that a broader base of data than has previously

HAMAMELIDACEOUS POLLENS ot

been available will facilitate interpretations of intra- and inter-familial
relationships and phylogenetic trends.

The genera illustrated here, and their distribution among the sub-
families and tribes of the family, for purposes of this work, are as
follows:

SUBFAMILY LIQUIDAMBAROIDEAE. Liquidambar, Altingia.

SUBFAMILY RHODOLEIOIDEAE. Rhodoleia.

SUBFAMILY EXBUCKLANDIOIDEAE. Exbucklandia (syn.: Bucklandia, Sy-
mingtonia), Mytilaria, Chunia.

SUBFAMILY DisANTHOIDEAE. Disanthus.

SUBFAMILY HAMAMELIDOIDEAE.

Tribe Hamamelideae: Maingaya, Ostrearia, Neostrearia, Trichocla-
dus, Dicoryhe, Hamamelis, Tetrathyrium, Loropetalum, Embo-
lanthera.

Tribe Eustigmateae: Eustigma.

Tribe Corylopsideae: Corylopsis, Fortunearia, Sinowilsonia.

Tribe Fothergilleae: Fothergilla, Parrotiopsis, Parrotia.

Tribe Distylieae: Sycopsis, Distyliopsis, Histylium, Molinadendron,
Matudaea.

MATERIALS AND METHODS

Pollen samples were taken from specimens collected in the field
or in botanic gardens, or from specimens in or on loan from several
major herbaria. Both pickled and dried materials were acetolyzed
(Faegri and Iversen, 1975). For light microscope examination part
of each sample was then mounted in glycerine jelly (Erdtman, 1952)
and set aside for a period of at least two weeks to allow the grains
to adjust to the mounting medium. Measurements were then made
under oil immersion of at least thirty grains in each sample. The
€quatorial and polar axes were measured as seen in equatorial view.
The ratio of the average dimensions of these axes (P/E) was then
used to determine the shape classification according to the designations
of Erdtman (1952). The magnification bars included in micrographs
of whole grains represent lengths of approximately 10 pm, while those
in higher magnification pictures of surface details represent approxi-
mately 1 um.

For scanning electron microscopy, acetolyzed grains were washed
in two successive distilled water washes, followed by two washes
of 100 percent acetone, the first for ten minutes and the second for
30 minutes. The pollen was then transferred to aluminum stubs a4
means of Pasteur pipettes. The stubs had been previously _ se
an acetone-tape adhesive (1 mm length of half-inch double-stick tape

as BOGLE AND PHILBRICK

partially dissolved in 10 ml of 100 percent acetone). The stubs were
then coated with palladium-gold on a Technics Hummer 2 sputter
coater and examined with an AMR 1000 scanning electron microscope
at the Museum of Comparative Zoology of Harvard University.

OBSERVATIONS

Size. Pollen size in the family ranges from very small grains,
averaging less than 20 pm along the largest diameter, to large grains
averaging close to 60 ym along the largest diameter.

In general, the smallest grains among our samples are concentrated
among the genera of the tribe Hamamelideae (subfamily Hamameli-
doideae), in which flowers are complete, often contain staminodia
and/or sterile disc lobes, and are presumably insect-pollinated. Con-
versely the largest grains appear among those genera which tend
toward, or have advanced to, the naked-flowered, wind-pollinated
state (Liquidambaroideae, Chunia among the Exbucklandioideae,
members of tribe Distylieae in the Hamamelidoideae).

In contrast to the small grains of the Hamamelideae are the very
large grains of the genus Eustigma (Eustigmateae, Hamamelidoideae),
which is generally considered closely allied to the Hamamelideae.
Although the flowers of Eustigma are complete, their petals are small
and their styles greatly elongated, with broadly expanded sitgmatic
surfaces, suggesting a specialized pollination mechanism which is,
as yet, unreported.

Pollen size can also vary significantly among the species of a genus
(e.g., Fothergilla, Corylopsis), or even among geographical races of
the same species (C. T. Chang, 1964).

Among the most morphologically distinct pollen forms in the family
are those of the large, nodding, red-flowered pseudanthia of Rhodoleia
(Rhodoleioideae), which are reported to be bird-pollinated. These
grains are small, with a very smooth surface marked only by minute
perforations.

Shape. Pollen grain shapes among our samples ranged from oblate,
through spheroidal, to prolate, with most of the taxa falling within
the subspheroidal categories. Wind-pollinated genera with porate or
polyporate grains, such as Altingia, Liquidambar and Sycopsis, exhibit
large spheroidal grains. Mytilaria appears to be consistently oblate
(C. T. Chang, 1964; Lee, 1969). However, shape designations are
based on averages and may not reflect the full range of variation
within a genus or species. Grains of the monotypic genus Parrotiopsis,
for example, vary from oblate to prolate in shape, although the average
form falls within the subspheroidal range. Furthermore, it is well-
known that grain dimensions may be affected by the mounting medium

ae

HAMAMELIDACEOUS POLLENS 33

used, so one cannot be certain whether differing reports of grain
shape are due to natural variation in the plant population, or to
technique. For example, the general form of grains of Loropetalum
chinense in our preparation (acetolysis, glycerine jelly) is subprolate,
but C. T. Chang (1964, Methyl green/Glycerine jelly method of
Wodehouse, 1935) reports grains of this species to be prolate, and
Lee (1969, unacetolyzed grains in aceto-carmine, poly-vinyl alcohol,
lactic-triacetin, or Calberla-basic fuchsin) reports them to be oblate-
spheroidal. Consequently reports of pollen grain shape should be
considered very carefully, and may be of little value from a taxonomic
point of view. The general shape classification for our sample of
each genus illustrated here is included in the plate caption.

Apertures. A large majority of hamamelid genera are tricolpate, with
apertures varying in length, sharpness or bluntness of terminations,
margin, and membrane characteristics among the genera. In general,
apertures are long, with regular margins and pointed terminations
approaching the poles in genera with complete flowers. Apertures
become shorter, with blunt to round ends, and indistinct, irregular
margins in those genera which tend toward incomplete or unisexual,
wind-pollinated flowers. Rugate grains appear in Chunia, Matudaea,
Sycopsis, and Distylium, and polyporate grains appear in Altingia
and Liquidambar. Pore shaped apertures also appear rarely in Chunia
(C. T. Chang, 1964), and with greater frequency (presumably through
modifications of colpi or rugae) in Distylium, Matudaea and Sycopsis.

It thus appears that rugate to polyporate grains have arisen within
three separate lineages within the family, in conjunction with a
transition to naked, bisexual or unisexual, wind-pollinated flowers
(Liquidambaroideae; Chunia among the Exbucklandioideae; Disty-
lium, Matudaea, Sycopsis among the Distylieae in the Hamamelidoi-
deae.

Tricolporate or tricolporoidate grains have been reported for a
number of genera. Simpson (1936) described some degree of pore
structure in the colpus membranes of twelve genera (Exbucklandia,
Corylopsis, Disanthus, Eustigma, Fortunearia, Fothergilla, Hamame-
lis, Loropetalum, Rhodoleia, Sinowilsonia, Tetrathyrium, Trichocla-
dus). Lee (1969) also reports tricolporate grains in Exbucklandia, and
C. T. Chang (1964) for Rhodoleia. Tricolporoidate grains are described
for Mytilaria and Exbucklandia by Erdtman (1952), Chunia and
Sycopsis by C. T. Chang (1964) and Fothergilla by Lee (1969). Among
the photographs included here there is the appearance of central
pore-like protrusions in the colpi of Fothergilla. This aspect of aperture
structure is in obvious need of detailed investigation.

Aperture margins are distinct in most genera and often differentiated
as a pronounced margo with even or uneven edges, but margins become

34 BOGLE AND PHILBRICK

indistinct in Fortunearia, Sinowilsonia, Parrotia, Distylium, Sycopsis,
Distyliopsis and Molinadendron. In Dicoryphe, and to a lesser extent
in Embolanthera, the margins of the apertures consist of the deep
muri of the exine reticulum, which, on invagination of the colpus,
close over and obscure the aperture.

Aperture membranes range from finely to coarsely granular. In some
genera the granules appear to be concentrated in a longitudinal band
in the center of the membrane, surrounded by a relatively smooth
border. In this sense the membranes of Disanthus bear deeply sculp-
tured opercula. In Parrotia the coarse granules of the membranes
are fused in irregular ornate or vermiform patterns.

Exine. Sculpturing of the exine surfaces is generally reticulate in
the family. The overall trend of modification is from a very coarsely
reticulate pattern, with deep muri borne on pronounced bacula (e.g.,
Dicoryphe), through reduction of the meshes of the reticulum and
depth of the muri, to very finely reticulate (Chunia, Parrotiopsis,
Distylium), foveolate (Altingia, Liquidambar, Matudaea), or scrobicu-
late (Rhodoleia) patterns, with correspondingly shallower muri or a
thinner tectum. In general this trend corresponds to the trends of
floral modification from complete and presumably insect-pollinated
types to incomplete, naked and/or unisexual forms, and is more or
less evident among the three genera of Exbucklandioideae (Exbucklan-
dia, Mytilaria, Chunia).

In a number of genera the exine reticulum is considerably finer
in the polar than in the equatorial regions. This condition is well-
developed in Eustigma, in which the polar areas may become scro-
biculate or partially psilate. The very smooth, scrobiculate grains of
Rhodoleia may also appear psilate in some cases.

The muri of a number of genera bear verrucae (Altingia, Liquidam-
bar, Fortunearia, Distylium, Distyliopis, Sycopsis, Parrotia). These
are mostly wind-pollinated taxa but the significance of this structural
feature in relation to the pollination mechanism is not known at present.

Unusually variable exine sculpturing, previously unreported, is

evident in our samples of Sinowilsonia and Molinadendron, and is
illustrated.

ACKNOWLEDGEMENTS

We wish to express our a
by the Central University R
Experiment Station (Project H-216)

tend special thanks to Dr. Otto T. Solbrig, Director of the Gray Herbarium,

nt early
Solbrig, Dr. Alice Tryon (Gray Herbarium)

HAMAMELIDACEOUS POLLENS 35

A number of individuals have very kindly assisted, either directly or indirectly,
in helping us to obtain material for this psd wc , and we — to express our
thanks to all of them, including: Dr. A. R. uchebute for flowers of Parrotiopsis,
Dr. . Gentry 208 Dr. Eizi Woo ny saree with collecting cawittes in Mexico;
Mr. Lau. Mr. Tang and Mr. r help with collections in Hong Kong, and Dr.
Masami hic abena for mbectal of Disanthus from Japa

The Directors of the very generous in providing specimens
or in Bowe us to draw on their living or tars eikaiws for pollen material:
Ar ; Hong Kong Botanic Garden; Morris Arboretum

e University of Pennsylvania; Museum de l’Histoire "Nauvelle. Paris; Plant

a additi
Esher valuable Eee from the Chinese literature, and» o Mr. A. Baranov

and Mr. Edward Seling, SEM Laboratory, Museum of Comparative Zoology, Harvard
University, cantebuted his expertise in making the photographs.

36 BOGLE AND PHILBRICK

PiaTE 1. Liquidambar L. (Subfamily Liquidambaroideae)
VOUCHER MATERIAL: Liquidambar orientalis Mill.; Bogle 973: from
cultivated tree in authors collection (NHA).

A. Somewhat angular, periporate grain with foveolate exine and
numerous granules of irregular size and shape on the pore membranes.
Note the minute supratectal verrucae scattered on the surface of the
tectum, x 2959. Shape: spheroidal. Size range: 32-55 pm. (Bar =
10 pm.)

B. Close up of grain surface showing a rounded aperture with a
somewhat irregular margin, the supratectal verrucae, and the verrucate
granules of the pore membrane, x 10,000. (Bar = 10 pm.)

A genus of three or four widely disjunct, wind-pollinated species
distributed in Southeast Asia (L. formosana Hance), southeastern Asia
Minor (L. orientalis Mill.) and southeastern North America and Central
America (L. styraciflua L.). The Central American specimens are
segregated by some authors as L. macrophylla Oerst. The genus has
been the subject of several palynological investigations aimed at
working out the evolutionary history and taxonomic relationships of
the extant species (C. T. Chang, 1958, 1959, 1964; Kuprianova, 1960;
Makarova, 1957)

Pore number, shape, diameter and margin characteristics appear
to be somewhat variable within and among the species. Pores range
from circular to elongate. Pore margins may be more or less even
to fissured (“cracked’’)

37

HAMAMELIDACEOUS POLLENS

38 BOGLE AND PHILBRICK

PLaTE 2. Altingia Nor. (Subfamily Liquidambaroideae)
VOUCHER MATERIAL: Altingia chinense Oliver ex Hance; Bogle 583: Hong
Kong Botanic Garden (NHA); A. excelsa Nor. Bogle 313: Mentigi
Forest Reserve, Cameron Highlands, Malaya (NHA). A. obovata Merr.
and Chun; H. Y. Liang 64734: Hainan, China (AA).

A. Periporate grain of Altingia chinense with foveolate exine; pore
membranes bearing granules of varying size and shape. Note the small
supratectal verrucae uniformly distributed over the surface of the
tectum and on the membrane granules, x 2780. Shape: spheroidal.
Size range: 35-58 ym. (Bar = 10 Lm.)

B-D. Close-ups of grains of three species to show sexine, pore
margin, pore membrane and membrane granules (some of which bear
verrucae). B, A. excelsa, X 10,000; C, A. obovata, X 10,000; D, A.
chinense, X 10,000. (Bar = 10 Lm.)

Altingia is a genus of one (Vink, 1957) to about 13 species (various
authors; see Index Kewensis, H. T. Chang, 1973). The flowers and
inflorescences of the genus are similar to those of Liquidambar.
Pollination is anemophilous. The pollen grains of Altingia in our
material have characteristically round pores which may vary slightly
in diameter, and have a distinct margin which is very nearly regular
or even. Figures B-D illustrate the variation, which is apparent also
in light microscope observations, of perforation size in the reticulum
of the tectum. In this respect A. excelsa (B) has many minute, rounded
perforations; A. obovata (C) exhibits rounded perforations of larger
diameter; and A. chinense (A, D) has perforations which may vary
in shape from rounded to somewhat elongate or irregular. From a
taxonomic point of view the significance of this variation is obscure.

a
Z,
(]
-
ra]
a
—
a
<2)
S
ec)
5
O
<
Q
=
-]
fx]
=
<
=
2
<
jo

40 BOGLE AND PHILBRICK

PLaTE 3. Rhodoleia Champ. ex Hook. (Subfamily Rhodoleioideae)
VOUCHER MATERIAL: Rhodoleia championi Hook. f.; Bogle 276: Klang
Gates, near Kuala Lumpur, Selangor, Malaya (NHA).

A. Oblique view showing relatively smooth, scrobiculate tectum
and two invaginated colpi, x 5800. Shape: subprolate. Size range
in equatorial view: equatorial axis, 16-26 wm; polar axis, 20-29 pm.
(Bar = 10 pm.

B. Slightly oblique polar view showing very small perforations
(lumina) of the tectum and granular sculpturing of invaginated colpus
membranes, X 5920. (Bar = 10 um.)

A genus of one variable species (Vink, 1957: Tardieu-Blot, 1965)
or about 7-10 distinct species (see Exell, 1935; H. T. Chang, 1973;
discussion in Tardieu-Blot, 1965) distributed from southern China
southward through Indochina and Malaya to Sumatra. The pollen
grains have been described as being “‘tricolpate-porate” by C. T. Chang
(1964), but a pore is not evident in our scanning electron micrographs.
Rhodoleia may be unique among hamamelids in being bird-pollinated
or at least partially so (D. van Leeuwen, cited in Vink, 1957). The
birds may be attracted to nectar reportedly secreted by a cycle of
glands inserted between the stamens and the ovary, and the sticky,
rather than powdery pollen may represent an adaptation toward
ornithophily.

HAMAMELIDACEOUS POLLENS

42 BOGLE AND PHILBRICK

PLATE 4, Exbucklandia R. W. Brown. (Subfamily Exbucklandioi-
deae)
VOUCHER MATERIAL: Exbucklandia populnea (R. Br. ex Griff.) R. W.
Brown; Bogle 314: Tanah Rata Village, Cameron Highlands, Selangor,
Malaya (NHA).

A. Equatorial view showing the coarse reticulum. Note the large
lumina of fairly uniform size but somewhat irregular, angular shape;
the margin of very small lumina bordering colpi; deep muri resting
on distinct bacula; coarse granules on surface of the nexine; granular
colpus membranes, x 3520. Shape: prolate spheroidal. Size range
in equatorial view: equatorial axis, 26-32 um; polar axis, 23-37 pm.
(Bar = 10 pm.)

B. Polar view showing coarsely reticulate exine with lumina not
differing markedly in size from those of the mesocolpium. Note the
colpi with acutely pointed ends reaching nearly to the poles; the

distinct bacula subtending the muri; coarse granules on nexine surface,
xX 3540. (Bar = 10 Lm.)

A genus of two (Vink, 1957; Tardieu-Blot, 1965) or three (H. T.
Chang, 1973) species ranging from the Sikkim Himalaya to central
China, and southward through Indochina and Malaya to Sumatra.
The mature flowers of Exbucklandia are incomplete, lacking a calyx,
and are rather inconspicuous. No published observations on the
pollination mechanism are known to us. The pollen structure is similar
in its coarsely reticulate exine and tricolpate condition to that of a
number of other hamamelidaceous genera with complete and presuma-
bly insect-pollinated flowers, including those of Dicoryphe (Madaga-
scar), and Maingaya (Malaya). The closely related genus Mytilaria
(Indochina) hasa finer reticulum. Simpson (1936) suggests the presence
of a pore in a fossil grain he attributes to Exbucklandia, while Lee
(1969) describes and illustates (Pl. I, Fig. 6; Pl. III, Fig. 2) a pore
in the colpus of Exbucklandia.

HAMAMELIDACEOUS POLLENS

44 BOGLE AND PHILBRICK

Pate 5. Mytilaria Lecomte (Subfamily Exbucklandioideae)
VOUCHER MATERIAL: Mytilaria laosensis Lecomte; Ko 55988: Kwangsi,
China (AA),

A. Slightly oblique equatorial view showing the coarsely reticulate
exine, with deep muri borne on bacula. Lumina more or less isodia-
metric to elongate; angular in outline; varying in size, but on average
smaller and more numerous per unit area than in the closely related
Exbucklandia, x 5550. Size range in equatorial view not available.
(Bar = 10 pm.)

B. Polar view of tricolpate grains. Note the less elongate lumina
in the apocolpium as compared with Exbucklandia; margo distinct,
with associated small lumina surrounding the colpi; colpus membranes
appear to be relatively smooth in marginal areas, but finely granular
towards the center, x 5830. (Bar = 10 p.m.)

A little known monotypic genus found only in southern China and
Indochina. It is almost indistinguishable from Exbucklandia and
Chunia vegetatively, but differs strongly in floral morphology. The
flowers are complete, with nearly inferior ovaries immersed in a fleshy
spike. Ten stamens with horned filaments and hooded antlers are
connivent in one cycle over minute styles and stigmas. Grains are
oblate according to C. T. Chang (1964) and Lee (1969), or sub-oblate
to oblate-spheroidal (Erdtman, 1946). There is no information on the
pollination mechanism.

HAMAMELIDACEOUS POLLENS 45

4

Ls sce

€
6
e
wee

€

he

5 OK

& Py he
r ‘
TOSS
|
; 4 ee :

46 BOGLE AND PHILBRICK

PLATE 6. Chunia H. T. Chang (Subfamily Exbucklandioideae)
VOUCHER MATERIAL: Chunia bucklandioides H. T. Chang; C. Wang
36075: Hainan, China (AA).

A. Slightly oblique equatorial view of a tricolpate grain, showing
the microreticulate sexine with small lumina of variable size and shape,
and granular membranes of the colpi. Note also the numerous small
supratectal verrucae on the surfaces of the muri, X 2840. Shape: prolate
spheroidal. Size range in equatorial view: equatorial axis, 26-37 pm;
polar axis, 29-38 ym. (Bar = 10 pm.)

B. Slightly oblique polar view of a tricolpate grain showing the
foveolate exine of the apocolpium, with lumina of reduced size and
regular shape. Note the distinct margo and bluntly rounded ends
of the colpi, the coarse granules of the colpus membranes, and the
verrucae of the tectum and margo, x 3820. (Bar = 10 pm.)

Chunia is another poorly known monotypic genus. It is endemic
to the island of Hainan, off the southern coast of China. Although
vegetatively similar to Exbucklandia and M ytilaria, its flowers differ
by being naked and clustered in short; fleshy spikes that present
a “ball” of stamens on elongate filaments at anthesis. The genus

in the number of apertures. Grains in our sample range from 3- or
4-colpate to hexarugate (see Plate 7). C. T. Chang (1964) reports 15
per cent hexarugate grains and some polyporate grains in his material
of Chunia. In our sample 81 per cent of the grains are tricolpate,
while 19 per cent have more than three apertures.

These modifications (naked flowers; exserted anthers; increase in
aperture number: relatively smooth, foveolate exine) run parallel to
similar tendencies in other members of the family which have pro-
gressed independently toward anemophily (e.g., Liquidambaroideae;
Distylieae of the Hamamelidoideae). However, the minute styles and
stigmas of the ovaries in Chunia might seem to contradict this list
of characteristics which often accompany anemophily.

S POLLENS

U

HAMAMELIDACEO

)
7s

Fa” ' te FF, ;
aie
—ve Fe wy

of To
pA aaa

we 8

cae)

48 BOGLE AND PHILBRICK

PLaTE 7. Chunia H. T. Chang (Subfamily Exbucklandioideae)
VOUCHER MATERIAL: same as Plate 6.

A. View of hexarugate grain, showing three colpi of shortened length
with coarsely granular membranes. Note the finely reticulate exine
of the mesocolpium and the reduced size of lumina in the apocolpium,
xX 3170. Size range for hexarugate grains, 32-43 ym. (Bar = 10 pm.)

B. Tricolpate grain (left) next to hexarugate grain. Note difference
in length of the colpi, x 2000. (Bar = 10 pm.)

49

OUS POLLENS

HAMAMELIDACE

50 BOGLE AND PHILBRICK

PLATE 8. Disanthus Maxim. (Subfamily Disanthoideae)
VOUCHER MATERIAL: Dianthus cercidifolius Maxim.; Bogle 1268, from
M. Mizushima, s.n.: cultivated plant, Japan (NHA).

A. Equatorial view of tricolpate grain, showing reticulate exine,
with relatively deep muri borne on short bacula. Note the generally
uniform size and angular outline of the meshes, and the occasional
small lumina scattered among the larger ones; the border of small
lumina along margins of the colpi; the strongly sculptured exine of
the colpus membrane, X 4300. Shape: prolate spheroidal. Size range
in equatorial view: equatorial axis 22-33 pm; polar axis, 22-33 pm.
(Bar = 10 pm.)

B. Polar view showing reticulate exine with slight reduction in
size of the lumina in the apocolpium; sharply pointed ends of colpi;
operculum-like islands of deeply sculptured exine on colpus mem-
branes. The exine frequently appears tectate along the margins of
the opercula, x 3900. (Bar = 10 pm.)

A monotypic genus (and subfamily) distributed in the mountains
of Japan and central China, but widespread in cultivation. The presence
of opercula on the colpus membranes is distinctive among hamame-
lidaceous pollens. The pentamerous flowers of Disanthus are reported
to bear nectaries on the inner face of the petals (Mizushima, 1968)
which attract flies as pollinators.

HAMAMELIDACEOUS POLLENS ol

as BOGLE AND PHILBRICK

PLaTE 9. Maingaya Oliver (Subfamily Hamamelidoideae; Tribe
Hamamelideae)
VOUCHER MATERIAL: Maingaya malayana Oliv.; Burkill 7594: Penang,
Malaya (AA).

A. Equatorial view of tricolpate grain, showing the coarsely reticulate
sexine, with deep muri borne on relatively long bacula. Lumina of
the sexine are irregular in shape and angular in outline. Note the
very small lumina along the smooth margin of the colpus; occasional
coarse granules on the nexine surface under the exine reticulum;
coarsely granular central portion of the colpus membrane, X 5960.
Shape: oblate spheroidal. Size range in equatorial view: equatorial
axis, 16-23 jm; polar axis, 13-21 pm. (Bar = 10 pm.)

B. Polar view, grain slightly indented at upper right. Lumina of
the reticulum reduced slightly in size in the apocolpium. Ends of
the colpi sharply pointed, x 5910. (Bar = 10 pm.)

Maingaya is a monotypic genus known only from a very few
collections in Perak and Penang, Malaysia.

LIDACEOUS POLLENS 53

54 BOGLE AND PHILBRICK

PLare 10. Ostrearia Baill. (Subfamily Hamamelidoideae; Tribe
Hamamelideae)
VOUCHER MATERIAL: Ostrearia australiana Baill.; Brass 20266: Queens-
land, Australia (AA).

A. Equatorial view of a tricolpate grain (partially collapsed), illus-
trating the moderately coarse reticulum of the sexine. The lumina
vary widely in size and shape. The tectum is supported on short
bacula. Note the elongate colpus, the distinct margo with numerous
small perforations along the margins, and the granular colpus mem-
brane, X 5700. Shape: prolate spheroidal. Size range in equatorial
view: equatorial axis, 17-20 wm; polar axis, 18-21 wm. (Bar = 1
um.

B. Oblique polar view. Reticulation of the apocolpium not differing
significantly from that of the mesocolpium; short bacula supporting
tectum visible in upper and lower left quadrants. Note acute ends

of colpi, colpus margins and granular membranes, X 5670. (Bar =
10 wm.

A monotypic genus endemic to the rain forests of northern Queens-
land, Australia, and together with Neostrearia the only members of
the family presently known from that continent. The pollen morphol-
ogy of the two genera is rather similar, the only obvious differences
in our samples being the degree of coarseness of the reticulum, and
in the distinctness and evenness of the colpus margins.

i>
Te)

HAMAMELIDACEOUS POLLENS

56 BOGLE AND PHILBRICK

PLaTE 11. Neostrearia L. S. Smith (Subfamily Hamamelidoideae;
Tribe Hamamelideae)
VOUCHER MATERIAL: Neostrearia fleckeri L. S. Smith; Brass 2140: Moss-
man River Gorge, Cook District, Queensland, Australia (AA).

A. Oblique equatorial view of a tricolpate grain showing the irregular
and moderately coarse reticulum of the exine. Note the elongate colpi
with blunt ends; the degree of irregularity in the size and shape
of the lumina; the uneven to “cracked” colpus margins, and the granular
colpus membranes, X 5650. Shape: oblate spheroidal. Size range in
equatorial view: equatorial axis, 22-25 ym; polar axis, 20-27 pm.
(Bar = 10 pm.)

B. Slightly oblique polar view. Lumina of sexine in apocolpium
not differing significantly in size from those of the mesocolpium.
Note the margins and membranes of the colpi, X 4830. (Bar = 10
em.)

A monotypic genus endemic to rain forests of northern Queensland,
Australia. Neostrearia is apparently closely related to Ostrearia. The
pollen in our sample of Neostrearia appears to differ from that of
Ostrearia in having slightly larger lumina in the reticulum, no distinct
margo, and more uneven margins along the colpi.

t~
uD

HAMAMELIDACEOUS POLLENS

58 BOGLE AND PHILBRICK

PLaTE 12. Trichocladus Pers. (Subfamily Hamamelidoideae; Tribe
Hamamelideae)
VOUCHER MATERIAL: Trichocladus crinitus Pers.; S. C. Troughton 180:
Swaziland (NHA).

A. Equatorial view of a tricolpate grain showing elongate colpi
and the moderately coarse reticulum of the exine. Note the numerous
small, rounded lumina scattered among larger meshes of irregular
shape; the outline of the lumina is not strongly polygonal or angular,
x 6000. Shape: prolate spheroidal to sub-prolate. Size range in
equatorial view: equatorial axis, 15-22 ym; polar axis, 18-25 pm.
(Bar = 10 pm.)

B. Polar view. The lumina of the apocolpium appear, on average,
to be slightly smaller in diameter and more rounded than in the
mesocolpium. In the distal portions of the mesocolpium the lumina
appear to exhibit a tendency to become aligned in short, curving
rows. Note the concentration of small lumina along the narrow margo
of the colpi, the even edge of the margin, and the coarsely granular
colpus membranes, x 6000. (Bar = 10 pm.)

Trichocladus consists of about five species distributed in eastern
Africa, from Ethiopia southward to the Cape of Good Hope, in South
Africa. Erdtman (1946) provides a brief description of the pollen of
T. crinitus, while Bakker (1959) gives descriptions of T. ellipticus
E. and Z. and T. grandiflorus Oliver.

HAMAMELIDACEOUS POLLENS 59

60 BOGLE AND PHILBRICK

PLaTE 13. Dicoryphe Thou. (Subfamily Hamamelidoideae; Tribe
Hamamelideae)
VOUCHER MATERIAL: Dicoryphe viticoides Baker; Perrier de la Bathie
239: Ankaratin, Madagascar (P).

A. Slightly oblique view of a tricolpate grain showing the coarsely
reticulate exine with deep muri resting on relatively long bacula. The
lumina are angular, ranging in shape from regular pentagonal to mostly
elongate or curving. Note the margin of the colpus; very small lumina
or perforations occur only occasionally along the margins, X 3030.
Shape: oblate spheroidal. Size range in equatorial view: equatorial
axis, 29-34 ym; polar axis, 27-35 um. (Bar = 10 pm.)

B. Slightly oblique polar view. The lumina of the apocolpium do
not appear to differ significantly in size from those of the mesocolpium.
The colpi are slightly shortened. Note that the colpus margins (lower
middle) consist of the undulating muri of adjacent large meshes. The
projecting angles and indentations of the opposing margins appear
to be complimentary, so that the colpi are obscured when invaginated,
Xx 2820. (Bar = 10 pm.)

Dicoryphe is a genus of about 13 species endemic to the island
of Madagascar. Little is known of this interesting group apart from
the morphological descriptions of the species provided in various
taxonomic accounts. Simpson (1936) attributed certain fossil grains
in Scottish lignites to Dicoryphe on the basis of their coarse reticulum
and colpus margin characteristics. Chang (1964) compares Dicoryphe
with Exbucklandia on the basis of their coarse reticula.

62 BOGLE AND PHILBRICK

PLATE 14. Hamamelis L. (Subfamily Hamamelidoideae; Tribe
Hamamelideae)
VOUCHER MATERIAL: Hamamelis virginiana L.; Bogle 771: Interstate
Park, Polk Co., Wisconsin (NHA).

A. Equatorial view showing the moderately coarse reticulum and
muri of moderate depth resting on short bacula, as shown in upper
right of photograph. The lumina are polygonal, with pronounced
angles. Note the occasional small, rounded perforations scattered
among the polygonal meshes. The peripheral region of the colpus
membrane appears unsculptured, the central portion granular, X 5710.
Shape: prolate spheroidal. Size range in equatorial view: equatorial
axis, 15-19 ym; polar axis, 18-22 ym. (Bar = 10 pm.)

B. Slightly oblique polar view, showing the elongate colpi with
acutely pointed ends. The reticulum of the apocolpium does not differ
significantly from that of the mesocolpium. Note the narrow margo
with associated small lumina, the even edge of the margin, and the
central granular and peripheral unsculptured portions of the colpus
membrane as shown in upper left of photograph, x 6000. (Bar =
10 pm

A genus of about nine species distributed in eastern North America,
the mountains of Mexico, in Japan and China. The pollen of Hamamelis
is similar to that of Loropetalum. The flowers of the two genera
are similar in appearance and share a similar tetramerous floral plan.
The two genera were once considered congeneric. Chang (1964) notes
minor differences but basic similarity among the species of Hamamelis
he examined, with the exception of H. japonica Sieb. et Zucc., which
he reports to be larger in size and to have a thicker exine. Ikuse
(1956) also provides brief descriptions of three species of Hamamelis
(H. japonica, H. mollis, H. virginiana) in Japan. The pollen of H.
vernalis and H. virginiana is “sticky,” with “pollenkitt” deposited
on the exine surface (Hesse, 1978). Small flies are the principal
pollinators of H. virginiana.

HAMAMELIDACEOUS POLLENS 63

64 BOGLE AND PHILBRICK

PLATE 15. Tetrathyrium Benth. (Subfamily Hamamelidoideae; Tribe
Hamamelideae)
VOUCHER MATERIAL: Tetrathyrium subcordatum Benth.; Bogle 586:
Bowen Road, Victoria, Hong Kong (NHA)

A. Equatorial view of a tricolpate grain showing a slightly expanded
colpus with central zone of granular sculpturing and relatively smooth
peripheral areas. Note the coarse reticulum of the mesocolpium with
irregularly shaped, angular lumina, the relatively deep muri on short
bacula, and the distinct but somewhat uneven margins of the colpus,
with scattered small lumina along the margin and occasional breaks
in the margo, X 6350. Shape: oblate spheroidal. Size range in equatorial
view: equatorial axis, 19-26 pm; polar axis, 16-25 ym. (Bar = 10

m

B. Slightly oblique polar view illustrating the reticulum of the
apocolpium; the lumina of the reticulum not differing significantly
in size from those of the mesocolpium, x 5800. (Bar = 10 pm.)

Tetrathyrium is a monotypic genus which is endemic to Hong Kong.
It was once thought to include Loropetalum, from which it differs
in its pentamerous rather than tetramerous floral plan, in its coarse
reticulum, and deeper muri.

HAMAMELIDACEOUS POLLENS

66 BOGLE AND PHILBRICK

PLaTE 16. Loropetalum R. Br. (Subfamily Hamamelidoideae; Tribe
Hamamelideae)
VOUCHER MATERIAL: Loropetalum chinense Oliv.; Bogle 776: cultivated
plant, U.S.D.A., Plant Introduction Station, Glenn Dale, Maryland
(NHA).

A. Slightly oblique equatorial view of a tricolpate grain showing
the moderately coarse reticulum of the exine, with muri of moderate
depth resting on short bacula, and two elongate colpi. Note the very
small, round lumina scattered among the larger, angular meshes, and
along the margin of the colpi, x 6090. Shape: subprolate. Size ae
in equatorial view: equatorial axis, 16-27 ym; polar axis, 19-23 p
(Bar = 10 pm.)

B. Slightly oblique polar view, showing general reduction in size
of the lumina of the apocolpium. Note the acutely pointed ends of
the elongate colpi, and the coarsely salen membrane of the colpus
(upper right), x 6000. (Bar = 10 pm

Loropetalum is a genus of about four species (Index Kewensis;
H. T. Chang, 1973) distributed in China, Hong Kong and westward
to the mountains of eastern India (Assam). Its tetramerous flowers
are similar in appearance to those of Hamamelis.

HAMAMELIDACEOUS POLLENS 67

68 BOGLE AND PHILBRICK

PLate 17. Embolanthera Merr. (Subfamily Hamamelidoideae; Tribe
Hamamelideae)
VOUCHER MATERIAL: Embolanthera spicata Merr.; Sulit 14791: Palawan,
Philippine Islands (AA).

A. Equatorial view of a tricolpate grain, showing the coarse reticulum
of the mesocolpium. Note the deep muri borne on short bacula; lumina
are of fairly uniform dimensions and angular outline; margins of
the colpi are distinct and even, containing a few very small, scattered
lumina. The colpus membranes appear smooth peripherally, but bear
a central zone of granular exine, X 4140. Shape: prolate spheroidal.
Size range in equatorial view: equatorial axis, 18-26 ym; polar axis,
17-25 pm. (Bar = 10 pm.)

B. Slightly oblique polar view. Lumina of the apocolpium appear
slightly smaller than in the mesocolpium. Note the margins an
membranes of the colpi, X 5250. (Bar = 10 pm.)

Embolanthera is a genus of only two species, one of which is known
only from the island of Palawan, in the Philippines, the other from
a single location in North Viet Nam (Lee, 1969; Merrill, 1909;
Tardieu-Blot, 1965). The genus is considered closely related to Main-
gaya and Loropetalum.

HAMAMELIDACEOUS POLLENS

70 BOGLE AND PHILBRICK

PLaTE 18. Eustigma Gardn. and Champ. (Subfamily Hamamelidoi-
deae; Tribe Eustigmateae)
VOUCHER MATERIAL: Eustigma oblongifolium Gardn. and Champ.;
Bogle 584: Victoria Peak, Hong Kong (NHA).

A. Oblique view of an invaginated grain showing the irregular
reticulum in the mesocolpium, with lumina varying from large and
angular to very small and rounded, grading to a scrobiculate or fully
tectate condition at the poles, X 3830. Shape: subprolate. Size range
in equatorial view: equatorial axis, 39-52 um; polar axis, 45-61 pm.
(Bar = 10 pm.)

B. Oblique polar view showing the reticulate mesocolpium and
foveolate to scrobiculate apocolpium. Note the margo, rounded ends,
even to slightly ragged margins, and granular membranes of the
elongate colpi, x 3440. (Bar = 10 pm.)

A genus of two species distributed in China, Hong Kong, Taiwan
(Eustigma oblongifolium) and Vietnam (E. balansae Oliver). Although
it shares many basic similarities in floral morphology with members
of the Hamamelideae, Eustigma is sufficiently different to merit
recognition as the sole member of a separate tribe. Chang (1964)
notes the basic similarity of the grains of the two species, but with
E. oblongifolium he found that pollen grains from plants of Kwangtung,
China, exhibit a thicker exine and a coarser reticulum than those
from the island of Hainan, off the southern coast of China.

HAMAMELIDACEOUS POLLENS

gee |

Bixoide) Ut
oes Vic wane
Te CS

5 ‘e. *

*

Vee 8

aha

i 2 BOGLE AND PHILBRICK

PiaTE 19. Corylopsis Sieb. and Zucc. (Subfamily Hamamelidoideae;
Tribe Corylopsideae)
VOUCHER MATERIAL: Corylopsis platypeta Rehd. and Wils.; Bogle 963:
from a cultivated plant, Arnold Arboretum (NHA).

A. Oblique equatorial view of a tricolpate grain illustrating the
coarsely reticulate exine with rather irregular, angular lumina, some
of which are highly elongate and curved or constricted, the moderately
deep muri borne on short bacula, the margo of the elongate colpus,
and the finely granular colpus membrane, x 5000. Shape: oblate
spheroidal. Size range in equatorial view: equatorial axis, 19-24 ym;
polar axis, 17-21 pm. (Bar = 10 Lm.)

B. Oblique polar view, illustrating the reduction in size of the lumina
in the apocolpium, and the minute lumina scattered along the margo
of the sharply pointed ends of the colpi, X 5640. (Bar = 10 pm.)

Corylopsis is a large and poorly understood genus of at least 36
described species distributed in Korea, Japan, Taiwan, China, and
the Himalayas of eastern India (see Index Kewensis; H. T. Chang,
1973). In a recent morphological survey of the genus Morley and
Chao (1977) reduced the number of species to seven, while acknowl-
edging that the genus needs intensive study. Lee (1969) reports
occasional syncolpate grains in two of the six species he examined;
he also failed to find pores in the colpi as reported by Simpson (1936).
The flowers are insect-pollinated in cultivation (e.g., honey bees, blow
flies, hover flies), but observations in nature are lacking (Morley and
Chao, 1977). Hesse (1978) describes the ultrastructure of the sticky
“pollenkitt” deposited on the exine surface.

HAMAMELIDACEQOUS POLLENS

74 BOGLE AND PHILBRICK

PLaTE 20. Fortunearia Rehd. and Wils. (Subfamily Hamamelidoi-
deae; Tribe Corylopsideae)
VOUCHER MATERIAL: Fortunearia sinensis Rehd. and Wils.; Bogle 778:
from a cultivated plant, grounds of U.S.D.A., Plant Introduction
Station, Glenn Dale, Maryland (NHA).

A. Slightly oblique equatorial view of a tricolpate grain. The
reticulum is moderately coarse, with very small, rounded lumina
scattered among larger lumina which vary in outline from rounded
to triangular or polygonal. The muri are verrucate. The colpi are
relatively short, with rounded ends, indistinct margins, and finely
granular membranes. The sexine appears here to become modified
into a finely perforate and verrucate margo around the colpi, x 3800.
Shape: subprolate. Size range in equatorial view: equatorial axis, 18-34
um; polar axis, 18-40 um. (Bar = 10 pm.)

B. Oblique polar view. The reticulum of the apocolpium does not
differ significantly from that of the mesocolpium. Note the finely
perforate and verrucate margo of the colpus, x 3900. (Bar = 10 pm.)

Fortunearia is yet another monotypic genus which is endemic to
China, with a very limited distribution in western Hupeh Province.
It is vegetatively similar to Sinowilsonia, but its pollen is closer to
that of Parrotia, differing in having a finer reticulum, broader muri,
and less coarsely granular membranes. Several other genera, however,
show basic similarities in having reticulate-verrucate exines (although
with increasingly smaller lumina) and relatively indistinct colpi, rugae
or pores with granular membranes, including Molinadendron, Disty-
lium, Sycopsis and Distyliopsis.

Te)
XS

HAMAMELIDACEOUS POLLENS

v-.%,

2 ot - A
; .. -
Li “ ‘e-
a : ’ 7
"s | Y
‘ ‘ ow i vA, -
+ "e * a Ce
”s ; ry »
# . *

76 BOGLE AND PHILBRICK

PLATE 21. Sinowilsonia Hemsl. (Subfamily Hamamelidoideae; Tribe
Corylopsideae)
VOUCHER MATERIAL: Sinowilsonia henryi Hemsl.; Bogle 970: from a
cultivated plant, Royal Botanic Garden, Kew (NHA).

A. Equatorial view of the interapertural area of a tricolpate grain
with a moderately coarse reticulum. The lumina range from more
or less isodiametric to angular, and show considerable variation in
size, with numerous very small lumina occurring singly or in small
clusters among the larger meshes. Absence of small segments of murus
results in exceptionally large or erratically shaped lumina. The colpi
are elongate, approaching the poles, x 2800. Shape: prolate spheroidal.
Size range in equatorial view: equatorial axis, 24-41 wm; polar axis,
28-43 xm. (Bar = 10 um.)

B. Oblique polar view of another grain, illustrating a reduction
in the size of the lumina in the polar area. The colpi exhibit a partial
margo around their rounded ends, while the lateral margins tend to
be less distinct, with numerous small lumina and a ragged to broken
ci The colpus membranes are coarsely granular, X 2600. (Bar =
10 pm.)

Sinowilsonia is a monotypic genus which, like Fortunearia, is
endemic to Hupeh Province in central China. The two genera are
somewhat similar vegetatively, but differ in their floral morphology,
and to a certain extent in their pollen morphology. The flowers of
Sinowilsonia are functionally unisexual, with the staminate and pistil-
late flowersin separate inflorescences, while the flowers of Fortunearia
are functionally bisexual. Chang (1964) describes the exine of Sinowil-
sonia as coarsely reticulate. His photomicrographs (op. cit., Plate XII,
Figs. 1-7) illustrate a reticulum that appears slightly finer than that

HAMAMELIDACEOUS POLLENS

78 BOGLE AND PHILBRICK

PLATE 22. Sinowilsonia Hemsl. (Subfamily Hamamiledoideae; Tribe
Corylopsideae)
VOUCHER MATERIAL: Same as Plate 21.

A. Coarse reticulum and aperture margin of the grain illustrated
in Plate 21B. Note the range of size in the lumina and the absence
of supratectal verrucae, x 10,000. (Bar = 1 pm.)

B. Close-up view of the reticulum of a variant grain, the exine

sculpturing varying from reticulate to vermiform. Note the ragged
margin and granular membrane of the colpus, x 10,000. (Bar = 1
um.)
C. Scrobiculate to vermiculate reticulum of another grain; the lumina
varying from small, round perforations to much elongated and errati-
cally curving and branching, surrounded by broad expanses of tectum.
Note the erratic colpus margin and granular membrane, x 10,000.
(Bar = 1 um.)

Chang (1964) does not mention variability of this type in the exine
of Sinowilsonia, but it is so prevalent in our sample that it seemed
worthy of note. Further investigation is needed to explain the range
of variation. Among the other genera included in this study only
Molinadendron approaches Sinowilsonia in the variability of its sculp-
turing.

79

HAMAMELIDACEOUS POLLENS

& 1a |

sa: “4 a

a a i |
a , E | ‘

a wv

%@
¢

yy,

x * m “i
ins ge
* 7 >

'

La

ye “ae
ae

_*
PY
.

SA)

J Gey
35

80 BOGLE AND PHILBRICK

PLATE 23. Fothergilla Murr. (Subfamily Hamamelidoideae; Tribe
Fothergilleae)
VOUCHER MATERIAL: Fothergilla major Lodd.; Bogle 929: cultivated
plant in the author’s collection (NHA).

A. Equatorial view of a tricolpate grain with elongate colpi. The
moderately coarse reticulum contains lumina of highly variable size
and shape, with numerous small round perforations scattered among
larger lumina of erratic shape. The bacula supporting the tectum are
visible to right and left. Note the concentration of small perforations
in the margo of the colpus (upper left), the even margin and the
sparsely and finely granular to almost smooth membrane. Note also
the expanded central region (pore?) of the upper colpus membrane,
3540. Shape: subprolate. Size range in equatorial view: equatorial
axis, 22-36 um; polar axis, 25-50 um. (Bar = 10 pm.)

B. View of a hexarugate grain. Note the reduced size of the lumina
in the apocolpium, x 2900. (Bar = 10 pm.)

A genus of two (—3) species (Weaver, 1969) distributed in the coastal
plains and mountains of southeastern United States. Rugate grains
were not seen in our light microscope preparation of this species.

HAMAMELIDACEOUS POLLENS

82 BOGLE AND PHILBRICK

PLATE 24. Fothergilla Murr. (Subfamily Hamamelidoideae; Tribe
Fothergilleae)
VOUCHER MATERIAL: Fothergilla monticola Ashe; Bogle 1270: cultivated
plant, Arnold Arboretum (NHA).

Fothergilla monticola is considered either as a distinct species or
as a variant of F. major (see Weaver, 1969, for a review of the genus).
The grain figured in (A) exhibits more erratic shape in the lumina
of the mesocolpium than is seen in F. major (Plate 23, A), through
interruption of some muri, producing stubs of muri which project
into lumina space. The “bottle brush” inflorescenses of Fothergilla
are very distinctive within the family. Pollinators are mainly bees
and bumble-bees (Endress, 1977). Lee (1969) describes the grains
of both F. gardenii and F. major as “‘tricolporoidate, pore-like
appearance about 4 um in diameter.” Chang (1964) does not describe
pores in his analyses of the same species.

cael
ee A ee
i oe ‘
= ea rs +! ‘ Pe
: CP tee
bk Lo | Bicie «

OLLENS

yes oy 4%
Rae aes
ae LE oy +5

A
n
ad
~
Pa
ea)
<
Q
=
J
ea)
=
2.
<
=
a
<a
—
am

84 BOGLE AND PHILBRICK

PLaTE 25. Parrotiopsis (Niedenzu) Schneid. (Subfamily Hamameli-
doideae; Tribe F othergilleae)
VOUCHER MATERIAL: Parrotiopsis jacquemontiana (Decne.) Rehd.; Bogle
964: cultivated plant in garden of Dr. A. R. Kruckeberg, Seattle,
Washington (NHA).

A. Slightly oblique equatorial view of a tricolpate grain. The
reticulum of the exine is fine; with minute perforations scattered among
larger lumina of irregular size and shape; the width of the larger
lumina is about equal to or only slightly larger than the width of
the adjacent muri. The surfaces of the muri are very weakly verrucate.
The colpus membranes are finely and uniformly granular, x 4840.
Shape: subprolate. Size range in equatorial view: equatorial axis, 19-28
4m; polar axis, 26-35 wm. (Bar = 10 wm.)

B. Polar view. Note the reduction in size of the lumina in the
distal portion of the mesocolpium and in the apocolpium. The colpi
narrow to acutely pointed or rounded ends. A distinct but narrow
margo with associated minute perforations and slightly rough margins
borders the colpus. (Bar = 10 Lm.)

A monotypic genus restricted in distribution to the Himalayan

Fa
=]
=
)
A
YL
=
ce
~
=
my
ea)
QO
<<
Q
=
—
ea
=
a.
<
=
a
ft
—_
=

ae
a @ ee
pe ape oes
*

86 BOGLE AND PHILBRICK

PLATE 26. Parrotia C. A. Meyer (Subfamily Hamamelidoideae; Tribe
Fothergilleae)
VOUCHER MATERIAL: Parrotia persica (DC.) C. A. Mey.; Bogle 952:
cultivated plant, Morris Arboretum, Philadelphia (NHA).

A. Equatorial view of a tricolpate grain with a moderately coarse
reticulum and relatively thin exine. The lumina vary from scattered,
small, round perforations to mostly large polygonal, elongate, or
irregular meshes with intruding segments of murus. Numerous supra-
tectal verrucae are borne on the muri. The colpi are of moderate
length, with bluntly rounded ends, and indistinct, ragged to broken
margins consisting of very finely reticulate or perforated exine. The
colpus membranes are covered with very coarse, isolated to ornately
fused, verrucate granules which in some areas become partially tectate,
x 2600. Shape: oblate to subspheroidal. Size range not available.
(Bar = 10 pm.)

B. Polar view. The lumina are slightly smaller in the apocolpium
than in the mesocolpium. Note the concentration of very small lumina
along the margins of the colpi, and the coarse sculpturing of the
membranes, x 2570. (Bar = 10 p.m.)

Parrotia is a monotypic genus which is narrowly distributed in
forests on the southern shores of the Caspian Sea, in northern Iran
and southern Russia (C. T. Chang, 1964). The verrucate reticulum

The similarity with Sinowilsonia cited by Chang (1964) is not apparent
in Our material. Lee (1969) reports the grains of Parrotia to be rarely
dicolpate, tetracolpate, or syncolpate. Nair (1965) states that ‘‘a lolon-
gate endocolpium is clearly noticed in some grains.” Hesse (1978)
describes and illustrates the ultra-structure of the pollen wall in Parrotia
with regard to “pollenkitt” deposition, and the powdery form of the
pollen in regard to its anemophilous pollination mechanism.

HAMAMELIDACEOUS POLLENS

88 BOGLE AND PHILBRICK

PLATE 27. Sycopsis Oliv. (Subfamily Hamamelidoideae: Tribe
Distylieae)
VOUCHER MATERIAL: Sycopsis sinensis Oliv.; Bogle 949: cultivated plant,
University of Washington Arboretum, Seattle, Washington (NHA).

A. Entire grain. The exine is finely reticulate, the lumina mostly
rounded. Supratectal verrucae are numerous on the muri and on the

aperture membranes appear to become fused and partially tectate in
places, x 2870. Shape: spheroidal. Size range: 34-55 pm. (Bar =

B. Close-up view of the reticulum, pore margin, and heavily sculp-
tured sexine on the membrane of the pore in middle-right of A (above),
x 12,000. (Bar = 10 Lm.)

Sycopsis consists of perhaps as many as nine species (cf. Index
Kewensis; H. T. Chang, 1973), or 13 if the segregate genus Distyliopsis

89

HAMAMELIDACEOUS POLLENS

90 BOGLE AND PHILBRICK

PLaTE 28. Distyliopsis Endress (Subfamily Hamamelidoideae: Tribe
Distylieae)
VOUCHER MATERIAL: Distyliopsis dunnii (Hemsl.) Endress; A. Kairo
44090: Yamap, Morobe District, New Guinea (AA).

A. Equatorial view of a tricolpate grain. The colpi are indistinct
and of moderate length, with broadly rounded ends, little or no
definition of the margins, and membranes covered with coarse verru-
cate granules. The exine is very finely reticulate, the lumina rounded
to slightly elongate in outline, and of approximately the same width
as the adjacent muri. Supratectal verrucae are borne on the muri,
x 3800. Shape: subprolate. Size range in equatorial view: equatorial
axis, 25-36 um; polar axis, 30-38 um. (Bar = 10 um.)

B. Polar view. The lumina of the reticulum are slightly reduced
in size in the apocolpium. Note the terminal portions of the three
colpi, X 3440. (Bar = 10 1m.)

Distyliopsis consists of about four species segregated from the genus
Sycopsis (Endress, 1970). The distribution of the genus ranges from
Taiwan and southeastern China westward in the mountains to Burma,
southward in an arc through the Philippines and Sabah, possibly
in Laos (see Tardieu-Blot, 1965), to Malaya and eastern Sumatra, and
erratically eastward in Celebes and in the mountains of New Guinea
(see map in Endress, 1970). The genus is closely related to Distylium
and Sycopsis. Pollination is anemophilous. Vink (1957) and H. T.
Chang (1973) reject Distyliopsis.

HAMAMELIDACEOUS POLLENS

*
«
+

SPAT ry
‘ :

e* We
v
oe

aos
4

92 BOGLE AND PHILBRICK

PLATE 29. Distylium Sieb. and Zucc. (Subfamily Hamamelidoideae;
Tribe Distylieae)
VOUCHER MATERIAL: Distylium racemosum Sieb. and Zucc.; Bogle 740:
cultivated plant, Dr. Graham’s garden, Seattle, Washington (NHA).

broadly rounded ends, indistinct margins, and uniformly granular
membranes, X 3000. Shape: spheroidal to oblate spheroidal. Size range:

B. View of a grain with about ten apertures ranging in shape from
rugae (right) to rounded and pore-like. Reticulum of the sexine finer
than that in (A) (above), the lumina smaller and perhaps a little more
irregular in outline. Aperture membranes coarsely granular, the gran-
ules also bearing verrucae, x 3030. (Bar = 10 pm.)

A genus of about 19 described species (see Index Kewensis; H.
T. Chang, 1973) ranging from southern Japan and Korea southward
and westward through central China to Assam, and southward through
Indochina to Malaya, Sumatra, Java and eastward to Flores. The naked
flowers are probably wind-pollinated.

93

HAMAMELIDACEOUS POLLENS

Bea)

La od

on

-

*
*

*

™

_*

~

-&

vé.

94 BOGLE AND PHILBRICK

PLatE 30. Molinadendron Endress (Subfamily Hamamelidoideae;
Tribe Distylieae)
VOUCHER MATERIAL: Molinadendron sinaloense (Standley and Gentry)
Endress; Bogle 860: Sierra Suratato, Sinaloa, Mexico (NHA).

A. Equatorial view of a tricolpate grain. The reticulum is fine,
consisting of numerous minute perforations interspersed among larger,
rounded to elongate lumina. The muri are complex, often appearing
multilayered; appearing to be made up of interwoven strands, some
of which produce erect processes which project above the surface
of the muri (compare enlargement in Plate 31, B-1), giving the outer
surface a verrucate to papillate (as in this figure) appearance. The
colpi are somewhat shortened, with bluntly rounded ends, indistinct
margins, and coarsely granular membranes, X 3630. Shape: subprolate.
Size range in equatorial view: equatorial axis, 22-32 um; polar axis,
26-41 wm. (Bar = 10 pm.)

B. Slightly oblique polar view of another grain, with processes
of the exine more or less elaborated; the shape and variation of the
lumina more apparent (compare close-up of exine in Plate 31, B-2),
x 3500. (Bar = 10 um.)

Molinadendron is a Central American genus consisting of three
species distributed in the mountains of Mexico, Guatemala and
Honduras. These taxa were originally d ibed p f Distylium,
but anatomical and morphological evidence support their segregation
as a separate genus (Bogle, 1970; Endress, 1969).

“

Loins =
vie“ ny Sh
We, «

]

oe

cat

HAMAMELIDACEOUS POLLENS

95

96 BOGLE AND PHILBRICK

PLATE 31. Molinadendron Endress (Subfamily Hamamelidoideae;
Tribe Distylieae)
VOUCHER MATERIAL: same as Plate 30.

A. A third variant form found in our sample. The lumina are much
reduced and rather irregular in shape; the exine appears scrobiculate
in some areas, particularly around the colpus, finely foveolate in other
areas. The overall appearance is of a more extensive and smoother
tectum than in the other grains of this species illustrated here, X
3550. (Bar = 10 pm.)

B. Close-ups of the exine sculpturing of the grains illustrated in
Plate 30, (A) and (B). B-1: note the extensive intertwining of the
exine strands in the muri, and elaboration of the supratectal processes.
The muri often appear double (or even triple) banded in width. The
exine grades into the granular pore membrane at left, X 12,500. B-2:
less extensive development of processes on the muri, and the more
typical conditions seen in our sample. Note the interwoven, over-and-
under appearance of the strands making up the muri, and in some
areas strands lying side by side in the muri, giving the surface a

channeled appearance. Processes less numerous and less protrusive,
x 12,780. (Bar = 1 pm

HAMAMELIDACEOUS POLLENS

98 BOGLE AND PHILBRICK

PLATE 32. Matudaea Lundell (Subfamily Hamamelidoideae; Tribe
Distylieae)
VOUCHER MATERIAL: Matudaea trinervia Lundell; Bogle 848: Behucos,
Nanchititla, Mexico, Mexico (NHA).

A. View of a rugate grain with six shortened colpi visible. The
reticulum of the exine is very finely foveolate. The aperture margins
are distinct but uneven, the aperture membranes are coarsely granular,
x 3000. Shape: spheroidal to subspheroidal. Size range: 32-45 wm.
(Bar = 10 pm.)

B. Another grain with irregular apertures, ranging from shortened
colpi to pores. The irregular pores at lower-middle and upper-middle
may represent medianly constricted colpi (compare Plate 33, A and
B, lower right grain), x 2360. Size range: 32-45 um. (Bar = 10 pm.)

Matudaea, like Molinadendron, is a Central American genus. It
contains only two species distributed in mountain forests of Mexico,
Guatemala, and Honduras. The number of apertures appears to vary
from four (tetracolpate) to as many as 12-rugate. The genus is possibly
allied with Distylium and Distyliopsis. Pollination is apparently
anemophilous.

HAMAMELIDACEOUS POLLENS

aScas
ss

ak,
ar
Rg? 2?
‘‘ Pu Pl)
ees:
Ne
J
Ye ryberl
as

Roe

ra

. » e
“ag SX a
S<<*

100 BOGLE AND PHILBRICK

PLATE 33. Matudaea Lundell (Subfamily Hamamelidoideae; Tribe
Distylieae)
VOUCHER MATERIAL: same as Plate 32.

A. Close-up view of the colpus of a rugate grain, showing an apparent
early stage in the median constriction of a colpus which, when carried
to the extreme, results in the formation of pore-like apertures. Note
the foveolate exine, irregular margins and coarsely granular membranes
of the colpus, x 12,500. (Bar = 1 Lm.)

B. A group of 12 grains, showing various aperture forms and
configurations, X 670. (Bar = 10 pm.)

HAMAMELIDACEOUS POLLENS 101

102 BOGLE AND PHILBRICK

LITERATURE CITED

Bakker, E. M. van Z. 1959. South ides Pollen Grains and Spores—Part III. A.
A. Ba hema, lbs South A

Brown, R. W. 1933a. Fossil plant sa he. aspen shale of southwest Wyoming. Proc.
U.S. lage heey sae 2): 1-10, 2 pls., 2 figs.

—————- A Cretaceous Sweet-gum. Bot. Gaz. 94:611-615, 1 fig

Bocte, A. re oe Floral morphology and vascular anatomy of the Baronclidecal
The aptalous genera of Hamamelidaceae. Jour. Arnold Arb. 51:310-366.

Cuan, C, T. 1958. The morphological characteristics of some recent and fossil pollens
of pees Altingia, Sycopsis, E ahaa and huge Acta Bot. Sinica
“Zlo— say pls. ee 11. (In Chinese with R sian Summ

— e pollen morphology of Titidsenbar L. rhe mea Nor. Bot.
Zhur. 44: eee pls. 1-5. (In Russian with English Summ

—————.. 1964. The € pollen Se giatens of ri families anol tncise and Altin-
giaceae. Acta Inst. Bot. Acad. Sci. USSR. 1. Fl. Syst. Pl. Vasc. 13:173-232. pls.

Cuanc, H. T. 1962. A new genus of Hamamelidaceae in the flora of China. Sunyatsen
Univ. Bull.: Nat. Sci. 1:34-44. (Semiliquidambar H. T. Chan
. 1973. A revision of the hamamelidaceous flor of China. Sunyatsen Univ.
Bull. 1 1:54-71.
Enprksss, P. K. 1969. eRe emerge eine neue Hamamelidaceen-Gattung aus Zentrala-
merika. Bot. one 89:353-358.

Ur Se | ,ih J tultAk

————. ,19 e Infloreszenzen der
iorsholee ihe und systematische sorrentah Bot. Jahrb. Syst. t. 90:1-54.
——_—___—_— on gee voiat: Trends in the Hamamelidales-Fagales- Group. Plant
Evol Suppl. 1:321-347.
EXELL, a W. 1933. A revision of the Genus Rhodolei a. Sunyatsenia 1:95-101.
Erprman, G. 1943. An Introduction to Pollen Analysis. Cheanics Botanica Co., Waltham,
ass.
———— » 1946. Pollen morphology and plant taxonomy. VII. Notes on various
families, Svensk Bot. Tidskr. 40:77-84. (Hamamelidaceae, pp. 82,83.
————— ; 1952. Pollen Morphology and shen sbrsereta a An Introduc-
tion to Palynology 1 Chronica Botani m, Mas
pine fies DT. IVERSEN. 1975. Tete. of - esc hasan “Ed. 3. Hafner Press,
ork
HEsseg, M. 1978. Entwicklungsgeschichte und Ultrastruktur von Pollen kitt und Exine
i nahe verwandten entomophilen und anemophilen Angiospermensippen: Ran-
unculaceae, Hamamelidaceae, hte und Phage Pl. Syst. Evol. 130:13-42.
Ikusr, M. 1956. Pollen Grains of Japan. Hirokawa Publ. C Tokyo
Kuprianova, L. A. 1960. P Pulsnctoiest ¢ dus contributing to eae “ota of Liquidambar.
Pollen et Spores 2:71-88.
Ler, K. Y. 1969. Some studies on the pollen morphology of Hamamelidaceae Lindl.
S. Thesis. University of Pennsylvania. 37 pp., 3 pls.
Makarova, Z. I. 1957. On the history of the genus Liquidambar L. Bot. Zhur.

42:1182-1195.
ae oe 1909. New or Noteworthy Philippine Plants. VII. Philip. Journ. Sci.
it

igre 24, 1968. On the flower of Disanthus cercidifolius Maxim. Jour. Jap. Bot.

Morey, B. anp J. M. Cu 1977. A review of Corylopsis (Hamamelidaceae). Jour.
Arnold ne 58:382-414, pl. 1
MULLER, J. 1970. Palynologea evidence on early differentiation of angiosperms. Biol.

ev. 45:417-
2g Oe K. esl 1965. Pollen Grains of Western Himalayan Plants. Asia Publ. House,
ew D

rarer Bianki, G. B. 7. De genera piarenettecud L. notulae systematicae. Not-
t. Leningrad 18: hae (In Russian.)

HAMAMELIDACEOUS POLLENS 103

Sears, P. 1930. Common fossil pollen of the Erie basin. Bot. Gaz. 89:95-1
Simpson, J: B. 1936. Fossil Pollen in Scottish Tertiary Coals. Proc. Roy. Soc. i.
56:90-108, -
pea Hamamelidaceae, pp. 75-116, in A. cape eet and M.
Tardieu- ae Pet du Cambodge de Laos et du Vietnam, No.
a W. 1957. Hamamelidaceae. In: C. G. G. J. van Steenis, Flora ished Hee : ne 379,
Aga R. E. Jr. 1969. pei 2 the North American genus Fothergilla fiiainamelida:
ae). Jour. ae Arb. 50:599-619.
tele R; P..1935.: Po tha panic Their structure, — and significance
in science and iin McGraw Hill Book Co., New Y

Contributions from the CL S4 9
GRAY iis
HERBARIUM

1982 No. 211

MITEANE, POT ANA

MAR 4 tog

i GARDEN i545 any

P hilip D. Cantino

Reed C. Rollins

SIBARA (CRUCIFERAE

EDITED BY Otto T. Solbrig
Kathryn Rollins

or in

Sray Herbarium will be con-_
he Guiaien e

Contributions from the
GRAY
HERBARIUM

1982 No. 211

A MONOGRAPH OF THE GENUS

Philip D. Cantino PHYSOSTEGIA (LABIATAE)

SPECIES OF DRABA, LESQUERELLA, AND

Reed C. Rollins SIBARA (CRUCIFERAE)

EDITED By Otto T. Solbrig
Kathryn Rollins

PUBLISHED By

THE GRAY HERBARIUM OF HARVARD UNIVERSITY
_ISSN: 0195-6094 ISSUED January 15, 1982

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE)
Puiutie D. Cantino!

INTRODUCTION

Species delimitation in Physostegia has long been a source of con-
fusion. The genus has never been monographed, and its treatment in
floristic works varies greatly. In Gray’s Manual (Fernald, 1950), for
example, seven species and two varieties are recognized, while Glea-
son and Cronquist (1963) accept only four species and do not recog-
nize any infraspecific taxa in their manual covering approximately the
same geographic area. Two of Fernald’s species are neither accepted
nor listed in synonymy by Gleason and Cronquist, and one species
recognized by them is similarly absent from Fernald’s treatment.
Thus, even in the part of North America that is best known floris-
tically, there is substantial disagreement about the taxonomy of Phy-
sostegia.

This derives in part from the lack of any thorough investigation of
the morphological variation to be found in the genus as a whole; all
Previous studies of Physostegia have been restricted to a limited geo-
Sraphic area (e.g., Lundell, 1959, 1969; Mohlenbrock, 1963) or a sin-
gle taxonomic subgroup (Boivin, 1966). As a consequence, there has
been little recognition of the magnitude of the geographic variation
found within some of the more widespread species. Accordingly, I
have conducted an extensive survey of the morphological variation in
the genus through the examination of some 5000 herbarium specimens
from 5] institutions, supplemented by three summers of field study
of natural populations. In addition, about 400 plants from 103 popu-
lations of eight species were grown together in the experimental gar-
den, and many of these were transplanted into growth chambers
Where various environmental parameters could be manipulated. In
this way it was possible to assess the plasticity of the morphological
characters, and hence their taxonomic value.

In an effort to approach the systematics of the genus with an un-
derstanding of its biology, I have investigated the growth cycle, floral
biology, habitat requirements, breeding system, and reproductive iso-
lating mechanisms of various species. Cytological study of root tips
has resulted in chromosome counts for 10 of the 12 species of Phy-
Sostegia, many not previously reported. One product of these studies
ls the realization that the factors to which taxonomic complexity is
Most often attributed in the more difficult groups of vascular plants
are not responsible for the taxonomic problems encountered in Phy-
Sostegia. There is no evidence of apomictic seed production in the

‘Current address: Department of Botany, Ohio University, Athens, Ohio 45701.

1

ys PHILIP D. CANTINO

genus, very little evidence of natural interspecific hybridization, and
few of the species are polyploid. Rather, the taxonomic complexity
of Physostegia can be attributed to two aspects of its variation pat-
tern—great interpopulational variability and a dearth of characters
unique to a single species, the latter factor necessitating the recog-
nition of species on the basis of combinations of characters.

In the classification presented herein I have employed a species
concept that is primarily phenetic. I have relied heavily on morphol-
ogy because it is easily studied and of no less adaptive significance
than any of the other sorts of characters that could be used. When-
ever possible I have considered ecological characters, and one phys-
iological trait has proven to be of taxonomic value (i.e., photoperiodic
requirements during inf] development). Its usefulness is
noteworthy because differences in photoperiodic sensitivity have rarely
if ever been employed for taxonomic purposes (Ornduff, 1978).

Although my species concept is primarily phenetic, it is not strictly
so. I have attempted to maintain a reasonable degree of consistency
in the magnitude of phenetic difference required, and in the amount
of overlap permitted in key characters, between taxa that I have rec-
ognized as species. However, in certain borderline cases, I have ul-
timately based decisions on non-phenetic criteria, such as evidence
concerning evolutionary history or isolating mechanisms. For example,
the phenetic difference between Physostegia ledinghamii and P. vir-
giniana is sufficiently low that, when compared to the interspecific
phenetic differences elsewhere in the genus, it is unclear whether the
two taxa should be treated as species or subspecies. However, the
evidence that P. ledinghamii is a tetraploid derivative of a hybrid be-
tween P. virginiana and P. parviflora swings the balance in favor of
treating it as a distinct species (Cantino, 198la).

In the assignment of rank to infraspecific taxa, I have followed ap-
proximately the usage of Du Rietz (1930); i.e., a subspecies is a wide-
spread segment of a species, while a variety is a local variant with
a small but discrete distribution, and a form is a sporadic variant with-
out a distinct distribution, usually distinguished by a single conspic-
uous character such as flower color. Many local variants are discern-
ible within Physostegia virginiana and a lesser number within some
of the other widespread species. Some of these have been described
at the varietal level in the past and there are others that could be.
Because these variants intergrade extensively, and many of them have
likely arisen independently in more than one place, I see little pur-
pose in giving them formal recognition. I have, however, recognized
two wide-ranging subspecies of P. virginiana.

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 3

ACKNOWLEDGMENTS

I would like to express my deep appreciation for the encouragement and guidance
received throughout the course of this research from Professor Reed C. Rollins. Others
have offered helpful criticism and advice include Christopher S$. Campbell, Mi-
chael J. Donoghue, Craig W. Greene, Walter S. Judd, Norton G. Miller, Otto T.

ohn Lewis of the British Museum (Natural
History) supplied me with photographs and descriptions of type specimens too valuable
to send on loan. Bernd Heinrich, Karen S. Vinson, and Ronald J. McGinley identified
my bee collections. William F. Mahler, Vernon L. Harms, and Joan Fryxell provided
me with live plant material for cytological investigations. To all of these people I am
deeply grateful.
My field studies were made possible through the support of the Fernald Fund,
arvard University. In the course of three summers of field work, I was aided in a
variety of ways by people too numerous to name. However, I would like to extend
my particular thanks to H. R. DeSelm for his hospitality and his invaluable aid in
locating populations of Physostegia in eastern Tennessee.
I am grateful to Michael A. Canoso, Manager of Systematic Collections of the Har-
ae oY Herbaria, for his help in obtaining specimen loans; I would also like

to thank the curators and staff of the lending institutions.

GENERIC AFFINITIES

Our understanding of intergeneric relationships in the Labiatae is
heavily based on the work of Bentham (1832-36; 1848; 1876). Treat-
ments of the family in modern floras are mostly patterned after the
system of Briquet (1895-96), which does not differ greatly from Ben-
tham’s classification (El-Gazzar & Watson, 1970). In Labiatarum Gen-
era et Species (1832-36), Bentham distributed the genera among
eleven tribes but did not further subdivide the family. The Stachyd-
fae, the tribe to which Physostegia was assigned, comprised a total
o¢ 26 genera. In Bentham’s second major treatment of the family
(1848), he reduced the number of tribes to eight but recognized sub-

Brazoria, Macbridea, and Synandra. In Bentham and Hooker's Gen-
°ra Plantarum (1876), Bentham added the then recently described
Senus Chelonopsis to the subtribe Melitteae but transferred Brazoria
to a different subtribe of the Stachydeae. Briquet (1895-96), in his
treatment of the Labiatae in Engler and Prantl’s Die Natiirlichen Pflan-
*enfamilien, included within his subtribe Melittinae the same five

4 PHILIP D. CANTINO

genera that composed Bentham’s Melitteae in Genera Plantarum,
namely Physostegia, Chelonopsis, Macbridea, Synandra, and Melittis.
Following Bentham’s lead, Briquet placed Brazoria in a different sub-
tribe of the Stachydeae.

Primarily on the basis of pollen morphology, but considering other
characters as well, Erdtman (1945) suggested that Brazoria belonged
with the Melittinae. A numerical phenetic study of the Labiatae and
Verbenaceae conducted by El-Gazzar and Watson (1970), while cast-
ing doubt on the naturalness of many of Bentham’s and Briquet’s
groupings, confirmed the close relationship between Brazoria and
three of the five genera of Bentham’s subtribe Melitteae (Physostegia,
Synandra, and Melittis); the other two genera were not included in
their study. El-Gazzar and Watson suggested, however, that the Mel-
itteae should be removed from the Stachydeae, the affinities of the
subtribe lying rather with the tribes Ajugoideae and Prasieae.

In the absence of any published studies of the intergeneric rela-
tionships within the Melitteae, a preliminary morphological survey of
the group has been carried out, based primarily on the collections of
the Harvard University Herbaria but supplemented by published de-
scriptions. The latter were necessarily relied upon heavily in the case
of Synandra, of which there was but a single specimen available in
the Harvard collections at the time of the study. The survey included
the six genera that have, at one time or another, been placed in the
Melitteae. Each genus was scored for 16 characters (Tables 1 & 2),
which were selected for their diagnostic value in distinguishing Phy-
sostegia from at least one of the other five genera in the subtribe.

When the differences between the genera are totaled (Table 3), it
is evident that Physostegia and Brazoria are much more similar to
each other than either is to any of the other genera in the subtribe.
Brazoria differs from Physostegia in its annual habit and strikingly
bilabiate calyx. In addition, three of the four species of Brazoria (the
exception being B. scutellarioides) have puberulent nutlets, bearded
anthers, and pubescent stems, whereas Physostegia has glabrous nut-
lets, glabrous to sparsely pubescent anthers, and stems mostly to com-
pletely glabrous below the inflorescence.

Of the remaining four genera, Macbridea resembles Physostegia
somewhat more closely than do Synandra, Melittis, and Chelonopsis.
The foliage of Macbridea, Physostegia, and Brazoria is very similar
in appearance and markedly different from that of Melittis, Chelon-
opsis, and Synandra. The leaves of the former group are firm in tex-
ture, usually glabrous or nearly so, and at least the upper (usually
most or all of them) are sessile. In the latter group, the leaves are
membranaceous, pubescent on both surfaces (often densely so), an

bo

im)

ol

aD

x1

oe

all are petiolate. A

. Leaves

Al
- Inflorescences
a. Fl

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 5

TABLE 1. CHARACTERS THAT DISTINGUISH PHYSOSTEGIA FROM RELATED GENERA.

. Stem Pubescence

a. Glabrous or nearly so below inflo-

resce
b. agg throughout much of its
en

. Leaf Texture

a. Firm
b. Membranaceous
Conspicuously Glandular-
punctate

a. Yes

b. No
Leaf Pubescence
a. Glabrous or at most ciliate on

margins
b. Pubescent on both surfaces
|

- Petioles

lower leaves petiolate
late

i
owers borne in bracteate ra-
cemes
b. ahaa borne in 1-3 tight capit-
u

c. Flowers borne individually in axils
of upper foliage leaves
d. Flowers borne in bracteate cy-
mose clusters in axils of upper
ave

: w, not hiding calyx
b. Broad enough to hide much or all
of calyx, but not leaf-like
¢. Leaf-like, but somewhat reduced
and different shape than foliage
aves

—
S

12.

14.

15.

d. Absent (i.e., flowers borne in ax-
ils of normal foliage leaves)

. Shape of Calyx

a. Campanulate to tubular-campanu-
te with 5 equal or subequal teeth

b. Bilabiate

c. 3-lobed

d. Shape irregular, with teeth of at

least 3 different lengths
Venation of Calyx at Anthesis
a. Obscurely veined
. Conspicuously veined

. Flower Color

a. White to shades of pink and pur-
le

b. Yellowish

Anthers Bearded

a. No (although may be slightly pu-
bescent)

Yes
. Anther Coherence

a. Anthers free from one another

b. Upper anthers coherent

Nutlet Shape

a. Trigon

b. Oblanceolate
strongly compre

. Obovoid, strongly compressed
d. Subspherical, somewhat flattened
one side, convex on the other

Nutlet Pubescence

a. Glabrous

b. Puberulent

Nutlet Sculpturing

a5 th

to obovoid, not
sse

. Smoo
b. With many irregular ribs running
parallel to long axis of nutlet

c. Verrucose

Ithough in the 16 characters considered in this

study, Macbridea has a slightly greater overall resemblance to Che-

nopsis than
sid

SOstegia lies in
Confined t.
whether
Subgroy

to Physostegia or Brazoria (Table 3), geographical con-

erations suggest that it is likely to be more closely related to the

tter two. Macbridea and Brazoria are both endemic to the southern
United States, the former to the Carolinas and Florida and the latter
to Texas and Oklahoma, and the center of species diversity of Phy-

the same region. Chelonopsis, on the other hand, is
0 eastern Asia. A cladistic analysis would help to determine
Physostegia, Brazoria, and Macbridea form a monophyletic
P of the Melitteae, as I suspect to be the case.

TABLE 2. DIAGNOSTIC TABLE: PHYSOSTEGIA AND RELATED GENERA (Use

with Table 1).

er of Published
Specimens Descriptions Characters
Genus Range Examined Used Lk 2.3 4 0.8. 7 829.1 13-14 18 36
Physostegia N. Am. 5000 Aaa Diao as A a se A eat a a a eae
Brazoria N. Am. 50 Correll & BS bea wba ke 6 oe Be ck a ae ce Ba ce
Johnston, 1970
Macbridea N. Am. 11 Radford et al., 1964 Roe * 6260 8 bb eb a a ae jb ae 8
Melittis Europe 24 Ball, 1972 ‘of = So 2S: ce ee BS) ae eh ce dd Ob
Chelonopsis Asia 43 Ohwi, 1965 a ob bab bed 6 ab bo ab b ae oa bb
Synandra N. Am. 1 Fernald, 1950; Coe 8 b. 8 oh aed bh bb eb ?

Gleason, 1952;
Radford et al., 1964

ONILNY)O ‘dc dI'IlHd

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 7

TABLE 3. SUMMARY OF DIFFERENCES (Based on Table 2).

Brazoria Macbridea Melittis Chelonopsis Synandra
Physostegia 2e; 3i Te; 3i 10e; li Qc; 2i 10c; li
Brazoria X Te; 4i 8c; 5i 8c; 5i 8c; 3i
Macbridea X x 10c; 2i 6c; 4i 8c; 2i
Melittis X xX Xx 6c; 3i 6c; Oi
Chelonopsis X X X xX 4c; 2i

¢: a consistent difference between the two genera concerned.
i; an inconsistent diff i.e., a character in which the state(s) that occur in one gi ti in the other.

In view of the kaleidoscopic interspecific variation pattern of Phy-
sostegia (species being distinguishable on the basis of combinations of
characters; see p. 44), it is interesting to note that, similarly, there
is no single character that will simultaneously distinguish Physostegia
from all of the other genera of the Melitteae; every character state
that is found throughout Physostegia occurs in at least one of the
other five genera. The trait that comes the closest to being unique
to Physostegia is the shape of the calyx (character 9, Table 2), but
some species of Chelonopsis have a similar calyx morphology. How-
ever, when the venation is taken into account, it is possible to dis-
tinguish Physostegia from the rest of the Melitteae on the basis of the
calyx alone; Physostegia is the only genus that has an obscurely veined
tubular-campanulate calyx with five teeth of approximately equal length.

DISTRIBUTION AND HABITAT

Physostegia is endemic to North America (Fig. 14-17). The center
of species diversity is in southeastern Texas and extreme southwestern
Louisiana, where seven of the twelve species are found. Garden forms
of P. virginiana have become widely naturalized in areas of the east-
ern United States where the genus is not native, as well as in at least
one locality in Europe, near Turin, Italy (Tosco, 1954).

The genus occurs in a great diversity of habitats. Native populations
Tange from sea level to at least 2300 meters in elevation. Most species
°ccupy relatively moist sites and several are facultative aquatics, able
° grow in up to a foot of water. In contrast, Physostegia virginiana
may be found in limestone barrens which are very dry during the
Summer months when it is in bloom. Physostegia also appears to tol-
erate a broad range of soil acidity. Physostegia virginiana is capable
or growing on nearly bare limestone, whereas the soils that support
the pine forests frequented by P. digitalis are moderately to strongly
acidic (Campbell, 1955). The distributions and habitats of the individ-

Species are discussed in the taxonomic section.

y
=
c
=
]
Q
>
eA
=
ee
oe)

iG. 1. Rhizome morphology of Physostegia virginiana ssp. virgin-
iana ny and ssp. praemorsa (b).

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 9

MORPHOLOGY AND ANATOMY

Rhizome and Root. Two fundamentally different kinds of rhizomes
are present in Physostegia. In P. correllii, P. intermedia, P. leding-
hamii, P. leptophylla, P. longisepala, and P. virginiana ssp. virginiana,
the primary rootstock gives rise to one or more elongate, horizontal
secondary rhizomes (Fig. la), which may be simple or branched and
up to 65 cm long. A perennating bud is borne at the apex of each
horizontal rhizome (Fig. 2p). In P. angustifolia, P. digitalis, P. pul-
chella, and P. virginiana ssp. praemorsa, no horizontal rhizomes are
produced. The perennating buds are borne directly on the primary
rootstock (Fig. lb) or, if it is deeply buried, at the ends of short,
vertical secondary rhizomes (Fig. 2m). The fundamental difference
between the two is in the directionality of the secondary rhizomes
rather than the length. Although the horizontal rhizomes are usually
much longer than the vertical ones, occasional plants have horizontal
rhizomes as little as 2 cm long. The form of the rhizome is among
the most taxonomically useful characters in Physostegia. Even in the
four species in which both forms occur, there is rarely variation within
populations. The single notable exception is P. purpurea, in which it
is common to find, within a single population, plants with horizontal
thizomes and those with perennating buds borne directly on the root-
stock. Except in very young seedlings, all roots are adventitious, aris-
ing from the nodes of the primary and secondary rhizomes (Fig.
2m,p).

Stem. The stem is quadrangular and slightly swollen at the nodes;
the angles are composed of collenchymatous tissue. In most species
of Physostegia, the base of the stem is only slightly thicker than the
middle and upper sections, but in P. intermedia the base is often
Srossly enlarged and hollow. The same is very rarely true of P. lep-
tophylla and is perhaps an adaptation related to the aquatic habit of

€se two species. :

Leaf. The leaves of Physostegia are universally glabrous, with a
Prominent midrib and obscure secondary venation. Leaf shape and

entation are extremely variable. Leaf outline ranges from linear in
P. godfreyi and P. purpurea to broadly elliptical, obovate, or ovate in
P. correllii. The leaf base ranges from attenuate to rounded or auric-
ulate and the apex from attenuate to obtuse. The margins may be
sharply serrate, bluntly dentate, repand, or entire (Fig. 3). In most
‘pecies the middle and/or upper leaves clasp the stem to some degree
(Fig. 4b-d), but in P. godreyi the leaves never clasp, and they rarely

in P. virginiana. Leaf shape and dentation vary little within pop-
ulations, but P. purpurea exhibits a remarkable degree of variation
i : i the
leaf shape both within and between populations (Fig. 13), and

10 PHILIP D. CANTINO

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 11

leaf margins in a few populations of P. virginiana and in one of P.
angustifolia range from sharply serrate to entire.

The degree of reduction of the upper stem leaves is a useful tax-
onomic character. In some species (e.g., Physostegia correllii, P. par-
viflora), the uppermost leaves below the inflorescence are scarcely
smaller than those borne on the middle of the stem, while in others
(e.g., P. angustifolia, P. purpurea) the top few pairs of leaves are
greatly reduced. The degree of reduction can be expressed quanti-
tatively as a ratio of the length of the leaves of the second pair below
the terminal raceme to the length of the internode above that pair.
It is necessary to specify the terminal raceme as the reference point,
rather than the inflorescence as a whole, so that plants with many
racemes can be meaningfully compared with those bearing only a sin-
gle raceme.

Both leaf surfaces are minutely pitted, the depressions occupied by
microscopic glands of unknown function. Easily studied by means of
longitudinal leaf sections and epidermal peels, these structures consist
of a multicellular cap borne on top of a single basal cell. The cap is
composed of from 4 (rarely 2 or 3) to about 25 cells, and its diameter
ranges from 20 to 70 ». When viewed from above, the basal cell is
concealed by the larger cap, its circular outline faintly visible through
the cap if the focus is properly adjusted (Fig. 5a,b).

In Physostegia virginiana and P. angustifolia the glands are of two
distinct size classes. The smaller (20-45 p dia.) has a cap composed
of 4-8 cells, with the cell walls all situated perpendicular to the pe-

cp eae

Fic. 2. Selected morphological features of Physostegia. A—M: P.
@ngustifolia (Cantino 1057). A, upper part of flowering plant, x 0.3;
B, flower with bract, X 1.8; C, flower in longitudinal section, x 2.7;
> apex of upper lip with stamens and style, seen from below, x 4.5;
E, anther from below, x 10.8; F, anther from above, X 10.8; G,
stigmatic lobes, x 10.8; H, ovary and nectary, X 10.8; I, calyces in

it, X 1.8; J, nutlet, adaxial surface, X 5.4; K, seed, x 5.4; L, vowel
"Yo, X 5.4; M, developing winter rosette, X 0.5. N-P: P. correllii
(Cantino 1064). N, fruiting calyx, x 1.8; O, vesture of calyx, showing
stalked glands, x 10.8: P, horizontal rhizome, the terminal perennat-
ing bud starting to expand, x 0.5. Q-R: P. godfreyi (Godfrey et al.
53473), Q, adaxial surface of nutlet, xX 5.4 (note small size and ver-
Tucose sur face); R, abaxial surface of nutlet, x 5.4. The plants of P.
@ngustifolia and P. correllii were grown in an experimental garden
from thizomes collected in the same natural populations as the voucher.
Vouchers at GH.

12 PHILIP D. CANTINO

Fic. 3. Variation in leaf margin. A, Physostegia virginiana SSP.
praemorsa (Cantino 918). B, P. purpurea (Cantino 1004). C and D,
P. leptophylla (Cantino 973 and 970). E, P. purpurea (Cantino 1004).
F, P. intermedia (Cantino 1065). G, P. correllii (Cantino 1064). H and
I, P. virginiana ssp. praemorsa (Cantino 918 and 916).

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 13

< a

Fic. 4. Variation in leaf base morphology. A, Physostegia virgin-
iana ssp. virginiana (Cantino 885). B, P. angustifolia (Cantino 874).
eg virginiana ssp. praemorsa (Cantino 946). The subamplexicaulous
leaves of this plant are atypical of P. virginiana, which nearly always
has the non-clasping base illustrated in drawing A. D, P. correllii
(Cantino 1064).

rimeter of the cap as seen from above (Fig. 5a,b). The larger type
(50-70 . dia.) has a cap composed of 15-20 cells, with many of the
cell walls running parallel to the perimeter of the cap (Fig. 5c). In
P. purpurea there is no such clear distinction between size classes.
Here the cap is composed of 4-14 cells, often with a few cell walls
running parallel to the perimeter when the number of cells is 8 or
more; the diameter of the cap ranges from 20—45 yp, with no obvious
Correlation between the number of cells and the diameter of the cap.

The microscopic glands are present on all photosynthetic parts of
the plant as well as on the corolla, and Junell (1937) has observed
Similar structures on the outside of the ovule in Physostegia virgin-
‘ana. Statements in the descriptions and keys that the calyx is or is

PHILIP D. CANTINO

ma da ee — bs é

Fic. 5. Glands on leaf surface in Physostegia virginiana. A, small
gland with four-celled cap, x 1150 (Cantino 944). Note circular outline
of basal cell visible through cap. B, small gland with eight-celled cap,
x 850 (Cantino 946). C, large gland, x 700 (Cantino 877). Note that
some of the cell walls run parallel with the perimeter of the cap.
Epidermal peels were obtained from greenhouse plants, grown from
es collected from same natural population as voucher. Vouchers
a

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 15

not glandular-punctate, or that the leaf surface bears glandular dots,
refer to the comparative conspicuousness of these structures, this per-
haps being a function of the relative frequency of the larger-sized
glands. The glands have a shiny, resinous appearance which is much
more noticeable in dried material. However, even when they are par-
ticularly abundant and conspicuous, the surfaces they are borne upon
are not sticky to the touch and have no distinctive odor.

Stomata are abundant on both leaf surfaces and are slightly elevated
above the general level of the epidermis. They are amphidiacytic (ter-
minology follows Dilcher, 1974), with three to four subsidiary cells of
variable shape. Nearly the entire range of variation in both the num-
ber and shape of the subsidiary cells can be found on a single leaf.

Inflorescence. The flowers are borne in pairs (or aberrantly in
whorls of four) in racemes, the uppermost terminating the shoot and
the others situated in the axils of the upper leaves. Floral density
varies greatly in the genus and shows a high degree of species-con-
stancy.

In Physostegia virginiana ssp. praemorsa, there is commonly a row
of empty bracts below those subtending flowers (Fig. 9a). The pro-
duction of these sterile bracts is under photoperiodic control (see p.
25). The friction of the stiff bract against the pedicel and the base
of the calyx is at least partially responsible for the phenomenon to
which the common name “obedient-plant” alludes; i.e., when the
flowers are rotated to the right or left in the raceme, they remain
where they are placed. This phenomenon, termed “catalepsy, re-
ceived considerable attention in the 19th century (Ventenat, 1801;
Vilmorin-Andrieux & Co., 1866; Bailey, 1882; Coulter, 1882; Robert-
son, 1888; Meehan, 1897). Linsbauer (1940), who explored the mech-
anism in depth, concluded that it is not only the rigidity of the bract
that prevents the flower from springing back to its original position,
but also the friction between the trichomes on the bract and those
on the calyx and pedicel. He aptly compared this phenomenon to the
friction between a pair of brushes.

Flower. The flowers of Physostegia range from 1 to 4 cm in length.

though flower length is extremely variable within P. purpurea, HS
low variability in many other species (e.g., P. intermedia, P. godfreyi,
P. parviflora) makes it a taxonomically valuable character. It must be
used with caution, however, because adverse environmental condi-
tions can lead to anther abortion accompanied by r educed flower size.
The length measurements in the keys and descriptions are based ne

ied specimens; flowers on live plants average several millimeters
longer,
The pedicels may be up to
orter, the flowers appearing nearly sessi

2.5 cm long but are usually much
le. The internal anatomy of

sh

16 PHILIP D. CANTINO

the pedicel of Physostegia virginiana has been studied by Miiller
(1933), who has documented the existence of a ring of specialized
parenchymatous tissue at its base that, because of its unusual capacity
for water absorption, provides the pedicel with the flexibility neces-
sary to endure repeated rotation of the flowers in the inflorescence
(i.e., the catalepsy mentioned above).

The calyx at anthesis is tubular-campanulate to campanulate, its five
short lobes equal in length or nearly so. It becomes somewhat inflated
as the nutlets develop (Fig. 2i,n). The corolla is illustrated in Figure
2 and described on p. 57. Although certain aspects of corolla mor-
phology are quite variable in Physostegia, most characters that vary
at all tend to vary a great deal within populations and are thus of
little taxonomic value. The one character that has proven useful is
coloration. Two species, P. angustifolia and P. digitalis, have consis-
tently pale flowers, the color ranging from pure white to very pale
lavender. In contrast, P. pulchella and P. longisepala have deep lav-
ender to reddish violet corollas. Although flower color is constant or
nearly so in the above four species, it is extremely variable in P. vir-
giniana, ranging from pure white to deep lavender, occasionally even
within a single population.

The four stamens ascend along the adaxial side of the corolla tube,
the anthers lying side by side beneath the upper lip (Fig. 2b,d), or
the outer pair of stamens slightly exceeding the inner. The stamens
are epipetalous, the filaments of the inner pair becoming free of the
corolla tube near its mouth, while the outer pair arises deeper within
the tube (Fig. 2c). The filaments are densely villous, the tangle of
trichomes causing the four stamens to cohere to one another and to
hold the style in a position between them. The retention of the style
in this position is an essential part of the pollination mechanism (Coul-
ter, 1882; Cantino, 1980, pp. 97-98). The two equal to subequal an-
ther sacs of each stamen are borne parallel to or slightly divergent
from one another around a small connective (Fig. 2e). There is a scat-
tering of multicellular glandlike structures of unknown function on the
abaxial surface. The dehiscence is longitudinal, a few tiny teeth usu-
ally bordering the opening at its proximal end (Fig. 2e) and sometimes
throughout its length. Delpino (1868) noted that these teeth facilitate
the release of pollen when brushed by an insect.

The ovary is deeply cleft into four equal lobes (Fig. 2h), and the
style is gynobasic. Lying adjacent to two of the ovary lobes and sur-
passing them in height is a single yellowish nectary (Fig. 2h). The
development of the ovule, embryo sac, and seed in Physostegia vir-
giniana has been studied by Billings (1909), Sharp (1911), and Junell
(1937). The ovule is anatropous and has a single massive integument,

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 17

there being one ovule per ovary lobe. The embryo sac is unusual in
shape, composed of two expanded regions, a downward-directed mi-
cropylar lobe and an upper lobe in which the endosperm later de-
velops, with the two lobes separated by a constriction.

Fruit and Seed. Under optimal conditions, four nutlets are pro-
duced per flower. The nutlets are trigonal, the surfaces smooth in all
but one species (Fig. 2j); the nutlets of Physostegia godfreyi are ver-
rucose over part or all of their surface (Fig. 2q,r; also see Cantino,
1979, for SEM photomicrographs). The single seed inside each nutlet
is obscurely trigonal (Fig. 2k), with a membranaceous seed coat. The
mature seed is nearly filled by the ovoid-lenticular embryo, which has
only a thin layer of endosperm surrounding it.

Trichomes. The vesture of Physostegia is composed of both glan-
dular and nonglandular trichomes. It is largely confined to the inflo-
rescence, the only exception being the sparse puberulence that is
sometimes present in the nodal regions of the upper stem. The non-
glandular trichomes are structurally similar throughout the genus but
vary in length and density. They are simple, erect to slanting or some-
what curved (but never appressed), and composed of two to five cells
in a single series. To obtain photomicrographs of the trichomes (Fig.
6), pieces of raceme axis from live plants were dehydrated by means
of immersion in a series of progressively stronger acetone:water mix-
tures (50%, 70%, 90%, 95%, and 100% acetone), 30 minutes in each.
The material was then mounted on aluminum stubs with double-stick
tape, critical-point dried with carbon dioxide, sputter-coated with
gold-palladium to a thickness of 200 A, and examined with an AMR
Model 1000 scanning electron microscope.

The striking difference in the length of the trichomes of Physostegia
angustifolia and P. virginiana ssp. praemorsa (Fig. 6a,b) is the most
reliable distinction between these two widespread and morphologically
variable taxa. Although the absolute difference in length is small (in
the order of 0.1 mm), the difference is significant in multiplicative
terms; members of P. angustifolia generally bear at least a few tri-
chomes twice as long as those of P. virginiana ssp. praemorsa. Tri-
chome length is also useful in distinguishing P. digitalis, which has
the longest trichomes in the genus, from P. purpurea. While trichome
ength shows a relatively high degree of species-constancy, the density
of the vesture varies greatly within species, some rare individuals
being nearly glabrous. One such variant was formally recognized by
Fassett as P, speciosa var. glabriflora.

Stalked glands, both the cap and stalk of which are multicellular
Fig. 6c), can sometimes be found interspersed with the nonglandular
trichomes in the inflorescence. When living material is examined, a

PHILIP D. CANTINO

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 19

droplet of glandular exudate can be seen on top of the cap. The func-
tion of the liquid has not been explored. The presence or absence
of these glands is among the most taxonomically useful characters in
Physostegia, three species always producing them (P. godfreyi, P. led-
inghamii, and P. parviflora) and six species always lacking them (P.
angustifolia, P. intermedia, P. leptophylla, P. longisepala, P. pulchella,
and P. purpurea). In the other three species that may or may not
produce them, there is very rarely any variation within populations.
The distribution of stalked glands on the plant is also of taxonomic
value. Whenever they are produced, they are found on the calyx and
the raceme axis and usually on the pedicel and floral bract, but in
P. parviflora the glands are practically always present on the corolla
as well, and in P. ledinghamii and P. correllii they are occasionally
borne there; their presence on the corolla is extremely rare in the
other three species that produce the glands.

Pollen. Erdtman (1945) has suggested that there are two major
groups of genera in the Labiatae which can be defined on the basis
of whether the pollen grains are (A) tricolpate and binucleate or (B)
hexacolpate and trinucleate. The pollen of Physostegia is tricolpate
and, according to Waterman (1960), binucleate; it is thus of type (A).
Waterman published a photomicrograph of an acetylated pollen grain
of P. virginiana and reported that grains obtained from three herbar-
ium specimens were subprolate to prolate spheroidal with reticulate
sculpturing. In equatorial view, the grains were 39-62 p long and 29-
59 » wide; the polar diameter ranged from 38 to 58 pH.

The sculpturing of the pollen of Physostegia has been examined
more closely by means of scanning electron microscopy (Fig. 7). Pol-
€n from greenhouse plants and herbarium specimens was mounted
on aluminum stubs with double-stick tape and sputter-coated with
gold-palladium. The pollen was not acetylated or pretreated in any
way. There was no striking difference in the sculpturing of the exine
among the ten species examined; although a little variation could be
observed in the size of the lumina of the reticulum, the differences
Were slight. The floor of the lumina of all species was found to be
minutely pitted when examined at a high magnification (Fig. 7f).
Length and width measurements in equatorial view (Table 4), ob-

Fic. 6. Trichomes and stalked glands on the raceme axis. A, Phy-
*stegia virginiana ssp. praemorsa (Cantino 883), white bar = 100 p.
B, P. angustifolia (Cantino 1057), white bar = 100 p. C, f. godfreyi
(Cantino 1054), white bar = 10 p. Plants grown from rhizomes col-

ed in same natural population as voucher. Vouchers at GH.

20 PHILIP D. CANTINO

Fic. 7. Pollen grains of Physostegia. A, P. virginiana ssp. prae-
morsa (Cantino 882). B, P. correllii (Cantino 1137). C, P. parviflora
(Eastham s.n, DAO 16106). D. P. digitalis (Cantino 1068). E, P. god-
freyi (Cantino 1054). A-E, white bar = 10 pw. F, P. leptophylla (Can-
tino 970), white bar = 1 w. Pollen obtained from greenhouse plants
grown from rhizomes collected in same natural population as voucher.
Cantino vouchers at GH

tained from SEM photomicrographs, were found to be roughly com-
parable to those obtained by Waterman. In Table 4, vouchers in the
Cantino series do not represent the actual plant from which pollen
was removed. In each case, pollen was taken from a greenhouse plant
grown from a rhizome collected in the same natural population as the
voucher. The vouchers in the collection series of other collectors reP-

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 21

TABLE 4. POLLEN MEASUREMENTS FROM SCANNING ELECTRON PHOTOMICROGRAPHS.

Equatorial View

Taxon Voucher Length (u) Width (p)

P. angustifolia Cantino 1132 (GH) 71 46
P. correllii Cantino 1137 (GH) 70 46
P. digitalis Cantino 1068 (GH) 70 50
P. godfreyi Cantino 1054 ( 71 51
P. intermedia Correll 20840 (LL) 51 41
P. ledinghamii Frankton & Bibbey 384 (DAO) 70 48
P. leptophylla Cantino 970 (GH) 68 48
P. parviflora Eastham, s.n., 8-VIII-1947 (DAO 16106) 59 36
P. purpurea Cantino 974 (GH) 64 49
P. purpurea antino 1027 (GH) 57 35
P. virginiana ssp. praemorsa Cantino 882 (GH) 69 49
P. virginiana ssp. praemorsa Cantino 946 (GH) 65 37
P. virginiana ssp. virginiana Cantino 910 (GH) 67 42
Measurements refer to single pollen grains; listed values do not represent means

resent the actual herbarium specimens from which pollen was re-
moved

CHROMOSOME NUMBERS

The chromosome numbers of five species of Physostegia have been
reported (Taylor & Brockman, 1966; Fedorov, 1969; Cantino, 1981a,
1981b). Three were found to have 19 pairs of chromosomes and the
other two had 38 pairs. In agreement with the published reports, sev-
eral specimens of P. virginiana in the herbarium of the Canadian
Department of Agriculture (DAO) have been annotated with a diploid
number of 38 by Wray M. Bowden (Bowden Cyt. No. 3030-C1, 3175-
Cl, 3029).

Using somatic tissues, I have determined the chromosome numbers
of ten species of Physostegia. Root tips were obtained from ceil’
house plants which had been raised from rhizomes collected previ-
Cusly in natural populations. The material was pretreated in 8-hy-

Xyquinoline and stained with aceto-orcein, according to a procedure
(see Cantino, 1980 for further details) similar to that outlined by B.
4s Smith (in Radford, et al., 1974: 251-252), originally adapted from

jo and Levan 1950).

The results of ae have substantiated the earlier suggestion
°y Taylor and Brockman (1966) that the base number of age
's 19. There are two tetraploid species with 76 chromosomes, an A e
other species investigated have a diploid number of 38. The c
“0some numbers of eleven of the twelve species are listed in ne
®: the only species not examined is P. longisepala. All but two ot
“Sunts in Table 5 are my own; permanent slides have been retaine
and are available upon request. The chromosomes of selected species

TABLE 5. CHROMOSOME NUMBERS IN PHYSOSTEGIA.

PHILIP D. CANTINO

Chromosome Voucher and/
Taxon Number (2n) Location or Reference
P. angustifolia 38 Louisiana: Tangipahoa Parish, Cantino 1132
0.4 mi W of oe
38 Missouri: Newt n Co., 2 mi Cantino 1133
N of R
38 Missour te, Co. 3:5 Cantino 1134
i SW of Neosho.
38 Missing: Pearl River Co., Cantino 1135
.5 mi NW of Picayune
38 Oklahoma: McCurtai Cantino 1136
ca. 6 mi S of Smithville
P. correllii 38 Louisiana: Cameron Parish, Cantino 1137
c i E of Grand Lake
P. digitalis 38 Louisiana: Vernon Parish, Cantino 1071
near Leander.
38 eo Cass Co., N of Queen Cantino 1076
P. godfreyi 38 Florida: Ae Co., 10 mi $ Cantino 1051
of tol.
38 lag ‘Gall Co., 12.4 mi S§ Cantino 1138
of Wewahitchka.
P. intermedia 38 Texas: elon Co., just W of Cantino 1139
Ora
P. ledinghamii 76 Canada: Saskatchewan, ca. 8 V.L. Harms
of Saskato 27623 (Cantino,
198la
P. leptophylla 76 Five localities from Virginia (Cantino, 1981b)
to Florida.
P. parviflora 38 Canada: British Columbia, J. A. Calder 37028
Lower Arrow Lake. (DAO) (Taylor
& Brockman,
P. pulchella 38 Texas: Navarro Co., 1 mi NE W. F. Mahler 8520
of Richlan
38 Texas: Brazos ey College =‘. Fryxell 171
Station.
38 Texas: Brazos Co., just E of J. Fryxell 172
College Station
38 : Brazos C mi ‘J. Fryxell 173
E of College Station
P. purpurea 38 Five localities from North (Cantino, 1981b)
Des Carolina to Fl
P. virginiana 38 Location unknown (Fedorov, 1969)
(ssp. unknown)
P. virginiana ssp. 38 West Virginia: pee Co., Cantino 915
virginiana W of Hinton.
38 Illinois: “ee Oo: ca. 6 mi Cantino 1144
W o
38 meee Patt ca. 4miSE Cantino 1145
P. virginiana ssp 38 Mino: "Vermilion Co., Fi- Cantino 847
praemorsa
38 Cantino 1146

can ‘Carolina York Co., ca.
3 mi NW of Smiths.

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 23

TABLE 5. CONTINUED.

Chromosome Voucher and/
Taxon Number (2n) Location or Reference
38 North Carolina: Transylvania Cantino 1147
Co., ca. 4 mi SW of Lake
Toxaway.
38 Arkansas: Craighead Co., Jo- Cantino 1148
nesboro.

have been photographed (Fig. 8). Unless otherwise stated, all vouch-
ers listed in Table 5 have been deposited in the Gray Herbarium.

GROWTH CYCLE

All species of Physostegia are perennial. The perennating buds may
be present at the time of anthesis, but they are frequently not pro-
duced until after the blooming period has ended. Shinners’ (1956)
erroneous assertion that some species of Physostegia are annuals was
probably based on a misinterpretation of specimens collected before
the perennating buds had developed.

These buds apparently always undergo a period of dormancy before
bolting to produce the flowering shoot, but the duration of the dor-
mant stage is enormously variable, even within species. Moreover,
both vernalization and photoperiodism appear to be involved in the
induction of bolting, the precise requirements being unknown. Gen-
eralization about this aspect of the growth cycle of Physostegia is
therefore difficult. During their dormancy, the perennating buds may

located at or below the soil surface. If at the surface, and thus
exposed to sunlight, a rosette of expanded photosynthetic leaves de-
velops; if the bud is below the surface, only whitish scale-leaves are
Sgaa It is common to find buds in both positions on a single
Plant.

tee that flow-

The induction of bolting does not necessarily guaran
n at least one

ering will occur. Experimental evidence indicates that i
Species of Physostegia there are subsequent photoperiodic ee
ments for flowering (Table 6). Representatives of six species were
grown from rhizome buds in Sherer Controlled Environment Cham-
= with photoperiods of 8, 12, and 16 hours. All p
Osette stage when placed in the chambers,
had received a eee cold treatment (1-4°C) to induce bolting (14
weeks for P. angustifolia and P. virginiana; 5 weeks for P. intermedia,
P. leptophylla, and P. digitalis). Lighting was supplied by both incan-
escent bulbs and fluorescent tubes and was of similar intensity tn
all three chambers (ca. 3000 foot-candles). The temperature was gen-

24 PHILIP D. CANTINO

Fic. 8. Chromosomes of root tip cells. Magnifications are approx-
imate. Vouchers at GH. A, Physostegia godfreyi, 2n = 38, X 3250
(Cantino 1138). B, P. pulchella, 2n = 38, x 2700 (Mahler 8520): upper
arrow—chromosome lying partly outside the plane of focus; lower ar-
row—two overlapping chromosomes appearing as a single long one.
C, P. leptophylla, 2n = 76, x 1800 (Cantino 1141): arrow—two over-
lapping chromosomes appearing as one. D, P. ledinghamii, 2n = 76,
x 2450 (Harms 27623): upper arrow—two overlapping chromosomes;
lower arrows—two chromosomes lying partly outside the plane of fo-
cus.

erally maintained at 21-25°C, but greater fluctuations were occasion-
ally caused by malfunctioning of the equipment.

Physostegia angustifolia, P. intermedia, and P. leptophylla flowered
normally in all three chambers and thus appear to be day-neutral.
Physostegia virginiana, on the other hand, appears to be a long-day

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 25

species; when grown in an 8-hour or 12-hour photoperiod, most plants
grew vigorously and produced an abnormal number of leaves, but no
inflorescence was formed. They eventually grew too tall for the growth
chambers and had to be removed, some with as many as 47 nodes;
the species rarely produces more than 30 nodes in natural situations.
It is not known whether inflorescences would eventually have devel-
oped if the plants had remained in the 8-hour and 12-hour chambers,
but a longer daylength is clearly necessary if flowering is to occur at
the normal time in the life of the shoot.

The photoperiodic requirements for flowering in Physostegia virgin-
iana appear to be complex. In most plants of ssp. praemorsa and in
one plant of ssp. virginiana, a variable number of empty floral bracts
were produced below or intermixed with those subtending normal
flowers when the plants were grown in a 16-hour photoperiod from
the time of bolting to anthesis (Table 7). Normal flowering appears
to require both an interval of long photoperiod and a subsequent in-
terval of shorter daylength. Few or no empty bracts were produced
when plants were transferred to a 12-hour photoperiod after four to
eight weeks in a 16-hour photoperiod. Figure 9(a,b) illustrates the
dramatic difference in inflorescence morphology that can be produced

TABLE 6. FLOWER PRODUCTION UNDER THREE PHOTOPERIODIC REGIMES.

No. n
producing flowers

toh Sample Size
8-Hour Photoperiod
P. angustifolia 9 plants/3 populations all
P. digitalis 1 plant all
P. intermedia 2 plants/1 population all
P. leptophylla 5 plants/3 populations all
: virginiana Ssp. virginiana 27 plants/8 populations none
P. virginiana ssp. praemorsa 25 plants/9 populations 2 (8%)
12-Hour Photoperiod
P. angustifolia 9 plants/3 populations all
P. correllii 1 plant (3 shoots/1 clone) all
t, intermedia 4 plants/1 population all
+ leptophylla 10 plants/3 populations all
P. virginiana ssp. virginiana 30 plants/8 populations 4 (13.3%)
P. virginiana Ssp. praemorsa 26 plants/10 populations 6 (23.1%)
P. 16-Hour Photoperiod -
; Ongustifolia 10 plants/4 populations
4 correllii 1 plant (3 shoots/1 clone) -
2 intermedia 4 plants/1 pop . =
P leptophylla 7 plants/2 populations
4 Virginiana ssp. virginiana 28 plants/8 populations “il

‘ virginiana ssp. praemorsa 19 plants/9 populations

26 PHILIP D. CANTINO

TABLE 7. PRODUCTION OF EMPTY FLORAL BRACTS IN 16-HOUR PHOTOPERIOD.

No. of plants with

Taxon Sample Size empty bracts
P. angustifolia 10 plants/4 populations None
P. correllii 3 shoots/1 clone None
P. intermedia 4 plants/1 population None
P. leptophylla 7 plants/2 populations None
P. purpurea 2 plants/1 population None
P. virginiana ssp. virginiana 28 plants/8 populations 1 (3.6%)
P. virginiana ssp. praemorsa 19 plants/9 populations 17 (89.5%)

in members of the same clone by growing them under different re-
gimes. The shoot in Figure 9a was grown to flower in a 16-hour pho-
toperiod (14 weeks); that in Figure 9b was grown for four weeks at
16 hours, followed by three weeks of a 12-hour photoperiod and six
weeks at a daylength of about 15 hours. The taxonomic significance
of empty bract production in P. virginiana is discussed on p. 92.

The early development of the inflorescence in plants that eventually
produced a large number of empty bracts was strikingly different from
that of plants that went on to flower normally. The newly formed
racemes of the former had a squat, flat-topped appearance (Fig. 9c)
which they generally retained for four to six weeks without any ap-
parent growth, before elongating suddenly and rapidly. This contrasts
with the usual pattern of inflorescence development in Physostegia,
in which the raceme remains cylindrical at all times and elongates
steadily without interruption (Fig. 9d).

Studies of seed germination in two species of Physostegia indicate
that at least a low germination rate can be obtained without stratifi-
cation or other pretreatment. Nutlets of P. angustifolia and P. virgin-
iana, collected in natural populations six months previously, were
planted in moist vermiculite, incubated in a growth chamber at 21-
25°C, and watered daily with quarter-strength Hoagland’s Solution.
The study was continued for two months, but no seed germination
occurred after 30 days. At least a few seeds from every population
germinated, but the percentage varied widely (Table 8). Although the
seeds in this study were six months old when tested, a single attempt

ee

Fic. 9. Variation in inflorescence morphology of Physostegia vit-

giniana ssp. praemorsa in response to different photoperiods. A, »:

offshoots of same rhizome grown under different photoperiods (see

text). C, early inflorescence development under 16-hour photoperiod.
D, normal inflorescence development.

“i, aes

ee ee

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE)

28 PHILIP D. CANTINO

TABLE 8. SEED GERMINATION WITHOUT PRETREATMENT.

% germination
days

Taxon and Voucher* No. of seeds after 30

Physostegia angustifolia

Cantino 868 (Missouri) 113

Cantino 872 (Missouri) 80 10.0

Cantino 873 (Missouri) 150 0.7
Physostegia virginiana ssp. praemorsa

Cantino 847 (Illinois) Gis 1.8

Cantino 869 (Missouri) 49 22.4

Cantino 892 (Indiana) 150 60.0
Physostegia virginiana ssp. virginiana

Cantino 852 (Indiana) oo 125

ach voucher represents population where seeds were obtained, but not the actual plant that supplied them. Vouchers
are deposited in Gu.

to germinate newly produced seeds of P. angustifolia was successful.
The nutlets were collected as soon as mature and planted in vermicu-
lite as just described. Of the 25 nutlets planted, the seeds inside 9
of them germinated within three weeks and those inside 3 others
germinated within another month.

The early seedling development of Physostegia angustifolia, P. pur-
purea, and P. virginiana has been examined in the greenhouse. Ger-
mination is epigeal, the photosynthetic, leaflike cotyledons remaining
on the seedling for four to eight weeks. The longest internode on the
seedling is invariably that between the cotyledons and the first pair
of foliage leaves; it is 3-8 mm long. The succeeding internodes are
extremely short, the early foliage leaves forming a tight rosette re-
sembling that of the perennating buds. The much branched primary
root is the principal absorptive organ during the initial month, but it
is soon dwarfed by the profusion of adventitious roots that arise from
the lower nodes. The first adventitious roots develop from the coty-
ledonar node as little as three weeks after germination. Developing
rapidly from the succeeding nodes, they generally form the bulk of
the root system by the time the seedling is two months old. The pri-
mary root is eventually lost; the entire root system of the mature plant
is adventitious.

As in the perennating buds, the rosette stage of the seedlings is of
variable duration and can be shortened by vernalization.. A few un-
vernalized seedlings of Physostegia virginiana grown in the green-
house bolted as little as two months after germination, but most re-
mained in the rosette stage at least four months, and many still had
not bolted eight months after germination. The seeds used in this
study were not subjected to any cold treatment previous to germ

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 29

nation. It is unknown whether vernalization of the seeds of Physo-
stegia will substitute for vernalization of the seedling, i.e., whether a
seed that has overwintered will develop directly into a flowering shoot
without an intervening rosette stage.

Because my study of natural populations of Physostegia was con-
fined to the flowering season, the timing of germination and seedling
development in nature is a matter of conjecture. In the garden, seeds
reached maturity two to four weeks after fertilization, and newly pro-
duced seeds of P. angustifolia germinated without pretreatment in one
to four weeks. It therefore seems likely that at least some seeds of
P. angustifolia, and perhaps of other species that bloom in the spring
or early summer, germinate late in the same season they were pro-
duced. Direct evidence from natural populations is lacking, but nu-
merous seedlings of P. angustifolia and/or P. virginiana were observed
in the experimental garden in early September, 1976. Because that
was the first year Physostegia was grown in the garden, the seedlings
had to have arisen from seeds produced earlier that summer. All seed-
lings were in the rosette state. In the species that bloom in the late
summer and autumn, the seeds probably mature too late in the season
for germination to occur until the following spring.

POLLINATION BIOLOGY AND BREEDING SYSTEM

At least 29 species of insects (Table 9) visit the flowers of Physo-
stegia, and hummingbirds are occasionally observed as well. Of these,
the primary effective pollinator is the bumblebee. Several of the other
bees and wasps (i.e., Apis, Anthophora, Megachile, Polistes, Vespula)
plus the soldier beetle (Chauliognathus) are large enough to effect
Pollination on a regular basis and may be locally important in popu-
lations of Physostegia where bumblebees are infrequent. Halictid bees
frequently visit the flowers of Physostegia, but they are so much
smaller than the mouth of the corolla that they are very ineffective
pollinators, They generally enter along the lower lip of the corolla,
eed on the nectar at the base of the flower, and then depart by the
same route without ever touching the anthers or stigma. However,
they do occasionally gather pollen and may contribute in a minor way
to pollination.

Physostegia suffers a considerable loss of nectar to carpenter bees
(Xylocopa virginica). They chew holes in the bases of the corollas
through which they remove nectar without effecting pollination. Once
a hole has been cut in the corolla, smaller insects such as syrphid
flies and halictid bees use the opening as a means of withdrawing
additional nectar as it is produced. Schneck (1891), who first noted

30 PHILIP D. CANTINO

TABLE 9. INSECTS OBSERVED AT FLOWERS OF PHYSOSTEGIA.

HEMIPTERA Unidentified member of
Pentatomidae subfamily Eumeninae
aff. Cosmopepla Colletida
Cydnidae (unidentified) Hylaeus sp.
Halictida

COLEOPTERA

e
Ponttiat Agapostemon virescens

Augochlora pura

antharis sp. :
Chauliognathus pennsylvanicus Augochlorella striata

Scarabaeidae sah sigh sp .
P illi . .
gra das xiao Megachilidae
LEPIDOPTERA Megachile sp.
Papilionidae (unidentified) Anthophoridae
Hesperiidae (unidentified) Anthophora sp.
Sp ingidae Nomada sp.
Hemaris sp. Xylocopidae
PTERA Xylocopa virginica
Syrphidae (unidentified) age lif
pis mellifera
HYMENOPTERA Bombus fervidus
Formicidae (unidentified) Bombus impatiens
Ves ic Bombus cf. pennsylvanicus
Polistes fuscatus Bombus terricola
Vespula spp. Bombus vagans

this phenomenon in Physostegia, commented that honeybees prefer
to take nectar through the holes left by carpenter bees and only enter
the mouth of the corolla if no opening has been made at its base.
The flowers of Physostegia are practically odorless. Corolla color
ranges from pure white to deep reddish violet. Although it is rela-
tively invariable within most species, nearly the entire range of color
can be found within some populations of P. virginiana. The movement
of bumblebees foraging in these populations seems to be independent
of flower color, and similar behavior was noted in the experimental
garden, where many species of Physostegia were growing together.
The protandry of Physostegia virginiana (Fig. 10) has long been
known (Delpino, 1868; Foerste, 1885). Although the genus is basically
protandrous, there is a great deal of individual variation in the relative
duration of the functionally carpellate and staminate stages, the rel-
ative lengths and positions of the stamens, style, and stigma lobes
during both stages, and the rate at which the process progresses. The
latter is also heavily influenced by weather conditions, all stages pro-
ceeding more rapidly at higher temperatures. For a generalized de-
scription of the flowering stages in Physostegia, including their tim-
ing, see Cantino (1980: 97). Physostegia is self-compatible. Members
of the following taxa were self-pollinated with the aid of a pair of fine
forceps and protected from external pollen contamination by enclosure

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 31

Fic 10. Stages of protandry in Physostegia virginiana. A, function-
ally staminate stage. B, functionally carpellate stage.

of the inflorescences in cheesecloth bags: P. angustifolia (8 plants/135
flowers); P. correllii (5 plants/75 flowers); P. intermedia (5 plants/63
flowers): P. leptophylla (6 plants/82 flowers); P. purpurea (6 plants/
101 flowers); P. virginiana ssp. virginiana (11 plants/103 flowers); F,
virginiana ssp. praemorsa (17 plants/208 flowers). After pollination,
the stigma of each flower was examined to ascertain that pollen ane
fer had been effected All plants produced seed, and in each species
and subspecies the number of nutlets reaching maturity was compa-
rable to or only slightly lower than that of intraspecific crosses in-
volving the same taxa.

Observations of pollinator activity in natural populati fs
the frequency of self-pollination may be quite high. Prolific asexual
reproduction in Physostegia via rhizome offshoots results in a situation
in which neighboring inflorescences are frequently members oa 2
gle clone. Foraging bumblebees are methodical and efficient; when

ons suggest that

392 PHILIP D. CANTINO

most or all of the flowers of a particular raceme have been visited,
a bee will usually proceed to one of its nearest neighbors. Flights
between clumps are much less frequent than between racemes in the
same clump. Since the clumps are usually clones, the incidence of
self-pollination (i.e., geitonogamy) is probably very high.

There is no evidence for agamospermy in Physostegia. Studies of
P. virginiana indicate that, at least in that species, the embryo sac
develops normally from one of the meiotic products of the megaspore
mother cell (Sharp, 1911).

Anther sterility occurs sporadically in Physostegia. In some cases it
is evidently induced by environmental conditions, as demonstrated by
the growth of plants with fertile anthers from the rhizomes of pollen-
sterile individuals. More frequently, anther sterility is a stable char-
acteristic of an individual and is retained when the plant is grown
under a variety of conditions. In Ohio, Indiana, and Illinois, the fre-
quency and regularity of such apparently genetically fixed anther ste-
rility is so great in P. virginiana ssp. virginiana that it qualifies as an
example of gynodioecy. This breeding system is more common in the
Labiatae than in any other angiosperm family (Darwin, 1897), so it
is not surprising to find it in Physostegia.

Because casual observation suggested that the pollen-sterile plants
in gynodioecious populations of Physostegia virginiana tend to have
smaller flowers than the hermaphrodites, and that the separation of
the stigma lobes to expose the receptive surface occurs earlier in the
former, a quantitative study of these floral characters was undertaken.
Four Ohio populations of P. virginiana were examined in August of
1977, three of them (A, B, & C) along the St. Mary’s River in Au-
glaize and Mercer Counties, and one (D) on the banks of the Sandusky
River in Seneca County, 4 miles north of Tiffin. The relative fre-
quency of pollen-sterile shoots varied greatly among the four popu-
lations. Only 5% of the shoots in Population B were pollen-sterile,
while the percentages in Populations A, C, and D were 49%,
and 18%, respectively. The relative frequency of pollen-sterile clones
may be somewhat different than that of pollen-sterile shoots. The
clones varied greatly in size, and it was not always possible to delimit
them on the basis of the subtle differences in their morphology.

I attempted to sample from every clone at each site. Within each,
one to four shoots were randomly selected for study; the number sam-
pled was roughly proportional to the apparent size of the clone. The
flower length was measured on each shoot, and the degree of sepa
ration of the stigma lobes in the most recently opened (i.e., the UP-
permost) flowers was recorded on a scale of zero to three, zero TeP-
resenting the stage shown in Figure 10a and three representing the
stage shown in Figure 10b. Whenever either character varied among

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) $3

the flowers of a given shoot, the midpoint of the range was recorded.
Because the separation of the stigma lobes progressed through the
day, I alternated between clones of the two sexual forms whenever
possible; thus the observed difference in mean stigma lobe separation
between hermaphrodites and pollen-sterile plants cannot be attributed
to the timing of the sampling.

The results are shown in Table 10. The statistical significance was
tested using a two-tailed t-test, unless the variances differed signifi-
cantly, in which case the “approximate t-test” (Sokal & Rohlf, 1969)
was used. In all four populations, the flowers of the pollen-sterile
plants were 2-4 mm shorter than those of the hermaphrodites, and
the separation of the stigma lobes occurred sooner after the opening
of the flower in the former. These differences are highly significant.

Although the pollen-sterile flowers are smaller than those of the
hermaphrodites, there is no obvious difference in the frequency of
pollinator visits to the two sexual morphs. As suggested by Arroyo
and Raven (1975) to explain a similar situation in Fuchsia, this may

TABLE 10. VARIATION IN FLOWER LENGTH AND STIGMA LOBE SEPARATION
IN GYNODIOECIOUS POPULATIONS.

Populations
B C

FLOWER LENGT
Pollen-sterile Plants
Sample Size

3 +2 10 412
Mean (mm) 19.38 19.92 19.30 17.71
Variance 2.84 2:31 273 0.52
Hermaphrodites
va i St. 40 ll 22
Mean (mm) 23.39 23.26 21.36 22.16
Variance 2.79 2.32 0.90 4.06
Difference of Means (mm) 4.01 3.34 2.06 4.45
Statistical Significance p < .001 p < .001 p < .O1 p < .001
STIGMA LOBE SEPARATION*
Pollen-sterile Plants
Sample Size 32 12 10
Mean 1.98 2.21 2.20 1.71
Variance 0.67 0.66 0.34 0.48
Hermaphrodites
Sample Size 7 40 11 22
Mean 0.69 0.71 0.05 0.55
Variance 0.55 0.35 0.02 0.19
Difference of Means 1.29 1.50 2.15 1.16
Statistical Significance p< .00l p<.00l p< .00l p< .001

*On a scale of 0 to 3 (see text).

34 PHILIP D. CANTINO

be due to the “flag effect”; i.e., pollinators are attracted by the show-
iness of the entire plant rather than by individual flowers.

At the time of the study, no mature nutlets had been produced,
but fruit development had begun on many plants. The inception of
fruit development was tallied by counting the number of ovary-lobes
that had begun to enlarge (each flower produces four one-seeded nut-
lets corresponding to the four lobes of the ovary). The nutlet incep-
tion of the two sexual forms was compared in a percentage form based
on the maximum possible, the latter figure being equal to four times
the number of flowers that were developmentally advanced enough
when sampled so that ovary enlargement would have been visible. As
recommended by Sokal and Rohlf (1969), the percentages were con-
verted to angles by means of the arcsine transformation. A two-tailed
t-test was carried out on the angular values to determine whether the
two sexual morphs differ in their mean nutlet inception.

There was no significant difference in the nutlet inception of the
sexual morphs in the St. Mary’s River populations (A, B & C). How-
ever, in the Sandusky River population (D), the nutlet inception of
the hermaphrodites was significantly greater than that of the pollen-
sterile plants (Table 11). These results appear to be in conflict with
the expectation that the pollen-sterile plants should have a higher av-
erage seed yield than the hermaphrodites if the gynodioecious breed-
ing system is to be maintained. However, inbreeding depression,
which is thought to be a primary factor in the development and main-
tenance of gynodioecy (Valdeyron, et al., 1973; Lloyd, 1975; Arroyo
and Raven, 1975), can be expected to reduce the viability of autoga-
mously produced offspring at all stages of their development, includ-
ing the period of seed maturation. Nutlet inception was scored rela-
tively early in the development of the seed; it would not be surprising
if, due to inbreeding depression, the eventual seed yield of the her-
maphrodites were lower than that of the pollen-sterile plants, even
though the amount of nutlet initiation is not.

TABLE 11. NUTLET INCEPTION IN GYNODIOECIOUS POPULATIONS.

Populations
B

Pollen-sterile Plants

Sample Size 28 10 7 ul

Mean Nutlet Inception 49.5% 27.7% 38.0% 35.9%
Hermaphrodites

Sample Size 32 28 7 a”

Mean Nutlet Inception 50.9% 24.4% 49.2% 67.2%

Statistical Significance n.s. n.s. n.s. p< Ol

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 35

The anther sterility in gynodioecious populations of Physostegia vir-
giniana is variable in its expression. Flowers were collected from 17
pollen-sterile clones in the four Ohio populations, and the anthers
were dissected in 1% acetocarmine. The anthers from a few plants
were severely shriveled and completely empty, but variously abnormal
pollen grains could be found in the anthers of most plants. In some
of these the grains were very small and irregular in shape, while in
others they were only a little smaller than normal but did not stain
normally. This variation in the nature of the sterile anthers and their
contents suggests that the normal course of pollen development is
interrupted at different points in different plants, in turn suggesting
that the genetic control of the phenomenon may not be simple. It is
complicated still further by the existence of partial anther sterility.
Although an infrequent occurrence, I have noted a few clones in gyn-
odioecious populations of P. virginiana in which a variable amount of
normal pollen is produced. The anthers of eight flowers from one
such clone were dissected in 1% acetocarmine, and the percentage
of stainable pollen was found to vary from 30% to 75%.

DISPERSAL

The nutlets of Physostegia are dry, 2-4 mm long, and lack projec-
tions that might facilitate their dispersal by wind or animals. They
remain inside the calyx until knocked out by raindrops or by the sway-
ing of the racemes in the wind. Because of their size and weight,
they are unlikely to be carried far by normal winds. Endozoochory
is unlikely to play a role in dispersal; the thin pericarp would be easily

roken and the seed almost certainly destroyed by a mammal’s teeth
or a bird’s gizzard.

Those species that grow along rivers are probably spread by flood-
waters. The abundance of Physostegia virginiana along certain rivers,
coupled with its complete absence from other apparently similar rivers
nearby, strongly suggests that these waterways are serving as dispersal
corridors. Most of the species of riverside habitats produce horizontal
thizomes, often in great quantity, which are easily broken from the
plant and will float. os

Man has played an important part in the spread of Physostegia vir-
giniana, cultivated forms of which often escape and persist in dis-
turbed sites. The entire northeastern segment of the modern range
of this species owes its existence to naturalization following escape
from cultivation (see p. 94). In addition, the native species that fre-
quent roadsides and railroad right-of-ways (e.g., P. purpurea and P.
angustifolia) are probably dispersed by mowers and other maintenance
equipment.

36 PHILIP D. CANTINO

ISOLATING MECHANISMS

Other than the very existence of Physostegia ledinghamii, which
appears to have had a hybrid origin (Cantino, 1981la), there is little
evidence of natural interspecific hybridization in the genus. This
being the case, it is pertinent to ask how hybridization is prevented.
The following discussion is organized according to Levin's (1978) scheme
for the classification of isolating mechanisms.

In Table 12, I have summarized what appear to be the principal
factors that restrict interspecific gene flow in Physostegia. Above the
diagonal, I have recorded the degree of distributional overlap for each
pair of species. For those pairs of species that are sympatric or para-
patric, and thus could potentially hybridize, I have summarized be-
low the diagonal my present knowledge of the mechanisms by which
gene flow is restricted. Of the 66 species pairs, 40 of them are al-
lopatric. Of the 26 pairs that are not strictly allopatric, 12 of them
have parapatric ranges; i.e., their distributions border one another
(H. M. Smith, 1965). Although not parapatric by the strictest defi-
nition, I have included in this category those species whose ranges
overlap very narrowly, as well as those that fall just short of meeting.

Of the isolating mechanisms recognized by Levin (1978), it is eco-
logical and temporal isolation that principally restrict gene flow in
Physostegia. The importance of habitat difference as an isolating
mechanism may be far greater than is indicated in Table 12, where
it is recorded in only the most dramatic cases. Less obvious differ-
ences in ecological requirements may contribute greatly to the re-

TABLE 12. GEOGRAPHIC AND REPRODUCTIVE ISOLATION IN PHYSOSTEGIA.

Ang Cor. Dig God int “Led “Leo lon Par Poul. fer Vir
Ang XEN. Fo PA S oth ok AL A ee
Cx tO MER OR Ce To a ae ee Ae
Dig T H XXK A 5S hk
God ee ee Gog UA Ae ee
it Co UT. AY Ue A S A S A S
led —- —- —- — a CK A S A A P
a = | eee A A Ss A
Lon — 9 TB es 8 ae a a
i os ae ee ee ee
Pul t (Se eee > oe sae as Oe, yt S
Pur — as ? oo oo ae ca ees: 6.0.
Vir To THR oP eee ee
Ang—P. angustifolia; Cor—P. correllii; Dig—P. digitalis, God—P. godfreyi; ve ——— Led—P. ledinghamii;
RoE : — Lon—P. longisepala; Par—P. parviflora; Pul—P. pulchella —P. purpure span virginiana

Vi my

A—Allopatric; P—Parapatric; S—Sympatric; C—Cross- incompatibility demonstrated in experimental garden
requirements — exclusive in zone = sympatry; T—Blooming periods do not pairs in zone i das vente
t—Blooming periods barely overlap in zone of sy mpatry (see text).

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) $7

striction of gene flow between the members of many other pairs. It
is significant in this regard that in twenty weeks of field study con-
ducted over a wide geographic area, I have only once observed two
species of Physostegia growing together (P. digitalis and P. angustifolia
in Bowie County, Texas). I am aware of one other instance of two
species occurring at the same site—P. pulchella and P. intermedia in
Denton County, Texas (Shinners 18830, 18831, SMU).

If the co-occurrence of two or more species of Physostegia is as
infrequent as my observations suggest, then the role of other isolat-
ing mechanisms may be minimal. However, temporal isolation may
be of importance in the rare instances of two species occurring to-
gether. Of the 26 sympatric and parapatric species pairs, the members
of 16 of them have nonoverlapping or barely overlapping blooming
periods in the area of sympatry or near-sympatry. (For this purpose
I consider blooming periods to be “barely overlapping’ if the period
of overlap constitutes no more than a quarter of the blooming period
of either species.) In a number of pairs, there is some overlap if the
entire range of each species is considered, but none in the zone of
sympatry. Temporal isolation appears to be the principal mechanism
preventing hybridization between P. digitalis and P. angustifolia in
Bowie County, Texas (Cantino, 1980: 125-127).

Of the postmating isolating mechanisms recognized by Levin, only
one (cross-incompatibility) is included in Table 12. Several postmating
mechanisms are clearly not operative in Physostegia (i.e., isolation by
differing reproductive mode; hybrid floral isolation), while others have
not been investigated adequately to draw any conclusions (i.e., hybrid
inviability or weakness; hybrid sterility; hybrid breakdown).

A study of cross-compatibility among six species of P hysostegia has
been conducted in the experimental garden of the Gray Herbarium.
Plants were grown from rhizomes collected during previous summers
from 64 natural populations. At the Onset of the crossing program,
cheesecloth enclosures were constructed around each inflorescence
(see Cantino, 1980, for design). The following morning, and each suc-
ceeding morning, the newly opened flowers were emasculated. The
second and succeeding mornings, pollen was transferred from the pa-
ternal parent (its inflorescence also. enclosed in cheesecloth) to the
stigmas of those flowers emasculated the previous day. The flat side

38 PHILIP D. CANTINO

examined with a 14x hand lens before pollination to ascertain that
there were no grains already present. The grains are easily visible at
that magnification.

For each intertaxon cross, 2-13 trials were run, each utilizing a
different set of parent plants (Table 13). An effort was made to use
plants from many different populations in the trials of a given cross,
but this was not always possible; all of the individuals of Physostegia
correllii used were offshoots of a single clone, and those of P. inter-
media originated from but two populations. The other five taxa in the
crossing program were each represented by plants from 9-16 popu-
lations. The number of flowers pollinated per trial was usually 10 or
more.

TABLE 13. CONTROLLED POLLINATION EXPERIMENTS: SUMMARY OF RESULTS.

Number of Total Number % Nutlet % Seed % Loss in
Cross* Trials of Flowers Inception Yield Maturation

Ang X Ang 4 54 87.1% 81.9% 5.9%
Ang X Int 8 76 63.5% 7.9% 87.6%
Ang X Lep 5 88 62.5% 1.1% 98.2%
Ang X Pra 4 87 73.6% 55.7% 24.3%

ng X Pur 5 72 57.3% 39.2% 31.6%
Cor Selfed 5 75 2 61.7% 24.8%
Cor X Ang 7 24 78.4% 53.2% 32.1%
Cor X Pra 5 110 4 34.8% 53.1%
Cor X Pur 3 49 46.4% 34.7% 25.2%
Int X Int 4 47 77.1% 68.1% 11.7%
Lep X Lep 5 93 73.9% 54.8% 25.8%
Lep X Ang 6 68 51.1% 0.0% 100.0%
Lep x Int 7 72 63.9% 0.3% 99
Lep X Pur 6 77 68.2% 0.0% 100.0%
Lep X Vir 7 119 64.1% 18.6% 69.6%
Pra x Pra 5 96 90.9% 84.9% 6.6%
Pra X Ang 4 60 90.2% 75.0% 7.4%
Pra X Cor 4 56 65.2% 40.2% 38.3%
Pra X Int 5 78 56.1% 0.0% 100.0%
Pra x Lep 4 46 78.8% 23.4% 70.3%

ra X Pur 3 49 75.5% 55.6% 26.4%
Pra X Vir 13 254 85.2% 54.9% 35.6%
Pur X Pur 5 60 60.4% 41.3% 31.6%
Pur X Ang 5 76 74.3% 65.5% 11.8%
Pur X Int 6 68 51.1% 29.4% 42.5%
Pur x Lep 5 57 81.6% 0.0% 100.0%
Pur X Vir 4 54 71.8% 52.3% 27.1%
Vir X Vir ll 166 79.4% 63.1% 28.4%
Vir x Ang 3 46 73.4% 72.3% 1.5%
Vir x Lep 2 38 75.0% 0.0% 100.0%
Vir < Pra 6 112 75.4% 65.0% 13.8%
Vir X Pur 6 46 66 34.8% 51.2%

*First listed taxon was pollen
Abbreviations: Ang—P. angustifo “eg Cor—P. correllii; Int—P. intermedia; Lep—P. leptophylla; Pra—P. virginiana nay
praemorsa; Pur—P. purpurea; Vir—P. virginiana ssp. virginiana.

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 39

A few days after the termination of each trial, nutlet inception was
scored by counting the number of ovary lobes that had begun to en-
large. After a maturation period of from two to four weeks, depending
on species and weather conditions, the mature nutlets were collected.
No attempt was made to record the number of nutlets initiated or
matured in individual flowers; all flowers from a given trial were
lumped.

Ten randomly selected nutlets from each trial were cut open and
the seeds examined. If the seed was found to contain a large, white
embryo, it was considered to be normal. In abnormal seeds, the em-
bryo is dark and/or shriveled or missing entirely. Nutlets that were
conspicuously smaller than their siblings usually lacked normal em-
bryos, but many nutlets that were full-sized and appeared normal in
every way also lacked good embryos, hence the need for dissection.
If the ten nutlets examined were found to have apparently normal
embryos, it was assumed that all other full-sized nutlets from that trial
did so as well. (Noticeably small nutlets were always dissected.) How-
ever, if some of the ten randomly selected nutlets contained aborted
embryos, then all the nutlets from the trial were cut open and the
embryos examined. Only nutlets containing healthy-looking embryos
were counted as “seed yield.”

The raw data were used to calculate a set of percentages for nutlet
inception and seed yield (Table 13). Since there are always four ovules
per flower in Physostegia, the percentages were based on a maximum
seed yield 4 times as great as the number of flowers used. Sokal and
Rohlf (1969) recommend that when dealing with percentages, some
of which are less than 30% or greater than 70%, they be converted
to angles by means of the arcsine transformation before proceeding
with the statistical analysis. This was done, and a one-tailed t-test was
then used to test the hypothesis that the mean of the angular values
calculated for the trials of a given cross is less than the mean of the
angular values of the appropriate control. The control consisted of a
set of intraspecific crosses between plants from different populations
of the maternal taxon in the interspecific cross.

The results of the crossing program (Table 14, Fig. 11) indicate that
interspecific fertility varies widely in the genus, although two species,
Physostegia leptophylla and P. intermedia, stand out by showing a
great reduction in seed yield in most of the crosses in which they
were involved. Figure 11 leaves one with the impression that P. pur-
purea is highly fertile when used as the maternal parent in interspe-
cific crosses. However, the absolute seed yield was usually rather low
(Table 13). The seed yield in the set of control crosses (P. purpurea
x P. purpurea) was so variable that only a complete failure to produce
seed in the interspecific cross (as in P. leptophylla x P. purpurea)

40 PHILIP D. CANTINO

constituted a statistically significant drop in fertility over the control.
A reduction or absence of seed yield in interspecific crosses may
be the result of incompatibilities operating before or after fertilization.
In this study, I have assumed nutlet inception to be an indication that
fertilization has occurred, thus ignoring the unlikely possibility of par-
thenocarpy. With this assumption, it is possible to calculate the pro-
portion of the fertilized ovules that abort before reaching maturity
([number initiated — number matured]/number initiated). This param-
eter, converted to percentage form, is tabulated in the right-han
column of Table 13. The statistical significance of the difference in
this parameter between the hybridizations and the controls (right-hand
column of Table 14) provides a measure of the importance of post-
zygotic incompatibility as a cause of low seed production. In 10 of the
17 crosses in which the seed yield was significantly lower than that
of the control, the reduction in seed yield was due primarily to seed
abortion after fertilization. In 4 others, seed abortion and a reduction
in the frequency of fertilization were about equally responsible for the
lowered seed yield. In only three crosses (Ang X Pra, Pra x Cor,

TABLE 14. CONTROLLED POLLINATION EXPERIMENTS: TESTS OF SIGNIFICANCE.

% ce? n
% Nutlet Inception % Seed Yield Maturation
Cross* Control Cross < Control Cross < ese Cross > Conk trol

Ang X Int Int x Int Not signif. p < .0005 p < .0005
Ang X Lep_ Lep X Lep Not signif. p < .0005
Ang X Pra Pra X Pra = p =~ .03 Not signif.
Ang X Pur Pur X Pur Not signif. Not signif. Not signif.
Cor X Ang Ang X Ang a Recs p =~ .03 p ~ .05
Cor X Pra Pra X Pra .005 = .01
Cor X Pur Pur X Pur Not sini Not signif. Not signif.
Lep X Ang Ang X Ang p < .0005 p < .0005
Lep X Int Int x Int Ne signif, p < .0005 p < .0005
Lep X Pur Pur X Pur a — p = .05 p ~ -Ol
Lep X Vir ‘Vir X Vir 04 p < .005 = .01
Pra X Ang Ang X Ang Not sien Not signif. Not signif.
Pra X Cor Cor X Selfed p= p ~ .05 Not signif.
Pra X Int Int X Int ~ ne p < .0005
Pra X Lep_ Lep X Lep Not signif. oad
Pra X Pur Pur X Pur Not signif. Not signif. Not signif.
Pra X Vir Vir X Vir Not signif Not signif. Not signif
Pur X Ang Ang X Ang Not signif. Not signif.
Pur X Int Int X Int p =~ .04 p =~ .005 go
Pur X Lep Lep X Lep Not signif.
Pur X Vir Vir X Vir Not signif, Not signif. Not signif.
Vir X Ang Ang X Ang Not signif. Not signif. ge signif
Vir x Lep Lep X Lep Not signif. p ps
Vir X Pra Pra X Pra p =~ .04 p =~ .03 Not signif.
Vir X Pur Pur X Pur Not signif. Not signif. Not signif.

"First listed taxon was pollen parent. Abbreviations as in Table 13.

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 41

P, angustifolia

P, leptophylla

P, correllii &

7 P, virginiana

Ssp, virginiana

oe

Fic. 11. Interfertility in Physostegia. Broken line—seed yield 10%

or less of maximum possible. Thin solid line—seed yield greater than

10% but significantly less than control crosses. Heavy solid line—no

significant difference between seed yield of experimental hybridiza-

tions and control crosses. No line signifies no cross attempted. Arrows
indicate direction of pollen transfer.

and Vir x Pra) was a reduced frequency of fertilization the principal
cause of the reduced seed yield.

Seed-incompatibility (Valentine, 1954) is thus the primary form of
cross-incompatibility in Physostegia. A very common phenomenon
among angiosperms (e.g., Levin, 1978, p. 241), it results from a dis-
harmonious interaction between embryo, endosperm, and maternal
tissues and may be expressed in terms of abnormalities in the en-
dosperm, the embryo, or both (Levin, 1978). Seed-incompatibility is
Particularly frequent in hybridizations between ploidal levels, even
when the taxa involved are very closely related (Levin, 1978). It there-
fore cannot be assumed that the infertility of crosses between the tet-
raploid P. leptophylla and the diploid species, P. purpurea and P. in-
termedia, both of which show a degree of morphological overlap with
P. leptophylla, indicates a lack of close affinities between the former
and either of the latter. Morphology is a better indicator of affinities

42 PHILIP D. CANTINO

than is interfertility in taxa that differ in ploidal level.

Although not as frequent a phenomenon in Physostegia as seed-in-
compatibility, prefertilization incompatibility appears to have been op-
erating in eight of the experimental crosses (Table 14). This may only
have been an artifact, however, in that nutlet inception was used as
the sole indicator that fertilization had occurred; if seed-incompati-
bility were to cause seed abortion at a sufficiently early stage of de-
velopment, the nutlet might never enlarge enough to be counted as
fertilized. If, on the other hand, prezygotic incompatibility really does
play a role in Physostegia, it is clearly a weak barrier to hybridization,
inasmuch as at least 46% of the ovules were fertilized (i.e., nutlets
began to enlarge) in every cross, when the data from all trials of the
cross were combined (Table 13). The variation in nutlet inception be-
tween trials of a given cross was often great, and figures below 30%
were fairly frequent, but in only 3 trials out of 172 conducted (over
all crosses) were there no nutlets initiated whatsoever.

Germination tests were conducted with the seeds obtained in some
of the experimental crosses. The nutlets were soaked in gibberellic
acid (500 ppm) for 24 hours and then placed on filter paper moistened
with distilled water. The statistical procedure described earlier in
relation to seed yield was also used to test the hypothesis that seeds
from interspecific and intersubspecific crosses have a significantly
lower germination rate than those obtained in the appropriate control
crosses (i.e., the set of crosses involving plants from different popu-
lations of the maternal taxon in the experimental cross). Seeds ob-
tained from only two crosses were found to have a significantly lower
germination rate than their respective controls (Table 15).

The seeds resulting from most crosses in which Physostegia pur-
purea served as the maternal parent had a low germination rate. This
was true of the control cross (P. purpurea X P. purpurea) as well as
the interspecific crosses. Although the overall germination percentage
of the former (all trials combined) was 40%, two of the four trials
resulted in no germination. The variance of the control was therefore
so high that the total failure of germination of the seeds resulting
from the three trials of P. virginiana ssp. praemorsa X P. purpurea
was not a statistically significant reduction over the control. Whatever
factors were responsible for the generally low and extremely variable
seed yield when P. purpurea was used as the maternal parent may
also have been responsible for the similarly low and variable germi-
nation rate.

Representatives of the F, generation of the crosses listed in Table
15 were grown to anthesis. The incidence of grossly abnormal plants
among the F,s was extremely low, nearly all flowered, and the anthers

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 43

TABLE 15. GERMINATION EXPERIMENTS WITH HYBRID SEEDS™.

Number of Total Number Percent
Cross Trials of Seeds Germination

Ang X Ang 4 159 86.2%
Ang X Pra 8 118 87.3%
Ang X Pur v4 98 13.3%
Ang X Vir 3 113 85.8%
Cor X Ang 4 100 35.0%**
Cor X Pra 4 78 59.0%**
Cor X Pur 4 39 48.7%
Pra X Pra 4 81.8%
Pra X Ang 6 110 62.7%
Pra X Pur 3 89 0.0%
Pra X Vir 8 171 62.0%
Pur X Pur 4 40.0%
Pur X Vir 5 120 68.3%
Vir X Vir 9 913 79.8%
Vir X Ang 3 117 83.8
Vir X Pra 2 26 76.9%

*Each trial involved seeds obtained from a di

ifferent set of parents. **Germination rate significantly lower than that of
control (see text). Abbreviations as in Table 13.

contained pollen. However, no quantitative study of pollen viability
was undertaken.

Because temporal and/or ecological differences form an effective
barrier to hybridization between most sympatric species of Physoste-
gia, the cross-incompatibility observed in the garden is probably of
only occasional importance as an isolating mechanism in nature. It may
reduce or prevent gene flow in the event of the breakdown of the
usual premating isolating mechanism (for example, if the habitats to
which two ecologically isolated species are confined were to occur
closely enough together for an insect to carry pollen from one to the
other), and it may serve as the principal form of reproductive isolation
in a few cases in which premating isolation is lacking. The broadly
sympatric species pair, P. angustifolia and P. intermedia, offers a pos-
sible example of the latter. These two species have overlapping
blooming periods, and, although I know of no instance of the two
actually growing together, I suspect there is some overlap in their
habitat preferences as well. When the two species were crossed eXx-
perimentally, seed yield ranged from 0 to 13% in seven trials and
reached 25% in one trial. Seed abortion was the principal cause of
the low seed yield.

There are six pairs of species whose members are sympatric or para-
patric, not ecologically or temporally isolated, and have not been ex-
Perimentally tested for cross-incompatibility (indicated by “?” in the
lower portion of Table 12). It would seem that the members of these

44 PHILIP D. CANTINO

pairs should be able to hybridize, yet apparently they rarely do. Three
of the pairs are parapatric, their members probably coming into con-
tact rarely if at all. The members of one of the remaining three pairs
(P. parviflora and P. ledinghamii) differ in ploidal level. As pointed
out earlier, this is often accompanied by a large reduction in cross-
compatibility; those hybrid offspring that are produced in spite of this
reproductive barrier will be triploid and hence largely sterile.

The remaining two pairs of sympatric species that theoretically
should be able to hybridize comprise (a) Physostegia intermedia and
P. longisepala and (b) P. intermedia and P. pulchella. Because P. lon-
gisepala is apparently quite rare, contact between it and P. intermedia
is probably infrequent. However, the two species of pair “b” are
broadly sympatric in eastern Texas and have been observed growing
together at least once (Shinners 18830, 18831, sMU). Physostegia in-
termedia produced few seeds when crossed experimentally with a
number of other species, including P. angustifolia (Fig. 11). On mor-
phological grounds, I strongly suspect that P. angustifolia, P. pul-
chella, and P. longisepala are closely related. If true, it would not be
surprising to find a similarly high degree of seed-incompatibility when
either of the latter two is crossed with P. intermedia. This I suspect
to be the principal barrier preventing gene flow between P. pulchella
and P. intermedia, but proof will have to await experimental work.
An additional isolating mechanism that may be involved in restricting
gene flow between P. pulchella and P. intermedia is floral isolation.
The two differ substantially in flower size (16-30 mm in the former
vs. 9-19 mm in the latter), leaving open the possibility that ethological
isolation may be operating to some degree.

VARIATION PATTERN

The interspecific variation pattern in Physostegia is best described
as kaleidoscopic, and the species are polythetic groups. A polythetic
group (Sneath & Sokal, 1973) is one in which the members share a
large number of character states, but not every member possesses
every one of the defining traits. A kaleidoscopic variation pattern
(Cullen, 1968; Stevens, 1980a) is one in which the taxa are distin-
guishable on the basis of different combinations of a relatively few
character states, none unique to a single taxon. The interspecific vari-
ation pattern of Physostegia is not entirely kaleidoscopic, in that there
is one two-state character in which one state is unique to a single
species (the verrucose nutlets of P. godfreyi) but lacking in a few
members of that species. The extremes of some continuously varying
characters are also confined to one species. There is, however, n0
character state that is both unique to a single species of Physostegia

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 45

and reliably present in every member of that species. There is also
no character of value in distinguishing any pair of species that doesn’t
vary within some species in the genus.

A kaleidoscopic variation pattern can result from at least three sorts
of evolutionary processes: (1) reticulate evolution involving hybridiza-
tion and usually polyploidy, (2) the divergence of numerous daughter
species from a variable and now extinct ancestral species in which all
of the character states that distinguish each of the daughter species
were present, and (3) parallel and reverse evolution within a collection
of closely related species. It is unlikely that reticulate evolution has
been a major cause of the kaleidoscopic variation pattern in Physo-
stegia. Two of the twelve species are polyploids, but only one of them
is clearly of hybrid origin. With this one exception, there is very little
evidence of natural interspecific hybridization in the genus. The sec-
ond process listed above, multiple divergence from a variable ances-
tor, may have contributed to the development of the kaleidoscopic
pattern in Physostegia. There is no evidence either for or against this
hypothesis. On the other hand, there is ample evidence (discussed
below) that there has been parallelism and/or reversal in many of the
taxonomically useful characters in Physostegia (i.e., those characters
whose states are constant enough within at least a few species to be
of diagnostic value). The third process listed above is therefore likely
to have been a major cause of the kaleidoscopic variation pattern in
the genus.

The absence of the clasping leaf base which is found in most species
of Physostegia is a useful diagnostic characteristic of P. virginiana.
However, in two widely separated localities (in Ohio and western
North Carolina), a few specimens of P. virginiana have been collected
in which the leaves do clasp the stem. Introgression is unlikely to be
involved inasmuch as no other species of Physostegia occurs near either
population. Nor is it likely that one of these variants evolved directly
from the other; they lie 400 miles apart in very different habitats, and
each resembles plants from nearby populations that lack clasping
leaves much more closely than they resemble each other. This appears
to be an example of parallel and/or reverse evolution. It is not pos-
sible to conclude which of the two processes has occurred without
knowing the intraspecific phylogeny of the forms involved. There is
no shortage of other examples. Indeed, there are probably few tax-
Onomically useful characters in Physostegia in which parallel and/or

parsimony method has been carried out (see Pp. 46), i
diploid species of Physostegia were included (i. cies
which there is no evidence of a hybrid origin). The results indicate

that at least half of the character changes involved in the evolution

46 PHILIP D. CANTINO

of the diploid species from the common ancestor were parallel with,
or reversals of, other character changes.

As an example of parallel evolution at a different taxonomic level,
it is interesting to note that in Brazoria and Macbridea, the two gen-
era that appear to be most closely related to Physostegia, the species
are distinguished by many of the same characters as in Physostegia.
The variation pattern of Macbridea is of particular interest because
of the remarkable parallels that can be seen in the morphological and
ecological differences between Macbridea alba and M. caroliniana,
and between Physostegia purpurea and P. leptophylla. All four species
occur on the Southeastern coastal plain. Macbridea alba and Physo-
stegia purpurea grow in moist, open pine woods and pine savannas,
while Macbridea caroliniana and Physostegia leptophylla are found in
marshes and wooded river swamps (habitat data for Macbridea taken
from Radford, et al., 1964; Ward, 1979). On the basis of a preliminary
study of a small number of specimens of Macbridea, it appears that
at least five of the six characters that help to distinguish Physostegia
purpurea from P. leptophylla also distinguish Macbridea alba from M.
caroliniana. The sixth character, form of the rhizome, remains un-
certain because none of the immediately available specimens of M.
alba include the underground parts. The pine flatwoods species, Phy-
sostegia purpurea and Macbridea alba, differ from the corresponding
swamp species in having very much more reduced upper leaves,
fewer petiolate lower stem leaves or none at all, leaves widest above
the middle of the blade (vs. at to below the middle), leaves that tend
to be obtuse to rounded at the apex (vs. acute), and paler colored
corollas. In most of these characters, the interspecific difference is
more marked in Macbridea than in Physostegia, suggesting that what-
ever environmental factors have provided the selective pressure be-
hind this parallel evolution have been acting on Macbridea for a
longer period of time. The rarity of Macbridea, and the confinement
of its two species to limited and widely disjunct areas, are also sugges-
tive of antiquity.

INTERSPECIFIC RELATIONSHIPS

There has been much recent interest in cladistic analysis as a means
of generating phylogenetic hypotheses. The theoretical groundwork of
Hennig (1950, 1966) and Wagner (1961, 1969, 1980) has provided the
basis for a variety of techniques, many of them adaptable to the
computer.

Several algorithms have been devised by Farris (1970, 1972) for the
generation of what he has named “Wagner Trees” and “Wagner Net-
works,” in reference to the initial development of the concept by W-

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 47

H. Wagner, Jr. Farris’ “Wagner 78” program was used in this study.
Wagner Trees are a subset of a more general category known as most-
parsimonious trees; they differ from other such trees in permitting
character reversal (Farris, 1970). A most-parsimonious tree is a cla-
dogram requiring a minimum number of evolutionary steps (i.e., char-
acter changes) to derive all extant taxa from the ancestor of the group.
A Wagner Network differs from a Wagner Tree in that it lacks evo-
lutionary direction; related taxa are grouped together, but the hypo-
thetical ancestor of the group is omitted.

A particularly critical and difficult step in any cladistic analysis that
is intended to produce a rooted tree (as opposed to a network) is the
determination of the evolutionary polarity of the character state trans-
formations. In my original phylogenetic analysis of Physostegia (Can-
tino, 1980), I generated a Wagner Network using Farris’ “Wagner 78”
program and then rooted the network in the vicinity of the taxon that
had the greatest number of presumed ancestral character states. This
approach was patterned after that used by Anderson (1972) for Crusea.
In the case of Physostegia, however, the determination of character
polarity was based almost entirely on the ingroup criterion (i.e., the
comparative frequency of occurrence of the alternative states of a
character within the group under study). This “common equals prim-
itive” criterion, although widely used, is based on faulty assumptions
and can be very misleading (Stebbins, 1974; Stevens, 1980b; Wiley,
1980).

Unfortunately, the rejection of this criterion leaves no other basis
for establishing the evolutionary polarity of most characters in Phy-
sostegia. Outgroup comparison, the most widely accepted criterion for
determining ancestral condition, is of little use here. Of the five gen-
era that are considered to be most closely related to Physostegia (see
p. 3), three differ so greatly from Physostegia in their foliage, inflo-
rescence structure, and calyx morphology that most of the characters
pertaining to these structures in Physostegia have no true homologue
in the related genera. Only in Brazoria and Macbridea is the foliage
Similar enough to that of Physostegia to permit a meaningful survey
f the leaf characters that distin-

guish the species of Physostegia, and only in Brazoria is the inflo-

of inflorescence characters. Brazoria an '
tute the only practical outgroup for the determination of character
polarities, All but three of the characters used to construct the Wag-
ner Network (see below) vary within or between Brazoria and Mac-
bridea or have no homologue in either genus. In two or these three
characters, the presumed apomorphic state (i.e., the state that does
not occur in the outgroup) occurs in only one species or subspecies

48 PHILIP D. CANTINO

of Physostegia. These two characters are therefore of no use in form-
ing phylogenetic groupings. Thus outgroup comparison establishes the
evolutionary polarity of but a single useful character, clearly an in-
sufficient basis for constructing a phylogenetic tree. Because of this
inadequacy of the data, only a network of relationships will be pre-
sented here, there being at this time no way to determine which por-
tion of the network approaches the ancestral condition of the group.

Cladistic analysis assumes strictly divergent (i.e., non-reticulate)
evolution. This is a serious drawback when one is dealing with an-
giosperms, a group in which a minimum of 30-35% of the species
are polyploids (Stebbins, 1971) and many if not most of these are of
hybrid origin (Grant, 1971). The best way to deal with this (Wagner,
1969, 1980) is to omit species from the data set used to construct the
tree or network if they appear likely to be of hybrid origin. They
may later be placed in the phylogenetic diagram with connections to
both putative parents. This procedure was followed in the phyloge-
netic analysis of Physostegia. The two tetraploid species, P. leding-
hamii and P. leptophylla, were omitted because of the likelihood that
they are of hybrid origin. The evidence in support of this premise
is far stronger in the case of the former, which almost certainly orig-
inated in a hybridization between P. parviflora and P. virginiana
(Cantino, 198la). The origin of P. leptophylla is uncertain. It is pos-
sible that it is an autotetraploid whose diploid ancestor is extinct or
undiscovered, but autopolyploid species are thought to be rare among
vascular plants (Grant, 1971); it is more likely that the ancestor of P.
leptophylla was a hybrid. Morphology and distribution point to P.
virginiana and P. purpurea as the most probable parents of such a
hybrid (Cantino, 1980: 253-256).

Of the 32 morphological characters that are of taxonomic value in
Physostegia, 16 of them are quantitative characters with more than
two alternative states. They present a problem in Physostegia because
of their frequently great intraspecific variability and the resulting in-
terspecific overlap of the ranges of their character states. The proD-
lem is compounded because the standard statistical parameters, mean
and standard deviation, cannot be used because the data were not
collected in an unbiased manner (Cantino, 1980, p. 204). The contin-
uously varying characters that could not reasonably be coded in a two-
state form were therefore omitted from the analysis. The resultant loss
of information is not as great as it might seem; most of the quanti-
tative characters are of use in distinguishing but a few pairs of species
or subspecies, the overlap between all other pairs of taxa being too
great for the character to be of use. Characters of this kind, although
sometimes useful in a key, are not likely to be significant indicators
of phylogenetic relationship.

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 49

Of the 16 remaining characters which could be coded in a two-state
form, 15 of them (Table 16) were used as the data base for the gen-
eration of an undirected network. The 16th character was omitted
because it was not sufficiently independent of two other characters
describing the same attribute (leaf shape). Even the coding of the
two-state characters was problematical, because for every character
there is at least one species or subspecies in which both states can
be found. It is possible to circumvent this problem if one regards
characters as being represented in taxa not by single states, of which
a given taxon is capable of producing only one of the two alternatives,
but rather by phenotypic tendencies. For any given character, some
species will have a tendency closely approaching 100% for the pos-
session of only one possible state, but others will have various tenden-
cies to possess either character state. The tendency for a particular
state can be roughly gauged by determining the proportion of spec-
imens in which that state is present.

The use of character state frequency as an indicator of tendency
within a taxon effectively converts a two-state character to a contin-
uous character whose extremes are zero and unity. If, for example,
80% of the specimens of a given species possess state A of a particular
character, while the other 20% possess state B, and if state A is ar-
bitrarily assigned the value 1.0 and state B is given the contrasting
value 0.0, then the species is scored with the value 0.8 for that char-
acter. The percentages were rounded to the nearest 10%. Thus the
characters, as scored, are neither two-state nor continuously varying,
but have 11 states ranging from 0.0 to 1.0 by intervals of 0.1.

TABLE 16. CHARACTERS USED IN CLADISTIC ANALYSIS.

Character 1, One or more leaves widest below middle of blade: , no; 1, yes.

2. At least one leaf widest near base of blade: 0, no; 1, yes.
3. Majority of larger leaves bluntly toothed to entire, 0; sharply serrate, 1..
4. One or more leaves clasp stem: 0, yes; 4, nO. —

5. One or more petiolate leaves present at anthesis: 0, no; 1, yes.

6. Empty bracts produced below flowers: 0, no; 1, yes.

7. Bas tem conspicuously swollen: 0, no; l, yes. os
8. Flowers crowded ‘acent calyces overlapping half or more of their

lengths: 0, yes; 1, no.
. Horizontal rhizomes produced, 0; all rhizomes vertical, 1.
. Stalked glands present on ee ee - no.
Stalked glands present on corolla: ©, no; l, yes.
Glandular dots calyx inconspicuous, few, or absent, 0; abundant and
Ghee tt te. 0; some cuspidate, 1
. te to attenuate, Y; reds
: | ioe as weak primary veins (besides midrib) at base of
blade: 0, no; 1, yes.
15. Nutlet surface smooth, 0; verrucose, 1.

te
NES ~o

a
wm OO
©
=]
io]
°
=
3
°

50 PHILIP D. CANTINO

A
PAR
DIG
COR
PUR LON
aie f_anae
INT PULO PRA
VIR

B

F IG. 12. Undirected networks for Physostegia. A, network based on
entire 15-character data set. B, network based on data set lacking
character no. 3. Taxon abbreviations as in Table 17.

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 5]

The Wagner Network shown in Figure 12A was based on the data
set in Table 17. The lengths of the branches are proportional to pa-
tristic distance, as defined by Farris (1967), i.e., the sum over all
characters of the change from point to point on the phyletic line. To
what extent are the relationships portrayed an accurate representation
of the true situation? One criterion that can be used to evaluate the
reliability of a network is its stability when characters are removed

om the data set used to generate it. If the most parsimonious network
produced when a single character is omitted is drastically different
from that produced using all characters, it is probable that the inclu-
sion of a new character, as additional data are collected, will also re-
sult in a different network.

As a simple test of the stability of the most parsimonious network,
15 additional networks were generated on the basis of data sets from
which a single character had been omitted, each of the 15 characters
being omitted once. Some of these networks were little different from
that based on all characters, but others differed to various degrees.
The network based on a data set lacking character 3 was among the
most divergent (Fig. 12B). The degree to which the most parsimoni-
ous network can be altered by the omission of but a single character
casts serious doubt on its reliability as an indicator of actual relation-
ships. However, there are a number of elements in it that remain
unchanged in most or all of the alternative networks generated through

a *
TABLE 17. CHARACTER STATE DISTRIBUTION IN PHYSOSTEGIA
Taxa**

Miincter. ané con ie Oe ot ee a

1 Os 6. 0 66 10 8S 10 00 06 GO OF
- 0.2 a0 0 : 60 on 06 10 00 05 GO 09
3 ie 10 oo oO ee ee ot 20 te eo
4 Os BO 66 fe ee oe Oe 2D Oe oe te
5 bs. oe on Be be ee oe ee Se ee CS
6 Oo: 06 600 ee on Oe OO Oe 08 oO Ce
7 60.60. an oe Ge ee oe, 00 oe oe. OP
8 re ae a 03
9 10-06 46 (0m on ee 0 10 10 OF 0.0
10 inh ae ae on ke 1 oo ee. Lt 0 5
oe O60 aa ho oa ee ee Oe. OO 0.0
oe bc on Ge 68 ae Oo ee ee 8S 0.0
13 Ok 06 Lo 60 60 68° GO 02 08 0.0 02
14 Oo: 16 62. ee. oo 00. Oe Of 88 0.0 0.0
15 ah an be pe oe. on 80. Oe. 00. O

- leaaageagees gaps alis, GOD, P. godfreyi; INT, P. intermedia;

LON, P. longisepala; PAR, P. parviflora; PRA, P. virginiana ssp
P. virginiana ssp. virginiana.

** Abbreviations are: ANG, P. angustifolia, COR, P. correllii, DIG, P. - Se adobe oh S ac a

52 PHILIP D. CANTINO

the removal of characters. The two subspecies of Physostegia virgin-
iana, not surprisingly, are grouped together in all 15 networks, and
in 14 they terminate a line. In every network, P. purpurea and P.
godfreyi are placed as nearest neighbors, and in 13 of them they ter-
minate a line. In 11 networks, P. purpurea, P. godfreyi, P. intermedia,
and P. longisepala are placed together in the cladistic configuration
shown in Figure 12A, and in 9 of them, the cladistic relationship of
P. pulchella to these four species is also as shown in Figure 12A. In
9 of the 15 networks, P. parviflora and P. correllii are nearest neigh-
bors and terminate a line. There is less agreement as to the placement
of P. digitalis and P. angustifolia. Physostegia digitalis is most often
associated with P. correllii, but not necessarily in the configuration
shown in Figure 12A. Physostegia angustifolia is always placed in a
relatively central position, but its nearest neighbors vary, P. pulchella
and P. digitalis being the most frequent ones.

In summary, this analysis suggests that the diploid species and sub-
species of Physostegia fall into three groups: one consists of the two
subspecies of P. virginiana; the second includes P. parviflora, P. cor-
rellii, and probably P. digitalis; and the third includes P. purpurea,
P. godfreyi, P. intermedia, P. longisepala, and P. pulchella. The affin-
ities of P. angustifolia within this scheme are uncertain.

A final point that can be made about the evolution of Physostegia,
on the basis of the most parsimonious network, is that parallelism an
character reversal have been extremely common in the genus. There
is no guarantee, of course, that evolution has followed the most par-
simonious pathway that would explain the distribution of character
states over taxa, but even the most parsimonious network that could
be devised by the Farris algorithm on the basis of the full 15-character
data set required 52% of the character changes to be parallel with,
or reversals of, other changes. This provides a measure of the mini-
mum amount of homoplasy that has occurred during the phylogeny
of Physostegia; the actual amount may be far greater.

TAXONOMIC TREATMENT

The specimen citations in this treatment are of two kinds. For the
relatively rare species, Physostegia correllii and P. longisepala, every
collection studied is listed. For the other species, only representative
specimens are cited. For a more complete listing, see Cantino (1980).
The first set of my own Physostegia collections has been deposited
in the Gray Herbarium. Unless otherwise stated, all cited type spe™
imens have been personally examined. Herbarium abbreviations fol-
i system used in Index Herbariorum (Holmgren & Keuken,

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 53

NOMENCLATURAL HISTORY

A survey of North American floristic works published during the
past 70 years testifies to the confusion and disagreement that has sur-
rounded the application of the name Dracocephalum to the genus now
correctly known as Physostegia. The situation has received consider-
able discussion (S. J. Smith, 1945; McClintock, 1949; Shinners, 1949;
Sealy, 1954; Hitchcock, et al., 1959; Mohlenbrock, 1963), the review
by Sealy being the most thorough. With the conservation of a Eur-
asian species, D. moldavica L., as the type of Dracocephalum (1961
International Code of Botanical Nomenclature), permitting the appli-
cation of the more recent name, Physostegia, to the North American
genus, the controversy should finally have been resolved. However,
the incorrect use of the name Dracocephalum in a number of rela-
tively recent floras (e.g., Radford, Ahles & Bell, 1964; Welsh &
Moore, 1973) suggests that the situation is still widely misunderstood.

The earliest published description of a representative of Physostegia
is probably that of Morison (1669), who referred to it as “Galeata &
verticillata, persicae foliis, digitalis aemula.” Morison did not provide
an illustration, but Boccone’s (1674) “Pseudo-Digitalis persicae foliis,”
under which Morison’s earlier name is listed in synonymy, is accom-
panied by an unmistakable picture of Physostegia virginiana.

The name Dracocephalon was first applied to the genus by Breyne
(1680), in fanciful allusion to the shape of the flowers, and again by
Tournefort (1700); it was modified to Dracocephalum by Linnaeus
(1737) in the first edition of Genera Plantarum. Within his circum-
scription of Dracocephalum, Linnaeus included not only the American
genus to which Breyne and Tournefort had applied the name Dra-
cocephalon, but also the Eurasian genus referred to by Tournefort as
Moldavica. After outlining the differences in calyx morphology by
which Tournefort’s two genera could be distinguished, Linnaeus dis-
missed these characters as too variable within the genera to be 0
diagnostic value; he finished by stating that the uniting feature of his
Dracocephalum was the shape of the corolla. Of the 11 species in-
cluded under Dracocephalum in Species Plantarum (1753), only the
first one listed, D. virginianum, was a member of the genus now
known as Physostegia.

Linnaeus’ broad circumscription of the genus did not long stand
unchallenged. Adanson (1763, Vol. 2: 187-194) distributed the species
included under Dracocephalum by Linnaeus into three genera, Dra-
cocephalon Tourn., Moldavica Tourn., and Rhuyschiana Amm., each
assigned to a different section of the family. He did not indicate types
for his genera, but, as pointed out by Sealy (1954), the type of Dra-
cocephalum L. emend. Adanson must be the species that Linnaeus

54 PHILIP D. CANTINO

named D. virginianum, because this species (under a different name)
was the only one placed by Tournefort in his genus Dracocephalon,
upon which Adanson based his Dracocephalon. Moench (1794) also
subdivided Dracocephalum L., distributing the species Linnaeus had
included within it among four genera, Dracocephalum, Moldavica,
Cedronella, and Zornia. Under Dracocephalum, Moench placed only
a single species, D. lancifolium Moench, a superfluous name for D.
virginianum L. because the latter was listed in synonymy.

These treatments were not widely accepted. Jussieu, for instance,
followed Linnaeus’ broader circumscription of the genus in his Genera
Plantarum (1789). It was not until Bentham’s treatments of the La-
biatae were published, first a synopsis of the family in the Botanical
Register of 1829-1830 (sub t. 1282, 1289, 1292) followed by an ex-
haustive monograph of the family, Labiatarum Genera et Species (1832-
36), that Linnaeus’ view was finally rejected once and for all. Recog-
nizing the disparate nature of the elements included by Linnaeus un-
der Dracocephalum, Bentham erected a new genus, Physostegia, to
comprise D. virginianum L. and its congeners, while distributing the
remainder of Linnaeus’ species among several other genera, the larg-
est of which he called Dracocephalum. Thus Bentham agreed with
Adanson and Moench that D. virginianum belonged in its own genus
distinct from the Old World species of Dracocephalum L., but unlike
these earlier authors he reserved the name Dracocephalum for the
larger Eurasian genus and provided a new generic name for the
American plants. Sealy (1954) has expressed the opinion that Bentham
reversed the earlier approach of Adanson and Moench for the sake
of convenience, choosing to rename the one American species as Phy-
sostegia rather than creating new combinations for more than 20 spe-
cies that would have had to have been transferred to Moldavica if the
name Dracocephalum had been reserved for D. virginianum L. The
name Physostegia is derived from the Greek, physa (bladder) and
stege (covering), in allusion to the calyx, which becomes slightly in-
flated when the plant is in fruit (Fernald, 1950).

Bentham’s treatment in Labiatarum Genera et Species was almost
universally accepted for nearly 80 years. Endlicher (1838), Meisner
(1839), Lindley (1846), Gray (1848, 1868), Bentham and Hooker (1876),
Baillon (1891), Briquet (1895-96), Britton and Brown (1898), Britton
(1901), and Small (1903) all applied the name Physostegia to the
American genus. However, in 1913, when Britton and Brown pub-
lished the second edition of their Illustrated Flora of the Northern
United States and Canada, they reversed their earlier usage and ap-
plied the name Dracocephalum to the American genus and Moldavica
to the primarily Eurasian genus, specifying D. virginianum L. as the
type of the former and D. moldavica L. as the type of the latter. No

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 55

ail was given for the change. While Sealy (1954) suggested that
ritton and Brown were simply following Adanson rather than Ben-
am, a more likely explanation was hinted at by Shinners (1949)
although he incorrectly attributed the reversal of Bentham’s usage io
Small rather than to Britton and Brown. The 1907 American Code of
Botanical Nomenclature (cf. Bull. Torrey Bot. Club 34: 167-178) stip-
ulated that the rule of priority in typification should apply to the
precedence of names within a given publication as well as to the dates
. different publications. Inasmuch as N. L. Britton was one of the
eee proponents of the American Code (Lawrence, 1951), it is
ikely that he chose Dracocephalum virginianum as the type of Dra-
ong ound because it was the first species listed by Linnaeus, rather
: an because Adanson had indirectly typified the genus in this way

y reference to Tournefort’s earlier usage.
: ~ Seaggabe assertion (1949) that Epling (1929) selected Draco-
ephalum virginianum as the type of Dracocephalum is incorrect; he
merely chose a lectotype for the species. However, Britton and Browns
nue of Dracocephalum with D. virginianum was supported by
cs ~ and Green (1929) in their list of proposed “standard-spe-
= ones genera. This list was incorporated as a supplement
ysstopde 5 and unofficial 1947 editions of the International Rules of
— ica nealailagtertoen Although the proposals contained in the list
. not inding, they were undoubtedly influential, with the result
= os floristic works published in the United States after 1930
“ . names Dracocephalum and Moldavica rather than Physo-
: . . and Dracocephalum for the American and Eurasian genera, re-
rely (e.g, Rydberg, 1932; Small, 1933; Kearney & Peebles,
i ; Davis, 1952; Hitchcock, Cronquist & Ownbey, 1959; Radford,
ee & Bell, 1964). Other floras published during the same period
the strictly American genus as

contusion.

oe the mid-1940’s, dissatisfaction with the
ormal proposals that one or another mem

nag (i.e., Moldavica L. emend. Adanson or Dracocephalum L.

5 end. Benth.) be conserved over D. virginianum as the type of
racocephalum, thus permitting the use of the name Physostegia for

: . American genus (Smith, 1945; Weatherby, 1947, as footnote

0 Hitchcock & Green’s supplement to the unofficial International

ules of Botanical Nomenclature [Brittonia 6:115]; McClintock, 1949).
(1954) that the name

~ culminated in a formal proposal by Sealy
racocephalum L. emend. Benth. (type: D. moldavica L.) be placed

situation led to several
ber of the Old World

56 PHILIP D. CANTINO

on the list of Nomina Generica Conservanda. This proposal was re-
ferred to committee at the 8th International Botanical Congress in
Paris. It was eventually endorsed but changed in form to agree with
Article 48 of the International Rules (Rickett, 1960); i.e., it was rec-
ommended that D. moldavica L. be conserved as the type of Dra-
cocephalum, rather than Dracocephalum L. emend. Benth. being con-
served over earlier circumscriptions of the genus. The proposal was
adopted as part of the 1961 International Code of Botanical Nomen-
clature, where Dracocephalum first appeared in the list of Nomina
generica conservanda et rejicienda.

Dracocephalum has thus now been typified, but Physostegia has
not. In connection with McClintock's (1949) proposal that Draco-
cephalum ruyschiana L. be selected as the type of Dracocephalum,
she suggested that D. virginianum L. be treated as the type of Physo-
stegia. The former was not a formal proposal and was never acted
upon by an International Botanical Congress. However, now that D.
moldavica L. has been conserved as the type of Dracocephalum, 1
propose that McClintock’s informal typification of Physostegia be ac-
cepted. In the protologue of Bentham’s original publication of the
name Physostegia (Bot. Reg. sub t. 1289. 1829), he stated that the
genus includes Dracocephalum virginianum L., D. variegatum Vent.,
D. denticulatum Ait., and probably D. cordatum Nutt. The first three
he synonymized under Physostegia virginianum in Labiatarum Genera
et Species (1832-36), and the lattermost he placed in a different ge-
nus, Cedronella (it is now known as Meehania cordata). If Bentham’s
inclusion of D. variegatum and D. denticulatum within Physostegia
virginiana is accepted (and it is in this treatment), then the only spe-
cies of Physostegia known to Bentham at the time he described the
genus was P. virginiana.

Physostegia Bentham, Edward’s Botanical Register 15: sub t. 1289.
1829.

, ser

ceme axis puberulent to tomentose throughout or glabrous towards the base, the ves-
ture sometimes including minute stalked glands; floral bracts lanceolate to ovate, acute
to attenuate; pedicels 0.5-2.5 mm long, densely pubescent, sometimes bearing 4 few
stalked glands. Calyx regular, campanulate to tubular-campanulate, very obscurely 10-
nerved, the exterior densely puberulent to pubescent (very rarely subglabrous) and
often glandular-punctate and/or stipitate-glandular, the interior glabrous or stipitate-
glandular; the 5 lobes equal or nearly so, deltoid to lanceolate, acute to cuspidate,

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 57

slightly galeate, horizontal or divergent; lower lip 3-lobed, the lobes divergent to re-

flexed. Androecium of 4 stamens, ascending parallel with or slightly divergent from

forming a thin layer around the embryo. CHROMOSOME NUMBERS: 2n = 38, 76. LECTOTYPE
SPECIES:Physostegia virginiana (L.) Bentham

ARTIFICIAL KEY TO THE SPECIES

k The polythetic nature of the species in Physostegia (see p. 44) greatly complicates
€y construction. A choice must be made between a key with extremely complex c
plets, in which each species appears only once, and one with

with some species keying out several times. I have elected to employ the latter be-

the point of carelessness. The statements in the couplets do not necessarily apply to
members of the species concerned, but only to the portion of the species that keys
out under that statement.
Illustrations of the following key characters have been provided: elongate, horizontal
thizomes (Fig. la); perennating buds borne irectly on rootstock (Fig. 1 ); leaves

; y raceme axi e with
10x hand lens. Those on the corolla in P. arviflora are smaller but clearly visible
ey a dissecting microscope. The latter are most easily seen on flower buds just be-
‘ore anthesis and near the tip of the upper lip in newly opened flowers.

: @endular pubenilence 6665 ee
. Calyx and rachis of inflorescence puberulent but lacking glands .------+----
“9 Leaves sessile or petiolate, but none clasping stem «-+--+--e0rt erect C.
. One or more leaves clasping stem, sometimes narrow y EEO ale pre
C. Nutlets 1.7-2 mm long, usually warty over part or all of s

toothed to entire, 2-8 mm wide; flowers loosely spaced, ectear" cis over-
. P. go

wider than 8 aiid: flowers usually tightly ked in inflorescence, adjacent ca-
lyces overlapping considerably; widespread .------+00077"" ae. F. virginiana.
D. Larger leaves 2.5—4 times as long as wide, only: conspicuously clasping stem;

58 PHILIP D. CANTINO

19-33 leafy nodes below inflorescence; flower hig with elve lobes
1-2.3 mm long; nutlets 3.1-3.5 mm long; — Co . Carolin

Bch oth tat Wee cet Megs ibaa ke. Gi. wisse bg Glk « 49 shee oe Po piratadonk

. Two or more characters not as above; Louisiana to Manitoba and
rin lik i ees

E. Flowers 21-41 mm long with calyx lobes 2-4 mm long; all major leaves
usually widest at to above middle of blade; Louisiana, Texas, norther
re as oie sf Beis Ss as Saw eS

Flowers 9-23 mm long with calyx lobes 0.7-2 mm long; some major leaves

nat ig nat below middle of blade, often near ee of blade; north-central

d northwestern U.S., central and etic Can ada PRE apres Sore tel ke c.

each terminating in a perennating i an ular dots pesca on so an

picuous on upper leaf surface of dried specimens .... 9. P. correllii.

t of short, vertic:
rhizomes, horizontal rhizomes lacking; glandular dots usually ab-
sent ae calyx and absent or inconspicuous on upper leaf surface of dried
IN 5g a SS ok ow hs i 8. ea italis.

upper pairs usually widest near base of blade; sometimes with majority
of stem leaves nie toothed; widespread in northwestern U.S. and
wig Poe ee ei ees : arvifl ra.

H. One or more na clasping st

H. Leaves sessile or petiolate, bee de none clasping stem a

J cond pair below terminal raceme longer than the internode above:
glandular dots present on calyx an conspicuous on upper leaf surface of dried
specimens; obits leaves seine te serrate and no more than 3 times as long as wide;
northern MeO) ie cic ae 9. P. correll
Two or more characters not as above; not occurring in Mexico .........-.

. Ro ng one to many elongate, horizontal secondary rhizomes, cach
terminated by a een age bud

K. Rerouting bud

Ce ee ee a a

ondary rhizomes, hora ont rhitoner lacking: 260.0. . Ss
i Flowe m long; all or most of eae stem leaves acute to attenuate
at apex; axis os raceme densely pubescent, always some (usually —_
trichomes 0.2-0.25 m long; west of Mississippi fo. ee es
Lik — smaller, or half or more of larger leaves obtuse at — or axis 0
raceme minutely oo few if any trichomes more than 0.15 mm —_
(mostly less than mm); espr read in setahsasty astern U.S.) 4 ses

; petiolate lower oe eo dihe present at anthesis; south-
he Louisiana aed hectormccncnta SOL eh es oP. loagiene®
ender to shia all leaves sharply serrate; petiolate
efore anthesis; southwestern —— and north-
ee Oe ee P. angustifolia.
N. At least one pair of upper stem leaves usually widest vs base of pape
flowering calyx be (1-)2—4 mm long; flowers always less than 20 m

rs v
“leaves usually Pareto b
est uisiana

ee

N. Upper stem leaves widest above to below middle of ick, a rarely
at ; flowering calyx tube 3—7(—8) mm long; flowers often longer than

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 59

mm; base of plant rarely conspicuously swollen, usually little thicker
ote central part of stem; Atlantic and Gulf coastal plains from Virginia
to Florid
F.

ee kik 6 + ke ee ee ee ee ae ee ee ee ee eee ee eee ee eee eee Oe wee ee ee

mor
than three-tenths as long as internode ~_ neip stem leaves

e r ow termin
larger than the floral pag those of second pair (1.5—) veohegedl cm

ng an ths
principal | stem leaves ae widest at to below middle of en
wo er swamps and fresh and brackish marshes, Fics gi grow-
ing in deep shade. |... 65.34 < ine P. leptophylla.

Q. Most or all of the larger leaves sharply serrate; largest ee on dried spec-
imens not more than 2.5 em wide (wider when fresh) and rarely less than 5
Deoes as long as Wide «5. ccc us kik ee ek ee?

Q. Most or all of the ore leaves bluntly toothed to os or largest leaves
more than 3 cm wide or less than 5 times as long as wide ..........--. S.
R. Corolla deep lavender to reddish violet; stem with 7 “10 nodes below in-

florescence; petiolate lower stem leaves frequently present at ie after an-
thesis;eastern T6xas ©... oe es ee oe 6. P. pulchella.

inflorescence (if in Texas, 11-20 nodes); pet tiolate ie stem scan ae
deciduous before anthesis; southwestern Georgia to Texas, north to Mis
Souri. Kansas 9. ee ee 7. a neusifolia,

S. Largest leaves more Sets pare | iho an. or, if concentrated
in lower third of stem, then most leaves of upper Phe of stem
apr toothed to entire; southeastern U.S., sahaline ie T.

of raceme densely pubescent to tomentose, the ferret mostly
0. 2-0, 3 mm long; siriesg Paces at anthesis generally 2-4 mm long, ona
or all of them attenuate or cuspidate; Louisiana and easter
Re ee P. digitalis.
T. Axis of raceme puberulent to pubescent, the trichomes rare rely as muc
as 0.2 mm long; calyx lobes at anthesis generally 1-2 mm long, mostly

merel te: Casale Georgia, Florida ....-..----- 1. P. purpurea.
U. All or most of larger eee sharply serrate and acute to attenuate at apex; leaves
of second pair spines inal raceme a fifth as long as to about equalling the in-

Pp

Mmode directly above ........:--::-::erttre ttt
Or more of eae ate bluntly toothed to entire and/or obtuse at apex, or
leaves of second pair below terminal raceme less than a fifth as long as or longer
the inieriade shoe oe ee lee ee mer eet ees iret ee W.

V. Axis of raceme densely pubscent, some (usually many) trichomes 0.13—0.25 m
long; sterile bracts eteee from inflorescence; usually blooming April to mid-
ul 7.7, 8 eg

» Axis of raceme minutely puberuleat, few if any trichomes more Oem, mm
long sterile bracts frequently present below the flowers; pony Nooming July
0 October P. virginia
W All or most of 1 l harply serrate ...---+++-+-> i: Fr. ited rr
arger leaves sharply s
W. Half or more of larger leaves bluntly tailed in entire 5. sae cee
X. Leaves of second pair below terminal raceme rar rarely more than three- a s
as long as the internode above; open pinelands, glades, and igi om
ta Sy Wil eres ee eek eee abe damien
X. poet of pond pair below terminal raceme a third as lon: om as as ce
long as the int above: swamps, marshes, and river and lake margins
from Quebec . Florida and west to Tennessee ...-.---++-+-+e0- 000" 7

ea eRe ee ae eee ee
a POR Cele Oe ace wie eo ee ee eee Pe

60 PHILIP D. CANTINO

Y. Lowest leaves present at or after anthesis usually petiolate, petioles often
more than 2 cm long, some petiolate leaves (if present) among largest leaves
on plant; ded swamps and fresh and — igi marshes of coastal plain

m

o Mary
land, south through mountains to northern Tennessee . . . 12. P. virginiana.

1. Physostegia purpurea (Walter) Blake

Prasium? purpureum Walt. Fl. Carol. 166. 1788. LECTOTYPE: Specimen labeled P. pur-
pureum Walt. by S. F. Blake on pg. 87 of the bound herbarium of Thomas Walter
BM, not seen; photo seen in GH).

Physostegia purpurea (Walt.) Blake, Rhodora 17: 134. 1915.

Dracocephalum purpureum (Walt.) McClintock ex Gleason, ene te 4; 24.

1952.
er obovatum me gat ig Bot. S. Carol. & G . 1821. HOLO-
: Georgia, St. Mary ep $.n. (CHARL, not seen; pir seen in GH).
Physo sateeie virginiana var. ee a (Ell) Gray, Synopt. FI. N. A

. 1878.
rare cep ig? um denticulatum var. ae Re (Ell.) Farwell, Pap. Mich. Acad. Sci. Arts
+ OF. 1923
Papeete alisoata (Ell) Godfrey ex Weath. Rhodora 44: 254. 1942.

Erect perennial herbs to 14 dm high, with (5—)7—11(—22) nodes below the inflo-

rescence, largest leaves pues <9 anions in lower third of stem, the inflores-

cence thus appearing subscapose. Primary rhizome unbranched or with few branches,

vertical or horizontal, up to 15 c heey jf pienn 1-6(-11) pairs of stem leaves petiolate

or all leaves sessile; petiole, hed present, up to 6.5 cm long; blades of lower and
ves c 4

middle stem 16 cm long, 0.2- e emely variable in shape, from
linear to spatulate to broadly obovate, oblong, elliptical, or pandurate, at to
above the middle of th blade, base attenuate, cuneate, or sli tly auriculate, east
a few leaves usually clasping the , apex obtuse to round ss frequently
acute, nd or bluntly toothed, the upper leaves sometimes sharply serrate
Upper stem leaves greatly reduced, often little larger than the floral bracts (those o
second pair below the terminal rac .4—3.2 cm long and rarely more than three-

pita as long as the internode above), linear to narrowly lanceolate, sometimes nar-
ni y oblanceolate or elliptical, apex acute, margin sharply serrate to entire. Flowers

a few oe, (0.8—)1-2.3(—3) mm he see t fruit maturity (4— )5-9(-11) eg
long. Corolla white to lavender, usually spotted oe streaked inside with purple, to-
eeeiclie to glabrous. Nutlets 2-3. 13.6) mm ne. trigonal with concave to slightly
convex sides, surface smooth. CHROMOSOME NUMBER: 2n = 38.

iy meet oat et Florida. Lee Co.: Fort Myers, Foocecwoein 690 (PENN). ve
py Aeon, E of Bithlo, Kral 6559 (14, DUKE, vpB, GH, FSU). Osceola he 1m
0 Kissimmee Fine on Fla-60, Lakela 25212 (vps, NCU, SMU. ets Georgi
5 mi N of Irwinville, Wilbur 3364 (ta, Ga, smu, Fsu). Lowndes Co.: 5 mi E of 8 F Valdosta
Godfrey & Houk 62758 (smu, sv). McIntosh Co. be of ridge on Bill S
mor 1091 (Ncv). North eee Columbus Co.: i SE of Old D math ee 5a
a CH). —— Co.: 3.5 mi SE of Wendell Radford 25198 (DAO, vDB, NCU). Onslow
ot mi N of Lolbysuhen: Ahles ¢> Haesloop 28215 (Ncu). South Carolina. Clar-
encion Co.: 2 mi SSE of St. Paul, Radford 24522 (Ncu, Fsv). Georgetown Co.: 4 mi

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 61

SW of Andrews, Godfrey & Tryon 149 (TENN, GH). Horry Co.: 1 mi NW of Loris,
Bell 13746 (GH, NCU).

DISTRIBUTION AND HABITAT (map: Fig. 14): moist openings and roadside ditches in pine-
lands from east-central North Carolina to southern Florida, west to southwestern Geor-
gia and adjacent parts of the Florida panhandle. The species occasionally occurs in
cypress savannas in southern Florida. A report of Physostegia purpurea from Tennes-
see (Wofford & Dennis, 1976) is based on a misidentification. The specimen concerned
(Wofford ¢& Dennis 51757, TENN) is a member of P. virginiana ssp. virginiana.

FLOWERING: early May through mid-August, except in the southern third of the Flor-
ida peninsula, where it may bloom at any time of the year.

NOMENCLATURAL NOTES: The application of the epithets purpurea and denticulata to
this species is discussed elsewhere (Cantino, 1981b).

Physostegia purpurea exhibits clinal variation in several characters,

€ most conspicuous of which is leaf shape (Fig. 13). At the northern
end of the range, in east-central North Carolina, the leaves are
broadly obovate to elliptical or oblong, frequently as much as 3-4 cm
wide. In southeastern North Carolina and eastern South Carolina, the
largest leaves are mostly 1-2 cm wide. Although leaves as much as
2 cm wide can be found as far south as southern Florida, they are
usually much narrower in that region, often as little as 0.2-0.5 cm
wide. The transition is gradual, with considerable variation both within
regions and within populations. For example, in a population in Pen-
der Co., North Carolina (Cantino 975), the widest leaf per plant
ranged from 1.4 to 3.1 cm in width. Near the other end of the cline,
in a population in Sarasota Co., Florida (Cantino 1006), the width of
the widest leaf ranged from 0.6 to 2.0 cm.

Another character that varies clinally in Physostegia purpurea is the
degree of crowding of the flowers. In southern Florida, the flowers
are always loosely spaced, adjacent calyces overlapping little if at all.
More tightly packed flowers occur with increasing frequency as one
moves northward through the range of the species, reaching a max-
imum in North Carolina, where over half of the specimens have
tightly packed flowers with much overlap between adjacent calyces.

At least three other characters exhibit a north-south pattern of vari-
ation in Physostegia purpurea, but in contrast to the clinal variation
in leaf width, there is for each of these characters a relatively abrupt
transition line, on one side of which the character is monomorphic
and on the other side polymorphic. In the Florida peninsula P. pur-
Purea rarely if ever produces horizontal rhizomes, while north of
about the 30th parallel, horizontal rhizomes may be present or absent,
with much variation within populations. Throughout most of the range
of the species, the axis of the raceme and the outside of the calyx
are densely puberulent to pubescent. However, in southern Florida
many specimens are only very sparsely puberulent, some approaching
a glabrous condition. Plants with the usual dense puberulence are also

62 PHILIP D. CANTINO

PHYSOSTEGIA PURPUREA

—<—S 7 —
cd a
= =

N. CAROLINA GEORGIA

GEORGIA om (2). i 4 S. FLORIDA

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 63

common in southern Florida, and there is much variation within pop-
ulations. The marked reduction of the upper stem leaves that is so
characteristic of P. purpurea reaches an extreme state in the east-cen-
tral and southern parts of the Florida peninsula (Volusia Co. to Collier
Co.), where a growth form occurs in which all of the larger leaves
are concentrated near the base of the plant, the raceme thus ap-
pearing almost scapose. This growth form is nearly unique within the
genus, occurring elsewhere only in a very few specimens of P. pul-
chella. It is not, however, a consistent characteristic of any population.

Flower size varies geographically in Physostegia purpurea. In south-
western Georgia and the adjacent part of the Florida panhandle, the
flowers of P. purpurea are among the smallest in the genus, ranging
from 11 to 23 mm long on dried specimens (a few millimeters longer
when fresh). Throughout the rest of its range the flowers are mostly
20-34 mm long, the only exceptions being a few specimens from east-
ern Georgia. The intrapopulational variation is great. In one popula-
tion in Sarasota Co., Florida (Cantino 1006), the flowers ranged from
22 to 33 mm long. I found ranges nearly as great (21-30 mm; 20-29
mm; 25-34 mm) in populations in Camden Co., Georgia (Cantino 990)
oe Flagler and Lake Counties, Florida, respectively (Cantino 1001,

).

In spite of the extensive morphological variation in Physostegia pur-
purea, there are no clearly delimited infraspecific taxa. The variation
in several characters is clinal, and of those characters in which there
is a more abrupt transition between character states, no two of them
have a geographically similar variation pattern. There is therefore not
enough correlation among the character states to warrant the recog-
nition of infraspecific taxa.

2. Physostegia godfreyi Cantino
Physostegia godfreyi Cantino, Rhodora 81: 415. 1979. HoLoTyPE: Florida, Gulf Co.,
Wet pine flatwoods, and in shallow water of ditches, 7 miles S of Wewahitchka,
18-VI-1958, Godfrey 57086 (GH). ISOTYPES: FSU, IA, USF.

Erect, slender, perennial herbs to 1 m high, with 7-13 nodes below the inflores-

Fic. 13. Geographic variation in middle leaves of Physostegia pur-
mre@: each leaf from a. different plant, each cluster from a single
Population. Vouchers at GH. Upper left—Pender Co., NC (Cantino
975). Middle left—Columbus Co., NC (Cantino 978-980). Lower
left—Giynn Co., CA (Cantino 989). Upper right—Camden Co., G
Cantino 995). Middle right—Sarasota Co., FL (Cantino 1006). Lower
"ight—Collier Co., FL (Cantino 1011-1014).

64 PHILIP D. CANTINO

ence. Rhizome usually unbranched, vertical or horizontal, up to 10 cm long. Lowest

1-3(-5) pairs of stem leaves petiolate or, less frequently, all leaves sessile; petiole,

when present, up to 3 cm long; blades of lower and oo stem leaves lacking any

visible secondary venation, 1.5-7.5 cm long, 2—6(-8) mm wide, linear to narrowly

ene, spatulate, or oblanceolate (rarely lanceolate), ea: somewhat falcate, base at-
s

subgla-

brous at the base, trichomes rarely over 0.1 mm long, stalked glands catered
throughout; floral bracts ovate, 2-3.5 mm long, ae 2 mm wide; flo owers

long, loosely spaced, adjacent calyces usually not overlapping at — se ahien no re
conspicuously glandular-punctate but aa stalked glands, tube pri esis 3-5.5
mm long, lobes acute, 0.6—-1.8 mm long; ae at fruit maturity 4-651 mm long. Corolla
pale lavender, spotted and streaked inside with purple, puberulent or tomentulose to
glabrous. Nutlets 1.7-2 mm long, trigonal, sides strongly convex when fully developed,
usually verrucose over all or part of surface. CHROMOSOME NUMBER: 2n= 38.

aeiovarr athe SPECIMENS (for a more complete list, - See 1979): Florida. Bay

: Calloway, 21-VI-1977, Athey s.n. (Fsu ae Calhou : 4 mi a of Blountstown,
Godfrey et al. as ge cH). Franklin Co.: 6 mi S of Age atra, Kral & Godfrey i
(vpB). Gulf C i W of Wewahitchka, Godfrey 71370 (Fsu, a Liberty C

mi N of Oran ne. “Meanie 4472 (Fsv).
DISTRIBUTION AND HABITAT (map: Fig. 14): moist pine savannas in the lower Apalach-
— region of the F lorida panhandle.
OWERING: mid-May through late August.

In my discussion of the morphological and geographical relation-
ships between Physostegia godfreyi and P. purpurea (Cantino, 1979),
I stated that the Ochlockonee River forms the napraesp ls between the
ranges of the two species in the Florida panhandle, P. purpurea oc-
curring strictly to the east of the river and P. godfreyi entirely or
nearly entirely to the west. It has since come to my attention that P.
purpurea has been collected west of the Ochlockonee River (Godfrey
65848, Fsu), in Liberty Co., within about 5 miles of a site where P.
godfreyi occurs. The specinier resembles many collections of P. pur-
purea from Wakulla Co., just east of the Ochlockonee River, and
shows no sign of introgression with P. godfreyi.

3. Physostegia leptophylla Small

a bg Shetages Small, Bull. N ard. 1: 286. 1899. LecToryPE: Florida,

he “ ae Xe VI- may Rete : 7A a no. er Garber s.n. (xy), here des-
ted. ISOLE

Deacccaphad: leseophidan, (Small) Small, Flora of Miami, 163. 1913.

Gemaed spine ath Small, Fl. Southeastern U.S. 1028, abho 1903. HOLOTYPE:

= G near Sunbury, LeConte s.n. (Ny?; missi ing).

racocephalum veroniiformis (Small) Small, Man. Southeastern Fl. 1156. 1933.

Physostegia aboriginorum Fern Rhodora me 459. 1943. LECTOTYPE: Virginia, Norfolk
0., margin o a Tacks n Creek, northeast of Northwest, 30-VI-1942, Fernald & Long
14397 (GH), here aseeied tse H, (PH, not seen).

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 65

arg
\
? |
<9
)
x* %, é
«
x
* * oe
*
* | 2°
td
S : : | . ;
x ok
ba |
> 7
* ‘ x
Wk ok oe
tiga “ee aa
¥ * Y
* * .
J
* re a
a
@ P purpurea
# P leptophylla
¥* P intermedia
A pgodfreyi

Fic. 14. Distribution map of Physostegia purpurea, P. leptophylla,
P. godfreyi, and P. intermedia.

Erect perennial herbs to 14 dm high, with 7-15 nodes below the inflorescence.
Primary rhizome branching to produce 1-many elongate, horizontal secondary rhizomes
up to 40 ¢ ak. Lowest gia pairs of stem leaves petiolate, frequently still
Present at anthes petiole to 6 cm long; blade 3-11 cm long, 1-3 cm wide, el-
liptical to lanceolate (rarely Slenoslada: base cuneate to attenuate (rarely rounded),
apex obtuse to acute, margin entire, repand, or crenate. Sessile leaves of central part

BS es.

a

Me nd clasping the stem, a apex acute to jatbasietts, margin bluntly toothed to repand,

occasionally entire or sharply serrate. Upper stem leaves little to Cael reduced

in size over central leaves bese of second pair apni ne terminal raceme [1.5—]2-
int

sPidate), 1-2.5(—3) m . calyx at fruit maturity 5-9.5 mm long. Corolla deep
lavender to reddish Sek eet aid streaked inside with purple, sparsely ROOTES

66 PHILIP D. CANTINO

puberulent or tomentulose to subglabrous. Nutlets 2.2—3.2 mm long, trigonal with
concave to slightly convex sides, surface smooth. CHROMOSOME NUMBER: 2n = 76.

[W?] of Gulf Hammock, Kral 6495 (GH, IA, DUKE, VDB, FSU). Wakulla Co.: Newport,
Godfrey 62831 (vpB, sMU, FsU). Georgia. Chatham Co.: Onslow Island, Duncan 21021
(miss). North Carolina. Jones Co.: 5 mi NE of Pollocksville, Sears C322 (Ncu). Onslow
Co.: 4.2 mi SE of Gum Branch on Half Moon Creek, Ahles ¢> Haesloop 28312 (NCv).
Pender Co.: along NC-210, 2.6 mi E of US-117, Ahles ¢ Haesloop 28054 (Ncu). South
Carolina. Colleton Co.: Edisto River at US-17, Leonard & Radford 1693 (miss, WVA,
GH, TENN, NCU, GA, CM, SMU, FSU, NO). Virginia. Southampton Co.: Nottoway River, Mon-
roe Bridge, Fernald & Long 13122 (GH, GA, SMU, WVA, TENN).

marshes from extreme southeastern Virginia to south-central Florida, west to south-
western Georgia and adjacent parts of the Florida panhandle.
FLOWERING: late April through early August.

Small stated in the protologue of Physostegia leptophylla that the
“original specimens” were collected by Garber (So. Fla. Fl. No. 10).
There are four sheets of this widely distributed collection at the New
York Botanical Garden, where Small was working at the time the
name was published. Of the four, one of them is too incomplete to
have served as the primary basis for Small’s comprehensive descrip-
tion, and one was transferred to Ny from the Princeton University
herbarium in 1945, long after the name was published. The choice
of one of the remaining two sheets as the lectotype is problematical.
Both were transferred to Ny from other institutions, one from Co-
lumbia College (previously transferred to Columbia from Franklin and
Marshall College, Lancaster, Pennsylvania) and one from the New
York College of Pharmacy. Small could easily have seen either, but
the Columbia College specimen has been selected because it is in
better agreement with the description in one detail (petiole length)
and because Small attended Franklin and Marshall College and re-
ceived his doctorate from Columbia College before joining the New
York Botanical Garden (Barnhart, 1938). He almost certainly would
have seen this specimen, and he may well have been responsible for
its transfer from Franklin and Marshall College to Columbia.

Physostegia veroniciformis Small appears from the description to be
a taxonomic synonym of P. leptophylla, but the small flower size of
the former is uncharacteristic of the latter. Unfortunately the type
specimen, listed by Small as having been deposited in the herbarium
of Columbia College (now part of the herbarium of the New York
Botanical Garden), is missing.

Two practically identical sheets of the type collection of Physostegia
aboriginorum Fernald are in the Gray Herbarium. Because both cor-
respond equally well to the description and both have been annotated

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 67

TABLE 18. DISTINGUISHING CHARACTERISTICS OF PHYSOSTEGIA LEPTOPHYLLA AND P. INTERMEDIA.

ophyl P. intermedia
Uppermost leaves am tet widest at Pppeseest leaves usually widest at base
base of blade

Bomrering calyx tube 3.5-6 mm long F ieeded calyx tube (1-)2-4 mm long
ase of plant rarely conspicuously swol- Base of plant frequently conspicuously
Bon swollen
Length + width of fruiting calyx tube 0.8 Length + width of fruiting calyx tube 0.5
to 1.5, rarely to 1

“Type no.” by Fernald, the selection of one as the lectotype was nec-
essarily arbitrary. Both sheets include leaves, rhizomes, roots, and
mature flowers, but only one includes mature nutlets; this specimen
was therefore chosen as the lectotype.

I have discussed elsewhere (Cantino, 1981b) the nomenclatural con-
fusion surrounding the epithet denticulata, which has been applied
incorrectly to Physostegia leptophylla by Fernald (1950) and others.

The morphological overlap between Physostegia leptophylla and the
sympatric but reproductively isolated species, P. purpurea, has been
discussed elsewhere (Cantino, 1981b). There is a roughly comparable
amount of morphological overlap between P. leptophylla and two
other species with which it is allopatric—P. longisepala and P. inter-
media. The differences between P. leptophylla and P. longisepala are
discussed under the latter species. Physostegia intermedia occupies
swamp and marsh habitats in the Mississippi Valley and westward,
Similar to those frequented by P. leptophylla on the Atlantic coastal
plain. There is no single morphological character that will, in itself,
reliably distinguish the two species, but there are four rp pee
characters that will, in combination, distinguish them (Table 18). I
"age P. leptophylla and P. intermedia differ in chromosome num-

ber (38 pairs vs. 19 pairs, respectively).

4. Physostegia intermedia (Nuttall) Engelmann & Gray

pe ehatom intermedium Nutt. Trans. Am. Phil. Soc. 5: 187. 1837. LecToTyPE: Red
er, Nuttall s.n. eg here designated.
P “ocnaltigs ra (Nutt.) Engelm. & Gray, Boston gee Nat. Hist. 5: 257. 1845.
Physostegia micrantha Lundell, Wrightia 2: 8. hares OLOTYPE: Texas, Titus Co., off
Hwy 49, about 1 mile SE of Mount Pleas in grees wet bottom land of Hart
Creek, 29-V-1958, Lundell 15075 (LL, not saul ISOTYPES: GH, NY.

Erect perennial herbs to 12 dm high, with 9—16(-20) nodes below the inflorescence.
Primary rhizome brane raged to Ls agp te l1-many elongate, horizontal secon: d ter
Hiary raging up “a 40 cm long. Stem often conspicuously sa at t base. ice

cm long; bade es of ee and middie stem leaves 3-14 cm 8 0.3-1.5 m wide, all
olate or some leaves oblanceolate to narrowly elliptical, base ae to cuneate
on lower Leia rounded to auriculate upwards, at least a few leaves clasping the

68 PHILIP D. CANTINO

stem, apex acute to attenuate, margin repand, entire, or bluntly toothed, teeth few
and widely spaced. Upper stem leaves little to moderately reduced in size over central
stem leaves (those of second pair below the terminal raceme 1.7-9[-12] cm long and
a third

tenuate, usually widest at or very near the auriculate-claspin s borne i
1-5(-10) racemes, raceme axis densely puberulent to pubescent throughout or sparsely
so to glabrous towards the base, trichomes frequently up long, rarely to

0.2 mm; floral bracts lanceolate or less frequently ovate, attenuate, (1.5-)2—5(-6) mm
long, 1-1.5(-2.5) mm wide; flowers normally 9-19 mm long (shorter if anthers aborted),

4 mm long, lobes acute (rarely a few cuspidate), (0.7—)1-2(-2.8) mm long; calyx at fruit
i la

puberulent to tomentulose. Nutlets 2-2.5(-2.9) mm long, trigonal with concave to
slightly convex sides, surface smooth. CHROMOSOME NUMBER: 2n = 38.

REPRESENTATIVE SPECIMENS: Arkansas. Ashley Co.: Mist, Demaree 15091 (OKL, SMU).
Craighead Co.: Lake City, Demaree 5089 (TENN, CH, TEX). Drew Co.: Tillar, Demaree
p ;

Taylor 4154 (smu, OKL). Texas. ison Co

Trinity River, Nixon 4036 (Ncvu). Rains Co.: 9.8 mi S of Point, Van Vleet 1374 (sMv).
Refugio Co.: ca. 2 mi E of Refugio, Jones 1803 (smu, Fsv). Waller Co.: entrance to
Austin State Park, near Sealy, Correll ¢ Edwin 16438 (GH, NCU, LL).

ISTRIBUTION AND HABITAT (map: Fig. 14): swamps, marshes, river bottoms, wet mead-
ows, and drainage ditches, from southeastern Missouri (one record each in Kentucky
and Illinois) south to the Gulf Coast of Louisiana, and west to central Texas and south-
eastern Oklahoma. The collection locality of the Kentucky record is unknown (“barrens

rt

>

f Kentucky,”
tral Illinois (He
the Field M

a
more likely by someone sorting or mounting his collection after its purchase.
FLOWERING: late March through late July.

In the protologue of Dracocephalum intermedium, Nuttall cited no
specimens but stated that the species occurs “on the prairies in moist
places, from Arkansas to Red river.” I have seen two specimens (BM,
PH) that might be considered as candidates for the lectotype, both of
which correspond well to the description. As’is characteristic of Nut-
tall’s specimens, they are accompanied by only the briefest collection
data, the specimen from the British Museum bearing the words “Red
River,” and the specimen from the Philadelphia Academy labeled sim-
ply “Ark.” Both labels are in Nuttall’s handwriting, and on both there

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 69

is an asterisk preceding the specific epithet. An asterisk, as Pennell
(1950) has pointed out, was Nuttall’s notation to indicate a new species
or genus.

Pennell (1936) related that until 1818, Nuttall kept few specimens
for himself, presenting “a complete series of his plants” to the Acad-
emy of Natural Sciences in Philadelphia, but that starting with the
Arkansas collections he reserved an increasing proportion of his better
specimens for his own personal collection, which he took with him
when he returned to England in 1842. Thus, Pennell went on to say,
“We may consider that his later types are in London, with isotypes
in Philadelphia.” Accordingly, I have selected as the lectotype the
specimen in the British Museum.

Although the label data on the lectotype is brief, it is possible to
obtain more precise information from Nuttall’s published account of
the journey (Nuttall, 1821; Pennell, 1936). Nuttall’s explorations of the
Red River were confined to the 15-mile stretch immediately upriver
from the mouth of the Kiamichi River in what is now Choctaw Co.,
Oklahoma. He collected in that area from May 23 through June 13
of 1819.

Physostegia micrantha Lundell is based on a single population in
Titus Co., Texas. It differs from P. intermedia in having very small
flowers (5-7 mm long) with aborted anthers, but it resembles P. in-
termedia in all other characteristics. Plants with aborted anthers and
an accompanying reduction in flower size are quite common in some
Populations of P. virginiana ssp. virginiana (see p. 32) and occur spo-
radically elsewhere in the genus. The specimens upon which P. mi-
crantha is based are without doubt simply another example of this
phenomenon. William F. Mahler has collected a series of specimens
from the type locality of P. micrantha (Mahler 6458 a—h, SMU), some
of which exhibit the floral characteristics of P. micrantha while others
have the larger flowers and fertile anthers of P. intermedia.

5. Physostegia longisepala Cantino sp. nov.

-Herba perennis erecta ad 1 m alta, nodis 9-15 infra inflorescentiam. Caudex rhi-
ia foliorum caulinorum

i on
folii petiolati 5-8 cm longa, 1-1.5 cm lata, elliptica, oblonga vel oblanceolata, bas

minuta, lanceolata vel elliptica, saepe prope basem laminae amplectem latissima :
mi 1-7, axe dense pubescenti, trichomatibus |
bracteae florales ‘eaceak Ga attenuatae, (3-)4—6(-7) mm longae, 1-2 mm latae; flores

70 PHILIP D. CANTINO

23-32 mm longi, rare vel dense positi. Calyx sub eset signee iy Sei pn tubo

8 mm longo, dentibus attenuatis vel cuspidatis, 2-3.5 mm longis. Corolla intense
Sandielocea vel asec , parte in rr seepiroah psig -purpureis. Nuc a 3-3.3
mm longa, trigona, lateribus laevibus. HOLOTYPUS: Lou Daven Parish, 2 miles

north of Edgerly, 18-V-1968, J. W. Thieret ‘28876 aint ISOTYPU US
Erect perennial herbs to 1 m high, with 9-15 nodes below eae inflorescence, Pri-

pair below the terminal raceme 2.7-6 cm long and three-tenths as long as to about
half as long as the internode directly above), lanceolate to elliptical, often widest near
the clasping base of the blade. Flowers borne racemes, raceme axis dense
pubescent, some (usually many) of the seiclibinies a Hs 25 mm long; floral bracts
lanceolate, attenuate, (3-)4-6(-7) mm lo ong, 1-2 mm wide; flowers 23-32 mm long,
loosely to tightly spaced. Calyx not conspicuously it rian lacking stalked
glands, tube at fy ies 4-8 mm long, lobes attenuate to cuspidate, 2-3.5 mm long;

o.
“<

#& P. PULCHELLA

Fic. 15. Distribution map of Physostegia angustifolia, P. digitalis,
P. longisepala, and P. pulchella.

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 71

va x at pee maturity 7.5-10 mm long. Corolla — lavender to reddish violet, spot-
te : ed inside with purple, puberulent or tomentulose to subglabrous. Nutlets
ew available) 3-3.3 mm long, surface smooth. Chr. romosome number unknown

4 hac EXAMINED: ayo — Parish: near Crowley, Debaillon 3 (Mo); near
. wle epee s.n. (BH owley, Dormon 3 (smu). Bienville Parish: grown
rN ete en (originally olla. near Crowley, Acadia Parish), Ewan 19233 (GH, No).

above four collections were apparently all from the same clump in Caroli
it ; garden, originally transplanted from Crowley by M Debaillo aie
— arish: 1.5 mi SW of Starks, Thieret 23545 (smu); 4 mi S of Gillis, Shinners
(SMU, ILL, NCU). Texas. Jasper Co.: near Evadale, Correll 32936 (LL).
enn AND HABITAT (map: Fig. 15): moist prairies, thickets, and ditches, in
uthwestern isiana and southeastern Texas; apparently rare

Loui
FLOWERING: mid-May through mid-June

The 12 specimens upon which Physostegia longisepala is based bear
ag resemblance to P. leptophylla, P. angustifolia, and P. pulchella,
ut the extent to which they differ from each of these species is too
great to permit their inclusion in any of them (Table 19). The most
distinctive characteristic of P. longisepala is indicated by the name I
ave chosen; the calyx lobes at anthesis are relatively long for Phy-
Sostegia. Every specimen has at least some calyx lobes 2.5 mm long

TABL
E 19. DIAGNOSTIC CHARACTERS OF PHYSOSTEGIA LONGISEPALA IN RELATION TO THE 3 MOST
SIMILAR SPECIE

‘ P. longise
ngest calyx lobes 2.5- a 5 mm long; fre-
"rele cuspidate
ngest trichomes on raceme axis 0.2—0.25
mm lon
Lower stem leaves usually oblanceolate

Stem leaves commonly sharply serrate

" P. longisepala

ngest calyx lobes 2. ries 5 mm long; fre-
Bes ently cuspidate

Orizontal secondary rhizomes produced
Lowest 4—

leaves

Stem with (9-)11- Ww
: 11-15 |
= eafy nodes belo

8 stem nodes bear petiolate

P. long la

- gisepala
i Nejad lavender to reddish violet
n condary rhizomes produced

Lo
west 4-8 stem nodes bear petiolate

lea
“esa some usually persistent through

P. leptophylla

Calyx lobes Balt over 2.5 mm long; in-
sexiest Se

s on raceme axis 0. 1—0,13
mm lon “4

Lower stem leaves elliptical to lanceolate
(rarely oblanceolate)

hei bluntly esi repand, or entire

y sharply serrate)

mm long;

—

P: ella
Calyx lobes rarely over 2.2(-2.5
infrequently cuspidate
Horizontal secondary rhizomes never pro-
duced; rhizomes short and vertical
a, 1-4 stem nodes bear petiolate

Pesos nih hae leafy nodes below
inflorescen
P. angustifolia
Corolla usually - as Siiilie to white
—— secondary rhizomes rarely pro-

iowa t 1-4 stem nodes bear pace
leaves; these usually deciduous by a
thesis

7 PHILIP D. CANTINO

or longer, and on most specimens many or all of them are cuspidate.
In P. pulchella and P. leptophylla they rarely exceed 2.5 mm in length
at anthesis and are infrequently cuspidate. However, this character
does not distinguish P. longisepala from P. angustifolia; calyx lobes of
the length found in P. longisepala are at the upper end of the range
of variation seen in P. angustifolia, and the cuspidate shape is fre-
quent in the latter species. Physostegia longisepala differs from P.
angustifolia in flower color, the form of the rhizome, and the number
of pairs of petiolate leaves (Table 19).

In a discussion of Physostegia pulchella, Lundell (1969) noted that
a collection from Jasper Co., Texas (Correll, Johnston & Edwin 22299),
which he included within P. pulchella, is notable for having “strong
lateral rhizomes.” He suggested that it might represent a distinct
taxon. I have not seen the specimen, but it was collected within a
few miles of the site where Correll 32936 was collected and, like the
latter specimen, probably represents P. longisepala.

6. Physostegia pulchella Lundell

Beale aah apg Lundell, Wrightia 2: 4. 1959. HOLOTYPE: Texas, Kaufman Co.,
$-175, ca. 1 mile E of Crandall, in bl Aopen land along ae
bed. 12- V-1959, Lundell 16026 (LL, not seen). ISOTYPE NY.

Erect perennial herbs to 14 dm high, with 7-10(-12) nodes below the inflorescence.
Rhizome unbranched or with few branches, strictly vertical, up to 6 cm long. —
1-4 pairs of stem leaves petiolate, some usu ly present at anthesis; petiole up t

m long; blade 2.5-11 cm pied 0.6-1.7 cm yg narrowly elliptical to ee or
obtuse, margin entire, repand, or re-

1-3 pairs of we primary ag (other than the midrib) a ing from base of blade,
7 2 cm wide, lanceolate to Msniielaes or nage atulate, base usually
auriculate-clasping, apex pies or obtuse, margin usually sharply serrate to base of

Calyx not conspicuously glandular. nt ei lacking stalked glands, tube at anthesis
(3-)4-6 mm long, lobes acute to attenuate (occasionally a few ae 1-2.2(-2.5)
mm long; calyx at fruit maturity 6-9 mm long. Corolla deep lavender to reddish violet,
purple, puberulent, occasionally tomentulo se or gla-

REPRESENTATIVE ete be oie ne ~ 9.6 mi NE of Lyons, Cory 51642
Ga mick). Collin Co.: 1.6 of Royse City, Cory 55813 (smu, No). Delta Co.:
mi NE of coger hives seaes (suv, FSU, deb Grayson Co.: N of Denison,

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 73

Gentry 1236 (smu). Grimes Co.: 5 mi from Navasota on FM-159, Massey 108 (sMv).
Hunt Co.: 2.1 mi S of Commerce, Shinners 28448 (GH, SMU, NO, FSU, BH). Kaufman Co.:
2 mi E of Terrell, Shinners 10087 (smu). Lamar Co.: 4.6 mi WSW of Paris, Shinner.
14843 (smu). Matagorda Co.: College Port, Demaree 61548 (OKL, sMU). Navarro Co.:
6 mi S of Richland, Cory 51540 (smu). Robertson Co.: Hearne, Lundell & Lundell
10378 (NCU, SMU).

DISTRIBUTION AND HABITAT (map: Fig. 15): moist meadows, river bottoms, and ditches
in eastern Texas. Thieret (1971) has reported P. pulchella from Louisiana on the basis
of two collections that I have included within P. longisepala (one of them the type
of the latter). I have seen no specimens from Louisiana that fit within my circum-
scription of P. pulchella.

FLOWERING: early April through early June (rarely to late June).

Physostegia pulchella shares many morphological characteristics with
P. angustifolia, and the two species frequent similar habitats; they are
probably closely related. The most conspicuous difference between
them is the color of the corolla, which is deep lavender to reddish
violet in P. pulchella and very pale lavender to pure white in P. an-
gustifolia. | am aware of a single specimen of P. pulchella with white
flowers (Fleetwood 9780, TEX), and Lundell (1969) mentions having
seen a specimen of P. angustifolia with reddish purple corollas, but
exceptions of this sort are rare.

In Physostegia pulchella there are usually 7-10 stem nodes below
the inflorescence (very rarely 12), and the blooming period lasts from
the beginning of April to the beginning of June. Plysostegia angus-
tifolia has 9-20 nodes and blooms from April through July; however,
in Texas where the two species are parapatric (Fig. 15), P. angustifolia
has 11-20 nodes and starts flowering in mid-May, when P. pulchella
is approaching the end of its blooming period. In P. pulchella the
petiolate lower stem leaves tend to persist longer than in P. angus-
tifolia, often up to or beyond the time of anthesis; they are therefore
frequently present on herbarium specimens of the former but usually
acking on specimens of the latter.

7. Physostegia angustifolia Fernald

Physostegia angustifolia Fern. Rhodora 45: 462. 1943. HOLOTYPE: Mississippi, Chickasaw
Co., roadside bank near Egypt, 18-V-1933, Weatherby & Weatherby 6318 (GH).
I PES: NY, TENN, GH.
Dracocephalum virginianum var. album Nutt. Trans. Am. Phil. Soc. 5: 187. 1837.
: “Arkansa,” Nuttall s.n. (BM).
Physostegia edwardsiana Shinners, Field & Lab. 19: 167. 1951. HOLOTYPE: tor
Blanco Co., between Johnson City and Dripping Springs, at Hey 300, ip marshy

Erect perennial herbs to 17 dm high, with 9-18(-20) nodes below the inflorescence.
Rhizome unbranched and strictly vertical, 2-4(-10) cm long, or (infrequently) branch-
ng to produce 1-many elongate, horizontal rhizomes up to 20 cm long. er and
middle stem leaves all sessile or lowest 1-4(-7) pairs petiolate, the petiolate leaves
usually early deciduous; sessile leaves 3-21 cm long, 0.3-2(-2.7) cm wide, lanceolate

74 PHILIP D. CANTINO

mewhat auriculate, usually clasping the stem at least slightly, apex acute to atten-
uate, the lowest leaves occasionally obtuse, margin sharply serrate, rarely bluntly
toothed or entire. Upper stem leaves usually much reduced in size over central stem
leaves (those of second pair below the terminal raceme (0.7-)1—5(-6) cm long and an
eighth as long as to two-thirds as long as [or rarely longer than] the internode directly
above), similar in shape to central stem leaves, but often entire or only remotely ser-

Flowers borne in 1—5(-8) racemes, raceme axis densely pubescent throughout,
some (usually many) of the trichomes 0.13-0.2 mm long, often to 0.25 mm long; floral

to oblanceolate, occasionally almost linear, base cuneate to narrowly truncate, often
so

3
sometimes as much as 12 mm long); flowers (18—)22—33(-36) mm long, tightly packed,
adjacent calyces at anthesis mostly overlapping a quarter or more of thei

tomentulose. Nutlets 2-3(-3.5) mm long, trigonal with a to slightly convex
sides, surface smooth. CHROMOSOME NUMBER: 2n = 38.

REPRESENTATIVE SPECIMENS: Alabama. Montgomery Co.: 2 mi S of Montgomery, Kral
& Demaree 30947 (vps, sMU, FsU). Sumter Co.: 3.3 mi S of Dancy, Kral 42992 (vDB).
Arkansas. Hempstead Co.: Fulton, Palmer 8028 (MO, PH). Yell Co.: Plainview, De-
maree 63819 (KANU, VDB, NCU, SMU, MIN). Georgia. Baker Co.: 1 mi NE of Newton,
Thorne 4372 (Ga, F, MO, CU). Kansas. Cherokee Co.: 1 mi N of Columbus, McGregor
15659 (NY, KANU, NCU, GH, SMU). Montgomery Co.: 3 mi E of Sycamore, McGregor 14380
(KANU, NCU, SMU). Louisiana. Caddo Parish: 3.2 mi E of Trees City, Shinners 26178

m

, 8} many Parish: to
Demaree 49985 (vPl, PH, sMU). Mississippi. Kemper Co.: E of Scooba, Jones 12542
MISS, Ga). Lee Co.: near Verona, Cooley & Ray 5247 (GH, VDB, NCU, FSU). Mis-
souri. McDonald Co.: roadside and in Blackjack-Post Oak savanna, Buck 255 (OKL).
Oklahoma. Ottawa Co.: 0.5 mi NE of Quapaw, Wallis 7260 (KANU, NCU, Ga, TEX, SMU,
OKL). Sequoyah Co.: 0.5 mi S of Gore, Wallis 7495 (KANU, NCU, TEX, SMU, OKL). Texas.
i Co.: 0.5 mi W of Hooks, Lundell ¢ Lundell 16039 (Ny, GH, LL). Burnet Co.:
mi W of Bertram, Rogers, Albers ¢> Barksdale 6870 (PH, MICH, F, TEX).
DISTRIBUTION AND HABITAT (map: Fig. 15): moist prairies, meadows, ditches, river bot-
toms, and marshy areas, from extreme southwestern Georgia (1 record) and eastern
Alabama to central Texas, north to Missouri and Kan

FLOWERING: early April through the third week in July (rarely the end of July).

Physostegia angustifolia is the archetype of the polythetic species.
It is easily recognized by a suite of correlated traits, not one of whic
is reliably diagnostic in itself. The leaves of P. angustifolia are usually
sharply serrate, but in one population in Oklahoma (Cantino 1079,
GH) some plants have entire leaves. The leaves usually clasp the stem
to some degree, but when they are particularly narrow the clasping
base is inconspicuous or lacking. The rhizome is usually short an
strictly vertical, but elongate horizontal rhizomes are produced by
many plants in a limited area of northwestern Louisiana and south-
western Arkansas (Lundell 16032, LL, GH, NY, TEX; Moore 6057, GH;
Cantino 1074, Gu). The corolla is characteristically very pale lavender
to white, but reddish violet floral variants are known.

A character that is highly useful in distinguishing Physostegia an-
gustifolia from the sometimes similar P. virginiana ssp. praemorsa is

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 75

the length of the trichomes that compose the raceme vesture (Fig.
6a,b). The distinction is quite evident at a magnification of 10x, but
a higher magnification (60x) is necessary to quantify it. Physostegia
angustifolia consistently has some (and usually many) trichomes 0.13
to 0.25 mm in length, while P. virginiana ssp. praemorsa rarely has
even a few over 0.1 mm long. Exceptions are extremely rare in both
taxa, accounting for at most 1-2% of the specimens. The exceptions
in P. virginiana ssp. praemorsa are widely scattered through its range,
whereas in P. angustifolia the exceptions are concentrated in north-
western Arkansas.

The distribution of Physostegia angustifolia (Fig. 15) is strikingly
discontinuous; it is composed of three discrete sections in which the
species has been abundantly collected, separated by an extensive area
where it appears to be rare. Surprisingly, there are no consistent
morphological differences between the plants of the three areas; the
interpopulational variation within the easternmost range segment is
at least as great as the variation between the three segments. Shinners
(1951) has recognized the plants from the Edwards Plateau region as
P. edwardsiana, asserting that they differ from P. angustifolia in being
slightly taller, having a later blooming period, and in having leaves
that are serrated only in the apical two-thirds of the blade (versus all
the way to the base in P. angustifolia). I have not found any of these
distinctions to survive a wider survey of P. angustifolia, and I agree
with Lundell (1969) who synonymized P. edwardsiana under P. an-
gustifolia.

A widely distributed collection made by Roland Harper north of
Tuscaloosa, Alabama is problematical (Harper 3531). The population
from which the specimens were collected in 1936 is now extinct due
to the damming of the North River to form Lake Tuscaloosa, but
enough specimens are in existence (13 plants on 8 sheets) to provide
4 sample of the morphological variation in the population.

Although the collection is more similar to Physostegia angustifolia
than to anything else, there are five characters in which three or more
Specimens have a character state atypical of the species, and one of
“iem, the presence of stalked glands in the inflorescence, is unknown
in P. angustifolia. The glands are present on three specimens (F, NY,
PH), where they are produced in abundance on the calyx and sparingly
on the axis of the raceme. In 4 of the 13 specimens the raceme ves-
ture consists of trichomes that are shorter than is usual in P. angus-
tifolia, few of them exceeding 0.1 mm in length. In 4 of the 13 spec-
‘mens, the flowers are arranged rather loosely in the raceme, a
Condition that is infrequent in P. angustifolia. At least the lower leaves
2” nearly all of the specimens are bluntly toothed to entire, and in
5 specimens the leaves do not clasp the stem. Both of these condi-

76 PHILIP D. CANTINO

tions occur infrequently in P. angustifolia, entire leaves much more
rarely than non-clasping leaf bases.

The only other taxon to which the Tuscaloosa plants bear any re-
semblance is Physostegia virginiana ssp. virginiana. Although the
presence of stalked glands in the inflorescence of some specimens
might suggest affinities with P. virginiana, in which such glands occur
frequently, the majority of the Tuscaloosa plants differ from P. vir-
giniana in having leaves that clasp the stem slightly and in having
longer trichomes (frequently over 0.1 mm) on the raceme axis, both
conditions characteristic of P. angustifolia and extremely rare in P.
virginiana south of the Ohio River.

Perhaps the most likely explanation for the high variability of the
population is that it represents a hybrid swarm, possibly a remnant
of a rather old hybridization event. Morphology points to Physostegia
angustifolia and P. virginiana ssp. virginiana as likely parents, but the
absence of the latter from Alabama (see Fig. 17) argues against a re-
cent hybridization. These two taxa might, however, have been sym-
patric in central Alabama during the Wisconsin glaciation, when the
range of P. virginiana probably shifted southward to some extent.
Another unusual specimen, collected along the Chattahoochee River
in Early Co., Georgia (Thorne 5395, Cu), also appears to be inter-
mediate between P. virginiana and P. angustifolia, in that it possesses
the stalked glands and non-clasping leaves of the former and the
longer trichomes of the latter; Wisconsin-age sympatry and hybrid-
ization could account for the characteristics of this collection as well
as the Tuscaloosa population.

In view of the extensive variation within the Tuscaloosa population
in characters that are usually diagnostic of Physostegia angustifolia,
I cannot state with certainty that the population represents that spe-
cies. For the same reason, it is not possible to describe a new taxon
at the specific or infraspecific level based on that population. There
are some specimens (e.g., the left-hand specimen on the sheet at wIs)
that do not differ in any way from normal P. angustifolia, while others
(F, Ny) differ in three to four characters. This interesting population
is best left without formal recognition.

8. Physostegia digitalis Small

Physostegia digitalis Small, Bull. Torrey Bot. Club 25: 613. 1898. LECToTYPE: Louisiana,
Hale s.n. (NY), here designated.

Erect, robust, perennial herbs to 2 m high, with 9-13(-16) nodes below the inflo-
cence. Rhizome strictly vertical and usually unbranched, up to 10 cm long. Lower
and middle stem leaves all sessile or the lowest 1-4 pairs petiolate, the petiolate leaves
early deciduous; sessile leaves often having 1-3 pairs of weak primary veins (other

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 77

than the midrib) arising from base of blade, 5-17 cm long, 1.5—7 cm wide, broadly
oblanceolate or obovate to elliptical, less frequently ovate, base rounded to cuneate,
usually somewhat auriculate and strongly clasping the stem, apex acute (to acuminate)

bluntly toothed (rarely a few leaves sharply serrate). Upper stem leaves much reduced
in size over central leaves (those of second pair below the terminal raceme 1—3[-6
em long and less than a fourth as long as to longer than the internode directly above),
oblanceolate to lanceolate, the sessile base cuneate and sometimes slightly clasping,
apex attenuate, margin sharply serrate, sometimes bluntly toothed or entire. Flowers
borne in 1-11 racemes, raceme axis densely pubescent to tomentose throughout, tri-
chomes mostly 0.2-0.3 mm long, a few stalked glands occasionally present; floral
, attenuate, 3-9 mm long, 1.5-4 mm wide; flowers 25-41
mm long, tightly packed, adjacent calyces at anthesis overlapping half or more of their
lengths. Calyx never conspicuously glandular-punctate but occasionally bearing stalked
glands, tube at anthesis 4-8.5 mm long, lobes mostly attenuate to cuspidate, 1.5-4
mm long; calyx at fruit maturity 7-13 i
usually spotted inside with purple, glabrous to subglabrous, occasionally spar
tulent. Nutlets 2—3(-3.3) mm long, trigonal, sides usually slightly concave, surface
smooth. CHROMOSOME NUMBER: 2n = 38.

REPRESENTATIVE SPECIMENS: Louisiana. Beauregard Parish: 8 mi W of Longville,

as NLU). De Soto Parish: S of Evelyn on La-177, Thomas 45729 (NLU).

Vernon Parish: 8.3 mi S of Leesville, Shinners 20597 (cH, sMU). Texas. Bowie :
. Hardin C Es

|
©

nett, Lundell & Lundell 14733 (us, LL, SMU). Upshur Co.: 8.5 m
ners 18965 (smu). Van Zandt Co.: 6.3 mi S of Canton, Van Vleet 1629 (sMv).

DISTRIBUTION AND HABITAT (map: Fig. 15): moist, open pinewoods in western Louisiana
d eastern Texas.
FLOWERING: mid-June through late July.

The two specimens that were cited in the protologue as being “the
original specimens” can be found, mounted together on a single sheet,
at the New York Botanical Garden. Both were collected in Louisiana,
one by W. M. Carpenter and one by Josiah Hale. Of the two, the
Hale specimen is the more consistent with Small’s description, the
leaves and floral bracts of the Carpenter specimen being smaller than
those described in the protologue. For this reason, I have selected

€ Hale specimen as the lectotype.

est in the genus (mostly 0.2-0.3 mm long). A few specimens of P.
Purpurea from eastern North Carolina (e.g., Ahles & Haesloop 30027,
NCU), which are at the broad end of the leaf shape gradient exhibited
by that species, bear a superficial resemblance to P. digitalis. How-

78 PHILIP D. CANTINO

ever, they are easily distinguished on the basis of the length of the
trichomes on the raceme axis, those of P. purpurea rarely reaching
and never exceeding 0.2 mm in length. In addition, the calyx lobes
of P. purpurea are rarely cuspidate and are generally shorter than
those of P. digitalis.

9. Physostegia correllii (Lundell) Shinners
Dracocephalum correllii Lundell, Wrightia 1: 165. 1947. HOLOTYPE: Texas, Val Verde
along fees near the International Bridge at Mig Rio, 26-VI- 1946, Cone

0
e& Correll 1 (LL; missing, not seen). ISOTYPE: SM
Pie ji (Lundell) Shinners, "edora 51: 120. 1949.

Erect, robust, perennial herbs to 13 dm high, with 10-24 nodes below the inflo-
rescence. Primary rhizome branching to produce an extensive system of elongate,
horizontal secondary and tertiary rhizomes up to ong t 3-6 pairs of

spicuously glandular-punctate when ot #30 ~~ conspicuous, with 1-3

m mm lon
calyces at anthesis usually btheirny ‘aaned half or more of ei = ene Calyx c war 65

SPECIMENS EXAMINED: United States. Louisiana. Cameron Parish: 9 mi E of Grand
e, Cantino 1064 (cu). St. Charles Parish: ditch along Mississippi River levee and
0). T San

near Monterrey, 14-VII-1888, Pringle s.n. (LL, Vt); Monterrey, Santa Catarina, Arsené
ore i US, MO) [=Abbon 82h; Biss Salinas et al. 16M14 (TEx). Sonora: Near Santa
PH).

t
DISTRIBUTION AND HABITAT fag Fig. 16): river bottoms and ditches, from southern

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 79

Louisiana to northeastern Mexico (1 record from Sonora); widespread but rare.
FLOWERING: the third week of June through the end of September.

Physostegia correllii is characterized by having rather broad leaves,
rarely more than four times as long as wide, at least the upper ones
with one to three pairs of weak primary veins (in addition to the
midrib) arising from the clasping base of the blade (Fig. 4d), by the
presence of conspicuous glandular dots on the calyx and upper leaf
surface (conspicuous in dried material only), and by the production
of elongate, horizontal rhizomes.

There are two rather odd specimens of a single collection (White
1781, GH, MICH) from near Monclova in Coahuila, Mexico, whose af-
finities are clearly with Physostegia correllii but which differ in having
entire to subentire leaves that are somewhat narrower than is usual
in the species. Since the leaves of P. correllii are usually serrate, a
case could be made for recognizing a new variety on the basis of these
specimens. However, two collections (Latorre 47, Marsh s.n.) from
Muzquiz, a town about 100 miles from Monclova, appear to be in-
termediate between the Monclova collection and the more usual forms
of P. correllii; although the leaves on these two specimens are as nar-
row as those from Monclova, their margins are remotely but sharply
serrate. With collections of P. correllii as scarce as they are, it would
seem unwise to recognize a variety when there is evidence that it may
represent the endpoint of a cline of variation, the apparent morpho-
logical gap between it and the rest of the species being possibly only
a collecting gap.

The distribution of Physostegia correllii exhibits a sizable disjunc-
tion (Fig. 16) owing to the existence of a single collection from north-
€rn Sonora, Mexico (Wright 1536, GH, PH). The Wright specimens
have slightly smaller leaves and less crowded flowers than is usual in
P. correllii, but they are not otherwise distinctive. Although the gap
between the site of the Wright collection and the nearest collection
site of P. correllii to the east stands out as a particularly large dis-
Junction, the species as a whole appears to be rather sparsely distrib-
uted over a large area, perhaps due to disruption of a formerly more
Continuous range.

Physostegia correllii is classed as “endangered” in the 1974 Smith-
Sonian report to the Congress of the United States and in the more
recent revision of that list (Ayensu & DeFilipps, 1978). It has recently

€en recommended that its status be changed to “threatened” (R. S.
Irving, personal communication). As rare as it is, and tending as it

°es to grow in habitats subject to human and natural disturbance
(e.g., roadside ditches, river bottoms), there can be no doubt that it

80 PHILIP D. CANTINO

Fic. 16. Distribution map of Physostegia correllii (stars), P. parvt-
flora (circles), and P. ledinghamii (triangles).

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 81

is in an extremely vulnerable position. Every effort should be made
to protect the few remaining populations.
10. Physostegia parviflora Nuttall ex Gray

Physostegia parviflora Nutt. ex Benth. in D.C. Prodr. 12: 434, 1848. As synonym.
Physo ae sia. ad Nutt. ex Gray, Proc. Am. Acad. 8: 371. 1873. et valid (no

description; cites P. parviflora Nutt. ex Benth., itself not a vali
Physostegia pariefion ra Nutt. ex brine Synopt. FI. 2: 383. 1878. Not racocephelin
utt. Gen. N. Pl. 2: 35. 1818. Lectotype: Columbia R., Nuttall
e design

(BM
Pe orca eatniens var. r oaivifions (Nutt. ex Gray) Boivin, Nat. Canad. 93: 575.

Dracocephalum —— Britton in Britt. & Brown, Ill. Fl., 2nd ed., 3: 117. _
Based rviflorum Nutt. ex Benth.; not superfluous because the com
Sieg Dr oaphalon parviflorum would be a later homonym of D. parvi Jicrun

ice nuttallii (Britt.) Fassett, Rhodora 41: 525. 1939.

Erect perennial herbs to 7 dm high, with 9-15 nodes below the inflorescence. Rhi-
zome branched or unbranched, vertical or more frequently horizontal, up to at least

al raceme 2-11 cm long and two-thirds as long a o twice as long

uate, mostly 2-4 mm long and 1. 5 2.5 mm wide (the lowest pair sometimes up to
mm long); flowers 9-16 mm long, —. packed, stone calyces at anthesis over-

pu
ally tomentulose usually bearing a few stalked glands. Nutlets 2.1-3.3 mm 1 Jon ng, ‘tri
Zo aia

ly
nal with flat to slightly convex sides, surface smooth. CHROMOSOME NUMBER: 2n =

REPRESENTATIVE SPECIMENS: Canada. British Columbia. 1 mi N of S end of Mar
Lake, Calder ¢> ported 11834 (pao, atta); Vanderhoof, Calder et al. 13539 (pao); pea
Arrow Lake, 16 mi of Nakusp, Calder 37145 (Dao). Manit toba. Virden, Scoggan
11196 (cx). haidaachisees — 17 July 1938, Bolton s.n. (scs); Qu ‘Appelle Vale

Cc ion, S of Ellisborough, s 214 (pao). United — Idaho.
MnP Bay, Lake Pend oie “Pennell 21281 (cu, PH). Canyon Co.: Falk's stor,
acbride 314 (xy, 14, cH, LL, RM, US). Montana. Pit Co.: Flathead

WTU, MO, I
somers, Muenscher ¢ Muenscher 1 1386 ‘(cu wtv). North Dakota. Benson Co.: otk
“er “es 1907, Lunell s.n. (xy, IND). Oregon. Multnomah Co.: Rooster Rock Park, just

Portland, Taylor ¢ Staudt 4264 (pao). Utah. Cache Co.: 2 mi W of Mendon,

82 PHILIP D. CANTINO

Smith 1889 (RM). Washington. Clarke Co.: mouth of Lewis River, 6 Sept 1892, Hen-
derson s.n. (WTU). Pend Oreille Co.: 3 mi SW of Usk, Kreager 320 (wrv). Wyoming.
Albany Co.: 12 mi S of Laramie, Porter 6021 (DAO, NY, GH, WTU, SMU, RM)

DISTRIBUTION AND HABITAT (map: Fig. 16): wet meadows, ditches, gravelly and marshy
areas along lake shores and river banks, from central British Columbia south to Or-
egon, northern Utah, and Wyoming, and from southeastern Saskatchewan to northern
North Dakota. A single specimen, collected in 1883 at Lake City, Minnesota (Manning

nin
virginiana ssp. virginiana. Neither specimen is in any way c
taxon to which it belongs. There are no additional collection data accompanying either
specimen to indicate whether or not they were collected at the same site. In the
absence of any other record of P. parviflora east of north-central North Dakota, I
suspect that the Manning specimen may have been collected elsewhere and misla-

eled.
FLOWERING: late June to mid-September.

Index Kewensis cites two references to the name Physostegia par-
viflora, but in neither instance was the name validly published. The
first publication was by Bentham, who attributed it to Nuttall but
merely listed the name in synonymy with P. imbricata Hook., without
referring to any earlier published description. The second reference
in Index Kewensis is to a use of the name by Gray which has no
nomenclatural status. It consists only of a reference to the earlier in-
valid publication of the name by Bentham and lacks a description.
The earliest valid publication appears to be that of Gray (1878). He
again attributed the name to Nuttall, citing Bentham’s initial publi-
cation of it as well as his own use of the name in 1873, but for the
first time a description was provided.

I have seen two specimens (BM, PH) that would undoubtedly be con-
sidered to be type material if Nuttall, himself, had published a de-
scription of the species. Both are labeled in Nuttall’s hand as being
Physostegia parviflora, and the collection data on both consist solely
of the name “Columbia R.” Because the first description to accom-
pany a publication of this name was supplied not by Nuttall, but by
Gray, the situation is more complicated. However, it seems best to
choose one of the two Nuttall collections as the lectotype inasmuch
as Gray and Nuttall are known to have been in frequent contact dur-
ing Nuttall’s later years in North America (Graustein, 1967). It is
highly probable that Gray saw either Nuttall’s specimens of Physo-
stegia parviflora or a manuscript based on them, and that his own
description was based directly on these materials. Bentham’s origin
citation of “P. parviflora Nutt.! mss.” in synonymy with P. imbricata
Hook. indicates that a manuscript description of some kind existed
at that time; Nuttall supplied many such descriptions to Torrey 2?
Gray (Graustein, 1967).

It is not surprising that of the two specimens of Physostegia par

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 83

viflora, the one in the British Museum is by far the better (see p.
69), the collection in the herbarium of the Philadelphia Academy
being but a fragment. The specimen in the British Museum fits Gray’s
description perfectly. It is not unlikely that Gray saw it, either before
Nuttall left for England or after Nuttall’s death, when his personal
collection was deposited in the British Museum. Gray spent a year
in England beginning in September of 1868 (Dupree, 1968), and he
would presumably have had the opportunity to examine Nuttall’s spec-
imens at that time. In consideration of these facts, I have chosen as
the lectotype of Physostegia parviflora the specimen in the British
Museum. According to Graustein (1967), Nuttall collected P. parvi-
flora during the midsummer of 1835 at The Dalles, a narrows of the
Columbia River in the region where it forms the border between
Wasco County, Oregon and Klickitat County, Washington.

Physostegia parviflora is characterized by its small and densely
crowded flowers, the presence of stalked glands on the calyx and co-
rolla, and the broadly clasping upper stem leaves, some of them
usually widest near the base of the blade. The only species with
which it could be confused is P. ledinghamii. The distinctions between
the two species and the probable hybrid origin of the latter are dis-
cussed elsewhere (Cantino, 1981a).

A cladistic analysis based on morphological characters (see p. 46)
suggests that Physostegia correllii is the closest extant non-hybrid rel-
ative of P. parviflora. Although not strikingly similar in overall ap-
pearance, the two species share several characters that are infrequent
in the genus as a whole. The most unusual characteristic of P. par-
viflora is the presence of stalked glands on the corolla. I have been
able to find at least a few on better than 95% of the specimens of
P. parviflora. These glands are also present on a quarter of the spec-
imens of P. correllii and a third of those of P. ledinghamii. This trait
is very rare elsewhere in the genus, although stalked glands are pres-
ent on the calyx and the axis of the inflorescence in several other
Species. A second uniting characteristic is the unusual leaf venation
that is universally present in P. correllii and occurs in about 90% of

€ specimens of P. parviflora. In addition to the midrib, there are
Sne to three pairs of weak primary veins that arise from the clasping
leaf base and ascend part way up the blade. This venation also occurs
Commonly in P. pulchella and P. digitalis and infrequently in several
other species, but it is more prominent in P. correllii and P. parvi-
flora than in the others. :
_ The distribution of Physostegia parviflora exhibits a notable dis-
Junction (Fig, 16), the two segments of the range separated by a gap

at is 350 miles wide at its narrowest point. It is probable that the
Species had a more continuous distribution shortly after the most re-

84 PHILIP D. CANTINO

cent glaciation and has since been eliminated from the intervening
zone by the advent of drier climatic conditions. This hypothesis is
supported by palynological evidence that the intervening prairie re-
gion was occupied by a spruce-dominated forest during the Wisconsin
glaciation. This spruce forest was replaced by prairie vegetation about
12,000 years ago in Kansas and Nebraska and about 10,000 years ago
in south-central Canada (Wright, 1970; Ritchie, 1976).

11. Physostegia ledinghamii (Boivin) Cantino
oe — Boivin ex Fraser & Russell, Annot. List Pl. Sask.: 36. 1953.

Bieccepholen ledinghamii (Boivin) Russell, Ledingham & Coupland in Frase
Rus we Seu List Pl. Sask.: 36. 1953. No description; cites only an cigublehd

bas
PE ame ig virginiana var. ledinghamii poy, Nat. Canad. 93: 574. 1966. First valid
obi n of the basionym. HOLOTYPE: Saskatchewan, Swift Current District, Ca-

illes au sigh opi re sablonneuse de la Saskatchewan du Sud,” 28-VII-
1952, Boivin & Alex 9978 (pao).

Physostegia ledinghamii ois Cantino, Rhodora 83: 111. 1981.

Erect perennial herbs to 1 m high, with 9-16 nodes below the inflorescence. Pri-

mary rhizome branching to produce 1-m any elongate, horizontal secondary and tertiary
rhizome middle stem leaves all sessile or the lowest 1-5 pairs petiolate,
the petiolate leaves usually early deciduous: sessile leaves sometimes having 1-3 pairs
of weak primary veins (other than the mi rib) arisin nde base of blade, 4-15 cm

i as
long, 0.5-3 em wi e, narrowly basta to lance eolat or percogienit' base cuneate

2.5 mm wide (lowest pair sometimes up to 11 mm long); flowers 14-23 mm long,

i packed, adjacent calyces at anthesis usually overlapping half or more of their

engths. Calyx not Soy glan belli but bearing stalked glands, tube
-5 mm long,

REPRESENTATIVE SPECIMENS: Canada. Alberta. Fort Saskatchewan, ng tone 4979 (ALTA);
1

near Manola uly 1968, write $.n. (ALTA); near Cl ep
> yde, ca. 45 mi N of Edmonton,

ee E2692 (axa). —— Le Pas, 21 July 1936, Howe s.n. pe a TRT, SCS)-
orthwest Territories toes 7 (DAO, ALTA, MO), Saskatchewan. Tisdale,

al
Breitung 1790 (pao, ALTA. es "Mae S of North Battleford in North cage
resol F baron 945 (Dao); Green Lake Village, Harms 16792 (DAO, GH). United $
- McLean Co.: Ft. Berthold Indian Reservation, Heidenreich 210 (ox.
phatase AND HABITAT (map: Fig. 16): low, wet woods and s swampy areas along

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 85

lake shores and stream banks, from northeastern Alberta to northeastern Manitoba,
south to North Dakota.
FLOWERING: early July through late August.

See Cantino (198la) for a discussion of the probable hybrid origin
of Physostegia ledinghamii and a table of the diagnostic characters by
which it can be distinguished from its putative parents, P. virginiana
and P. parviflora.

12. Physostegia virginiana (L.) Bentham

Erect perennial herbs to 18 dm high, with 10-34 nodes below the
inflorescence. Primary rhizome branched or unbranched, vertical or
horizontal, up to 65 cm long. Lower and middle stem leaves all sessile
or the lowest 1-7 pairs petiolate, the petiolate leaves usually early
deciduous; sessile leaves 2-18 cm long, 0.2—4.4(-5.5) cm wide, most
often elliptical to oblanceolate, varying to lanceolate, ovate, obovate,
spatulate or linear, base attenuate to cuneate, less frequently rounded,
rarely clasping the stem, apex acute to attenuate (occasionally the
lower leaves obtuse and rarely all leaves obtuse), margin most often
sharply serrate, less frequently bluntly toothed and rarely entire. Up-
per stem leaves scarcely to greatly reduced in size over central stem
leaves and similar in shape, those of the second pair below the ter-
minal raceme 0.8—-11 cm long and a third as long as to four times as
long as the internode directly above. Flowers borne in 1—16(—20) ra-
cemes, raceme axis densely puberulent to pubescent throughout or
sparsely so to glabrous towards base, nonglandular trichomes rarely
more than 0.15 mm long with stalked glands sometimes intermixed;
floral bracts lanceolate to ovate, attenuate, mostly 2-8 mm long and
1-2.5 mm wide (the lowest bracts occasionally larger and intergrading
with foliage leaves); sterile floral bracts often present below the flow-
ers; flowers (13-)14-37 mm long, tightly to loosely spaced, adjacent
calyces overlapping or not. Calyx conspicuously glandular-punctate or
not, sometimes bearing stalked glands, tube at anthesis 2.56-8 mm
long, lobes acute to attenuate (occasionally cuspidate), (0.6—)0.8—3 mm
long; calyx at fruit maturity 4-10(-11) mm long. Corolla reddish violet
to lavender to white, usually spotted and streaked inside with purple,
densely puberulent or tomentulose to glabrous. Nutlets 2.1-4.2 mm
long, trigonal with concave to convex sides, surface smooth. CHRO-
MOSOME NUMBER: 2n = 38.

ARTIFICIAL KEY TO THE SUBSPECIES (NATIVE PLANTS ONLY)

This key is not intended to be used for cultivated forms of Physostegia ghia
these are variable in morphology and may in some cases represent artificial intersub-

ch an :
Specific hybrids. Garden escapes frequently key to ssp. praemorsa at couplet D. In

86 PHILIP D. CANTINO

asmuch as the vast awed of the specimens that actually represent ssp. praemorsa
will key out at couplet A those keying to this subspecies at couplet D should be
viewed with suspicion, ee they were not collected in the region specified
in the first lead of the latter cou

Rhizome aetione are Hieaatieed in Fig. 1, leaf bases in Fig. 4(a,c), and sterile
bracts in Fig.

A. thesia buds borne directly on rootstock or at ends of short, vertical secondary
rhi , the clones forming tight clumps. .... . P. virginiana ssp. praemorsa.
A. Perennating buds borne at ends of elongate, horizontal secondary rhizomes origi-

B. Leaves 2.5—4 times as cae as wide, the largest “ae m long, some — stun
uf

slightly; Transylvania Co., N. Carolina. ..... virginiana ssp mors
B. — either longer or narrower than the rising all of them pitty prs
ng OES SRE NES UES Se SG ST nee eng re nny ee C

elit on dried specimens 13-24 mm long (longer when fresh).
Te a ee ee 12a. P. virginiana ssp. virginiana.
C. Flowers on dried specimens 25-35 mm long (longer when fresh). D.
e priismecpnee usually bearing 3 to many pairs of sterile bracts below flow-
rs; prairies, cedar glades, limestone barrens, and occasionally along streams;
haeed. prieacers Tennessee, northern parts of Sih ~~ Alabama, east-
ern Arkansas, perhaps in Kentucky Kat soathers Illin
Pole eM ae ee ate igh. P. set ssp. praemorsda.
D. Inflorescence rarely bearing more than 2 pairs of sterile bracts below
flowers; margins of rivers, ra lakes, and bays; Lake Erie to southern
Quebec, south commonly to Maryland and West Virginia, sparingly to cen-
tral Virginia and northeastern Tennessee
: RGN Moorea icles CRT E ee we ‘La. P. virginiana ssp. virginiana.

12a. Physostegia virginiana (L.) Benth, ssp. virginiana

parca BASED ON CULTIVATED FORMS OF PHYSOSTEGIA VIRGINIANA, INCLUDING NATURALIZED ES-
APES.

Dracocephalum virginianum L. Sp. Pl. 2: 594. 1753. Lectotype: Linn. Herb. Cat. no.
-1 (LINN, not seen; photo seen at GH), designated by Epling, Journ. Bot. 67:

1929.
Physi virginiana (L.) Benth. Bot. Reg. sub tab. 1289. 1829. Not validly Varner
bec — e the combination is not explicitly proposed (Art. 33.1, 1978 Int. Code Bot

Physostgia petal easton (L.) Benth. Lab. Gen. et Sp. 504. 1834. First valid publication

oO

Dracocephalum denticulatum Ait. Hort 2: 317. 1789. HOLOTYPE: BM (not seen);
photographs seen at GH. solic feel of the type are not absolutely certain
patient 1b).

Physostegia a sie! Benth. Bot. Reg. sub tab. ae 1829. Not validly pub-

lished because the combination is not explicitly pro

agony etn virginiana var. denticulata (Ait.) Gray, Synopt. FL. N. Am. 2: 383. 1878,
t nl

Physostegia dentition (Ait.) Britton, Mem. Torr. Bot. Club 5: 284.
—— virginiana forma denticulata (Ait.) Benth. ex Sieb. & wae wie Blum.

gi cited in :
Dracocephalum speciosum Swan Br. Fl. Gard. 1: tab. 93. 1825. Not D. speciosum
siat. nknown

i: 65. 1830. T u
Physostegia speciosa (Sweet) Sweet, Hort. Brit., ed. 3 406. 1830.

Poon antes lento sitio Meth. Pl. 410. 1794. Superfluous name (D. vir-
nianu my)

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 87

Physostegia virginiana var. speciosa (Sweet) Gray, Synopt. Fl. 2: 383. 1878.
Physostegia virginiana forma speciosa (Sweet) Benth. ex Sieb. & Voss, Vilm. Blum.
1895.

1: 856.
Dracocephalum virginianum var. speciosum (Sweet) Farwell, Pap. Mich. Acad. Sci.
Art 1

Se 1

praorteeio eons ook in Curt. Bot. Mag., new ser., 9: tab. 3386. 1835. TYPE:
n the Hooker Herbarium (k).

aaa aa louisianum Hort. in Lem. Hort. Univ. 4: 298. 1845. Type unknown.

fe roehelem rhea aha Hort. in Karsch, Phan. Fl. Prov. Westf. 797. 1853. Type

Dra Se alos regelii Hort. ex ec mg Voss, Vilm. Blum. 1: 856. i 95. As nye:
Physostea latidens House, Bull. . St. Mus. 176: 38. 1915. HOLOTYPE: New York,
ong peer and in a field pas of State Road, just south ‘of Utica, 10-VIII-
12, Haberer 3084 (nys).
Physostegia nivea + Lande Wrightia 5: 70. 1974. HOLOTYPE: Texas, Dallas Co., culti-
ar plants from Sune? asa San Francisco, California. 10-
VI-1974, Lundell 18837 (LL). ISOTYPE:

NAMES BASED ON WILD REPRESENTATIVES OF P. grag te SSP. VIRGINIANA.
Physostegia formosior Lunell, Bull. Leeds Herb. . 1908. LEcToTyPE: North Dakota,
Ward Co., in the timber along Souris Nag a Minot, 22-VIII- oe Lunell 883
(MIN), i mag by Elizabeth McClintock but never published. IsOLECTOTYPES: MIN(2).
1.

urrprephe mi onnoneie rage Rydberg, Brittonia 1: 95. 1931. Not po reicionga
rmosum Gontsch. Not. Syst. Herb. Inst. Bot. Acad. Sci. URSS 7: 101. 1938
P hysostegia virginiana v. - Lunell) Boivin, Nat. Canad. 93:

r ( 574. 1966.
P hysostegia s speciosa var. age le. Fassett, Rhodora 41: 377. 1939. HOLOTYPE: Wis-
consin, Racine, Hale s.n. (wis).
P, hysostegia granulosa Fassett, Rhodora 41: 377. 1939. HOLOTYPE: Canada, Prov. Que-
on Ca ; ue greves estuariennes, 21-VIII-1928, Marie-Victorin 28178 (wis).
TYPE:
P eathteets virginiana var. granulosa (Fassett) F sen Rhodora 45: 464.
Dracoc cephalum virginianum var. granulosum (Fassett) Core, Castanea se 301. "1972.
P a egia formosior aap! alba J. W. Moore, Rhodora & 2: 58. 1950. HOLOTYPE: Min
nesota, Roseau Co. wing in wet ground 1/4 95 ‘poctlewent of Wareed: 2.
VIII-1939, Moore & oe 11399 (MIN).
Physostegia vi; lia forma alba (Moore) Boivin, — eo 93: 574.
Ppvaeteaia giniana var. elongata Boivin, Nat. Canad. 573. 1966. HOLOTYPE:
Canada, oat Quebec, Ile Perrot, marecages a ye rd roa eaux, 1-IX-1927, Marie-
Victoris & Rolland-Germain 29005 (DAO). ISOTYPES: DAO

Primary rhizome usually branching im ae an extensive system of elongate, hor-

izontal = terti rhizomes up to 65 cm long. Central stem leaves 0
4.3(-5.5) wide, elliptical to nocclas. blaccsbbate: or spatulate, rete sharply
serrate, hs tk esa bluntly toothed, rarely entire. Flowers borne in 1-16(-20)
racemes, raceme axis cadcdanant or puberulent, nonglandular trichomes acs up
to 0.15 mm ia ng, rarely to 0.20 mm; sterile floral at usually not present
flowers; — (13-)14-28 mm long, tightly to loosely spaced, adjacent calyces over-
lapping or not. Calyx tube at anthesis 2.3-6(-7) mm long, lobes (0.6-)0.8-2.2(-3) m

ng; calyx at fruit hie esis 4-9.5(-11) mm long. Nutlets 2.1-4.2 mm long. CHROMO.
SOME NUMBER: 2

meal SPECIMENS (native range of ssp. virginiana): Canada. Manitoba. Bran-
don, Stevenson 552 Dao); Morden, Scoggan 11486 (GH, MIN, ALTA). Ontario. We
Co.: P chr as. Glowenke 2926 (BH, PENN). York Co.: Center Island, Toro a Watson
190 (rer), Quebec. Montmorency Co.: Ile d’Orleans, 2 mi E of Ste-Pétronille, Perras
‘1-758 (os, SCs). Quebec Co.: Cap Rouge, Marie-Victorin 21570 (DAO, GH, PH). United
States. Illinois. Peoria Co.: N of Averyville, Peoria, Chase 3690 (11). Indiana. Allen
Co.: § of Fort Wayne, Dian 2552 (IND). Hamilton Co.: just N of Noblesville, Deam

88 PHILIP D. CANTINO

12141 (IND). Iowa. Cedar Co.: 0.5 mi SE of Rochester, Fay 1283 (1a). Johnson Co.:
Lake McBride State Park, 12 Aug 1956, Pfeifer s.n. (1A). Kansas. Douglas Co.: 2 mi
S of Lone Star, McGregor 615 (KANU). Kentucky. Jefferson Co.: Islands at Falls of
Ohio, 1840, Short s.n. (cM, PH). Maryland. Montgomery Co.: Stubblefield Falls, Killip

Aes k. Essex Co.: Lake
Champlain, Mullen Bay, Muenscher, Manning & Maguire 503 (Ncu, CU). North Da-
kota. McHenry Co.: Towner, 12 Aug 1908, Lunell s.n. (Us, MIN, NY, PH, MO). Pembina
Co. omas, Larson 3522 (KANU). Ohio. Ottawa Co.: Bay Twp., Winous Point,
Lowden 766 (0s). Van Wert Co.: 3.5 mi N of Delphos, Stuckey 6327 (os). Pennsyl-
vania. Erie Co.: Presque Isle, 12 Aug 1879, Guttenberg s.n. (cM). Lycomin
branch of Susquehanna River opposite Jersey Shore, Westerfeld ¢ Wahl 3009 (oKL,
SMU, FSU, DAO, WVA, DUKE, NCU). South Dakota. Vermillion (near Big Stone Lake), John-
son 90 (IA, NY, GH, MICH, WTU). Vermont. Chittenden Co.: Burlington, Lake Champlain,
Charette 695 (DAO, VT, FsU, SMU). West Virginia. Barbour Co.: Arden, 22 June 1973,
Bush s.n. (wva). Preston Co.: Erwin, 23 July 1959, Bartholomew & Vail s.n. (US, LL,
MO, SMU, FSU, DAO, MSC, NY, GH, DUKE, TENN, NCU, GA). Wisconsin. Buffalo Co.: Mississippi
River bottoms opposite Wabasha, Minnesota, Fassett ¢& Hotchkiss 3456 (MIN, GH).

Fig. 17): river and stream banks, lake and bay shores,
drainage ditches, marshes, estuaries, and other moist sites, from Quebec to Manitoba,
south to northeastern Kansas, southern Illinois, northern Tennessee, and eastern Vir-

nia

FLOWERING: mid-June through mid-October.

The type specimen of Dracocephalum virginianum L. appears to
be a cultivated plant. Epling (1929), who selected one of the two
specimens in the Linnean Herbarium, believed it to be of garden
origin. After examining a photograph of the specimen, I have no rea-
son to disagree with his interpretation. Although the species is based
on a cultivated plant, the application of the specific epithet to one
of the two native subspecies does not present a problem.

Physostegia virginiana was cultivated in Europe at least as early as
1674 (Boccone, 1674), and the cultivars grown there at the time of
Linnaeus probably were the offspring of more than one introduction.
It seems reasonable to assume that most if not all of the introductions
before 1753, when Linnaeus described the species, originated from
the eastern seaboard of North America near the few major settlements
within the range of P. virginiana. The more southerly subspecies 0C-
curs from western Virginia and central North Carolina to northeastern
Mexico, north to central Missouri, northern Illinois, and western Ohio
(Figs. 17 and 18). The northern subspecies occurs naturally within 60
miles of Philadelphia, within 40 miles of Baltimore, and along the St.
Lawrence River in the vicinity of both Montreal and Quebec City-
Because the latter subspecies probably gave rise to the plants culti-
vated in 18th century Europe, of which the lectotype of Dracoceph-
alum virginianum is a representative, it is the most reasonable can-
didate to bear the specific epithet.

Physostegia virginiana is the most widespread and variable species

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 89

Fic. 17. Distribution map of Physostegia virginiana. Ssp. virgin-
‘ana (closed circles); ssp. praemorsa (triangles); garden escapes (open
circles); uncertain subspecific affinities, probably mostly garden es-
Capes (stars).

in the genus. Infraspecific taxa have been described in piecemeal fash-
ion, but Boivin (1966) has made the only attempt to apportion the
variation within the entire species into a limited number of precisely
defined varieties. While his approach is laudable, his attempt to de-
fine his eight varieties so that they are mutually exclusive has pro-
duced a collection of rather artificial taxa. Most of the varieties Boivin
Tecognizes correspond to real morphogeographical entities, but the
sharply defined limits he provides are not a true reflection of the
natural situation, where there is considerable morphological overlap.
Illustrative of the problem is the degree of intergradation that exists
een two varieties of Physostegia virginiana that are among the
Most distinctive and widely accepted—var. granulosa and var. for-
mosior. The former ranges from the St. Lawrence River and Lake
Champlain to West Virginia and northern Tennessee, and the latter

90 PHILIP D. CANTINO

/
{
|
pris
YW

|
|

Zz
Pg
Ys

|
|
\\

¢ Y ssp. virginiana
S& SSP. praemorsa

< ncertain affinities
\ jets. b
sf ~ * y
Y \
\ i “} zone of b ic intergradation
“i Bs sympatry without int i

2 ‘ergradation i

Fic. 18. Subspecific Ssympatry and intergradation in Physostegia

virginiana. Plants of “uncertain affinities” are probably garden escapes
(see text). Definite garden escapes have been omitted.

from Manitoba to Michigan, south to northeastern Kansas and south-
central Ohio. Not only is there overlap in all the characters that dis-
tinguish the two varieties when they are delimited in such a way that
they are strictly allopatric, but they intergrade through a series of
populations connecting their geographic ranges. The plants that fre-
oe the shores of Lake Erie and Lake Ontario resemble var. for-
mosior in some characters, var. granulosa in others, and are inter-
mediate in others yet (Table 20). Specimens from the western end 0
La e Erie, nearest to the range of var. formosior, resemble that va-
riety more closely than do those from Lake Ontario or the eastern
end of Lake Erie. My observations of populations at both ends of
Lake Erie suggest that intrapopulational variation is not great and that
they are probably not hybrid swarms. The intergradation between the

two varieties is more plausibly a case of simple clinal variation rather
than the result of hybridization.

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 91

TABLE 20. INTERGRADATION OF PHYSOSTEGIA VIRGINIANA VAR. “GRANULOSA”

P. virginiana
var. “granulosa”

AND VAR. “FORMOSIOR”

P. virginiana of
Lake Erie, Ontario

P. virginiana
var. “formosior”

Plants 2-10 dm high
Widest leaf 0.5—1.8(—2.2)
cm wide

Leaf serrations frequently
confined to apical half
of blade

Apices of leaf teeth blunt
or sharp

Some stem leaves usually
widest above middle of
blade

Flowers loosely or tightly
spaced

Leaf length of second pair
below terminal raceme
+ length of internode
above = 0.3-1.8(-2. 1)

Fruiting calyx (5-)6-9.5
(-11) mm long

Length + width of fruit-
ing calyx = 1-2

Plants 4-14 dm high
Widest leaf 0.5-2.0 cm
wide

Leaf serrations confined to
apical half of blade, or
not

Apices of leaf teeth sharp

Stem leaves widest above
middle of blade, or not

Flowers tightly spaced

Leaf length of second pair
below terminal raceme
+ length of internode
above = (0.7—)0.9-2.3
(-3.1)

Fruiting calyx 5—8(-10) mm
long

Length + width of fruit-
ing calyx = 0.8-1.7

Plants 3-16 dm high

Widest leaf (0.5—)1.2—4.2
(-5.5) cm wide

Leaf serrations rarely con-
fined to apical half of
blade

Apices of leaf teeth sharp

Stem leaves rarely widest

above middle of blade
Flowers tightly spaced

Leaf length of second pair
below terminal raceme+
length of internode
above = 1.2-3.8

Fruiting calyx (4—-)4.5-7
mm long

Length + width of fruit-
ing calyx = 0.7-1.2

There is little point in formally recognizing varieties that intergrade

= completely as do var. granulosa and var. formosior. However, there
Is too much geographic variation in the species to ignore completely.
Instead, I have recognized two wide-ranging subspecies. They are
easily distinguished throughout most of their range but intergrade in
one of their two regions of sympatry (Fig. 18). Because the two sub-
Species do not intergrade in the other (larger) zone of sympatry, and

cause the number of cases of intergradation is much less than when
4 system of numerous varieties is used, the proposed classification
Involving two subspecies is preferable on both theoretical and prag-
Matic grounds.

The best distinction between the subspecies is the form of the rhi-
“me, P hysostegia virginiana ssp. virginiana nearly always produces
elongate, horizontal rhizomes (Fig. la), each terminated by an over-
wintering bud. Subspecies praemorsa usually lacks horizontal rhi-
zomes; its perennating buds are borne either directly on the rootstock
(Fig. 1b) or at the ends of short, vertical rhizomes that branch off

92 PHILIP D. CANTINO

from the lower portion of the rootstock and ascend directly to the
surface.

Exceptions occur in both subspecies. Depauperate individuals of
ssp. virginiana may not produce any horizontal rhizomes. I have seen
few examples of this in the field, but it occurs commonly enough in
the greenhouse. In addition, there is a population of plants in Hardy
Co., West Virginia (Cantino 908, GH) that resemble the local race of
ssp. virginiana but lack horizontal rhizomes. Throughout much of the
range of ssp. praemorsa, horizontal rhizomes are never produced. It
is primarily in the eastern half of Tennessee and adjacent parts of
Georgia, Alabama, and North Carolina that members of some popu-
lations that otherwise appear to be ssp. praemorsa possess horizontal
rhizomes. The two subspecies intergrade in this region, a situation
that is discussed below.

Garden transplant experiments indicate that intraspecific variation
in the form of the rhizome is not merely a plastic response to edaphic
conditions. Thirty plants from 10 populations of ssp. virginiana and
70 plants from 18 populations of ssp. praemorsa were grown together
in the garden. With the exception of a few depauperate individuals
of ssp. virginiana, they all produced the same rhizome morphology
as that produced in their natural habitat.

A second character correlated with the short, vertical rhizome of
ssp. praemorsa is the production of empty floral bracts below the
flowers in response to a long photoperiod (see p. 25). There are two
requirements that must be satisfied if the empty bracts are to be
produced: first, the plant must have the genetic potential to respond
to the proper photoperiod; and second, the proper photoperiod must
be provided. Because the latter requirement is not always satisfied
under natural conditions, an observed absence of empty bracts is of
no taxonomic significance, but the presence of the bracts is a reason-
ably good diagnostic character for ssp. praemorsa.

The determination of the degree of correlation between the two
defining characters of Ssp. praemorsa requires experimental manipU-
ation. Nineteen plants (from 9 populations) that lacked horizontal rhi-
zomes and 28 plants (from 8 populations) that possessed them were
grown under a 16-hour photoperiod (see p. 25 for further details of
the study). Seventeen of the 19 plants (89% ) that lacked horizontal
rhizomes produced empty bracts, whereas 27 of the 28 plants with
horizontal rhizomes (96%) failed to produce empty bracts. Although
the sample size was small, the plants tested were taken from widely
scattered populations, so the results should be reasonably represen-
tative of the degree of correlation that exists in nature. Only speci-
mens from natural populations were included in this study; empty
bract production is quite common in the cultivated forms of Phys0-

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 93

stegia virginiana, all of which have horizontal rhizomes.

e subspecies are sympatric in parts of Illinois, Indiana and Ohio
(Fig. 18), where they occupy somewhat different habitats, ssp. prae-
morsa most often growing in prairie vegetation or on open wooded
slopes, while ssp. virginiana is most frequently found in swampy areas
along rivers, lakes, and man-made ditches. In this region the two sub-
species differ in several additional characters that are diagnostic locally
but do not hold true throughout the range of the species (i.e., degree
of reduction of the upper stem leaves, leaf texture, and flower
length). They have been treated as distinct species in at least two
floristic works of the region (Deam, 1940; Jones, 1963), and hybrid-
ization between them is probably infrequent because of their differing
habitats. There are a number of collections from the Chicago area that
may be hybrids, but outside of that area there is very little evidence
of intersubspecific hybridization. If ecological isolation is indeed pre-
venting hybridization, it would not be surprising to see a breakdown
of isolation in areas where human disturbance is extensive.

There is a smaller zone of subspecific sympatry in southeastern
Kentucky and northeastern Tennessee (Fig. 18), where the two taxa
intergrade both ecologically and morphologically. Subspecies virgi-
niana is confined to streamsides in this area; ssp. praemorsa occurs
along streams occasionally but is found more frequently in open
woods, limestone barrens, and prairie-like sites. The foliar and floral
characteristics that distinguish the two subspecies farther north do not
hold true in Kentucky and Tennessee, and even the primary char-
acters, rhizome morphology and the potential to produce empty floral
bracts, are less well correlated in this region than elsewhere.

The most interesting aspect of the intraspecific variation pattern of
Physostegia virginiana is the existence of two separate regions of sub-
specific sympatry, in one of which the subspecies intergrade, while
in the other they remain ecologically and morphologically distinct.
This situation is most adequately explained by a hypothesis of circular
overlap (Mayr, 1963). Subspecies praemorsa and “var. formosior” of
‘SP. virginiana, the two infraspecific taxa of Physostegia virginiana
that co-occur without intergradation in the Lower Great Lakes States,
are connected by a chain of intermediates. “Variety formosior” inter-
Srades with the eastern race of ssp. virginiana (“var. granulosa’) via
4 group of morphologically intermediate populations along Lake Erie
(Table 20 and related text). “Variety granulosa,” in turn, intergrades
with SSP. praemorsa in eastern Tennessee and parts of adjacent states
(the “zone of subspecific intergradation” in Fig. 18). This pattern of
morphological variation can be explained by hypothesizing a diver-
Sence of ssp. virginiana from ssp. praemorsa (or the divergence of
both subspecies from a common ancestor) in or near the “zone of

94 PHILIP D. CANTINO

subspecific intergradation,” followed by their northward migration
along separate pathways, concomitant with continued morphological
and ecological divergence. By the time they came into secondary con-
tact in the Lower Great Lakes States, they had attained both ecolog-
ical isolation and a high degree of morphological distinctness. The
hypothesized route of the migration and its timing in relation to the
Wisconsin glaciation are discussed elsewhere (Cantino, 1980: 256-268).
The frequent naturalization of cultivated Physostegia virginiana pre-
sents a problem when one is identifying seemingly wild specimens
to subspecies. The cultivated forms apparently always produce the
horizontal rhizomes characteristic of native ssp. virginiana, but some
cultivars bear a closer resemblance to ssp. praemorsa in other aspects
of their morphology. Although the cultivated specimen upon which
Linnaeus based the species was almost certainly a member of the sub-
species that, for this reason, must be referred to as ssp. virginiana,
there is no reason to assume that all modern cultivars are purebred
descendants of the forms cultivated in 18th century Europe. Some of
the showier forms of ssp. praemorsa may have been collected by hor-
ticulturalists and included in breeding programs. This would explain
e much more frequent production of empty bracts in cultivated
forms of P. virginiana than in wild forms of ssp. virginiana, as well as
the resemblance of some cultivars to ssp. praemorsa in above-ground
vegetative morphology. Because the genetic background of modern
cultivars is unknown, they cannot reasonably be placed in either sub-
species and should not be identified below the species level.
Physostegia virginiana commonly escapes from cultivation, and it is
probable that even some populations in undisturbed sites owe their
origin to the escape of garden plants. The recognition of garden es-
capes has presented the most serious problem in parts of the south-
eastern United States, where there are no indisputably native popu-
lations with which to compare. Subspecies praemorsa occurs commonly
across the northern part of Alabama, Georgia, and South Carolina,
but I have seen only a scattering of specimens of Physostegia virgin-
iana from south of the Appalachian section of these states and from
neighboring Mississippi and Florida (“uncertain affinities” in Fig. 18)-
Many have horizontal rhizomes and the remainder lack underground
parts. Because of the relative rarity of collections from that region
and the association of many of them with disturbed habitats, I suspect
that most such collections represent garden escapes. However, some
of the collections come from seemingly natural habitats, and one of
these (Thorne 5395, CU) is unlike any cultivated form I have seen.
There is agreement among authors of floristic works that most rec-
ords of Physostegia virginiana from New England, eastern New York,
New Jersey, and eastern Pennsylvania represent escapes from culti-

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 95

vation (e.g., Britton, 1889; Graves, et al., 1910; Taylor, 1915; Hoff-
man, 1922; Wagner, 1943; Schaeffer, 1949; Fernald, 1950; Seymour,
1969; Domville & Dunbar, 1970; Harris, 1975). This is supported by
herbarium label data indicating that the majority of the specimens
from this region were collected along roads, railroads, and in other
disturbed areas. Physostegia virginiana is absent from many of the
older floras dealing with the region, substantiating this view.

The recent spread of Physostegia virginiana in the Philadelphia area
is particularly well documented. There is no mention of it in Barton’s
(1818) listing of the indigenous and naturalized plants within a 10-mile
radius of Philadelphia. A more recent flora of Philadelphia and vicin-
ity (Keller & Brown, 1905), which covers much of southeastern Penn-
sylvania, records the species only from the shores of the Susquehanna
River, where it is apparently native. In Bucks County, north of Phil-
adelphia, P. virginiana was unknown in 1876 (Thomas & Moyer, 1876)
and known from but a single locality in 1932 (Benner, 1932). It has
been collected from at least four additional sites in Bucks County
since then, as well as from numerous localities in nearby Berks, Le-
high, Montgomery, and Philadelphia Counties.

A second example is provided by the Boston area, where Physo-
stegia virginiana is somewhat less common than in southeastern Penn-
sylvania. The species is not mentioned in early 19th century floristic
works dealing with Boston (Bigelow, 1824) or Massachusetts (Hitch-
cock, 1835; Dewey, 1840). In Middlesex County, just northwest of
Boston, Dame and Collins (1888) recorded it from only one locality,
where it was “probably an escape from cultivation.” I have seen more
recent specimens from three other sites in Middlesex County, and it
now occurs sporadically in most of the counties in eastern Massachu-
Setts.

On the basis of a similar historical approach, it can be stated with
near certainty that Physostegia virginiana is native nowhere in New
Jersey, southeastern New York, Pennsylvania east of the Susquehanna
River, or New England, except along the shores of Lake Champlain
and possibly the Kennebec and Penobscot Rivers in Maine. Less cer-
tain is the status of collections from central New York, central and
western Pennsylvania, and eastern Ohio, but it is likely that most rec-
ords from these areas also represent garden escapes. In Ohio, New-
berry (1860) recorded the species from the central and western parts
of the state only. Transeau and Williams (1929) mapped it as occurring
in eastern Ohio, but only in the counties adjoining Lake Erie. The
scattering of more recent collections from nonlacustrine eastern Ohio
are probably all escapes from cultivation. In New York, House (1924)
states that P. virginiana is native from Lake Champlain and Oneida
County southward and westward, but 19th century floras dealing with

96 PHILIP D. CANTINO

central and western New York, except the Buffalo-Niagara area where
it is apparently native, either do not mention the species (Paine, 1865;
Beckwith & Macauley, 1896; Clute, 1898) or indicate that it is known
only as an escape from cultivation (Burgess, 1877; Dudley, 1886). In
Pennsylvania west of the Susquehanna River, it appears to be native
only along the Allegheny-Monongahela-Ohio river system, including
several tributaries.

The ostensibly greater abundance of garden escapes in the north-
eastern states than in the rest of the range of the species is very likely
due to the preference of collectors for uncommon or unfamiliar plants.
Where Physostegia virginiana is native and abundant, the occasional
naturalized garden plant will go unnoticed, whereas in the northeast
and parts of the southeast, where native P. virginiana is rare or ab-
sent, garden escapes are much more likely to be collected as a nov-
elty.
The preparation of the distribution map (Fig. 17) has been difficult
because the structure of the greatest diagnostic value in distinguishing
the two native subspecies—the rhizome—is missing from many her-
barium specimens, and because naturalized garden forms of the spe-
cies are frequently intermediate between the native subspecies. The
usual problem is not one of determining to which of the two native
subspecies a particular specimen belongs, but whether it represents
a native population or a garden escape.

There is no character that is universally useful in distinguishing
native plants from garden escapes. However, there are character
states frequent enough among the cultivated forms to be of use in
certain limited regions, where these traits are absent from native pop-
ulations. The elongate, horizontal rhizome of the cultivated forms fa-
cilitates the recognition of garden escapes in regions where only ssp.
praemorsa is native. The frequent presence of sterile bracts below the
flowers in garden plants helps to distinguish them from native ssp-
virginiana. The leaves of cultivated plants are usually sharply an
deeply serrate, in contrast to the shallowly and bluntly toothed leaves
of the Appalachian race of ssp. virginiana. Cultivated forms often have
five or more racemes, whereas Ssp. praemorsa and the Appalachian
race of ssp. virginiana usually do not have more than three. The up-
per leaves of garden plants are usually not much smaller than the
middle leaves, while those of ssp. praemorsa and the Appalachian
race of ssp. virginiana are often much reduced in size. Native ssp-
virginiana from Wisconsin and Iowa northwestward appears similar to
some of the frequently encountered cultivars, but the native plants
tend to have shorter calyx tubes and longer trichomes on the axis of
» . ee ae most garden forms. If the calyx tube is 4 mm oF

axis of the raceme includes some nonglandular tri-

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 97

chomes more than 0.1 mm long, then the specimen is almost certainly
native. However, the lack of trichomes over 0.1 mm long or the pos-
session of a longer calyx tube does not necessarily indicate that the
plant is an escape. Because of their unreliability, these characters
must be used with caution.

The proposed infraspecific classification of Physostegia virginiana
is not without problems. Although the majority of the collections from
most parts of its range can be assigned to a subspecies, there are a
number of specimens from the southeastern United States that can-
not. Indeed, if the southern Appalachian “zone of subspecific inter-
gradation” is, in fact, occupied by the ancestral complex from which
the two subspecies diverged, it may not even be reasonable to try to
assign specimens from that area to one or the other of the subspecies.
The subspecific classification may be similarly inapplicable in Ala-
bama, Mississippi, and Georgia if the specimens from there are relics
of the Wisconsin-age range of the ancestral complex.

There is no perfect solution when one is attempting to subdivide
a species in which there is circular overlap. If no infraspecific taxa
are recognized, the classification will be rejected by those familiar
with the region of overlap, where there will be two morphologically
and ecologically distinct taxa going by the same name. If the two
overlapping “arms” of the species are given formal recognition, as I

ave chosen to do, the classification will be satisfactory in the region
of overlap but problematical in the source area where the two lines
diverged. I can only offer the pragmatic suggestion that, in the case
of Physostegia virginiana, the two subspecies be recognized where
they are distinct but that no effort be made to distinguish them in
the southern Appalachian region where they intergrade.

12b. Physostegia virginiana (L.) Benth.
ssp. praemorsa (Shinners) Cantino comb. nov.

Physostegia praemorsa Siapncr Field & Lab. 19: 166. 1951. HOLOTYPE: Texas, Fannin
Co. Ma 6 mi W of Honey Grove, chalk outcrop, larger plants from ditch bank,
sm cay ri dry chalk, "6. X-1949, Shinners 11980 (SMU). ISOTYPE: SMU

D rctephalum variegatum Ventenat, Descr. oy No om Jard. C wii tab. 44. 1801. Su-

uous name (Prasium incarnatum Walt. cited in synony

P hysostegia variegata (Vent.) Benth. Bot. Reg. sub tab. ae 1899. Not validly pub-

sh ecause the combination is not explicitly propose

preset virginiana forma candida Benke, Am. Midl. Nat. 16: 423. 1935. HOLOTYPE:
Illinois, Cook Co., Arlington Heights, 95-VIII- 1934, Benke 5681 (F). ISOTYPES: GH,

Physostegia oo Shinners, Field & Lab. 24: 1956. HOLOTYPE: Louisiana, Cal-
eu Parish, clay ditch bank on east sae of ‘ake Charles, 9-X-1955, Shinners

be MU). ISOTYPES: GH, ae DUKE, MIC _ BH, FSU, SMU.
Physostegia Lod oe var. re a Boivin, Nae nay 93: 572. 1966. HOLOTYPE: In-
diana, Co., prairi area along R.R., 1/2 mile east of Lake

oaks oh:
Cicott, 36 IX. 1936, Friesner 10132 ‘aaa ISOTYPES: NY, GA, OKL, SMU, CU, ILL

98 PHILIP D. CANTINO

Rhizome short, unbranched, and strictly vertical, or (infrequently) branching to pro-
duce elongate, horizontal secondary rhizomes up to cm long. Central stem leaves
0.2-3(-4.4) cm wide, elliptic to oblanceolate, varying to obovate, ovate, spatulate,
lanceolate, or linear, the margin sharply serrate. Flowers borne in 1-8 racemes, ra-
ceme axis puberulent, nonglandular trichomes usually not more than 0.1 mm long
(rarely to 0.13 mm); sterile floral bracts (up to 40 pairs) frequently present below the

owers; flowers (16—-)18—37 mm long, usually tightly packed, adjacent calyces at an-
thesis usually overlapping a quarter to half or more of their lengths. Calyx tube at
anthesis (3.5-)4—8 mm long, lobes (0.8-)I-3 mm long; calyx at fruit maturity 6-10
(-11) mm long. Nutlets (2.5-)2.9-3.8(-4) mm long. CHROMOSOME NUMBER: 2n = 38

REPRESENTATIVE SPECIMENS: Mexico. Coahuila. Mcpo. de Muzquiz, Rincon de Maria,
Wendt et al. 1265 (cu). United States. Alabama. Colbert Co.: S side of Littleville,
Kral 44026 (vps). Morgan Co.: Lacey's Springs, Kral 48532 (vps). Arkansas. Lonoke
Co.: Grand Prairie, Demaree 22473 (sMU, MIN, NY. MO). Prairie Co.: Hazen, Demaree
54730 (SMU, NCU). Illinois. Vermilion Co.: Fithian, Gates 2170 (MICH). Indiana. Porter
Co.: 1 mi E of Crisman, Deam 21260 (IND). White Co.: 3.2 mi S of Reynolds Center,
Webster & Webster 7129 (NCU, DUKE, MSC). Iowa. Muscatine Island, 6 Sep 1895, Ref-
fert s.n. (IA). Kentucky. Madison Co.: Big Hill, McFarland 30 (bu, F, GH, NY, PH, IND,
US, OKL, WVA, PENN, DUKE, TENN, MICH, MIN, SMU, MO, Wis). Louisiana. Calcasieu Parish:
Lake Charles, 3.1 mi S of McNeese campus, Thieret 27966 (Gu, FsU). Missouri. Benton

o.: 8 mi NE of Warsaw, Stephens 36317 (KANU). Wright Co.: 3 mi:N of Manes,
Steyermark 25084 (F). Nebraska. Richardson Co.: NE corner of sec. 33: RI5E, T3N,

ico. : rk of Big Canyon, E scarp
of Guadalupe Mts., Wendt & Lott 2126 (cH). North Carolina. Ashe Co.: Bluff Mt.,
Radford et al. (Bozeman et al. ) 45258 (NCU, GA, VDB, TENN, SIU, GH, NY, WVA, ‘MISS, VPI, FSU,
MIN, SMU, CM, TEX, LL, WIS, WTU, IND). Jackson Co.: 10 mi E of Cherokee, Correll &
Correll 22 3 (LL). Ohio. Adams Co.: 2 mi E of Lynx, Terrell 1034 (os). Marion Co.:
E of Marion, 28 Aug 1939, Fullerton s.n. (os). South Carolina. York Co.: just E of
York, Bell 10097 (ncu). Tennessee. Blount Co.: near Rich Gap, Greene 4054 (TENN).
Rhea Co.: tween Dayton and Pikeville, Rogers 44236 (VPI, TENN). Texas. ‘Jefferson
Co.: 9 mi W of Beaumont, Cory 50021 (Ny, GH, MICH, US, SMU. RM). Liberty Co.: 3.5
mi E of Moss Hill, Correll 34000 (LL). Virginia. Tazewell Co.: S side of US-19, 1 mi

PI).

0.:

). .
, including prairies, limestone glades and barrens, calcareous out-
crops, open w ands, stream margins, and roadside ditches.
FLOWERING: June through October, rarely to mid-December.

gree of reducti
stalked glands in the inflorescence, and the size and color of the flow-

i most distinctive race comprises a cluster of populations in
— — Louisiana and southeastern Texas, separated by 2
miles from their nearest consubspecific neighbor and by 300 miles

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 99

from the main body of the subspecies. Shinners (1956) based his Phy-
sostegia serotina on representatives of this race, recognizing their af-
finities to P. praemorsa Shinners, but stating that “P. serotina is a
larger plant with larger, much deeper-colored corollas which have a
more pronounced basal tube, the limb flaring well above the calyx
when fully expanded.” In addition, the race is characterized by having
rather narrow, crowded leaves and a great many empty bracts below
the flowers. All of these character states occur commonly elsewhere
in the subspecies, and in rare instances they are found in combination
in geographically distant populations that are very unlikely to form a
monophyletic group with “P. serotina Shinners.” For example, a few
collections from calcareous cedar glades in northern Alabama are very
similar to the type collection of P. serotina (e.g., Kral 44026, vps;
Kral 48532, vpB), as are some from the same habitat in Missouri. In
northern Alabama, I have seen plants that resemble the type of P.
serotina intermixed in populations with others that are of the usual
Appalachian form of P. virginiana ssp. praemorsa. Parallel evolution
within the subspecies is the most plausible explanation for this situ-
ation.

A second distinctive, but very local race occurs at two sites near
the Horsepasture River in Transylvania Co., North Carolina (Bannis-
ter & Anderson 298, DUKE; Hardin 2297, Fsu, GA; Rodgers 62330b,
DUKE; Cantino 946, GH). The rather short, broad leaves of these plants
are unusual in Physostegia virginiana but can be found in a few pop-
ulations of ssp. praemorsa in Polk Co., Arkansas (McWilliam 589, GH,
wis; Backholz 273, w1s). More unusual is the slightly clasping leaf base
of this variant (F ig. 4c), a trait that I have observed in only two other

TABLE 21. DISTINGUISHING CHARACTERISTICS OF PHYSOSTEGIA ANGUSTIFOLIA AND P. VIRGINIANA
. PRAEMORSA.

P. angustifolia P. virginiana ssp. praemorsa
Flowering April-late July Flowering (June-) July-December (Au-
gust-December in zone of sympatry)

Sterile bracts never present below flow- Sterile bracts frequently present below
ers flowers
Longest nonglandular trichomes on ra- Longest nonglandular trichomes on ra-
ceme axis 0.13—0.2(—0.25) mm long ceme axis not more than 0.1(—0.13) mm
long

Stalked glands never present on raceme Stalked glands frequently present on ra-
axis ceme axis

Nutlets 2.0-3.0(—3.5) mm long Nutlets (2.5-)2.9-3.8(-4) mm long
Leaf base frequently clasping stem Leaf base rarely clasping stem (never in
zone of sympatry)

100 PHILIP D. CANTINO

specimens of P. virginiana (Wilkinson s.n., Us, CU, from Mansfield,
Ohio).

A barely discernible race of Physostegia virginiana ssp. praemorsa
occurs in prairie habitats in Illinois, Indiana, and Ohio, and in a few
sites in the extreme eastern parts of Iowa and Missouri. The leaves
of these plants tend to be a bit narrower and the upper ones more
reduced than is usual in the subspecies. A similar form occurs in iso-
lated prairie patches in Arkansas and Tennessee. The strong resem-
blance of this prairie ecotype of P. virginiana ssp. praemorsa to P.
angustifolia has caused confusion in floristic works. The two taxa can
be reliably distinguished on the basis of the length of the trichomes
on the axis of the inflorescence (see discussion of this character under
P. angustifolia). A number of other distinguishing characters, Bape
less reliable but more readily observable, are listed in Table

EXCLUDED OR DUBIOUS NAMES

Physostegia Lo. Benth. Lab. Gen. et Sp. 505. 1834. =Brazoria truncata (Benth.)
elm. & Gra
ips pont rahe ae var. denticulata Chapm. Fl. South. U.S. 325. 1860. Not P. v
gin ar. denticulata (Ait.) Gra ay, 1878. Apparently not bors on race eohiale
den ade ‘oi Ait.; type unknown; description insufficient to determine affinities.
Pune incarnatum bsdanet er, Fl. Carol. 165. 1788. Pr obably based on a member 0:
Physostegia, but eile rte oo aa to determine specific affinities. Blake's
lectotypificatio on of the name wit pecimen of Physostegia virginiana in the
erbarium” ilbeead be failed because the specimen disagrees with the
device, (Cantino, 1981b).

LITERATURE CITED

ADANSON, M. 1763. Familles des Plantes. 2 vols.
ANDERSON, W. R. 1972. A monograph of the ine mea (Rubiaceae). Mem. N. Y.
t 128.

Bot. Gard.
ArRoyo, M. T. K. & P. H. Raven. 1975. The evolution of subdioecy in morphologically
gynodioecious species of Fuchsia sect. Encliandra (Onagraceae). Evolution 29:
-51l1.
Avensu, E. S. & R. A.’ DeFrurrs, 1978. Endangered and Threatened Plants of the
—— States. Rett Institution and World Wildlife Fund, Inc., Washing-
403

~ D

aoe. L. H. 1949. Manes! of — igre Macmillan Co., New York. 1116 pp.

BaILEy, W. W. 1882. Notulae exiguae i Gay 7:

BAILLON, H. 1891. Histoire des Plantes. val n, Monographie des Labiées, Verbénacées,
Ericacées, et Ilicacées. Hachette & C , Paris.

BALL, - » 1972. Melittis, oe 143 in T. C. ‘Tutin; . al. (eds.), Flora Europaea, vol.

mbridge Univ. Pre
Serbs ay J. H. 1938. The ie Sa of Doctor Small. Jour. N. Y. Bot. Gard. 39: 73-

Banton, pat C. 1818. Compendium Florae Philadelphicae. Carey & Son, Philadel-
phi Pp.

BECKwitH, F. & M. MAcnuLEr. 1896 Plants of Monroe County, New York, and
adjacent teritory. Prec ochester Acad. Sci. 3: 1-1

BENNER, W. M. 1932, a bit or Bucks Cc P. ] Publ. by author,
Philadelphia. ren seo enaiey) vania. Pu ig

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 101

BENTHAM, fe 1832-1836. Labiatarum Genera et Species. Ridgway & Sons, London.
783 p

"184 18. Labiatae, pp. geod) in A. de Candolle, Prodromus Systematis Naturalis
Regni Vegetabilis, Vol.
. Labiatae, pp. 60 "1223 in Bentham & Hooker, Genera Plantarum.
Vol. 2. Reeve & Co., Lon
BIGELOW, J. 1824. Florula Boatoainen Cummings, Hilliard & Co., Boston. 423 pp.
sage AL H. 1909. ae nutrition of the embryo sac and embryo in certain Labiatae.
s Univ. Sci: 7-83.
a nag P 674. ‘alii et Deseriptioes Rariorum Plantarum Siciliae, Melitae, Gal-
liae, et Italiae. Oxford.
Boivin, B. 1966. Les sno pe Physostegia a aati Nat. oo 93: 571-575.
BREYNE, J. 1680. Prodromous Fasciculi Rariorum Plan
BRIQUET, J. 1895-1896. La PP. ae in Eudler "be neg ‘Die Natirlichen
Pflanzenfamilien. Teil 4
Britton, N. L. =, oo tg Sia Found in New Jersey. John Murphy Co.,
Trenton, NJ. 64
———. 1901 Maen of the Flora of the Northern States and Canada. Holt & Co.,
New York. 1080
& A. Brown. 1898. An Illustrated F lora of the Northern United States, Canada
and the British Possessions. Vol. 3. Scribner's Sons, New York.
1913. An Biase Flora of the Northern a States, Canada
and the British Possessions. 2n Scribner's Sons, New Y
Sices E. S. 1877. The Chautau a Flora. Clinton, NY. 37 p
CAMPBELL, R. S. 1955. Vegetational changes and management in Mie cutover longleaf
pine-slash wages area of the Gulf Coast. Ecology 36: 29-34.
CanTINO, P. 1979. Physostegia godfreyi (Lamiaceae), a new species from northern
Florida. iddes 81: 409-417.
. 1980. The Le pat and Evolution of the Genus Physostegia (Labiatae).
-D. thesis, Harvard University. 454 pp.
. 1981a. a. Change of status be Pe 4 agate var. ledinghamii (Labiatae)
and evidence for vs id o . Rhodora
981b. The Phascs tegia tires: lepton dent problem: taxonomic
and nomenclatural clarificatio on. Rhodora 83:
Cio. W. N. 1898. The Flora of the Upper Gauhan and its Tributaries. Clute
& Co., Binghamton, NY. 142 pp.
CorrELL, D. S. & M. cand bec Lelie Manual y the Vascular Plants of Texas.
Texas Research raaast tion, Ren 81 p
COULTER, J. M. 1882. Some notes on , Phisoeeea cirginiana. Bot. Gaz. 7: 111-112.
CULLEN, J. 1968. Botanical problems of numerical taxonomy, pp. 175-183 in V. H.
Heywood Poy. Modern Methods in Plant ae rae ._ Academic Press, London.

ME, L. L. & F. S. Coins. 1888. one of Middlesex County, lie se Mid-
dlesex Institute, Malden, MA :
Darwin, C: 1897 i? ye oa Bish ra Flowers on Plants of the Same Species.

Appleton & Co., p-
sta : a ier “Flore ne dake. W. C. wn Co., Dubuque, Iowa. pp
Dea 0. Flora of Indiana. ati . eet of Conservation, Te lis

we
DELPINO, F. 1868. Ulteriori Osservazioni e cone Sulla Dicogamia Nel Regno
G. Bernar Mila:

egetale. . G. Mila:

Dewey, C. 1840. Reports on the Aes Plants and on the Quadrupeds of M
sachusetts. Comm. Zool. & Bot. Survey of the State. Folsom, Wells & ice.
Cambridge, MA. 277 pp. é :

Divcuer, D. L. 1974. Approaches to the identification of Angiosperm leaf remains.

Domvitte, M. & H. F. DunsaR. 1970. The binge 4 Ulster County, New York. John

urroughs ee Hist. Soc. Bull. No. 8.
DupLey, gee 886. The Cayuga Flora. oa eral Univ. (Sci.) 2: 1-132.

102 PHILIP D. CANTINO

Dupree, A. H. 1968. Asa Gray. Harvard Univ. Press, Cambridge, MA. 505
DuRtETz, G. E. 1930. The fundamental units of biological taxonomy. Sida Bot.
Tidskr. 333-428.
EL-GazzarR & L. Watson. 1970. A taxonomic study of Labiatae and related genera.
New Pith 69: 451-4
me eee S. 1838. Genera Plantarum Secundum Ordines Naturales Disposita. Fr.
eck, Vienna. 1483 pp.
Perse C. 1929. Notes on the Linnean types of American Labiatae. Jour. Bot. 67:
1-12.

sogegneer i G. 1945. Pollen morphology and fe pope IV. Labiatae, Verbenaceae,
vicenniaceae. Svensk Bot. Tidskr. 285.
rain J. S. 1967. The meaning of Giignke, alt taxonomic procedure. Syst. Zool.
16:

_ 1970. Methods for computing Wagner Trees. Syst. Zool. 19: 83-92.
1972. Eiticasting phylogenetic trees from distance matrices. Amer. Nat. 106:

64
FASsETT, N. Cc. 1940. A Manual of Aquatic Plants. McGraw-Hill Co., New York. 382

pp.
FEporov, A. A. (ed.) 1969. Khromosomnye Chisla Tsvetkovykh Rastenii. Nae
Numbers of Flowering Plants.) Acad. of Sci. of USSR, Komarov Bot. , Len-

ingra
FERNALD, M. L. 1943. Virginian botanizing under restrictions. Rhodora 45: 445-480.
———. 1950. Gray’s Manual of Botany, 8th ed. American Book Co., New York. 1632

Pp. ,
Foerste, A. F. 1885. The fertilization of Physostegia virginiana. Amer. Nat. 19:
168—

GLEason, H. A. 1952. The New Britt tton and Brown Illustrated Flora of the North-
—- eae: States and Adjacent Canada. New York Botanical Garden, Bronx,
Y. 3 volum

noice. 1963. Manual of Vascular Plants of Northeastern United
States and Adjacent Canada. Van Nostrand Co., Prine
Grant, V. 1971. Plant Speciation. Columbia Univ. Press. New

: ork.
Beene: J. E. 1967. Thomas Nuttall, Naturalist. Harvard Ate ae TRE cae
p-

Graves, C. E. H. Eames, C. H. BISssELL, L. ANDREws, E. B. esr aes & C.

WEATHERBY, 1910. Ca eat e of the Flowering Plants anil Ferns of Cea a

fe ii Bisa — Gulrivanees. Bull. 14, Conn. Geol. & Nat. Hist. Survey, Hart-
or

Gray, A. ae q on of the
a. meg 710

Manual Vof Bs fisted of the ince United States. 5th ed. Ivison,
Phinney, Blikem Yor!

. 1878. Sraceteak Flora of a Mee Vol. 2, part 1. Ivison, Blakeman,
Ta ylor & ee Bye Yor

Harris, S. K, ae Flora of foc County, Massachusetts. Peabody Museum,

Botany of the Northern United States. Munroe &

nde einer Th d tik. Deut.
Zentralverlag, Berlin. eorie der phylogenetischen Systema

——.. 1966. a abn pene elas of patie Press, ee 263 p
Hitcucock, A. S. & tandard-species of Lin pioggatit era Me, ” Pha-
in prrcnads iy Briti ish Botanists (for 1930 Int.
“ . Ww = & ge ndon
mencock, C. L., A. Cronguist & M. Ownsey. 1959, Vascular Plants of the Pacific
se Pegg fe: Part 4: Ericaceae through Coruna ‘Gaby. of Washington Press,
attle

case age E. 1835, Catalogues Ad Animals and Plants of Massachusetts. J. 5-
. Adams, Amherst, MA.

rc
Sa)
(21
4
_

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 103

HOFFMAN, R. 1922. F ia of Berkshire County, Massachusetts. Proc. Boston Soc. Nat.

HOLMGREN, P. K. & KEUKEN. 1974. Index alc i 6th ed. Utrecht. 397 pp.
House, H. D. 1924. he List of the Fern ag Flowering Plants of New York
f New Yor

JONES, G. N. 1963. Flora of Illinois. 3rd ed. Univ. lf Notre Dame Press, Notre Dame,
IN. 401 vi
GD: - ULLER. 1955. Vascular Plants of Illinois. Univ. of Illinois Press,
Urbana. 593
JUNELL, S. “1937, Dic Samenentwicklung bei einigen Labiaten. Svensk Bot. Tidskr. 31:
78-110.

pend A. L. DE. ger Genera Plantarum. Herissant & Barrois, Paris. 498 pp.
Kearney, T. H. & R. H. PEEBLES. 1951. Arizona "Flora. Univ. of it lt Press,

pp.

KELLER, I. A. & S. Brown. 1905. — of the Flora of Philadelphia and Vicinity.
=e are Fee ‘Glob:

LAWRENCE, aie . M. 1951. i ro Vascular Plants. Macmillan Co., New York.

LEVIN, D. . 1978. The origin of isolating mechanisms 4 flowering plants, pp. 185-
n M. K. Hecht, et al. (eds.)., Evol. Biol., Vol. Plenum Press, —" York.
tf 1846. The Vegetable Kingdom. wd adbu ury & gir London
sean C. 1737. Genera Plantarum. oi
1753. Species Plantarum. Stockho
LINSBAUER, re 1940. Der “Starrkrampf” feo piceanre Die Gartenbauwissen-
schaft 1 press 357
LLoyp, D. Ce S. The sneeica of gynodioecy and androdioecy in Angiosperms.
Genetica ey ise
LUNDELL, C. L. 1959. ieee of Physostegia—I. New species and observations on
others. Wrightia 2: 4-12.
————. 1969. - pa of Texas. Vol. 2. Texas Research Foundation, Renner, 1X, 417

Pp. .
Mayr, E. 1963. Animal Species and Evolution. Harvard Univ. Press, Cambridge, MA.
797

McCuintock, E. 1949. A proposed retypification of Dracocephalum L. Leafl. West.
Bot. 5: 171-172.

MEEHAN, T. oa rege asc virginiana. Meehan’s Monthly 7

MEISNER, Ko FB, Plantarum Vascularium Genera. nyc santa nia, Leipzig.

Moencn, K. 1704, reeraer ag Plants Horti Botanici et Agri Marburgensis. Nova Li-

braria —— Marburg. pp. Fl
MonLENBROCK, R. 1963. Abed abi to an Illinois Flora I. The genus Physostegia.
Rh odora 65: Sgro
pOnpeN., R. 1669. sp whee Regius Blesensis. London A
MULLER, L. 1933. r den Bau und die Entwi icklung des Bewegungsmec anismus
n P, “earn einginiana Pl. Arch. Wissensch. Bot. 18:

Neweznar, J. S. 1860. sap of pe Flowering Plants and Forms of Ohio. R. Nev-
ins, Columbus. 41 i

Mra, T. 1821. Sas of Travels Into the Arkansa Territory During the Year 1619.

- Palmer, Philadelphia. 296 pp.

Onwi os ‘1965. Flora of Japan. Smithsonian Institution, Washington, “ogee bos “sg

et al R. 1978. Reproductive characters and taxonomy. oi ae 3 - :

PAINE, J. A , JR. 1865. Catalogue of plants found in Oneida County and oe n
ae of the Regents of the University of the State of New York, March 22,

1865. O pp. -
esl H. N. 1874. A List of Plants Collected 8 ~ "ead of Oquawka, Hen-

der Oquawka, I
‘ 876. ses a and og Cryptogamous Plants of

Hlinois, — Print, Oquawka, IL. 54 pp.

104 PHILIP D. CANTINO

unl F. W. 1936. Travels and scientific collections of Thomas Nuttall. Bartonia
18; 1-51.

1950. Historic botanical collections of the American he age oie Bae and
the Academy of Natural Sciences of Philadelphia. Proc. Amer. Phil. . 94: 137-

poteae A. E ABLES & °C: R: 1964. Manual of the Vascular Flora of
the Carotins, Un: of North ating reas Chapel Hill. 1183
KISON, J. R. Massey & C. R. BELL. 1974. Vascular Plant Systematics.
er @ ta New York. 891 pp.
Roker, H. W. 1960. Report of the committee for Spermatophyta: conservation of
ric Sse II. Taxon 9: 14-17.
Riri, 3 C. 6. The late-Quaternary bee; arom history of the Western Interior
anada. Clnat J. Bot. 54: 1793-1818.
ane C. 1888. Effect of the iad: on bees and flowers. Bot. 13: 33-34.
weer : A. 1932. Flora of the Prairies and Plains of Contial ett Atnerica: New
Bot. Gard., Bronx. NY. 969
bine peek RE. Fu 1949. The Vascular Flora of Northampton County, Pennsylvania.
-D. thesis, Univ. of Pennsylvania. 702 pp.
SCHNECK, J. 1891. Further notes on the mutilation of flowers by insects. Bot. Gaz.
16: 312-313.

SEALY, J. R. 1954. Proposal n
SEYMOUR, : C. 1969. The Pica oe se pea: a E. Tuttle Co., Rutland, VT.

596 p
SHARP, L EW. 1911. The embryo sac of Physostegia. Bot. Gaz. 52: 218
gape ion L. H. 1949. Physostegia Correllii (Lundell), comb. nov. cake 51; 120-

166 pee Two new Texas species of Physostegia (Labiatae). Field & Lab. 19:

: i.

56. Physostegia or tne Neespaual a new species from coastal Louisiana
Texas. Field & Lab. 24:

SMALL, J ‘) Rie Flora of the fre see United States. Publ. by the author, New

1933, MeriGl of the Southeastern Flora. Publ. by the author, New York. 1554

pp.
SmitH, H. M. 1965. More evolutionary terms. Syst. Zool.

SmiTH, S. J. 1945. Contributions to a Flora of Central ah ae NY State Mu-
seum Bull. no. 338, cr 74 p

- H. A. . SOKAL. 1973. Numerical Taxonomy. Freeman & Co., San
Francisco. hot

SOKAL, R. R. & F. 7 ROHLF. 1969. Biometry. Freeman & Co., San Francisco. 776

Pp.
STEBBINS, G. L. 19
nan on ading, M

F ering Planks Evolution Above the Species Level. Harvard Univ.
ess, 5, Cambri e.

Pre
STEVENS, ae 1950. Handle of North Dakota Plants. North Dakota Agr. College:
argo. P-
STEVENS, P. ~ 198 sie a revision
case J. Arn. : 117
1980b. eibete aabacy “ol character states. Ann Rev. Ecol. Syst. U:

(oe oo Evolution in Higher Plants. Addison-Wesley
Pp.

: ns Old World species of Calophyllum L. (Gut

STEYERMARK, PA 1963. Flora of Missouri. I 683
owa State Univ. Press, Ames. 1725 pp:
Tatton: B N. be " ‘Flora of oer vicinity of New York. Mem. NY Bot. Gard. 5:
YLOR & R. P. Brockman. Chromosome teins of some western Ca-
: Canad. er 44: 1093-1108,
Mead isiana flora. Castanea 36: 219-222.
P Moyer 1876. The meng ‘Bids are — of Bucks County,
ennsylvania. W W. W. H. Davis, Doylestown, PA.

A MONOGRAPH OF THE GENUS PHYSOSTEGIA (LABIATAE) 105

Tyo, Ph H. & A. LEVAN. 1950. The use of oxyquinoline in chromosome analysis. Anales
e la Estacion Expt. de Aula Dei 2: 21-64.
eg U. 1954. Brevi note di floristica piemontese. Nuov. Giorn. Bot. Ital. 61:
704—706.

TOURNEFORT, fs F. situps piace Rei Herbariae, Editio Altera. Pari

TRANSEAU, E. . WituiaMs. 1929. Distribution maps of certain ats in Ohio.
Ohi aa mtg 7. 181-217 (B Bull. 20).

ean G., B. Dommée & A. VALDEYRON. 1973. Gynodioecy: Another computer
simulation model. Amer. Nat. 107: 454—45)

VALENTINE, D. H. 1954. Seed incompatibility. Proc. 8th Int. Bot. Congr., sect. 9, pp.
170-171

VENTENAT, E. P. 1801. Descriptions des Plantes ee a et Peu Connues Cultivées
dans cai de J. M. Cels. Crapelet, Paris.
VILMORIN a L.F.L.)-ANDRIEUX & Co. 1866. Les Fleurs fe Pleine Terre. 2nd ed., Paris.

WAGNER, P R. 1943. The Flora to Schuylkill County, Pennsylvania. Ph.D. thesis,
Univ. of Pennsylvania. 230
Wacner, W. H., Jr. 1961. Problems in a faa ahi of ferns, pp. 841-844 in
Recent Advances in Botany. Univ. of Toronto Press, Toronto
1969. The construction 7 a attition @ pp. 67-90 in C. G. Sibley (chair-
man), lest Biology. Nat. Acad. Sci. Wash. Publ. 1692.
9 ~ philosophy of the Sa ee method of cladic-

tics. Sos. or 3:
Warp, D. B. (ed.). om. ‘aac ‘and renee Biota of Florida. Vol. 5: Plants. Univ.
Presses of Florida, Gainesville. 175
WATERMAN, A. H. 1960. Pollen grain sh of the Labiatae of Michigan. Webbia 15:
—415.
WELSH, S. L. lai tian 1973. Utah Plants. Tracheophyta. Brigham Young Univ.
Press, Provo.
Witey, E. — “980 Pee systematics and vicariance biogeography. Syst. Bot.
5: 194-2
WILLIs, J. C. gen A Dictionary of the Flowering Plants ta Ferns. 8th ed., revised
y H. K. Airy Shaw. Cambridge Univ. Press. 1245
easy B. E. & W. M. pevinmes 1976. State records Lee other recent noteworthy
i ae 119-121.

WricuT, H. E., . 1970. recut history of wes entral Plains, pp. 157-172 in
W. Dort, Jr. & J. K. Jones, eds.), Pleistocene ni tee t Environments of
the Central Great Plains. Univ. Press of Kansas, Lawren

SPECIES OF DRABA, LESQUERELLA AND SIBARA
(CRUCIFERAE)

REED C. ROLLINS

Vigorous field activity in western North America, particularly in
connection with rare and endangered species programs, has brought
to light new and interesting taxa in the family Cruciferae. Often sev-
eral trips to the populations of these new finds were necessary to
obtain appropriate material for an adequate study. I am indebted to
Barbara Williams of the Shasta-Trinity National Forest staff, Redding,
California, for making two trips back to the site where she first col-
lected specimens of Draba carnosula. She provided significant notes
on the plants as they occur in the field. My thanks to Dr. Thomas
Van Devender of the Arizona Natural Heritage Program, Tucson,
Arizona, who first sent material of Sibara grisea. Also, I am grateful
to Dr. Richard Spellenberg of New Mexico State University, Las
Cruces, New Mexico, and his wife Marie. They made special collec-
tions of Sibara grisea, including flowers fixed for morphological stud-
ies and buds for chromosome counts. Unfortunately, we did not get
positive results from the fixed buds. Research reported in this paper
was supported by National Science Foundation Grant DEB78-08766.

DRABA

The seeds of the large genus Draba are almost universally without
wings. Among North American species, only two have, in the past,
been known that possess winged seeds. These are D. pterosperma
Payson and D. asterophora Payson, both with relatively limited geo-
graphical ranges. Draba pterosperma appears to be present only in
the Marble Mountains of Siskiyou County, California and D. astero-
phora occurs in the mountains of extreme western Nevada and lim-
itedly in the Sierra Nevada Mountains of California. A third species
with winged seeds (Fig. 1), not previously known in fruit or with
mature seeds, has recently been found near Mt. Eddy in Trinity
County, California. This has prompted me to examine the winged
nature of the seeds in these three species in relation to the seeds of
other species of Draba. Murley (1951) states that a distal wing tip is
occasionally present on the seeds of D. glabella Pursh var. mega-
sperma Fern. & Knowlt. However, I have examined over one hundred
seeds of this taxon without being able to confirm this observation.

In most species of Draba, the seeds are plump and oblong to oval.
Often they are tapered toward the funicular attachment and thus take
the general shape of a tear-drop. All seeds of these shapes are without
the slightest suggestion of a wing. Usually they are slightly com-

108 REED C. ROLLINS

Fic. 1. Draba carnosula O. F. Schulz. Photograph of Williams 386
(GH). Photo by A. Coleman.

SPECIES OF DRABA, LESQUERELLA, AND SIBARA (CRUCIFERAE) 109

pressed perpendicular to the backs of the cotyledons, which are in
an accumbent position with respect to the radicle. The very different
winged seeds of the three species mentioned above are similar in that
they are strongly flattened, with winged margins that vary from 0.5
to 1.5 mm. wide toward the distal end of the seed. The wings, vary-
ing in width, border the seeds all the way around except for the funic-
ular area. The wing is an extension of the outer epidermal layer which
forms a sac-like structure enveloping the seed as well as extending
beyond it in three directions. In D. asterophora and D. pterosperma,
the epidermal layer forming the wing is loose and often wrinkled over
the seed surface but in D. carnosula the epidermal layer is very tight-
fitting over the seed. Also, the cells of the wing are smaller and the
wings themselves are more rigid than those of the other species.

In view of the substantial differences between the winged seeds of
the three species under review and those of the bulk of Draba, it
is pertinent to ask whether the winged seed character has any impli-
cations for the generic limits of Draba. Winged seeds of the type
seen in these species are common in Arabis. Were it not for the yel-
low flowers and scapose habit, D. carnosula could possibly be accom-
modated in Arabis. This is not true of Draba asterophora and D.
pterosperma. They are definitely Draba-like in habit and in every
other character except for the winged seeds. We have not found it
difficult to accept both winged and wingless seeds in the genus Arabis
(Rollins, 1941) and it is my conclusion that the same position must
be adopted with respect to Draba. Payson (1917) noted that D. as-
terophora and D. pterosperma are not closely related but rather ap-
pear to be independently related to other species of Draba. In the
case of D. carnosula, there is no apparent close relative among the
North American species of Draba. Certainly it is not to be associated
with D. howellii.

Draba carnosula O. E. Schulz (1927, p. 82) was described with only
a few lines. Subsequently it was treated by Hitchcock (1941) as D.
howellii var. carnosula, again with only a brief characterization. Be-
cause of this a full description is given.

P
straw-colored; leaves rosulate in a small dense cluster, entire, obovate, 0
4—8 mm. long, 3-4 mm i

; inflo-

usually tapering above and below, 13-22 mm. long, 4-6 mm. wide; terminal silique
usually erect on an axis up to 12 mm. long, an aborted flower pedicel scar is on the

110 REED C. ROLLINS

axis; valves with a strong nerve from base to narrowed apex that subtends the style;
styles slender, 2-3 mm. long; stigmas nearly entire, only slightly larger in diameter

an styles; septum entire, translucent, central nerve absent or indistinct; ovules 4-6
in each loculus; seeds flat, orbicular, 3-4 mm. in diameter including wings, wings 1-
1.5 mm. wide, cotyledons accumbent.

SPECIMENS STUDIED: California. Trinity Co.: on a dry, rocky, open hillside in ultra-
basic soil at 7,900 ft. elevation, T40, R5W, Sec. 18, Weed Quadrangle, Shasta-Trinity
National Forest, near Mt. Eddy, 27 Aug. 1980, Barbara Williams 386 (cu); along trail
out of basin abov
1979, Barbara Williams 270 (uc).

Draba carnosula is an unusual species in several ways in addition
to having widely winged seeds. The fruits are so large in comparison
to the size of the basal rosettes and the stems so slender that the
plants look top heavy (Fig. 1). The leaf rosettes are scarcely more
than 1 cm. across. The siliques are few in number on a given scape.
They are remote from each other giving a distinctive appearance to
the infructescence.

LESQUERELLA

In our treatment of Lesquerella (Rollins and Shaw, 1973) we called
attention to the disjunct populations of L. wardii which differed
somewhat from each other but not in a way that seemed to require
nomenclatural recognition. For several years, I have been particularly
interested in plants of the populations on the Kaibab Plateau of north-
ern Arizona which have shorter styles and on the average a higher
number of ovules than Utah material of the species. On the same day
in August, 1957, I collected late season plants at two different sites
between Jacob Lake and the entrance to the north rim of Grand Can-
yon National Park. A few days earlier, I had collected typical L. ward-
ii on the Aquarius Plateau in Garfield County, Utah. In making com-
parisons in the field, I was impressed with the differences between
these disjunct populations. Another chance to look for the Arizona
population with the hope of field study came in June, 1979. To my
surprise, the plants we found were white-flowered, not yellow as had
been expected. Correlated with the unusual white-flowered feature
is a very different trichome type than is present in L. wardii. This
material represents an undescribed species. Apparently we were too
early in the season to find the Arizona populations of L. wardii that
we had sampled in 1957.

Lesquerella kaibabensis Rollins, sp. nov.

Perennial with a thickened sj

: mple or scent,
trichomes similar throughout; : rarely branched caudex, densely pube

stems prostrate, arising below a terminal cluster of

SPECIES OF DRABA, LESQUERELLA, AND SIBARA (CRUCIFERAE) 111

leaves that form a flat rosette, less than 1 dm. long; radical leaves entire (1—)2—4(-5)
cm. long, (5—)8—41(-18) mm. wide, petiole slender, 1-3 times as long as the blade,
blades broadly ovate to nearly orbicular, abruptly narrowed to petiole; foliar and cau-
line trichomes with 3-5 primary radial unfused branches, minutely granular on surface,
branches usually forked; cauline leaves cuneate at base, lower sometimes nearly pet-
iolate, narrow to broadly oblanceolate, 5-10 mm. long, obtuse to somewhat acute,
usually less than 6 per stem; inflorescences congested, short, few-flowered; pedicels
ivaricately ascending; straight to slightly sigmoid with young fruit, 5-8 mm. long;
—6 mm. long, ca. 1.5 mm. wide, outer pair
ate, white, erect, 8-10 mm. long, 2.5-3.5 mm.
siliques elliptical to

o plane of septum,

pals narrowly ob aped,

slightly saccate at base; petals spatul.
wide; paired stamens 4-5 mm. long, anthers ca. 1 mm. long;
t

: u

attached to septum by ca. 1/3 their lengths, mature seeds unknown.
erba perennis pubescentibus; trichomatibus radiis 3-5 non coalescentibus autem
probe basin furcatis; caulibus prostratis 5-10 cm. longis; foliis radicalibus rosulatis pet-

iolatis integris (1-)2—4(-5) cm. longis, (5—)8—14(-18) mm. latis, laminis late ovatis vel
orbicularibus; foliis caulinis cuneatis anguste vel late oblanceolatis 5-10 mm. longis;
inflorescentiis congestis; petalis erectis albis spathulatis 8-10 mm. longis 2.5-3 5 mm.
latis; pedicellis adcendentibus vel divaricatis 5-8 mm. longis; siliquis ellipticis vel late
oblongis parum obcompressis 6-8 mm. longis; stylis pubescentibus 1-2 mm. longis;
ignotis.

Holotype in the Gray Herbarium, collected on a limestone-clay knoll, 18.6 miles
south of Jacob Lake on road to the north entrance of Grand Canyon National Park,
Coconino County, A
(Isotypes to be distributed.)

The short inflorescences, prostrate stems, and long petioled radical
leaves, together with the nonbranching caudex and elongated, some-
what compressed siliques, are features shared by Lesquerella kaibab-
ensis and L. wardii. Undoubtedly these two species are closely re-
lated. But the trichomes with only three to five primary branches,
which are forked and with ascending tips, that characterize L. kai-

abensis are strikingly different from those of L. wardii where the
5-7 rayed trichomes are more massive toward the center and the tips
lie flat on the leaf surfaces. The siliques are larger and the styles
shorter in L. kaibabensis than in L. wardii except that in the Arizona
populations of both species the styles are nearer to the same length.
In L. kaibabensis, the silique-valves are peculiarly uneven, with
depressions and slightly raised portions. This was noticed in the field
so it is not a result of drying, but the exact meaning of this feature
cannot be assessed without further collections to establish its con-
Stancy in the species as a whole. An important difference between
the species is in the flower color, white in L. kaibabensis and yellow
in L. wardii.

SIBARA

The species group to which Sibara grisea, newly described below,
belongs occurs in Mexico and west Texas, primarily in the lower Rio

112 REED C. ROLLINS

Grande Valley and adjacent areas. Up to now, we have not seen Sib-
ara from farther north in the Texas-New Mexico region except for
the more widespread and different S. virginica which occurs in central
and southern Texas eastward in the southeastern states, and with a
skip to southern California.

Sibara grisea Rollins, sp. nov.

Annual, greyish green, branched beginning at base, glabrous throughout, slightly
fleshy, some branches erect, others ascending, 1-2(-3) dm. tall; leaves monomorphic,
pinnately lobed, petiolate, not auriculate, 3-7(-20) cm. long, 1.5-3(—4) cm. wide, lat-
eral lobes broadly oblong to ovate, entire to shallowly sinuate-dentate, terminal lobe
triangular, shallowly dentate, larger than lateral lobes; flowers minute; sepals broadly
oblong, spreading at anthesis, nonsaccate, green then turning purplish, 2 m

long, slightly more than 1 mm. wide; petals white, spatulate, narrowed to a slender

nees 5-15 cm. long, usually extending nearly length of stem; pedicels divaricately

ascending, rarely more widely spreading, straight, 5-10 mm. long, reduced upward,

scarcely expanded at summit; siliques straight, divaricately ascending, compressed par-
m. wi

allel to septum, 2 g, 1.5-2.0(-2.5) mm. wide, one-nerved from base to apex
with a very short but noticeable gynophore; styles less than 0.5 mm. to 1.5 mm. long;
stigmas entire to very faintly lobed, smalle equal in diameter to style tip on fruits;

winged, ca. 1.5 mm. long, ca. 1
ent.

nnua ramosa griseo-viridis glabra plus minusve crassa 1-2 dm. alta; foliis
pinnatifidis lobatis petiolatis nonauriculatis 3-7 cm. longis 1.5-3.0 cm. latis; floribus
iridi mm. longis, petalis albis anguste-
- longis nonunguiculatis; pedicellis rectis divaricatis 5-10 mm.
longis, siliquis linearibus rectis 2—3 cm. longis, 1.5-2.0 mm. latis; stylis ca. 0.5 mm
vel 1.5 mm. longis, seminibus oblongis ca. 1.2 mm. latis; cotyledonibus accumbenti-

us,
Holotype in the Gray Herbarium, abundant in limestone cliff areas, Marble Canyon,
Sacramento Mountains, Otero County, New

t 3. ‘
70 through Alamogordo, in Marble Canyon of W slope of Sacramento Mts., ca. 5,000
ft., 11 May 1980, Richard and Marie Spellenberg 5500 (cH); same area,
U.S. Hwy. 70, ca. 6,000 ft., 11 May 1980, Richard and Marie Spellenberg 5507 (GH).
Texas. Hudspeth Co.: limestone soil, oak-juniper community, Pine Canyon, Guadalupe
Mountains National Park, 3 June 1974, Larry C. Higgins 8535 (wTs).

The nearest relative of Sibara grisea is S. runcinata (Wats.) Roll.
which occurs in southwestern Texas and Mexico. In my treatment of

SPECIES OF DRABA, LESQUERELLA, AND SIBARA (CRUCIFERAE) 113

Sibara (Rollins, 1947), S. viereckii (Schulz) Roll. was recognized but
this is now referred to S. runcinata as a synonym. Sibara grisea is
completely glabrous whereas S$. runcinata is hirsute with simple
spreading trichomes toward the base of the stems, more sparsely so
on the leaves. The styles on the mature fruits of S. runcinata are
much longer than in S. grisea and the cauline leaves are sessile with
auricles usually clasping the stems instead of being petiolate as in the
latter species. In the more northerly populations of S. runcinata, the
auriculate cauline leaves are more definite than in the more southerly
material. The problem is that often there is but a single cauline leaf
present on a plant. If this occurs near the base of the stem, the leaf
usually resembles the basal leaves which are uniformly petiolate and
have no auricles. If a cauline leaf occurs near the middle or on the
upper part of the stem, then it is auriculate. The unreliability of this
feature to mark off a taxon has led us to abandon the recognition of
S. viereckii.

Sibara grisea grows both on cliffs and at the base of cliffs or oth-
erwise nearly barren clayey soil. In the open, the plants are erect and
if in a sunny location, they are grey-green. Plants growing on the
cliff-face are often appressed to the cliff and in shady places they are
less gray and more green than those growing in the open. The plants
tend to be fleshy and very brittle. There is a considerable range in
the size of the individual plants as is characteristic of many annual
species.

LITERATURE CITED
Murer, M. R. 1951. Seeds of the Cruciferae of Northeastern North America.

Midl. Nat. 46: 1-81.
Payson, E. B. 1917. The Perennial Scapose Drabas
4: 253-267

Amer.

of North America. Amer. J. Bot.
ROLLINs, R. Cc. 1941. A Monographic Study of Arabis in Western North America.

ora 43: 289-325, 348-411, 425-481. .
1947. Generic Revisions in the Cruciferae: Sibara. Contrib. Gray Herb. no.

33-143
AND E. A.

SHAW. ' ruciferae) in North America.
1-288. Harvard University Press, Cambridge.

The Genus Lesquerella (C
d

Contributions from the pe
CRAY C5494
HERBARIUM

1982 No. 212
2. Brac aL
GARDEN ;,

~ Bi

BRARY } Why,
EBcabeth A. Sh AUGUSTUS FE ers COL UiLO.

sipkiegiad NEW MEXICO, 1846-1847 (QS
VY 7
XRAB:

Ra atin |S

EDITED BY Otto T. Solbrig
Kathryn Rollins

Contributions from the

GRAY
HERBARIUM

1982 No. 212

c.. AUGUSTUS FENDLER’S COLLECTION LIST:
h A. Shaw NEW MEXICO, 1846-1847

Reed C. Rollins THELYPODIOPSIS AND SCHOENOCRAMBE

(CRUCIFERAE)
|
|
Reed C. Rollins STUDIES ON ARABIS (CRUCIFERAE)

OF WESTERN NORTH AMERICA II

EDITED By Otto T. Solbrig
Kathryn Rollins

ev ep BY

Iss ae HERBARIUM OF HARVARD UNIVERSITY
ue 95-6094 Issued November 20, 1982

2
pri

AUGUSTUS FENDLER’S COLLECTION LIST: NEW MEXICO,
1846-1847

ELIZABETH A. SHAW

INTRODUCTION

“My needle is slow to settle—varies a few degrees and does not
always point due southwest, it is true, and it has good authority for
this variation, but it always settles between west and south-southwest.
The future lies that way to me, and the earth seems more unex-
hausted and richer on that side.” The words were written by Henry
Thoreau (1862); the feelings had long been shared by many Ameri-
cans, Asa Gray among them. There had developed awareness of the
west and conviction that the United States soon must extend to the
Pacific—the “manifest destiny” of the nation.

Gray’s interest in the west developed early. Even before his first
trip to Europe, in 1838-1839, Gray had had a taste of this barely
known flora through seeing the plants collected by Lewis and Clark,
by Edwin James, by Bradbury, and by Nuttall which he had studied
in Philadelphia and in Torrey’s herbarium. But when he saw in Glas-
gow the herbarium of William Hooker, and in England the herbaria
of Lambert and others, there was spread before him an array of new
and fascinating things from other collectors working in the northern
Rockies and in Oregon.

Gray now was hooked. He was totally committed to North Amer-
ican botany and he was acutely aware of the botanical terra incognita
of the southern part of the Rocky Mountains. Soon after his return
to the United States, Gray met George Engelmann, a German phy-
sician and botanist practicing medicine in St. Louis. Engelmann was
well situated, at the entry to the west, where he could watch the
flood of people headed out over the prairies, and sign up for his own
causes, those who seemed interested in plants. Gray, back in Cam-
bridge, had the resources of books and of herbarium which allowed
him to work on those new plants which would be coming in, and the
Gray-Engelmann partnership soon was established.

Their first efforts were not successful. Engelmann was 1n touch with
John Charles Frémont, but the plants collected by Frémont in 1842
in the Wind River Mountains of Wyoming and in the South Pass of
Colorado went to John Torrey. Torrey forwarded the Compositae to
Gray who found them interesting enough to write to Torrey (5 Dec.,
1842). “| | | Lieut. F. must be indoctrinated & taught to collect both
dried spec. & seeds. Tell him he shall be immortalized by having the

Senecio called S. Fremonti. . _” The next summer Frémont
led an expedition up the Arkansas and north to the Columbia River,
but he seemed always reluctant to take along any scientific assistant,

1

g ELIZABETH A. SHAW

and was noticeably unwilling to share either the work or the glory
which might result. So Frémont again did the collecting of plants
himself, but this time much of the material was lost, some by the fall
of a pack mule into a chasm, some soaked in a flood on the Kansas
River. Back in St. Louis, Frémont, to Engelmann’s annoyance and
disappointment, refused to allow these water-logged parcels be opened
and the plants dried, so many specimens were irreparably damaged.
Those remaining again went to Torrey.

Frémont was not the only string to the Gray-Engelmann bow. In
Silliman’s Journal for April-June, 1843, Gray advertised that, “. . .
three enterprising botanists are now engaged in exploring the most
interesting portions of the far West, and that their collections of dried
plants will be offered to subscribers, in sets, as they come to hand.”
Two of the three were Karl Geyer and Friederich Liiders, young
Germans whom Engelmann knew at St. Louis, who were in the field
with a party led by a wealthy Scot, Sir William Stewart. Liiders lost
his plants when his canoe capsized on the Columbia. Geyer had better
luck and came eventually to Fort Vancouver on the Columbia. How-
ever, from here Geyer sailed to England and offered his plants to
Hooker, an act of perfidy which much annoyed Engelmann and Gray,
especially Gray who had finally to buy a set of the plants for himself.

The third collector was Ferdinand Lindheimer, also a friend of
Engelmann. Lindheimer started in 1843 to collect in eastern Texas
with the thought, at least on the part of Gray, that he would work
westward and go eventually to Santa Fe. Plants did indeed come in
from Lindheimer, but rather than pushing to the west, Lindheimer
married and settled in the German community at New Braunfels.
Gray's frustration now was acute. He wrote on 8 April, 1846, to En-
gelmann, “What is Lindheimer about? Why is not his last year’s col-
lection yet with you? We had just got things going, and we can sell
fifty sets right off of his further collections, and he can go on and
realize a handsome sum of money if he will only work now! And he
will connect his name forever with the Texan flora!”

But it was 1846 and changes were coming to the southwest. On 13

the Government Expedition. If I were not so tied up, I would go
myself. Have you not some good fellow you can send? We could prob-

AUGUSTUS FENDLER’S COLLECTION LIST 3

ably get him attached somehow so as to have the protection of the
army, and if need be I could raise two hundred dollars as an outfit
. . .. Someone must go into this unexplored field.” Luck was at last
with the partnership. Four weeks later, Engelmann could tell Gray,
“T believe I have found a young man. . . His name is, if I am right,
A. Fendler.”; and on 3 July, “He has now concluded to go, accom-
panied by his brother; his name is Augustus Fendler.

Compared to his contemporaries in the southwest—to Lindheimer
or to Charles Wright—Augustus Fendler is little known. The bare
facts of his life are given in “An Autobiography and Some Reminis-
cences of the Late Augustus Fendler, ” edited by William Canby, and
published in the Botanical Gazette in 1885.

Fendler was born in 1813 at Gumbinnen in East Prussia, now
Gusev in the Russian S.$.R. He came to the United States in 1836
and after working in Philadelphia and New York, Fendler wandered
west in 1838, to Texas and then to St. Louis. In 1844 Fendler re-
turned to Germany and at Kénigsberg met Ernst Meyer. Meyer sug-
gested to Fendler, who had by then worked at various trades, that
sets of dried plants from the American west could easily be sold in
Europe. Fendler found this idea attractive, and back in St. Louis,
turned to Engelmann for advice, assistance, and books. Around St.
Louis, Fendler “practiced” collecting and preparing specimens, but
Engelmann dissuaded him from trying to sell in Europe sets of plants
already familiar there and suggested that Fendler gain more experi-
ence and then turn toward Santa Fe.

Fendler thus was on the scene, but even by the summer of 1846,
he had little experience in handling plants, and Engelmann was mod-
est in telling Gray about him. On 25 June, Engelmann wrote to Gray,

. . . he cannot yet be called a botanist, nor even 4 collector.”, but
Engelmann believed him to be sound and thought that plans for Fen-
dler’s trip should at once be made. Gray was to write to Secretary
of War Marcy to get permission for Fendler to travel with a military
party and to arrange for his transportation. Engelmann would provide
him with paper, lend him one hundred dollars ( relying on your
wealthy friends in Boston etc. to refund the money ), and do what
he could to smooth Fendler’s way with the authorities at Fort Leav-
enworth. Engelmann saw clearly, and coldly, the problems —
could arise, “The money advanced would be refunded after the eis
of the plants, or by sending plants and seeds, etc. But in case of is
death or of the loss of his collections, or (which I do not fear) of =
ping false [a memory of Geyer], that advanced money would be
eet”

arkable speed. By the end of July

Preparati with rem
P ons were made reached Fort Leavenworth and

a letter from Secretary Marcy had

4 ELIZABETH A. SHAW

Fendler was ready. He and his brother left St. Louis on 1 August for
the fort, armed with fifty dollars from W. S. Sullivant, one hundred
and twenty dollars from John Amory Lowell, a supply of paper from
Engelmann (as well as “full written instructions”), and the promise
that they would travel free of charge with the army. Engelmann was
pleased that Fendler was on his way after only six weeks of prepa-
ration and began, “. . . to hope a little more from this country for
science. ”

FENDLER’S ROUTE: THE ROAD TO SANTA FE

The “Army of the West” followed the Santa Fe trail which by 1846
was a well-defined road, for the wagons of traders had been traveling
for twenty-five years over its length of more than eight hundred
miles. A very brief account of Fendler’s route was provided by En-
gelmann for the preface to Gray’s “Plantae Fendlerianae Novi-Mexi-
canae’ (1849), but there are many other sources of information on the
trail and thus about Fendler’s collecting localities. An early one, pop-
ular at the time, is Josiah Gregg’s “Commerce of the Prairies or the
Journal of a Santa Fe Trader,” published in 1844 and still fine read-
ing. Also valuable is Wislizenus’ “Memoir of a Tour to Northern
Mexico” (1848) which has a good map showing most of the localities
cited by Fendler. Useful, too, is Emory’s “Notes of a Military Re-
connoissance” (1848) which has a general botanical appendix by John
Torrey, a treatment of cacti by Engelmann, and a report on natural
history by James Abert with another list of plants prepared by Torrey.
This, too, has a map. These two reports on the trail are of particular
interest for both Wislizenus and Emory traveled over it only eight
weeks before Fendler did. Among recent books, I have found most
useful Riddle’s “Records and Maps of the Old Santa Fe Trail” (1963)
and Stocking’s “The Road to Santa Fe” (1971). And there is the chap-
ter on Fendler in McKelvey’s (1955) incomparable “Botanical Explo-
ration of the Trans-Mississippi West 1790-1850.”

The Fendlers left Fort Leavenworth, Kansas, on 10 August, 1846,
with Colonel Sterling Price’s Second Missouri Mounted Volunteers.
Twelve of the fourteen companies in this command went to Santa Fe
by the Cimarron cutoff, across the Oklahoma panhandle, and into
northeastern New Mexico. Two companies, and the Fendlers, were
detailed to travel up the Arkansas River to Bent’s Fort, near La Junta,
Colorado, there to meet a commissary train which was to be escorted
to Santa Fe (Berry, 1964).

In August there was no real road between Fort Leavenworth and
the trail, about twenty-five miles to the south. Fendler’s party seems
simply to have headed south from the fort and to have picked up the
trail near Olathe, Kansas. The first locality on the trail which is cited

AUGUSTUS FENDLER’S COLLECTION LIST 5

by Fendler, on 12 August, is Lone Elm, south of Olathe in Johnson
County. For ease of reference I shall use both modern place names,
whether or not they were in use in Fendler’s time, and the names
used by Fendler, many of which are no longer in use.

The trail ran west-southwestward across Kansas through Council
Grove, past McPherson, to Great Bend where travellers reached the
Arkansas River and turned upstream. Above Great Bend there were
several crossings of the river; a “lower crossing” near Ford (Ford
County), two “middle crossings” between Cimarron and Charleston
(Gray County), and the “upper crossing,” near Lakin in Kearny
County. From each of these a trail ran across the Cimarron Desert,
the “Waterscrape” of travellers, to the waterholes on the Cimarron
River in southwestern Kansas, joining finally in Grant County and
continuing as the Cimarron cut-off into New Mexico. This was the
shortest and driest route to Santa Fe, taken by most of the companies
under Price’s command.

There was also the mountain route, an arm of the trail which con-
tinued up the Arkansas into Colorado, The Fendlers went this way
to Bent’s Fort. The arrival of the troops was noted by Francis Park-
man, “On the 6th [September] two companies of volunteers of : Price's
rgt. came up to our camp—a set of undisciplined ragamuffins.” (Park-
man, ed. Wade, 1947).

From Bent’s Fort the trail ran southwestward, nearly over route
350, to Trinidad, Colorado, thence over the Raton Pass into New
Mexico. About fifteen miles south of Raton, New Mexico, the moun-
tain route forked, the eastern road soon forking again, so that there
Were three routes through Colfax County, New Mexico. The west-
ernmost road ran through Cimarron, thence south through Rayado.
The middle road ran from a point north of Maxwell southwestward
to Rayado and there joined the western road. The eastern branch ran
farther to the south, then turned to the west and joined the others
about twelve miles west of Colmor. Fendler’s party probably used the
middle road. The mountain route then ran south to Watrous where
the Cimarron cutoff came in from the northeast. It was @ single trail
= this point and, nearly following route 85, continued into Santa

e.

Fendler was in Santa Fe from 11 October,
1847, He arrived too late in the season to d
in the spring of 1847 he started work along t
on the nearby plains. Fendler never got
from his lack of money and from his fear of Indians. ee
‘wo trips, on 8 May and on 24 and 25 May, 1847, to “Rio del No ;
the Rio Grande, about twenty miles west of the town. coins

€ got no more than ten or twelve miles into the mountains east 0

6 ELIZABETH A. SHAW

Santa Fe. Standley (1910) pointed out that, “. . . if he had gone fur-
ther he would have found hundreds of plants not in his collections.
Fifteen or 20 miles away he would have found a subalpine flora that
would have been rich in plants then undescribed.”

Fendler would, no doubt, have stayed longer in Santa Fe, but he
had run out of money. Presence of the troops had driven prices up
quickly. As early as 8 November 1846, Fendler wrote to Engelmann
that prices were two to three times higher than in St. Louis. In this
same letter Fendler suggested to Engelmann that he and his brother
could manufacture matches, very expensive in Santa Fe and in short
supply, and asked Engelmann to send raw materials and instructions,
but this scheme seems to have come to nothing. During the spring
of 1847 Gray tried to raise money among his usual donors in Boston,
but these efforts came also to nothing, and by summer Fendler must
have been very short indeed of money.

The Second Missouri and the Fendlers left Santa Fe for Fort Leav-
enworth on 9 August, 1847, by the shortest route. Fendler thus re-
traced his route of the year before to Watrous, New Mexico, where
the Cimarron route turned to the northeast. From Watrous the trail
ran across the southeastern corner of Colfax County, then through
Union County a few miles north of Sofia, Mt. Dora, and Seneca. It
crossed into Oklahoma just to the south of Mexhoma. In Oklahoma
the trail continued north to Cold Spring, east of Kenton (Cimarron
County), on the Cimarron River; and then followed the Cimarron into
Kansas. Still following the Cimarron, the trail continued into southern
Grant County; from here one could turn northward for the crossings
of the Arkansas. Fendler’s party took the trail to the middle crossings
and then followed the river down to Great Bend. From here the trai
to Fort Leavenworth was nearly that of a year before. Fendler reached
the fort on 24 September, 1847, and left the next day for St. Louis.
His last cited locality is “bank of the Missouri River, about Washing-
ton [Missouri]” on 27 September, 1847.

FENDLER’S COLLECTIONS AND THE COLLECTION LIST

On his return to St. Louis, Fendler worked with Engelmann to
order his plants into families, beginning with Ranunculaceae, and fol-
lowing the Candollean sequence through “Polypetalae,” “Monope-
talae,” “Apetalae,” gymnosperms, monocotyledons, and vascular cryp-
togams. Fendler also prepared from his notes kept day-to-day a list
of the collections, arranged by family, and numbered from “1,” “Cle-
matis ligusticifolia,” through “1026,” referred by Fendler to Chara-
ceae. A copy of this list, packed with information on places and dates

_ collection, on plant habits and habitats, went with the plants to
ray.

AUGUSTUS FENDLER'S COLLECTION LIST 7

The plants came to Gray over several months. On 20 December,
1847, he wrote to Engelmann that the families up to Rosaceae had
arrived, “The specimens are perfectly charming! So well made, so full
and perfect. Better never were made.” The rest were in Gray's hands
by the next summer and he then worked over the whole collection.
By November, Gray had prepared for communication to the American
Academy of Arts and Sciences, “Plantae Fendlerianae Novi-Mexi-
canae,” including families from Ranunculaceae through Compositae,
published in February, 1849, with the promise, “to be continued.”
But that was the end of it. As happened so often, Gray prepared for
publication those families through the Compositae, a favorite of his,
but there were then other demands on his time and no more family
treatments of Fendler’s plants appeared, although Gray did, of course,
eventually describe those things he considered to be new to science.

The plants and the publication have always presented problems—some
of Fendler’s making, others attributable to Gray. The Compositae, the
last family treated in the publication, fall about half way through the
dicotyledons in the Candollean sequence. As a result, only half of

distribution seems to have added other information.

Then in compiling the collection list, Fendler often included two
or more collections—which he thought to be the same—under one
number. In some cases, these had been made at the same place.
Thus, number 37 of the list has these data: “Santa Fe; upper part of
the creek valley, foot of mountains. 7 July-10 August, 1847.” But
other numbers in Fendler’s list comprise collections made at different
places. Number 92 of the collection list has this entry: “15-26 August,
1847. Between Vegas and Moro River [New Mexico] and a few miles
east of Cold Spring (Cimarron) [Oklahoma]. ”

The sets offered for sale were made up by Engelmann and Fendler,
not by Gray. In a few cases they lumped two or three of Fendler's

numbers, h haps including two or more individual col-
oe _ Thus material distributed

luded number 91 of the list, col-

as “Fendler 93”, Linum perenne, inc
just mentioned above, made up

lected at Santa Fe, and number 92,
of plants collected in New Mexico and Oklahoma. Grays copy of
Fendler’s list is neatly written in a now faded ink with Fendler s own
numbers lightly pencilled in. In the families up to Compositae, the
final distribution numbers, under which these plants are cited, were

8 ELIZABETH A. SHAW

added in red ink by Engelmann before sending the list to Gray, and
the lumpings thus are easily noted. The numbers in red ink are those
used in “Plantae Fendlerianae” and run through “295,” the first of
the Compositae, distributed as “Vernonia noveboracensis.”

Engelmann felt that he had neither the expertise nor the time to
arrange the Compositae and following families, so these he left to
Gray (Engelmann to Gray, 15 Feb., 1848). He wrote again to Gray
on 27 February, 1848, “Fendler is so situated [for money] that he
cannot wait with the distribution of his plants; he will therefore dis-
tribute them now with my red numbers (in your Catalogue [copy of
the collection list]) as far as they go and then with his own numbers
in pencil, so that the subscribers can alter them afterwards so as to
correspond with the printed Catalogue numbers. As far down as Ver-
nonia you could then adopt in the printed Catalogue the red numbers,
but farther on you arrange the plants as you think best, and add to
the end of each notice in brackets: (Fendler’s Catalogue Nro 714 or
whatever it may be) or (F.C. 714) something like Hookers Catalogue
of Geyers’ plants.”

Gray explained this in a footnote on page 61 of “Plantae Fendler-
ianae,” “From this onward, the numbers inclosed in parentheses, and
usually placed after the habitat, are those under which the specimens
have been distributed.” Thus in the published treatment of the Com-
positae the numbers which enumerate the taxa run sequentially from

295, “ “Vernonia noveboracensis, var.” through “462,” “Sonchus as-
per,” but these numbers have only rarely been used in citing the
specimens. Of the specimens of Fendler’s Compositae which I have
seen, only a few have on the label both the original distribution num-
ber, that is, the number which js used in the collection list and cited
in parentheses in “Plantae Fendlerianae”; and the taxon number used
in “Plantae Fendlerianae.” Many subscribers seem not to have added
this latter number.

Fendler’s presentation of dates and of localities can also be trou-
blesome. In her chapter on Fendler, McKelvey (1955) concluded that
the collection list does not provide an itinerary, “since the plants are
arranged according to families and the dates and localities supplied
are covering ones. For example his no. 1 bears the notation: ‘Santa
Fe, 1st-19th July, 1847. Moro River 15th August. Creek bottom land,
near the water climbing to the tops of trees and shrubs. Also the Ric
de los Animos between Bent’s Fort and Santa Fe. Fruit: 15th Au-
gust-16th January.’” That the dates are “covering ones” may be s0;
I suspect that in most cases they are not. Thus it is likely that

Ist-19th July, 1847” refers only to collections made on those two
days, not to some unknown number of collections made between
those dates. In fact it makes no difference to the user in most cases.

AUGUSTUS FENDLER’S COLLECTION LIST 9

It is no easier and no more helpful to refer any individual plant to
a collection made on one of two days than to try to refer it to some
one of nineteen days. I do not agree that Fendler’s itinerary cannot
be worked out. That is easily done; the dates and localities are there;
they are easily tied together and the localities are shown on contem-
porary maps.

The plants eventually were offered for sale as “Plantae Novo-Mex-
icanae.” Gray had been much impressed by them and hoped that
Fendler would return to the west, but this was not to be. On his
return to St. Louis, Fendler was sadly short of money—the financial
problems from Santa Fe remained, the sets did not sell as quickly as
Gray had hoped, and some of the purchasers were slow in paying.
The buyers paid ten cents per number; thus, John Amory Lowell paid
for his set of 513 numbers, fifty-one dollars and thirty cents. The
money went to Fendler, but he had large debts owed to Engelmann,
and he had to support himself and his brother. Fendler complained
to Engelmann and Gray about the lack of profit in plant collecting,
but he was, nevertheless, willing to go again into the field.

Early in the summer of 1849 he set out with a military party from
Fort Leavenworth towards the Platte River, but the group was led
by an inexperienced officer. As a result of this man s mismanagement,
Fendler’s wagon with paper and other supplies was overturned in a
flooded creek. Fendler, his equipment lost or badly damaged, was
forced to turn back to St. Louis where he found that all of his re-
maining possessions had been lost in a fire which destroyed much of
the waterfront of the city. Fendler was now very much discouraged.
A few months later he left St. Louis for Panama and although he later
spent many years in the United States, he had nothing more to do
with the west. ae

The collection list is presented so that Fendler’s plants, which in-
clude the first collections of importance from Santa Fe, and oni in
Many cases are the types of names published by Gray and by ot ers,
can be more easily used. This published list is not an exact aparind
tion of Gray's copy, for I have in many entries modified Fen is
erratic and inconsistent use of marks of punctuation, and I have e-
leted “st” and “th” from dates, but there are no other changes in
Fendler’s own wording.

SUMMARY OF DATES AND LOCALITIES,
EN ROUTE TO AND FROM SANTA FE

1846. Kansas. Aug. 11: 8, 10, 15, 18 miles south Fort Leavenworth;

2, 4, miles south of Kansas River. Aug. 12: few miles = - euie
ferry of Kansas River; 20 miles from the ferry; Lone Elm. Aug. te:

10 ELIZABETH A. SHAW

Lone Elm; 60, 80 miles west of Independence [distances cited for
Aug. 13-17 as “west of Independence” are exaggerated by 20-40
miles]. Aug. 14: Black Jack Point; Hickory Point; 90 miles west of
Independence. Aug. 15: 100, 117 miles west of Independence. Aug.
16: 117, 120 miles west of Independence; 30 miles east of Council
Grove. Aug. 17: 118, 120 miles west of Independence; 110 Mile
Creek; few miles east of Bluff Creek; 30, 25, 20, 15, few miles east
of Council Grove. Aug. 18: 15, 12, 10, 5 miles east of Council Grove.
Aug. 19-26: Council Grove. Aug. 27: 30 miles west of Council Grove.
Aug. 28-30: no collections listed. Aug. 31; between Cow Creek and
Little Arkansas River; Cow Creek. Sept. 1: between Cow Creek and
Walnut Creek. Sept. 2-3: Walnut Creek. Sept. 3: Walnut Creek. Sept.
4: no collections listed. Sept. 5: Arkansas River. Sept. 6: 15 miles east
of ford of Arkansas River. Sept. 7-9: no collections listed. Sept. 10:
5 miles south of ford of Arkansas River. Sept. 11-15: no collections
listed. Colorado. Sept. 16: 20 miles east of Big Sand Creek. Sept. 17:
10 miles east of Big Sand Creek; few miles east of Big Sand Creek.
Sept. 18: no collections listed. Sept. 19: 42, 32 miles east of Bent’s
Fort. Sept. 20: 6 miles east of Bent’s Fort. Sept. 21-22: no collections
listed. Sept. 23: 3 miles above Bent’s Fort; Bent’s Fort. Sept. 24-26:
no collections listed. Sept. 27: 20, 25, 27 miles south of Bent’s Fort;
15 miles northeast Rio de los Animos. Sept. 28: Rio de los Animos.
New Mexico. Sept. 29: Raton Mountains. Sept. 30: Raton Creek; be-
tween Raton and Bermejo Creek. Oct. 1: between Bermejo and Rio
Colorado; 3 miles north of Rio Colorado; 8 miles south of Rio Colo-
rado. Oct. 2-3: Poi Creek. Oct. 4: Poni Creek; Rayado Creek; few
miles south of Rayado Creek. Oct. 5-8: no collections listed. Oct. 9:
between San Miguel and Santa F e; between San Miguel and Pecos.
Oct. 10: few miles east of Pecos.

1847. New Mexico. Aug. 10: 8 miles east of Santa Fe; between
Santa Fe and Pecos; Pecos; from Santa Fe to San Miguel; between
Santa Fe and Ojo de Bernal. Aug. 11: between Pecos and San Miguel.
Aug. 12: between Pecos and San Miguel; San Miguel. Aug. 13: 18
miles west [east?] of Pecos; Ojo de Bernal. Aug. 14-17: San Miguel
to east of Rock Creek. Aug. 14: between San Miguel and Vegas; 15,
12, 10, 8, 7, 6, 5, miles west of Vegas; Vegas; 2 miles east of Mora
River [prob. wrong date—at Mora on Aug. 15]. Aug. 15: between
Vegas and Mora River; Mora River. Aug. 16: Mora River. Aug. 17:
Mora River; 2 miles east of Mora River; east of Rock Creek [wrong
date—at Rock Creek on Aug. 22]. Aug. 18: 2, 5, 6, 8 miles east of
Mora River. Aug. 19: no collections listed. Aug. 20: Ocate Creek; Rio
Colorado; Rabbit Ear Creek [wrong date—at Rabbit Ear Creek on
Aug. 25]; Middle Spring of the Cimarron [impossible for this date].
Aug. 21: between Ocate Creek and Rio Colorado; Rio Colorado; be-

AUGUSTUS FENDLER’S COLLECTION LIST ll

tween Rio Colorado and R
7 miles west [east?] of Ler Senge a
“Se aie er between Rock and Rabbit Ear Creeks. Aug. 95: Rab-
2 . ; few miles west of McNees’ Creek; between Rabbit Ear
eal om ea McNees’ Creek; east of McNees’ Creek. Okla-
.. “a 6: between McNees’ Creek and Cold Spring; Cold
. aneree east of Cold Spring. Aug. 27: Upper Spring of
. . ug. 28: few miles west of Cimarron; crossing of the
ae a Cimarron crossing and Willow Bar; Willow Bar.
a. ansas. Aug. 29: Willow Bar, between Willow Bar and
i... Spey Middle Spring; 10, 16 miles east of Middle Spring.
. - ug. 30: 16 miles east of Middle Spring; 18 miles west of
. sige Aug. 31: few miles west of Lower Spring; Lower
= el ept. mi Sand Creek. Sept. 2: 40 miles southwest of crossing
age Ps iver; Pawnee Fork [wrong date—at Pawnee Fork on
Sh ‘i ept. 3: 4 miles south of the Arkansas River crossing; Walnut
Po rong date—must have passed Walnut Creek on Sept. 8].
ps ai be collections listed. Sept. 5: 3 miles west of Fort Mann. Sept.
ae s — listed. Sept. 7: between Coon Creek and Pawnee
i. ee : Pawnee Fork; between Pawnee Fork and Cow Creek.
ie ae a Creek. Sept. 10-14: no collections listed. Sept. 15: Tur-
. s - Cottonwood Creek. Sept. 16: Cottonwood Creek; few miles
=e eesepets Creek. Sept. 17: 16 miles west of Council Grove;
a pk Sept. 18: Council Grove; Bluff Creek. Sept. 19: be-
A. uff Creek and 110 Mile Creek. Sept. 20: few miles west, few
voto of 110 Mile Creek. Sept. 21: 18 miles east of 110 Mile
aaa ferry of Kansas River. Sept. 22: upper ferry of Kansas
. miles north of upper ferry. Sept. 23: between Kansas River
a se Risaeclo pe: Sept. 24: few miles south of Fort Leaven-
ie issouri. Sept. 25: bed of i
ie Sept. 26: banks of Missouri
nks of Missouri River above Washington.

LOCALITIES CITED BY FENDLER

ere were at least four important

Arkansas River crossing or ford: Th
g” was at Mulberry Creek near

= of crossing. The “lower crossin
‘ ah of Ford, Ford Co., Kansas. There were two ‘middle cross-
ngs” west of Cimarron, Gray County; the easternmost between Cim-
arron and Ingalls, the western one tween Ingalls and Charleston.

ing (1 to be either the

ty. The “upper crossing was near

12 ELIZABETH A. SHAW

Bent’s Fort (mountain route): This was Bent’s old fort or Fort Wil-
liam, used from 1828 to 1853. It stood on the north bank of the Ar-
kansas River about four miles east of La Junta, Colorado.

Big Sand Creek (mountain route): Now Big Sandy Creek, this creek
enters the Arkansas a few miles east of Lamar, Prowers Co., Colo-
rado.

Bermejo Creek (mountain route): Now spelled “Vermejo,” this creek
enters the Canadian River a few miles south of Maxwell, Colfax Co.,
New Mexico. Fendler’s party probably crossed the Vermejo six or
seven miles west of Maxwell.

Black Jack Point: In Fendler’s time a grove of oaks, the locality is
in Douglas Co., Kansas, about eight miles southwest of Gardner.

Bluff Creek: The trail crossed this creek about three miles north
of Bushong, Lyon Co., Kansas.

Cimarron route: Ran from the Arkansas crossings southwestward to
the Cimarron River in southwestern Kansas and the Oklahoma pan-
handle, then across the northeastern corner of New Mexico to Wa-
trous and Santa Fe. This direct route was taken by Fendler on his
return to St. Louis (August, September, 1847).

Cimarron River (Cimarron route): The crossing or ford of the Cim-
arron was near Willow Bar, about ten miles north northwest of
Keyes, Cimarron Co., Oklahoma. For the Cimarron springs, see the
individual entries.

Coon Creek: drains into the Arkansas near Kinsley, Edwards Co.,
Kansas.

Cold Spring (Cimarron route): about twelve miles northwest of
Boise City, Cimarron Co., Oklahoma.

Cottonwood Creek: crossed by the trail about one mile from Dur-
ham, Marion Co., Kansas.

Council Grove: on the Neosho River, Morris Co., Kansas.

Cow Creek: crossed the trail just west of Lyons, Rice Co., Kansas.

Fort Leavenworth: on the Missouri River, just north of Leaven-
worth, Kansas.

Fort Mann: a short-lived military post west of Dodge City, Ford
Co., Kansas.

Independence: just east of Kansas City, Missouri, this was for many
the starting point of the road to Santa Fe.

Kansas River: The “lower ferry” was perhaps Grinter’s ferry, just
west of Kansas City; the “upper ferry” may have been Pappan’s ferry
at the site of Topeka, Kansas.

Liberty: Clay Co., Missouri.

Little Arkansas River: crossed by the trail a few miles southwest of
Windom, McPherson Co., Kansas.

Lone Elm: about four miles east of Gardner, Johnson Co., Kansas.

AUGUSTUS FENDLER’S COLLECTION LIST 13

Lower Spring: about two miles west of US 270 near the
Grant—Stevens county line, Kansas.

McNees’ Creek (Cimarron route): crossed by the trail a few miles
west of Mexhoma, Oklahoma; just into Union Co., New Mexico.

Middle Spring (Cimarron route): about two miles west of the bridge
carrying Kansas route 27 over the Cimarron, Morton Co., Oklahoma.

Mora River: crossed by the trail near the town of Watrous, Mora
Co., New Mexico.

Ocate Creek (Cimarron route): crossed by the trail about ten miles
east of Colmar, Union Co., New Mexico.

Ojo de Bernal: about five miles southwest of Tecolote, San Miguel
Co., New Mexico.

One Hundred and Ten Mile Creek: crossed by the trail just east of
Scranton, Osage Co., Kansas. The name refers to the distance from
Fort Osage, Missouri.

Mountain route: Ran upstream from the Arkansas crossings to Bent's
Fort, thence over the Raton Pass into New Mexico where it met the
Cimarron route at Watrous, Mora County. Fendler’s party took this
route to Santa Fe (September, October, 1846).

Pawnee Fork: crossed by the trail near Larned, Pawnee Co., Kan-
sas

Pecos: western part of San Miguel Co., New Mexico. :

Poti Creek (mountain route): now Ponil Creek; crossed by the trail
a few miles east of Cimarron, Colfax Co., New Mexico. —

Rabbit Ear Creek (Cimarron route): crossed by the trail between
Grenville and Mt. Dora, Union Co., New Mexico.
: Raton Creek (mountain route): crossed by the trail near Raton, Col-
ax Co., New Mexico.

Raton Mountains (mountain route): Colorado—New Mexico. ay
trail crossed by the Raton Pass, between Starkville, Las Animas ©o.,
Colorado and Raton, Colfax Co., New Mexico.

‘ Rayado Creek (mountain route): crossed by the trail at Rayado,
olfax Co., New Mexico.

eg Creek (Kansas): crossed by the trail about five miles east of
ouncil Grove, Morris Co., Kansas. :

; Rock Creek (New Mexico—Cimarron ne sees < tetiaiees
ew miles northeast of Sofia, Union Co., New Mexicr-

Rio C aes satel: River. Fendler cited the Rio — :
a collecting locality on 1 Oct., 1846. The party probably stags oon
Springer and Colfax, New Mexico on the mountain route an

the Rio Colorado on 21 Aug., 1847, this time on the Cimarron route
which forded the Canadian a few miles south of Tay]
0., New Mexico.

14 ELIZABETH A. SHAW

Rio de los Animos (mountain route): the Purgatoire River; i.e., Rio
de las Animas Perdidas en Purgatorio. The trail crossed the river
about twelve miles northeast of Trinidad, Las Animas Co., Colorado.

Rio del Norte: the Rio Grande, about twenty miles west of Santa
Fe.

Sand Creek (Cimarron route): crossed by the trail seven to nine
miles southwest of Ulysses, Grant Co., Kansas.

San Miguel: a few miles south of US 85 in the southwestern part
of San Miguel Co., New Mexico.

Turkey Creek: crossed by the trail near McPherson, McPherson
Co., Kansas.

Upper Spring (Cimarron route): about two miles west of the point
at which US 287 crosses the Cimarron River, Cimarron Co., Okla-
homa.

Waggon Mound: Wagon Mound, Mora Co., New Mexico.

Walnut Creek: crossed by the trail a few miles west of Ellenwood,
Barton Co., Kansas.

Washington: Franklin Co., Missouri, about forty miles west of St.

uis

is.

Willow Bar (Cimarron route): about 214 miles south of Col-
orado—Oklahoma state line; about ten miles north of Keyes, Cimar-
ron Co., Oklahoma.

Vegas: Las Vegas, San Miguel Co., New Mexico.

NOTES ON THE COLLECTION LIST

1. Each entry represents plants which Engelmann and Fendler re-
ferred to a single taxon, but an entry can include one or more sep-
arate collections.

2. Material enclosed in quotation marks consists of annotations writ-
ten into the manuscript by Engelmann or by Gray.

3. The first number cited is always that originally used in the enu-

meration.
4. For numbers 1-327 of the list, the number immediately follow-
ing and enclosed in parentheses is a distribution number assigned,
and written into the manuscript, by Engelmann. In a few cases, €--
(50) and (71), the distribution number includes two (rarely three) of
Fendler’s numbers; i.e., (50) includes “55” and “57;” (71) includes
77 and “78.” These distribution numbers, which run from (1) through
(295), are those used by Gray in “Plantae Fendlerianae” to enumerate
species in families Ranunculaceae through Valerianaceae, plus “Ver-
nonia noveboracensis var.,” the first taxon in the Compositae. For
these groups it is these numbers which have always been used in
citing the collections.

AUGUSTUS FENDLER’S COLLECTION LIST 15

5. F rom this point on, Engelmann turned over to Gray the task of
assigning distribution numbers. For the remaining Compositae (the
last family published in “Plantae Fendlerianae”) Gray assigned num-
bers which actually are the running numbers in Fendler’s manuscript
In the Compositae the numbers which enumerate taxa run scone:
tially from 295 (“Vernonia noveboracensis var.”) through 462 (“Son-
chus asper”), but Gray cited in the text the numbers “under which
the specimens have been distributed,” that is, Fendler’s numbers
and these are used in citing collections. N.B.: “295° is a distribution
number assigned by Engelmann; “296” and following numbers to the
end of the Compositae are merely sequential numbers. Starting with
the Lobeliaceae, Gray continued to use the manuscript numbers for
distribution numbers.

Thus in the following pages: (a) in the Compositae the first number
for each entry is the running number in the manuscript and is also
the number under which the collection was distributed; the second
number given is that used in the sequential listing of taxa in “Plantae
Fendlerianae;” the name cited is that used in “Plantae Fendlerianae’ ;
in a few cases the collection was named—by Gray's annotation in the
collection list—but was not cited in “Plantae Fendlerianae —these
are so indicated; (b) from the beginning of the Lobeliaceae to the end
of the list, the number given is both that used in the manuscript and
the distribution number.

COLLECTION LisT: 1846 —1847

(1) Santa Fe, 1-19 July, 1847. Moro River, 13 August.
Creek bottom land, near the water, climbing to the tops of trees and
shrubs. Also Rio de los Animos between Bent’s Fort and Santa Fe.
: = No. 1. Fruit. 15 August-16 January.

(3) Santa Fe, 12 May—3 June. Foot and sides of steep
rocks in shady places near the creek. Climbing over shrubs.
: (2) Council Grove, 21 August, 1846. Creek bottom
land. = 5,
i A few miles east of Council Grove, 17 August, 1846.
: igh bank of a dried-up creek.

(4) A few miles east of Santa Fe, creek valley in the
mountains. Shady places, foot of rocks near irrigating ditch. 16-24
June, 1847.
K “Anemone cylindrica—in f-uit” A few miles south of
<Sam River. 22 September, 1847.
e Santa Fe, eastern moun
i spoil hills at the foot of mountains, on gent
Clivities; tolerably good soil:

tain range. 17-26 April.
ly sloping northern de-

16 ELIZABETH A. SHAW

6) Sunny margin of the creek, 6 miles above Santa Fe
(in the mountains). Fertile soil. 3 June, 1847.
9 7) Santa Fe, 2 May-19 October. Creek bottom, wet
places. Form large patches by sending off runners in all directions.
10 Santa Fe, 26 April-16 June. Creek bottom, moist
places. Stem upwards of 18 inches high.
11 (9) Santa Fe, creek valley in the mountains; fertile soil
between rocks; also, sunny side of mountains. 26 April—3 June.
12 Poni Creek, between Bent’s Fort and Santa Fe.
Bottom land, near the creek, amongst willows. 3 October, 1846.
13 (11) Banks of Moro River. Low fertile soil. 15 August,
1847.
14 (12) Santa Fe, creek bottom, between the mountains;
damp shady places, fertile soil. 20 May—3 June in flower. 8 July—in
fruit. Sepals 4, white; petals none; stamens 23 or 24.
15 (13) Santa Fe, creek valley, shady places, margins of
irrigating ditches at the foot of perpendicular rocks. 13 June—1 July
in flower. 19 July in fruit. Flowers dioecious.
16 “I find no specimen with this number, but suppose that
a form of the last number, Thalictrum, is meant by it.” Bottom land
of Moro River, 15 August, 1847.
BERBERIDACEAE
17 (14) 26 April-15 May, 1847. Upper part of Santa Fe
creek valley. Sides of mountains, near the top as well as the foot of
them; also on the creek bottom. Evergreen leaves, most of them
partly eaten by animals. Fruit ripens by the 28th of July.
(15) Santa Fe Creek, foot of steep and rocky banks, not
far from the water. Shrubs of from 3—4 feet high. 28-31 May, 1847.
Some specimens have the lower leaves sharp serrate.
PAPAVERACEAE
19 “One of the varieties of Argemone mexicana; I find no
specimen with this number.” 3 September, 1846. Walnut Creek,
rather low prairie. Valves of the capsula, 4, with prickles, the points
of which are directed towards the apex.
20 “Argemone mexicana, a less hispid form” 27 August,
1847. Upper Spring (Cimarron River); prairie. Valves of the capsula
3 and 4 with prickles, the points of which are directed away from the
apex.
21 (16) 7 June-28 July, 1847. Santa Fe. Low, level, sandy
places, in the neighbourhood of fields. From upwards of 2 feet high.
Socially in great numbers.
FUMARIACEAE
22 (17) 15 April-20 July, 1847. Santa Fe, creek valley, be-
‘tween rocks, generally near the creek: also, moist gravelly parts 0

AUGUSTUS FENDLER’S COLLECTION LIST 17

the creek bed, exposed to occasional inundations of the creek.
CRUCIFERAE

23 (38) On the lesser hills west of Santa Fe, sterile pebbly,
and dry soil, gently sloping declivities (both sunny and northern)
amongst stones. Fruit: 29 April—7 June.
23 (39) Santa Fe, dry gravelly hills, 24 May, 1847.
24 (40) Santa Fe, lesser hills, 2 May, 1847. A fruiting spec-
imen, very different from those of no. 23; also some differences in
the leaves.
25 (41) 15 April-1l1 May, 1847. Santa Fe; creek bed,
parts exposed to inundations; gravelly moist soil; petals 4, white.
26 (42) 8-24 May, 1847. Rio de Norte, sunny side of ra-
vines. Silicles more crowded and somewhat dotted; stem nearly smooth.
27 (46) Mountain valleys, from Santa Fe east to Rabbit Ear
Creek. 10-20 August, 1847. There seem to be 2 different forms of
this plant.
28 (44) 19 March-5 June, 1847. Santa Fe; creek bottom in
the mountains; also foot of mountains. Petals 4, white.
29 (18) 8 May—17 August, 1847. Rio Las Norte, flat, sandy
pk washed by the water; east to Mora Rive

(19) 25 May—23 June, 1847. Santa Fe: creek bottom, low
oe places.
31 (21) 29 June 2 July, 1847. Santa Fe Creek, low wet
places and margin of the c
32 (45) 16 June- 7 Tad ey Santa Fe; eastern mountain
range, near the irrigating ditch of a
33 (20) Santa Fe; wet pee and moist places of the
creek ies 27 May-29 June, 1847.
4 = “Nasturtium = 33” Santa F

cipal ae ditches
“os (27) 30 March-15 May, 1847. Santa Fe, creek valley.
d to occasional in-

e, 12 July, 1847. Moist soil

Moist, sandy soil in parts of the creek bed, expose
undations, Rarely on hills between rocks.

(24) 13-18 July, 1847. Santa Fe;
amongst rocks. ae
= (22) 7 July-10 August, 1847. Santa Fe, upper part o
a creek valley, foot of mountains.

= 27 23) 7-28 July, 1847 :
= is) 26 ik 1847. Santa Fe Creek, margin of the water.
Flowers minute.
: (24) 11 eT August, 1847.
aa Santa Fe to Veg

(25) 6 June-28 July,
isa margin of the creek;

sunny side of a hill,

Mountainous region

1847. Santa Fe, creekbottom in the
also in shady moist places.

18 ELIZABETH A. SHAW

418 (26) 24 June, 1847.

42 (32) 24 April-19 May, 1847. Santa Fe, dry gravelly hills;
petals 4, yellow; plant rather socially.

43 (33) 14-17 August, 1847. San Miguel to east of Rock
Creek. Spots of fertile and loose soil on high prairies.

44 (34) 15 May-12 July, 1847. Santa Fe Creek valley, level
somewhat damp ground, foot of mountains.

44B (35) Rio del Norte, 24 May, 1847. On a ledge of rock
high up on the steep mountain—bank of the river.

45 (28) 15 May—16 June, 1847. Margin of Santa Fe Creek
in the mountains.

(43) 8 May—28 July, 1847. Santa Fe Creek, shady steep
declivities and foot of mountains. More rarely in the creek bottom
and low banks of the creek.

47 (37) 21 June-26 July, 1847. Santa Fe, near irrigating
ditches not far from fields.

(29) 15 August, 1847. Bank of Moro River.

49 (31) 16 June-28 July, 1847. Santa Fe, margin of the
Creek, 5 miles east of Santa Fe, in the mountains.
50 (30) 16-24 June, 1847. Santa Fe Creek, foot of moun-
tains.

CAPPARIDACEAE
51 (36) 27 September, 1847. 15 miles north east of Rio de

los Animos, between Bent’s Fort and Santa Fe.

5 (48) 21-28 August, 1847. From Rio Colorado to the
crossing of the Cimmarron River. Rather low places.

53 (47) 28 August, 1847. Between crossing of the Cimmar-
ron River and Willow Bar. Deep sandy soil.

54 (49) 7 June-19 October, 1846-47. Santa Fe, low level
places (somewhat sandy) near the town, also near the Creek. East as
far as Cow Creek. Annual herb.

548 Flower white, the same.
VIOLACEAE
55 (50) 8 September, 1847. Pawnee Fork, steep bank of

the river, loose soil.

(51) 15 May-28 June, 1847. Santa Fe, creek bottom in

the mountains; shady damp places, fertile soil; sometimes margin of
the Creek.

57 (50) 26 April-28 June, 1847. Santa Fe, creek valley.
Margin of the creek and low level places.
CISTACEAE
58 (52) 13 August, 1846. 60 miles west of Independence.
HYPERICACEAE
59 (53) 16 June-28 July, 1847. Santa Fe, creek bottom in

the mountains, margin of the creek and moist places.

AUGUSTUS FENDLER’S COLLECTION LIST 19

a CARYOPHYLLACEAE
ss (63) 12 May-25 June, 1847. Santa Fe, eastern moun-

of creek bottom, near the creek, between stone piles. Sepals 5;
4 als 5 (by abortion 4), cleft, ([small sketch here] inside of a petal

owing the 2 little elevations near the base of the blade), alternate;
._" 10 (5 higher than the rest). ,
i oe nf May, 1847. Santa Fe Creek valley, foot of
ns gently sloping declivities betw t “
BR oetsks whe een stones (only in one re
a. (55) Santa Fe, Creek valley in the mountains, open
A vel tract, where cattle and horses were always grazing. Also between
— Fe and Pecos. 28 July—10 August, 1847.
i, (61) 20 May-16 June, 1847. Santa Fe, creek bottom in
“s mountains, near the creek. Petals 5, ([small sketch here] cleft)
- j of floral diagram here]. Plant viscid, especially the upper part
. (62) 16 June, 12 July, 1847. Santa Fe, Creek valley in
€ mountains between rocks near the creek.
4... (54) 2 May—3 June, 1847. Santa Fe, Creek valley, moist
evel soil. Flowers minute; petals white.

(65) 3-19 July, 1847. Santa Fe, Creek valley in the

* margin of the creek.

(64) 15-28 July, 1847. Santa Fe, Creek bottom in the
mountains.
57) 14 August, 1847. Prairies, 5 miles west of Vegas.
Flowers white.
. h (56) 14 August, 1847. 8 miles west of Ve
. the mountain region, shady places. Differs from no.
Shape of its bracts.

71 (66) 14 August, 1847.
f (58) 14 August, 1847. 2 mi
of high rocks.

(67) 24 September, 1847. A

gas, woodland
62 in size and

Woodland 6 miles west of Vegas.
les east of Moro River, at the

few miles south of Fort

eavenworth,
: (60) 10 August, 1847. A few miles east of Santa Fe.
74 ILLECEBRACEAE
(68) 21 September, 1847. Creek bottom about 20 miles
ig of 110 Creek.
an (69) 17 September, |
pr . (between crossing of the Arkans
hia gravelly soil. Also: 14 August-
7 Zuel to Council Grove; loose soil.
(70) 25 August, 1847. A few miles west of McNeess

846. A few miles east of Big Sand
as River and Bent's Fort). High
17 September, 1847. From San

Creek :

20 ELIZABETH A. SHAW

PORTULACACEAE
or (71) 28 July, 1847. Santa Fe, Creek bottom, in the
mountains, on a level tract of grazing ground.
78 (71) 25 August, 1847. Between Rabbit Ear and Mc-
Nees’s Creeks. Flower lilac.
79 (72) 3 September, 1847. Sand hills 4 miles south of the
crossing of Arkansas River. Flower fine red, the size of a dime.
80 (73) 3 September, 1847. Crossing of the Arkansas River,
sand hills south of that river. Flower red, near the size of half a
ime.

(74) 3 August, 1847. Santa Fe, common in waste places

about town. Flower, deep yellow.

MALVACEAE
82 (75) 13 June-8 August, 1847. Santa Fe, yards and waste
garden places.
83 (76) 31 August, 1847. A few miles west of Lower Spring

= 85 (76) “the same as last” 8 September, 1847. Banks of
Pawnee Fork.

(77) 24 & 25 August, 1847. Low moist places at Rabbit
Ear and McNees’s Creeks.

(79) 30 June—31 July, 1847. Santa Fe, moist meadows.
87 (80) 24 June-28 July, 1847. Santa Fe, Creek bottom,
near the creek.

(81) 8 June—30 July, 1847. Santa Fe, neighbourhood of
fields and waterducts.

89 (78) 7 June—4 August, 1847. Santa Fe, wet meadows
and near the walls of gardens and fields.
9 (82) 24 May-3 September, 1847. Santa Fe, plains and
foot of the lower hills, near ditches. East to Rock Creek. Also Poni
Creek, between Bent’s Fort and Santa Fe.

LINEACEAE (sic)
91 (83) 23 May, 1847. Santa Fe, level rather fertile tracts;
petals 5, blue.
92 (83) 15-26 August, 1847. Between Vegas and Moro River
and a few miles east of Cold Spring (Cimarron).
93 (85) 4 May-18 July, 1847. Hills about Santa Fe, gravelly
dry soil. Flowers fugacious in the highest degree.

(85) 3-11 June, 1847. Santa Fe, in an old dry irrigating
ditch. This plant seems to differ from the preceeding one in having
the claw of the petals of the same colour of the blade, while the
preceeding one has the upper part of the claw brown. “This 85 may

a mere variety of 84 or a peculiar species which could be named
L. puberulum; probably the first.”

AUGUSTUS FENDLER’S COLLECTION LIST 21

95 (85) 29 August, 1847. Between Willow Bar and Middle
Spring (Cimarron).

(85) 16-26 August, 1847. Moro River to Cold Spring
(Cimarron River).

(84) 29 August, 1847. 16 miles east of Middle Spring
(Cimarron River).

(86) 15-20 September, 1847. Prairie, between Turkey
Creek and 110 Creek.

GERANIACEAE

99 (87) 15 April-20 October, 1846-1847. [i.e., April, 1847]
Santa Fe, Creek bottom and foot of hills not far from the creek.
Found a plant of this kind in flower even as late as 21 December
when the greater part of the creek valley was covered with deep
snow.
100 (88) 3 June-28 July, 1847. Santa Fe, Creek valley, shady
places, moist soil near the creek. I never found this plant growing in
very dry localities.
101 (89) 31 May-28 July, 1847. Santa Fe Creek, near irri-
gating ditches at the foot of mountains.
02 (90) 18 August, 1847. 5 miles east of Moro River, bot-

tom land.
208 (89) 18 August, 1847. Seems to differ from no. 101 only
In its color. 6 miles east of Moro River.

OXALIDACEAE
i (91) 7 July-8 August, 1847. Santa Fe, margin of irri-
gating ditches and in low wet places.
es “Marsilea without fructification!” 24 May, 1847. Rio del
Norte. Leaves floating on the surface of the water in shallow places.

(92) 7 September, 1847. Between Coon Creek and Paw-

nee Fork. Road side.

BALSAMINACEAE
a (93) 14-24 August, 1847. 90 miles west 0
dence to Council Grove, creek bottom, rich soil.

ZANTHOXYLACEAE (sic)

- (94) 25 May, 1847. Rio del Norte. Steep and rocky
banks (trap formation). Shrubs from 7 to 10 feet high, leaves of an

f Indepen-

agreeable odor. ee
i (95) 8 May, 1847. Rio del Norte. Deep ravines in the
Mountains.

(96) 21 September, 1847. Kansas River, bottom land, a

few miles ferry.
south west of the upper terry Creek valley. steep rocky

1]
(97) 8 July, 1847. Santa Fe,
Mountain sides. . Mn 1847. Fruit specimen, from the preceeding

year, in the same locality.

22 ELIZABETH A. SHAW

111 (98) 11-24 June, 1847. Santa Fe, Creek valley in the
mountains, shady bank. Upwards of 4 feet high. Socially.
112a (99) 25 May, 1847. Fruit specimens. Rio del Norte. Pre-

cipices of the mountains and steep rocky banks of the river. Shrub
from 3-8 feet high, possesses a peculiarly disagreeable odor. The ber-
ries of this shrub leave greasy spots on the paper in being dried. Also:
16 August, 1847. Moro River; rocky hillside; ripe fruit.
112b (99) 8 May, 1847, Rio del Norte. Flower specimens
from the same regions as no. 11a.

ZYGOPHYLLACEAE
113 (100) 24 May-6 August, 1847. Rio del Norte and Santa
Fe. Open level places, loose soi] about town, near fields and waste
places; adnate to the ground. The leaves of this plant when wet im-
part to the paper a yellow dye.

ACERACEAE

ll4a (102) male specimen, 22 April, 1847. Santa Fe, eastern
mountain range, creek valley, near the margin of the creek and in
low situations at a small distance from the same. Large trees, lower
part of the trunk generally very knotty, which seems to arise from
the many wounds the tree receives early in spring, in order to draw
the sap from it. This sap is gathered in holes cut in the trunk of the
tree at a little distance from the wounded places.
ll female specimens.

114c leave [sic] and fruit specimens. 3 April-12 May, 1847.
115 (102) 8 September, 1847. Pawnee Fork, creek bottom.
116 (101) 12-18 May, 1847. A shrub 15 feet high, of slender

growth, in shady places near the creek, eastern mountains, Santa Fe.
31 May-6 June, fruit.

CELASTRACEAE
117 (103) 17 August, 1846, 95 miles east of Council Grove.
118 (104) 20 September, 1847. Shady bottom land of 110
Creek. Somewhat twining.
119a (106) flower specimen: 24 & 25 June, 1847. Santa Fe,

eastern mountains, sunny side. fryit specimen: 28 July, 1847, in fruit.
A shrub about 11, feet high and 2 feet in diameter.

119b (105) 15 May-3 June, 1847. Santa Fe, Creek valley in
the mountains. Foot of mountains, shady steep declivities, in soil con-
sisting of decomposed rocks intermixed with vegetable mould. An
evergreen shrubby plant about 1 foot high. Their roots consisting for
the greater part in the older branches which by coming in contact
with the ground send down rootlets and are therefore but slightly
covered with soil.

TACEAE

VI
120 (108) 17 August, 1847. 2 miles east of Moro River, at the
foot of high rocks.

AUGUSTUS FENDLER’S COLLECTION LIST 23

121 (108) 28 July, 1847. Santa Fe, creek valley, between
rocks.
122 (107) 25 May, 1847. Rio del Norte. Steep, desolate places

of its high rocky bank, near the water, with their tendrils fastened
to other shrubs.
POLYGALACEAE

123 (109) 13 August, 1847. Low prairie, a few miles west of
Vegas.
124 (110) 25-28 August, 1847. From Rabbit Ear Creek to
Willow Bar. Road side.

LEGUMINOSAE
125 (111) 1 October, 1846. Between Colorado and Bermejo

Creeks, level prairie, good soil.

1 (111) 3 June-10 July, 1847. Santa Fe, creek valley, foot
of mountains.
127 (113) 8 May, 1847. Rio del Norte. Shady places, in deep
ravines between rocks.

128 (111) 7 July, 1847. Santa Fe, creek valley.

129 (111) 17 August, 1847. 2 miles east of Moro River.

130 (112) 25 June, 1847. Santa Fe, neighbourhood of a field.
131 (115) 8 May-25 July, 1847. Santa Fe, creek valley; most

abundant near irrigating ditches in the lower part of the valley. Flow-
ers of different colors, but most generally of a deep crimson which

turns into deep blue in drying. This plant I have never observed to
e eaten by any domestic animal. In fruit, 30 June.

132 (116) 31 May-3 June, 1847. Santa Fe, lower part of
mountainsides. Flower of a rusty color.

133 (117) 5 September, 1847. 3 miles west of Fort Mann
(near the Arkansas River). Banks of a creek.

134 (119) 28 August, 1847. Willow Bar, low wet bank of the
creek.

135 (121) 18 August, 1846. Low places in the prairies, 12

miles east of Council Grove. Also: Santa Fe, 23 June—4 July, 1847.
Creek valley and in waste places about town and near fields. Plant
viscose and leaves a print of its shape in green color on the paper.

136 ‘ (122) 14 August-9 September, 1847. From San Miguel
to Cow Creek, high priaries.

137 (123) 29 August, 1847. Between Willow Bar and Middle
Spring (Cimarron River). -

138 (124) 14 October, 1846. Santa Fe prairie.

139 (125) 14 August, 1847. 7 miles west of Vegas, in wood-
land. Plant 2-3 feet high.
140 (126) 17 August, 1846-8 September, 1847. 120 miles

west of Independence to Pawnee Fork. Shrub from 3 to 5 feet high.
14] (127) 10 July-15 September, 1846-1847. n.b.: 15 Sep-

24 ELIZABETH A. SHAW

tember, 1846 Santa Fe, margin of irrigating ditches, near fields. Also:
low prairies near the Arkansas River.

142 (132) 19 May-19 June, 1847. Santa Fe. Dry gravelly hills,
sunny sides. A shrub of dwarfish growth, 1/2 to 1 foot high; stem and
branches very crooked and confusedly entangled in each other. Fruit
specimen: 21 June, 1847.

143 (133) 21 May-25 August, 1847. Santa Fe, sunny sides
and foot of stony gravelly hills. East of McNees’s Creek.

144 (131) 27 August-3 September, 1847. Upper Spring (Cim-
arron River) to Walnut Creek. Prairie.

145 (129) 15-27 August, 1847. From Vegas to Upper Spring
(Cimarron). Prairie.

146 (130) 29 August, 1847. Willow Bar (Cimarron), sandy soil.
147 (128) 28 August, 1847. Between the Crossing of the Cim-
arron River and Willow Bar; low sandy places on the prairie.

148 (127) 20 September, 1847. A few miles west of 110 Creek.
Prairies.

149 (134) 28-29 August, 1847. Between the Crossing of the
Cimarron and Middle Spring. Sandy soil.

150 (135) 11-26 July, 1847. Santa Fe, near fields. 14 August,

1846. About 90 miles west of Independence; steep side of a rocky
bluff. 20 August, 1847. Middle Spring (Cimarron).

151 (136) 21 September, 1847. Prairies a few miles southwest
of the upper ferry of the Kansas River.

152 (137) 14 August, 1847. 5 miles west of Vegas.

153 (138) 13-29 August, 1847. Between Willow Bar and Mid-
dle Spring (Cimarron). Sandy soil. Also: Ojo de Bernal.

154 (139) 30 August, 1847. 18 miles west of Lower Spring
(Cimarron). Prairie.

155 (140) 3 October, 1846. Poni Creek (between Bent’s Fort

and Santa Fe). Also: 15 June, 1847. Santa Fe, creek bottom, near the
water.

156 (141) 7 June and 19 October, 1846-1847. [i.e., October,
1846 & June, 1847] Santa Fe, low somewhat wet places about town.
157 (142) 23 June, 1847. Santa Fe; enclosed field, amongst
weeds and cornstalks.

158 (143) 2-6 September, 1847. Between Fort Mann (Arkan-
sas River) and Walnut Creek; high prairie.

159 (146) 24 April-9 May, 1847. Plains about Santa Fe, red

sandy soil; socially, of rather low depressed growth, but in patches
from 1-2 feet diameter. I never observed that this plant had been
eaten by any animal whate

1 (147) 10-15 viene 1847. Together with the preceeding
one, flower white.

AUGUSTUS FENDLER’S COLLECTION LIST 25

161 (171) 15-26 August, 1847. From Moro River to Cold
Spring (Cimarron). Prairie.
162 (148) 22 April-28 May, 1847. Santa Fe, gravelly hills and

lesser mountains amongst rocks.
(149) 8-25 May, 1847. Rio del Norte and Santa Fe.
Stony hills.
164 (151) Santa Fe, 28 May, 1847. Dry gravelly hills.
)

165 (156) 25 May, 1847. Rio del Norte. Near its bank.

166 (152) 26 August, 1847. Between McNees’s Creek and
Cold Spring (Cimarron).

167 (153) 13 August, 1847. Neighbourhood of Ojo de Bernal.
168 (154) 5 August, 1847. Moro River, prairie.

169 (155) 25 August, 1847. Between Rabbit Ear and Mc-
Nees’s Creeks.

170 (150) 15 August, 1847. Moro River, prairie.

171 (144) 21 August, 1847. Between Colorado and Rock Creek.
Low Prairie.

172 (157) 10 August, 1847. Between Santa Fe and Pecos.
Woodland in the mountains.
173 (161) 8-25 May, 1847. Rio del Norte; loose sandy soil,

not far from the river, especially around and between low shrubs.

Flower sometime white.

iA (159) 19 April-28 June, 1847. Santa Fe; sunny and steep
declivities of stony hills.

: (159) 28 May-28 June, 1847. Santa Fe.

156) 24 May, 1847. Santa Fe.

(
ale l lv. 1847. Santa Fe, hills.
tees nny hillside, between

178 160) 3-13 June, 1847. Santa Fe; su
rocks, under protection of shrubs.
ey (164) 16 September, 1847. Cottonwood Creek to Kansas
River: dry Aa,

; prairies.
ay (165) 24 September, 1847. 8 miles south of Fort Leav-
enworth; gullies along the road side.
od (166) 8 May 30 June, 1847. Santa Fe. Level places on

the neighbourhood of fields and irrigating ditches; rather socially.
181b

Fruit: 30 June-2 August, 1847.
an (167) 21 Ree 1847. Santa Fe. Flower light rose ert
~ (168) 24 June-8 July, 1847. Santa Fe, creek valley in the
Mountains, foot of mountains.

: (169) 3 October, 1846. Poni Creek, between Bent’s Fort
and Santa Fe: low prairies.

: (170) 2-May-3 June,
above Santa Fe. Foot of mountains, an
: (172) 18 September, 1847.

1847. Santa Fe, creek valley 5 miles
d in level moist places.
Council Grove; creek bottom.

26 ELIZABETH A. SHAW

187 (173) 28 August, 1847. A few miles west of the ford of
Cimarron River and at Willow Bar, road side.

188 (174) 31 August, 1847. A few miles west of the Lower
Spring (Cimarron).

189 (180) 14 August, 1847. Between San Miguel and Vegas.
190 (179) 4 October, 1846. A few miles south of Rayado

Creek (between Bent’s Fort and Santa Fe), hillside, gravelly soil. 10-
25 August, 1847. A few miles east of Santa Fe to McNees’s Creek.

191 (178) 14 August, 1846. Black Jack Point; low prairies. 29
August, 1847. Between Willow Bar and Middle Spring (Cimarron).
192 (177) 21 August, 1846. Council Grove, grassy hillside.
193 (181) 27 August, 1847. Upper Spring (Cimarron), hillside.
194 (114) 8 August, 1847. Santa Fe, cultivated.

195 (145) 11 August, 1846. 2 miles south of Kansas River;

woods, roadside. 15 August, 1846. About 100 miles west of Inde-
pendence; on the more elevated places of the creek bottom.

196 (175) 21 August, 1846. Council Grove, low places. Plant

from 4-6 feet high.

197 (118) 24 August, 1846. Council Grove, sloping bank of

the creek and in the bottom land.

198 (120) 24 August, 1846. Council Grove, creek bank.

199 (162) 16 August, 1846. 117 miles west of Independence;

bank of a creek and in bottom land.

200 (163) 16 August, 1846. 117 miles west of Independence;

bank of a creek and in bottom land.

200° (176) 7 June, 1847. Santa Fe, gravelly hillside, socially.
ROSACEAE

201 (182) 22 August, 1847. Rock Creek; creek bottom.

202 (186) 17 August, 1847. 2 miles east of Moro River; foot

of high rocks.

203 (183) Santa Fe, creek valley; at no great distance from

houses. Trees of about 30 or 40 feet high, branches spreading very
much. 20 April, 1847. in flower. 20 May, 1847. in leaf.

204 (185) 28 May-3 June, 1847. Santa Fe Creek; steep rocky
banks, at the margin of the creek. 15 July in fruit.

205 (187) 2-16 June, 1847. Santa Fe Creek, foot of hills. A
shrub about 1 foot high.

205° 28 July, 1847, in fruit.

206 (190) 28 May-12 July, 1847. Santa Fe Creek, bottom
land, margin of the water; fertile soil.

207 (192) 16 June, 1847. Santa Fe, margin of the creek, in
the mountains.

208 (193) 24 May—4 October, 1846-7. Rio del Norte and Ray-

ado Creek (between Bent’s Fort and Santa Fe). Sides of ravines and

AUGUSTUS FENDLER’S COLLECTION LIST rigs

steep rocky banks of the river. Shrub fi
rom 2 to 5 feet high. [N.B.:
Rayado Creek = 4 October, 1846; Rio del Norte = 24 ‘aes be

209
—— (195) 3-28 July, 1847. Santa Fe, creek bottom, near the
210
eh (196) 11 August, 1846. 18 miles south of Fort Leaven-
21 (191) 17 A j
a ) ugust, 1846. 15 miles east of Council Grove;
21
a - (197) 24 June-15 July, 1847. Santa Fe, creek valley; sunny
i mountains between rocks, 11 miles above Santa Fe. Plant vis-
2 (201) 16 August, 1847. Moro River, hillside.
a (202) 25 June-8 July, 1847. Santa Fe, creek valley in the
“a ntains, 9 miles above Santa Fe; foot of mountains, sunny side.
of hill (199) 1-15 July, 1847. Santa Fe, creek bottom; also foot
an ills not far from the creek, sunny side.

(203) 22 June—26 July, 1847. Santa Fe, creek bottom,
_ of the creek.
a (198) 16-25 June, 1847. Santa Fe, creek bottom, moist
2
18 (200) 15 May—24 June, 1847. Santa Fe, creek bottom and
eg meadows.
a (197) 19 October, 1846. Santa Fe, damp places.
ne (204) 3 June, 1847. Santa Fe, creek bottom, at some dis-
on from the water.
th (205) 16 June-15 July, 1847. Santa Fe, creek valley in
oo mountains 9 miles above Santa Fe, foot of mountains.

(206) 26 April—3 June, 1847. Santa Fe Creek. Sunny de-

ly at the foot of mountains,

cliviti
ivities of the mountains; more frequent
and intermixed with

— in the valley not far from the creek. Socially

8 following one.

| (207) 22 April-15 May,
ilar situations with the preceding one,

1847. Santa Fe, creek valley. In
though generally not so high

ig the mountains.

shad (208) 11-26 June, 1847. Santa Fe, bank of the creek,
peg y places. 28 July, in fruit.

. Fe, creek valley in

. (209) 31 May-24 June, 1847. Santa

are ountains, shady place near an irrigating ditch. Petals 5, alter-

295 perpendicular and at right angles to
(210) 16-24 June, 1847. Santa

the sepals.
Fe Creek, near its mar-

sie (211) 22 August, 1847. Rock Creek, creek bank amongst
ig shrubs.

(212) 22 August, 1847. Rock Creek, bottom land.

28 ELIZABETH A. SHAW

228 (212) 15 August, 1847. Banks of Moro River.

229 (213) 7-8 July, 1847. Santa Fe, creek valley in the moun-
tains, 9 miles above Santa Fe, sunny side of mountains.

230 (184) 30 April, 1847. Santa Fe, neighbourhood of houses.
2307 (188) 15 July, 1847. Santa Fe, upper part of the creek;
steep sunny sides of mountains. Shrub about 4 feet high.

230° (189) 21 August, 1847. Rio Colorado, thin layers of soil
upon and between ledges of rocks, near the river bank.

2304 (194) 3-15 July, 1847. In fruit. Santa Fe Creek valley in

the mountains; sunny side of steep rocky mountains. A shrub 8 to 10
feet high.

LYTHRACEAE
231 (214) 25 August, 1846. Council Grove, creek bed.
232 (214) 18 September, 1847. Bluff Creek bottom, about 16

miles east of Council Grove. 20 September, 1847. Bottom of 110

233 (215) 21 August, 1846. Council Grove.

234 (215) 11 August-21 September, 1846-7. 18 miles south
of Fort Leavenworth, and prairies a few miles southwest of the upper
ferry of Kansas River. [N.B. 11 August, 1846 = 18 miles south of
Fort Leavenworth. |

ONAGRACEAE
235 (216) 24 June—28 July, 1847. Santa Fe, Creek valley in
the mountains, in ravines; flowers purple.
236 (217) 16 June-30 September, 1846-7. Santa Fe, creek

bed, gravelly wet places near irrigating ditches, grassy margin of the
creek. Also: Raton Creek (between Bent’s Fort and Santa Fe) creek
bottom in the mountains. [N.B. 30 September, 1846 = Raton Creek;
16 June, 1847 = Santa Fe.]

237 (230) 24 May-18 July, 1847. Santa Fe, sunny hillside;
also: Rio del Norte, bank of the river; flower of a sulphur color, which
turns to be red in a few days after opening and in drying.

238 (222) 28 May-12 July, 1847. Santa Fe, creek valley. In
different localities on soil that was dug up in the spring to elevate the
sides of an irrigating ditch. Also: on the sunny side of a small hill in
places where the soil was full of ants. Petals white, but turning into
red when drying.

239 (223) 8-10 May, 1847. Rio del Norte, gravelly soil near
the river. Petals white, showy. Also: Santa Fe.

40 (218) 24 June, 1847. Santa Fe, creek valley in the moun-
tains; foot and sunny sides of mountains. Plant 2-3 feet high.

241 (219) 19 September, 1847. Prairie between Bluff Creek
and 110 Creek.

AUGUSTUS FENDLER’S COLLECTION LIST 29

a (220) 12 August, 1846. Dry prairies, 20 miles south of
8 ower ferry of the Kansas River. Plant 4 feet high.
(223) 19 June-28 July, 1847. Santa Fe. In and near fields

a ae elevated places, Flowers white.
a the same with white flowers. 12 July, 1847. Santa Fe.
oe (221) 3-28 July, 1847. Santa Fe, hillside near a field.

(230) 28 August, 1847. A few miles west of the ford of
the Cimarron River.
246 (230) = 237, foliis angustioribus. 23-31 August, 1847. 7

miles west of Rock Creek to a few miles west of the Lower Spring
(Cimarron River).

(228) 16 June-19 July, 1847. Santa Fe, creek bottom.
ae (229) 25 August, 1847. McNees’s Creek. Sandy hillside.
(229) 21-28 August, 1847. Low prairies from Rio Colo-
5 to the ford of the Cimarron River.
= (229) 13-15 August, 1847. Ojo de Bernal to Vegas.
(224) the same as 243, fol. integr. 23 June-18 July, 1847.

ag Fe, near irrigating ditches. Also: 28 August, 1847. Between the
ord of the Cimarron River and Willow Bar. Sand.

i (224) the same as last, more canescent. 29 August, 1847.
gg 10 miles east of Middle Spring (Cimarron).

(225) 21 September, 1847. Creek bottom, about 20 miles
east of 110 Creek.
(225) 17 August, 1846. 20 miles east of Council Grove,
ry prairies.
= (226) 24 August, 1846. Council Grove, low places.

(227) 1 September, 1847. Sand Creek, margin of a low
ate place.

; (231) 17 May-2 August, 1847. Santa Fe, near fields. Se-

light rose color changing to deep

pals 4, reflexed; petals 4, alternate,
hanged to yellow. 3 Sep-

_ June, 1847. Santa Fe. Petals white c
:. er, 1846, Walnut Creek, rather low prairie.
dl (232) 28-29 August, 1847. Between Willow Bar and Mid-
a Spring (Cimarron).

for (233) 12 August, 18
ae of the Kansas River. Bottom land.

(234) 27 August, 1846. 30 miles west of Council Grove;
land of Moro River. Plant up-

46. A few miles south of the lower

ae bank. 15 August, 1847. Bottom

as of 15 feet high at Willow Bar.

nut C (234) 1 September, 1846. Between Cow Creek and Wal-

261 reek, low prairie. Flower white.
(235) 21 September, 1847. 10 miles southwest of the up-

per ferry of Kansas River, prairie. Plant 4-5 feet high.

30 ELIZABETH A. SHAW

262 (236) 18 September, 1847. Bluff Creek, bottom land,
near the water.
263 (237) 12 August, 1846. A few miles south of the lower

ferry of the Kansas River.
264 (238) 15 May-7 June, 1847. Santa Fe, in the water of a
pond near its margin.

LOASACEAE
265 (239) 1 August, 1846. Council Grove, on hills between
rocks. [N.B.: date must be incorrect; Fendler left Fort Leavenworth
only on 10/11 August. |

266 (242) 21 June-19 July, 1847. Santa Fe, near fields. Flower

very seldom open in the daytime.

267 (240) 21 August, 1847. Between Rio Colorado and Rock

Creek, low prairie. Stem 3-4 feet high, much branching.

268 (242) 17 September, 1846. Gravelly and pebbly hillsides,

from the ford of the Arkansas River to Bent’s Fort.

269 (243) 17 August, 1847. 2 miles east of Moro River, low

prairies

270 (240) 25 August, 1847. McNees’s Creek. 2 September,

1847. Pawnee Fork; prairies. Stem 4 feet high, flower vespertile.
CACTACEAE

271 (244) 8 July, 1847. Santa Fe. To cactus n. 1. Eastern

mountain range at a moderate height and near the foot of mountains,
between pieces of rock. Also: frequently on the road from Bent’s Fort
to Santa Fe, on hills and high prairies. Plant not more than 2 inches
above the ground. Subglobose or obovate with the upper part hem-
isphaerical. Either solitary or in pairs. There are some in which the
root sends up several subterranean branches which ramify again and
again in elongating, until they reach the surface, where each ramifi-
cation bears a globose plant at its apex. Spines straight, radiating at
right angles to the axes of their respective mammaeform or conical
tubercles, white, thin, more like fine bristles with the exception of
from 2-4 stronger ones, which are mostly of a red color and make
larger angles with the axes of their respective tubercles.

272 (247) 15 May, 1847. To cactus no. 3. Santa Fe. Higher
regions of the mountains, steep and rocky sunny declivities. Also: on
elevated plains somewhat sandy. Plant generally growing in large clus-
ters; the young plants growing from the sides of the old one, in the
dividing line of the subterranean part of the stem from that part
above the ground, that is, close to the soil. Sometimes in clusters of
14 individual plants. Alternate ribs and furrows, the ribs beset with
radiating spines, (generally 9 in number), nearly in a tangential plane
to the apex, and a tenth one in the center, erect, much longer than
the rest and of a darker color. Ribs 10, rarely 11.

AUGUSTUS FENDLER’S COLLECTION LIST 31

val (249) 9-21 June, 1847. To cactus no. 4. Santa Fe, high
leve somewhat sandy prairie. Young plants growing from the old ones
in the same manner as in the preceding no. Plant: 10 ridges, (gen-
erally) 7, radiating, tangential rather stout spines, with a larger some-
what curved and oblique standing one in the center and pointing
more or less towards the apex of the plant. The younger spines are
= or less dark colored, of a lighter shade near the apex. Many of
e spines are more or less dark colored, of a lighter shade near the
a Many of the spines are flattened.
a (248) 31 May-13 June, 1847. To cactus no. 5. Santa Fe.
igher regions of the mountains, steep and rocky sides. Also: on el-
: broken, somewhat sandy plains. 7 ridges, mostly 3 to 5, some-
. curved, stout, flattened spines with a sharp edge or ridge run-
: ng on the inside from the base to the apex; length of spines about
inch; no central one.
275 (251) 21 June, 1847. To cactus no- 6. Santa Fe. 3 to 8

miles southwest of the town. Elevated plains, somewhat sandy soil;
I never found them

obovate.
P02) 10 June-20 me 18 i
¢ of the mountains, on moderate hills and high prairies.
ot 6 feet high and the lower part of the stem in some instances 5
inches diameter. Spines sheathing. Fruit lateral as well as terminal,
either dry and hard, or soft and yellow. 92 December, 1846. The
yellow fruit adheres to the branches as strong @° ever, and has not
changed in its appearance in the least since the 1 November. This
nd of cactus seems to stand the cold much better than the flat
jointed kinds, no. 7 and 8. They are sometimes used on t
mud walls, for the same purpose as broken glass bottles are used in
other countries.
. (246) 4 May-9 June, 1847.
ern mountain range, sunny declivitie

47. To cactus no. 9. Santa Fe.
Upwards

To cactus no. 11. Santa Fe.

s of the smaller mountains,

between rocks.

ha (245) 15-21 May, 1847. To cactus no. 12. Santa Fe, val-

ey of the lesser hills; loose sandy clay, seemingly good soil.
GROSSULARIACEAE

ee (256) 16 August, 1847. Moro River, rocky hillside.

32 ELIZABETH A. SHAW

281 (256) 30 May, 1847. 5 miles southwest of Santa Fe. In
a dry valley between the lesser hills. Leaves viscose and of a peculiar
odor.

282 (257) 15-22 August, 1847. From Moro River to Rock
Creek, bottom land and near the banks. 3 to 5 feet high.
283 (254) 25 May, 1847. Rio del Norte. Steep and rocky
banks of that river. Shrub about 4 feet high.
284 254) = 283. 8-25 May, 1847. Rio del Norte, ravines.
285 254) 29 May, 1847. Santa Fe. Shrub 3 feet high.
286 255) 12 May, 1847. Santa Fe, mountains.
287 253) 11-28 May, 1847. Santa Fe, margin of the Creek.
288 (255) 2-12 May, 1847. Santa Fe Creek, shady bank, a
few steps from the creek. Shrub 4 feet high.

CUCURBITACEAE
289 (258) 14 August, 1846. Hickory Point, about 80 miles
west of Independence.
290 (259) 22 August, 1847. Rock Creek, bottom.
291 (260) 21-30 June, 1847. Santa Fe, margin of a pond and
near fields in hollows.
292 (261) 19-31 July, 1847. Santa Fe, fields.

CRASSULACEAE
293 (262) 3 August, 1847. Santa Fe, sides of rocks in the

mountains near the Creek. Damp rich soil. Sepals 5, white; petals
none; stamens 10? Flowers minute; leaves thick, fleshy.

294 (263) 18 September, 1847. Bluff Creek, bottom.
SAXIFRAGACEAE

295 (264) 16 June-28 July, 1847. Santa Fe; northern decliv-

ities of mountains, on rocks. Leaves green throughout the winter.

296 (265) 12 May-24 June, 1847. Santa Fe Creek. Steep

shady rocky banks.

297 (266) 24 June-8 July, 1847. Santa Fe Creek, sunny steep

sides of the mountains between rocks. 11 miles above Santa Fe.

297? (267) 6-22 June, 1847. Santa Fe Creek bank near the

water, where the creek is walled in on both sides by high rocks.
UMBELLIFERAE

298 268) 18 September, 1847. Bluff Creek, bottom.

299 269) 18 September, 1847. Bluff Creek, bottom.

300 270) 15 August, 1847. Moro River bank.

301 271) 8 August, 1847. Santa Fe Creek bottom.

302 272) 2-25 June, 1847. Santa Fe Creek, fertile soil, mar-

gin of

303

water.
(273) 16 June-8 July, 1847. Santa Fe Creek, margin of
er.

304 (274) 24 April-30 May, 1847. Santa Fe. Gently sloping

AUGUSTUS FENDLER’S COLLECTION LIST oo

declivities amongst the gravelly hills. Rather good soil.
305 (275) 24 April, 1847. Santa Fe. Socially and together with
the preceding one, found only in one locality.
306 (278) naturalized around Santa Fe [N.B.: entire entry in
hand of A. Gray. ]
307 (276) 26 April-3 July, 1847. Santa Fe Creek, sunny de-
clivities, foot of mountains.

8 (277) 17 August, 1847. 2 miles east of Moro River. El-
evated rocky region.

CORNACEAE
309 (279) 20 September, 1847. Bottomland of 110 Creek.
310 (280) 28 May-13 June, 1847. Sandy, steep and rocky
bank of Santa Fe Creek, close to the water. Shrub about 12 feet high.
LORANTHACEAE
311 (281) 4 November, 1846; 17 March, 1847. Neighbouring

hills about Santa Fe and elevated plains. I found this plant growing
only upon the branches of the two kinds of shrub cedar that grow
about Santa Fe, and on one other tree. It occurs sometimes in clusters
of over 1 foot diamter and 3/4 foot high. Wherever they are found,

berry is globose of a light champaig
dried, glossy. Berries on short branchlets, 1,
lines of dilation of the tops of any pair 0
angles to each other. Branches more wrinkle
312 (282) 2 March, 1847. Santa Fe, mountain range; decliv-
ities of the lower mountains. Only on Coniferae no. 830. in much
smaller and thinner clusters than the preceding one. Branches some-
what quadrangular. Berries smaller, on styles. The branches of the
tree, on which this plant grows, seem to be much injured by it.
nS (283) 7 March-20 April, 1847. Santa Fe, mountain range-
Lower part of the mountains in ravines. Only found on Coniferae no.
831. The branches of the tree, on which this plant grows, are likewise
injured by it.

CAPRIFOLIACEAE
314 (284) 8 September, 1847. Pawnee Fork. Creek bottom.
oe (285) 2 October, 1846. Poni Creek (between Bents Fort
cha Santa Fe). Creek bottom
- ORAL) 7 [uly Ie? anata Re ices CO
317 ms Per: foie 1847. Santa Fe Creek, margin of the

water. A shrubby plant about 4 feet high.
RUBI

ACEAE

318 (287) 16 June-7 July, 1847. Santa Fe Creek, shady bank
near an irrigating ditch.

319 (088) 15-28 July, 1847. Santa Fe Creek valley. Sunny

34 ELIZABETH A. SHAW

side of high mountains. Petals 4, pale yellow.

320 (289) 16-29 June, 1847. Santa Fe; wet places near irri-
gating ditches.

S21 (290) 12-23 August, 1846. From Lone Elm west to Council
Grove. Low prairie. also: near a pool of standing water.

322 (291) 17-30 May, 1847. 7 miles southwest of Santa Fe.
Foot of dry gravelly hills, tolerably good soil. Flower bluish red.
323 (292) 26 August, 1847. Between McNees’s Creek and

Cold Spring (Cimarron). Prairie.
324 (292) 19 August, 1846. Council Grove, on hills, in crev-
ices of rocks.

VALERIANACEAE
325 (293) 15-18 August, 1847. Vegas to Moro River. Elevated
rocky region.
326 (294) 22 April-20 May, 1847. Santa Fe Creek; foot of.

mountains, shady steep declivities; rich soil in corners of the rocks.
Flower branchlets each pair at right angles to the next pair below or
above. Inflorescence centrifugal. Each flower branchlet is subtended
by : bract. Petals 5, light rose colour, nearly white. 3-6 June, 1847,
in fruit.

COMPOSITAE
ont 295. Vernonia noveboracensis var. 3 September, 1847.
Ford of the Arkansas River. Bottom land.
328 299. Liatris punctata. 14-28 August, 1847. San Miguel

to Willow Bar (Cimarron). Prairies.

isi : 299. Liatris punctata. 14 August, 1846. Hickory Point;
prairies. Leaves larger and more narrow; bract of the flowers more
pointed.

328¢ 299. Liatris punctata. 30 August-15 September, 1846-7.
Walnut Creek to Cottonwood Creek; dry prairies. Flower spikes more
crowded; bracts of the flowers more acuminate. In one specimen
flower white.

329 300. Liatris spicata. 30 August, 1847. Middle Spring
(Cimarron). Low prairie. Plant 4—6 feet high.
330 301. Liatris pycnostachya. 11 August, 1846. 18 miles

south of Fort Leavenworth; dry prairies.

331 . Liatris squarrosa § intermedia. 15 September, 1847.
Prairies between Turkey Creek and Cottonwood Creek.

332 297. Liatris squarrosa. 8 miles south of Fort Leaven-

worth.

333° 302. Liatris scariosa. 10 August-16 September, 1847-46.
Region about Council Grove and to Fort Leavenworth; high prairies.
333b 302. Liatris scariosa. Council Grove; high prairies. In-

florescence different from the preceding one.

AUGUSTUS FENDLER’S COLLECTION LIST 35

334 445. Tetradymia inermis. 26 July, 1847. Santa Fe; sunny
oo. of a hill. Shrub 3 feet high, the only one shrub of this kind seen
y me.

335 306. Brickellia grandiflora. 22, 23 August, 1847. Near
Rock Creek; steep rocky side of a ravine.

336 identified in the collection list as Eupatorium agera-
toides, but not cited in “Plantae Fendlerianae.” 18 September, 1847.
Council Grove; creek bottom.

337 311. Eupatorium perfoliatum. 24 September, 1847. 8
miles south of Fort Leavenworth; low prairie.
338 identified in the collection list as Eupatorium purpu-

reum, but not cited in “Plantae Fendlerianae’ 23 September, 1847.
Creek bottom, north of Kansas River.
339 309. Clavigera brachyphylla. 17 August, 1847. 2 miles
east of Moro River; foot of high rocks.

304. Kuhnia eupatorioides. 17 August, 1846. 18 miles
east of Council Grove.

352. Linosyris graveolens. 23 September,
Fort, gravelly dry hills. A shrub 2 to 4 feet high.
342 305. Kuhnia eupatorioides gracilis. 15 August, 1847.
Bottom of Moro River. 2 to 3 feet high.
343 identified in the collection list as Kuhnia eupatorioides,
» 7 § September, 1847. Be-

1846. Bent’s

but not cited in “Plantae Fendlerianae
tween Cow Creek and Pawnee Fork. Prairies.
344 309. Eupatorium purpureum. 15 August, 1846. About
100 miles west of Independence; creek bottom.
310. Eupatorium ageratoides. 15 August, 1846. 100 miles
west of Independence; creek bottom.
308. Brickellia californica. 1
hillside between rocks. Also: 18 August,
River, bottom land.
al so Gciebelin fen ee ee fm Fe re
11 miles above Santa Fe, foot of mountains, sunny side.
4 q 335. Diplopappus ericoides. 10 May-19 June, 1847. Santa
e, side of ravines in very dry places.
349 306; 1 paps B papposa. 19 April-10 May,

1847. Santa Fe, gently sloping declivities of hills, less frequently on

grassy places.
0 May-2 August, 1847. Santa

6 August, 1847. Moro River;
1847. 8 miles east of Moro

37. Townsendia fendleri. 11

3

Fe, hillsides

: , gravelly soil.

. 338. Townsendia strigos4. 4-15 May, 1847. Santa Fe,
gravelly hills.

352 312. Dieteria coron

opifolia. 10 August-10 September,
1846-7. Walnut Creek to the ford of Arkansas River

[Sept., 1846];

36 ELIZABETH A. SHAW

loose somewhat sandy soil. Also: a few miles east of Santa Fe [Aug.,
1847].
353 340. Townsendia eximia. 28 June—-15 August, 1847. Santa
Fe Creek, sides of high mountains, to Moro River; prairies.
354 identified in the collection list as Aster novae-angliae,
but not cited in “Plantae Fendlerianae” 22, 23 September, 1847.
Shady creek bank and low a. a few miles north and south of
oi upper ferry of the Kansas Rive

316. Aster sericeus. 16- 19 September, 1847. Prairies 20
ae west of Council Grove to Fort Leavenwort

356 315. Aster sagittifolius. 17 Santember, 1847. Council
Grove; creek bottom
357 315. Aster sagittifolius. 17 August-23 September, 1847.

15 miles east of Council Grove, and a few miles north of the upper
ferry of the Kansas River; creek bottom.

358 not identified and not cited in “Plantae Fendleriane” 23
September, 1847. Prairies and creek banks, between Kansas River and
Fort Leavenworth.

359 “I do not find this.” 28 Sept., 1846. Rio de los Animos
(between Bent’s Fort and Santa Fe), bottom land.

360 320. Aster simplex. 3 July-3 August, 1847. Corners of
a walled in pasture.

361 “the same as last,” but not cited in “Plantae Fendler-
ianae” 15 August-3 October, 1846-7. Banks of the Moro River and
bottom of Poni Creek (between Santa Fe and Bent’s Fort).

362 313. Aster laevis. 28 July, 1847. Santa Fe Creek valley

in the mountains, 9 miles above Santa Fe, foot of mountains.

363 326. Aster divaricatus. 3 September, 1847. Ford of Ar-

kansas River, south side, > lan

he 323. Aster n. sp.? 17 Aagase 1847. 2 miles east of Moro
iver.

365 322. Aster miser y diffusus. 22 Sept., 1847. Creek banks

and bottoms a few miles north of the upper ferry of Kansas River;
5-6 feet high.
366 314. Aster azureus. 21 Sept., 1847. Prairies, 12 miles
southwest of the upper ferry of Kansas River.

367 319. Aster multiflorus y commutatus. 22 Aug., 1847.
Rock Creek, bottom land.

368 318. Aster multiflorus. 16-21 Sept., 1847. Cottonwood
Creek to the upper ferry of Kansas River.

369 324. Aster oblongifolius. 17 Sept., 1847. Prairies and
creek bottoms about Council Grove.
370 321. Aster carneus. 3 Sept., 1846. Banks of Walnut

Creek.

AUGUSTUS FENDLER’S COLLECTION LIST 37

371 325. Aster oblongifolius. 23 Aug., 1847. 7 miles east of
Rock Creek.
372 317. Aster fendleri. 20 Aug., 1847. Ocate Creek and Rio
Colorado.
373 314. Egletes humilis. 25 May, 1847. Santa Fe Creek val-
ley, foot of a hill.

374 333. Erigeron cinereum. 11 April-11 May, 1847. Santa
Fe; sunny declivities, foot of hills near irrigating ditches, socially.

37 332. Erigeron canum. 17 May-3 June, 1847. Santa Fe;
gravelly hills; also: foot of mountains, dry places.

376 = 384 331. Erigeron macranthum. 28 July, 1847. Santa Fe Creek,
valley, sides of high mountains 10 miles above Santa Fe.

377 327. Erigeron candense. 12 July-20 Oct., 1846-7. Santa
Fe Creek bottom, margin of the creek.
378 363. Conyza subdecurrens. 28 July, 1847. Santa Fe, foot

of mountains; high up the creek. ee
379 identified in the collection list as Boltonia glastifolia, but
this species is not mentioned in “Plantae Fendlerianae 16 Septem-
ber, 1847. Prairie, a few miles east of Cottonwood Creek.

380 333. Erigeron cinereum. 24 May-11 June, 1847. Rio del
Norte, low sandy bank. Also: Santa Fe, creek bottom in the moun-
tains.

381 334, Erigeron flagellare. 20 May-22 June, 1847. Santa
Fe Creek bottom, low moist places.

382 328. Erigeron divaricatum. 7 Sept., 1847. Between Coon

Creek and Pawnee Fork, roadside.

3 329. Erigeron hellidiastrum. 28 Aug-, - pA —
west of the ford of Cimarron River and Willow Bar; ged s = “a
384 = 376 330. Erigeron macranthum. ee 1847. san
Creek, high mountains, 11 miles above Santa ©.

38 333. Erigeron cinereum. 31 May-19 July, 1847. Santa Fe
Creek valley, near irrigating ditches.

386 357. Prionopsis ciliata. 28 Aug =1 Sept» 1847. Between
Willow Bar and Cow Creek, low prairies. :
387 346. Solidago speciosa B angustata. Santa Fe Creek val
ley, sides of high mountains.

387b 350. Solidago incand. Q7 July, 1847. eer cee ick
388 342. Gutierrezia euthamiae. 23 niger ;

Also: 8 July-10 Aug., 1847. Santa Fe, gravely ©

3 347. Solidago rigida. 22 Aug., 1846. Council Grove,

prairies.

= 358. Grindelia squarros4. 24 pees ee a

Fe Creek bottom; also: moist meadows and near inion * s ee
360. Chrysopsis villosa. 11 June-6 Aug.

38 ELIZABETH A. SHAW

Creek valley, sunny hillsides between rocks. Also: 10 Oct., 1846. A
few miles east of Pecos.

391 361. Chrysopsis canescens. 12 July—1 Sept., 1847. From
Santa Fe to Sand Creek, dry prairies.

391 362. Chrysopsis canescens var. nana. 17 Aug., 1847. 2
miles east of Moro River. Elevated rocky region.

392 359. Heterotheca scabra B nuda. 29-31 Aug., 1847.
Middle Spring to Lower Spring (Cimarron); low prairies.

393 356. Haplopappus gracilis. 19 Oct., 1846. Creek bottom,
Santa Fe.

394 354. Haplopappus spinulosus var. glaber. 16 miles east
of Middle Spring, prairies.

395 355. Haplopappus spinulosus var. canescens. 24 May,
1847. Between Santa Fe and the Rio del Norte.

396 364. Eclipta erecta. 26 Sept., 1847. Banks of the Mis-
souri River, below Liberty.

397 369. Melampodium cinereum. 17 Sept., 1846. 10 miles
east of Big Sand Creek, high prairies, gravelly soil. Also: 8 May, 1847.
Rio del Norte.

398 365. Silphium integrifolium. 21 Sept., 1847. High prai-
ries, a few miles southwest of the upper ferry of Kansas River.

399 366. Silphium perfoliatum. 23 Sept., 1847. Creek bank
5 miles north of Kansas River.

400 381. Zinnia grandiflora. 7 June—4 Oct., 1846-47. Santa
Fe. Dry gravelly hillsides. Also: Rayado Creek 4 Oct., 1846 (between
Bent's Fort and Santa Fe), foot of a hill: gravelly soil.

401 367. Engelmannia pinnatifida. 29 Sept., 1846. Raton
Mountains.

402 = 410 372. Ambrosia coronopifolia. Santa Fe, near garden walls,
moist soil. 2 August, 1847.

403 identified in the collection list as Ambrosia coronopifolia,
but not cited in “Plantae Fendlerianae” 16, 17 Sept., 1847. Region
about Council Grove.

404 identified in the collection list as Ambrosia coronopifolia,
but not cited in “Plantae Fendlerianae” 26 July, 1847. Santa Fe
Creek, valley near the creek.

405 373. Ambrosia longistylis. 26 July-10 Aug., 1847. Santa
Fe Creek bottom and other low level places. Branches mostly decum-
bent, upwards of 3 feet long. |
406 378. Franseria tenuifolia. 13 July, 1847. 3 Oct., 1846.
Poni Creek (between Bent’s Fort and Santa Fe), low prairies. Also:
Santa Fe, foot of hills near an irrigating ditch.

407 374. Ambrosia longistylis, var. 6 Aug. 1847. Santa Fe
Creek bottom.

AUGUSTUS FENDLER’S COLLECTION LIST 39

408 376. Franseria hookeriana var. 20 Sept., 1846. Dry Creek,
6 miles east of Bent’s Fort, sandy bed of the creek. 28 August, 1847.
Willow Bar, deep sand.

375 bis. Franseria hookeriana. 19 Oct., 1846. Santa Fe,
near the creek.

410 = 402 identified in the collection list as Ambrosia coronopifolia,
but not cited in “Plantae Fendlerianae” 13 Aug., 1846. dry prairie at
Lone Elm camp. 8 Sept., 1847. Pawnee Fork.

411 377. Franseria discolor. 21 Aug., 1847. Between Rio
Colorado and Rock Creek.

412 379. Franseria tomentosa. 2 Sept., 1846. Walnut Creek,
high part of the bank.

413 370. Iva ciliata. 1-3 Sept., 1847. Sand Creek. Low wet
places in the prairie, east of Fort Leavenworth.

375. Ambrosia aptera. 22 Aug. 1847. Rock Creek, bot-
tom land. 22 Aug., 1846-2 Sept., 1846. Council Grove to Walnut
Creek, 8 to 10 feet high.
415 371. Cyclachaena ranthiifolia. 8 Aug.—17 Aug., 1847.
Santa Fe—2 miles east of Moro River, bottom land. 8 to 10 feet high.
416 371. Cyclachaena xanthiifolia. 10-22 Aug., 1847. Santa
Fe to Rock Creek. Bottom land, 6 to 10 feet high.

417 384. Rudbeckia laciniata. 3 June-19 July, 1847. Santa Fe

Creek bottom, close to the water.

418 identified in the collection list as

but cited in “Plantae Fendlerianae” 17 Aug.

east of Council Grove, creek bottom. :

419 385. Rudbeckia subtomentosa. 11 Aug., 1846. 15 miles

south of Fort Leavenworth.

420 368. Berlandiera lyrata. 11-12 Aug., 1847. Between Pe-

cos and San Mi woodland on the mountains.

421 ig si encelioides 6 cand. 30 May-29 cm

i Santa Fe, waste gardens and about —— foot of hills.

Also: Rio de los Animos, Raton mountains, v@ ey.

: 387. Lepachys columnaris B pulcherrima. 13 Aug., 1847.

miles we alley.

os reece lewore 10 Aug., 1847. Between Santa
€ and Pecos. .

424 388. Lepachys columnaris B pulcherri?- ‘ peat ee

1846-47. Santa Fe [7 July, 1847], waste fields and level trac so

waterducts. Ojo de Bernal and Poni Creek [3 Oct. 1846], ees

D 393. Helianthus — - sitos oto
ry prairies between 110 Creek an¢ *° ;

426 392. Hidewien rigidus. 3 July, 1847. Santa Fe, waste

garden place.

Rudbeckia laciniata,
1846. About 20 miles

40 ELIZABETH A. SHAW

427 394. Helianthus maximiliana B asperrimus. 3 Sept., 1847.
South side of ford of Arkansas River, bottom land.
428 389. Helianthus lenticularis. 23 June—28 July, 1847. Waste
a and Santa Fe Creek valley, 3 miles above Santa Fe.

390. Helianthus petiolaris. 29 Aug., 1847. Middle Spring
es low places on the prairie, 8 to 12 feet high.
430 382. Heliopsis laevis. 15 Aug., 1847. Bottom land of
Moro River.
431 395. Heliopsis tracheliifolius. 16 Aug., 1846. Creek bot-
tom, about 30 miles east of Council Grove.
432 396. Heliomeris multiflora. 23 Aug., 1847. 7 miles east
a — Creek, bottom land.

383. Heliopsis laevis B gracilis. 26 Aug., 1846. Council
lay creek bed near the water.
434 391. Helianthus orgyalis. 14 Aug., 1846. Hickory Point,
low prairie.

435 402 bis. Bidens connata. 25 Sept., 1847. Dry part of the
bed of Missouri River, a few miles above Liberty.
436 408. Lowellia aurea. 26-27 Aug., 1847. Between Cold

and Upper Spring (Cimarron).

437 identified in the collection list as Bidens, but not cited
in “Plantae Fendlerianae” 26 Sept., 1846. Dry part of the Missouri
River bed, a few miles below Libe

438 identified in the collection list as Bidens frondosa, but
not cited in “Plantae Fendlerianae” 25 Sept., 1847. Banks of Missouri
River near Liberty.

439 402 bis. Bidens connata. 17 Sept., 1847. A few miles west
of Council Grove, low places.

440 not identified: in the collection list; not cited in “Plantae
Fendlerianae” 10 Aug., 1847. Between Santa Fe and Pecos.

441 397. Coreopsis tinctoria. 17 Aug., 1847. a few miles east
of Moro River, low places.

442 397. Coreopsis tinctoria. 5-9 Sept., 1846-47. Between

Coon Creek and Pawnee Fork, shallow hollows in the prairie (said to

be made by the buffaloes in wallowing).

443 402. Bidens frondosa. 3 Sept., 1847. South side of the

ford of Arkansas River, bottom land.

444 398. Coreopsis involucrata. 20 Sept., 1847. A few miles

east and ne of 110 Creek, hollows in the prairie.

445 osmidium gracile. 24 May-1 Sept., 1846-47. Santa Fe,

— of nea near irrigating ditches. Also: Rio del Norte-Cow Creek.
identified in the collection list as Cosmidium gracile, but

. cited in “Plantae Fendlerianae” 1 Oct., 1846. 3 miles north of

Rio Colorado; valley in the prairies.

AUGUSTUS FENDLER’S COLLECTION LIST 4]

447 399. Cosmos bipinnatus. 14 Aug., 1847. 10 miles west
of Vegas, woodland.

448 399. Cosmos bipinnatus. 20 Oct., 1846. Santa Fe Creek,
bottom.

449 401. Bidens tenuisecta. 3 Oct., 1846. Poni Creek (be-
tween Santa Fe and Bent’s Fort), margin of the creek bank.

450 412. Gaillardia pinnatifida. 13 June-26 July, 1847. Santa
Fe, foot of dry stony hills.

451 413. Gaillardia pinnatifida var. 23 Aug., 1847. 7 miles

east of Rock Creek. Bracts, subtending the heads of flowers, much
longer than on the preceding one.
452 414, Gaillardia pinnatifida var. 24 May, 1847. Rio del
Norte, foot of hills.
453 410. Gaillardia lanceolata. 1 Sept., 1846. Between Cow
and Walnut Creeks.
54 411. Gaillardia pulchella. 28 Aug-8 Sept., 1847. Ford
of the Cimarron River and Pawnee Fork.
415. Palafoxia hookeriana B subradiata. 25 Aug.—20 Sept.,

1847-46. Between Rabbit Ear Creek and Willow Bar, deep sand.
Also: sandy bed of Dry Creek (between ford of Arkansas River and
Bent’s Fort).
4 419. Hymenopappus luteus. 5 May-I1 July, 1847. Santa
Fe, sloping sides of dry hills.
457 421. Actinella arge
Fe, stony gravelly hills, sunny as well as n
socially.
458 416. Schkuhrie neo-mericana. 7 July-6 Aue 1047 pene
Fe, margin of fields, near walls.
459 identified in the collection list as Dysodia chrysanthe-
a but not cited in “Plantae Fendlerianae’ 10 July, 1847. Santa
e Creek, bottom and margin of the creek.
460 423. Actinella F hendesan var. fi loribunda. 8 June-13

ul Il as plains and creek bottom.
July, 1847. Santa Fe, rocky hills as we p ducal

10 July, 1847. A monstrosity of this plant, showing t

a compositae flower plainly.
461 409. Riddellia tagetina. 22 June-31 July, oe oy
Creek, valley, foot of hills near irrigating ditches. a a
Between San Miguel and Santa Fe, woodland, near t e. ne em
462 identified in the collection list as Melampo pa : it
cited in “Plantae Fendlerianae” 10 Aug: 1847. From Santa Fe to
Miguel.
417. Hymenopappus flavescens. 28 ANE. fake
miles west of Willow Bar (Cimarron River), deep san 2 aeneen
417. Hymenopappus flavescens. 14 Aug., 1547.

San Miguel and Vegas.

ntea. 19 April-7 June, 1847. Santa
orthern declivities; rather

42 ELIZABETH A. SHAW

465 418. Hymenopappus tenuifolius. 13-23 Aug., 1847. Ojo
de Bernal to Rock Creek, prairies.

466 422. Actinella scaposa var. mutica. 11, 12 Aug., 1847.
Pecos to San Miguel, woodland.

467 424, Helenium autumnale. 17 Aug., 1847. A few miles
east of Moro River; bottom land.

468 424. Helenium autumnale. 20 Sept., 1847. From 110
Creek to Fort Leavenworth; low places in the prairies.

469 420. Bahia oppositifolia. 10-26 Aug., 1847. 8 miles east

of Santa Fe to Cold Spring (Cimarron River), low places and sides
of ravines.

470 441. Senecio longilobus. 17 May—4 Nov., 1846-47. Santa
Fe, mountain sides and foot of hills. 2 to 3 feet high.

471 = 460 identified in the collection list as Actinella richardsonii var.
floribunda, but not cited in “Plantae Fendlerianae” 8 June, 1847.
Santa Fe, hills, amongst rocks. On leaves and stems there are scat-
tered bright, shining dots.

472 442. Senecio longilobus. 18-26 July, 1847. Foot of a hill,
Santa Fe.
473 442. Senecio longilobus. 22 June—25 July, 1847. Santa Fe

Creek, valley. Also: 19 Oct., 1846.

474 440. Senecio filifolius B fremontii. This number is so
identified in the collection list, although “(474)” is not cited in “Plan-
tae Fendlerianae” 8 Sept., 1847. Pawnee Fork. 27 Sept., 1846. 27
miles south of Bent’s Fort.

47 443. Senecio eremophilus. “(475)” is pencilled into the
GH copy of “Plantae Fendlerianae” in the hand of Sereno Watson. 20
Oct., 1846 & 22 June, 1847. Santa Fe Creek, bottom, not far from
the creek.

476 438. Senecio aureus. This number is so identified in the
collection list, although “(476)” is not cited in “Plantae Fendlerianae”
22 June, 1847. Santa Fe Creek bottom.

477 437. Senecio exaltatus. “(477)” is pencilled into the GH
copy of “Plantae Fendlerianae” in the hand of Sereno Watson. 26
April—3 June, 1847. Santa Fe Creek valley, foot of mountains 10 miles
above Santa Fe.
478 444. Senecio fendleri. Santa Fe Creek valley; foot of
mountains 11 miles above Santa Fe.

47 439. Senecio aureus y borealis. “(479)” is pencilled into
the GH copy of “Plantae Fendlerianae” in the hand of Sereno Watson.
rea June, 1847. Santa Fe Creek, foot of hills.

444. Senecio fendleri. 3 June, 1847. Santa Fe, foot of
mountains, 12 miles above Santa Fe.

AUGUSTUS FENDLER’S COLLECTION LIST 43

481 436. Erechtites hieracifolia. Creek banks, 5 miles south
of Fort Leavenworth.

482 380. Xanthium echinatum. 4 Aug., 1847. Santa Fe, waste
places.

483 identified in the collection list as Xanthium echinatum,

= not cited in “Plantae Fendlerianae” 3 Sept., 1846. Bank of Walnut
reek,

probably 448. Cirsium altissimum; “(584)” in “Plantae
Fendlerianae,” but that number in the collection list falls in Scro-
phulariaceae. 17 Aug., 1846. High bank of a creek.
485 446. Cirsium canescens. 31 May-Il7 June, 1847. Santa
Fe, sunny side of mountains, lower part.
486 447. Cirsium ochrocentrum. Santa Fe, mountainsides.

487 455. Macrorhynchus purpureus. 16 June-28 July, 1847.
Santa Fe Creek bottom, grassy places.
488 454. Crepis ambigua. 2 June-10 Aug., 1847. Santa Fe

Creek bottom, level grassy places. From the root, when broken, there

is exuding a white, milky juice.
489 450. Hieracium longipilum. 21 Sept. 1847. High prairies
about 12 miles southwest of the upper ferry of the Kansas River.
453. Malacothrix sonchoides. 94 May, 1847. Rio del Norte,

river bottom, low sandy bank.
9 457. Lactuca elongata ys
miles east of Moro River.
9 459. Mulgedium pulchellum. 21
Rio Colorado and Rock Creek. Also: 3 Sept.,
49 451. Nabalus asper. 17 Sept.
Council Grove, high prairies. Plant 2-3 feet high.

4 460. Mulgedium floridanum. 92 Aug.,

Grove, creek bottom.
495 456. Taraxacum palustre var. latifolium. 15 May, it

anguinea. 17 Aug., 1847. A few

Aug., 1847. Between
1847. Walnut Creek.
1847. 16 miles west of

1846. Council

tween Bent’ Fe), low rairies.

497 ent's Fort and sani ®) A . 12 July, 1847. Santa
Fe, in and about fields. The flowers consisting generally of 5 ligules,
which at the lower part form tub
flower: stamens and pistil), all th
a From this plant when broken exu
which in coming on the atmosphere soon
erties of Indian ms: Enveloping scales of the flower i ene’
than the pappus. Lower leaves entire, linear. Stem an ae
striated. Flower red and light bluish.

44 ELIZABETH A. SHAW

498 449. Stephanomeria runcinata. 9 June-1 July, 1847. Santa
Fe, foot of dry hills, although not far from irrigating ditches or fields.
499 353. Haplopappus spinulosus. 30 May—10 Oct., 1846-47.
Santa Fe, sandy and gravelly very little inclined places, generally not
very far from irrigating ditches. Also: about Pecos and Walnut Creek.
500 456. Taraxacum palustre var. latifolium. 15 May—25 June,
1847. Santa Fe Creek bottom, near the bank.
501 456. Taraxacum palustre var. latifolium. 15-20 May,
1847. Santa Fe Creek bottom, moist fertile soil.
502 462. Sonchus asper. 7-13 June, 1847. Santa Fe, waste
places about town and along garden walls.
503 450. Hieracium longipilum. 13 Aug., 1846. 80 miles west
of Independence, stony hill.
504 458. Lactuca graminifolia. 3 July, 1847. Santa Fe Creek
valley, 12 miles above Santa Fe, high up on the mountains. No ex-
panded flowers by which I could tell its color.
505 458. Lactuca graminifolia. 8 July, 1847, Santa Fe Creek
valley; flower yellow.

459. Mulgedium pulchellum. 8-25 July, 1847. Santa Fe
Creek, hillside facing the creek, loose gravelly soil. Flowers blue.
507 461. Sonchus oleraceus. 23 June-10 July, 1847. Santa
Fe, along the inside of field walls, near irrigating ditches.
508 461. Sonchus oleraceus. 11-23 June, 1847. Santa Fe Creek

valley.

509 426. Achillea millefolium. 16 June—20 Oct., 1847-46.
Santa Fe Creek bottom, moist meadows, also foot of mountains.
509b 426. Achillea millefolium. identified in the collection list

but not cited by number in “Plantae Fendlerianae” 13 July, 1847.
Santa Fe. Flowers rose color.

510 428. Artemisia canadensis. 17 Aug., 1847. 2 miles east
of Moro River, rocky elevated place. Also: 1 Oct., 1846. 8 miles south
of Rio Colorado (between Bent’s Fort and Santa Fe).

427. Artemisia dracunculoides. 15 Aug., 1847. Bottom
of Moro River.

512 433. Artemisia frigida. 3 Oct., 1846. Poni Creek, prai-
rie. 23 Aug. 1847. 7 miles east of Rock Creek.

ce 431. Artemisia ludoviciana. 26 July, 1847. Santa Fe Creek
valley

514 431. Artemisia ludoviciana. 26 Aug.—3 Oct., 1846. Coun-
cil Grove to Poni Creek, low prairie.

515 431. Artemisia ludoviciana. 17 Sept., 1847. Prairies, a

few miles west of Council Grove.

516 429. Artemisia filifolia. 24 Aug., 1847. Between Rock
and Rabbit Ear Creeks.

AUGUSTUS FENDLER’S COLLECTION LIST 45

se 430. Artemisia discolor. 31 July, 1847. Santa Fe, foot of
a hill.

518 432. Artemisia vulgaris. 15 July, 1847. Santa Fe Creek
bottom close to the creek.

519 434. Gnaphalium sprengelii. “(519)” is pencilled into the

GH copy of “Plantae Fendlerianae” in the hand of Sereno Watson.
10-14 Aug., 1847. Between Santa Fe and Vegas. Mountain woodland,
low places.
520 identified in the collection list as “Gnaphalium too poor
to say” 23 Aug., 1846. Council Grove, hill.

521 435. Antennaria dioica. 21 May-3 June, 1847. Santa Fe,
hills at the foot of the higher mountains (northern declivities).

522 435. Antennaria dioica. 6 May-3 June, 1847. Together
with the preceding one.
523 435. Antennaria dioica. 15 May-3 June, 1847. Santa Fe,

hills at the foot of the higher mountains (northern declivities); to-
gether with no. 521.

524 435. Antennaria dioica. 26 April-3 June, 1847. Santa Fe
Creek; hills at the foot of mountains.

525 349. Solidago incana. 23 Aug., 1847. 7 miles east of
Rock Creek.

526 407. Dysodia chrysanthemoides. 14-16 June, 1847. Santa
Fe Creek bottom, margin of an irrigating ditch.

527 343. Gutierrezia sphaerocephala. 97-29 Aug., 1847. Up-

per Spring (Cimarron) to Middle Spring, low prairie.
as 344. Amphiachyris dracunculoides. 21 Sept., 1847. High
prairies a few miles southwest of the upper ferry of
a 351. Solidago lanceolata. 3 Sept., 1847. Ford of Arkan-
sas River (south side) wet bottom land. 17-21 Sept., 1847. Council
Grove to the upper ferry of Kansas
a 348. Solidago ulmifolia.
miles southwest of the upper ferry of Kansas River. _ .
e! 345. Solidago speciosa. 21 Sept., 1847. High prairies, 4
ansas River.

Solidago rigida, but not
1847. The same

River.
21 Sept., 1847. Prairie, about 10

few miles southwest of the upper !etTY of K
532 Identified in the collection list as
— by number in “Plantae Fendlerianae’ 21 Sept.
ocality as the preceding one.

a 339. Townsendia grandiflora. 91 Aug., 1847. Between
cate Creek and Rio Colorado, low prairie.
534 405. Heterospermum tagetinum. 14 Aug., 1847. 12 miles

west of Vegas; woodland.
296. Pectis angustifolia. 10 Aug.,
Fe and Pecos.

1847. Between Santa

46 ELIZABETH A. SHAW

536 406. Flaveria angustifolia. 30 Aug., 1847. 18 miles west
of Lower Spring (Cimarron).
537 425. Amauria dissecta. 14-17 Aug., 1847. A few miles
east of Moro River.
538 404. Sanvitalia aberti. Between Santa Fe and Pecos;
woodland.

LOBELIACEAE
539 20 September, 1847. Creek bank, a few miles west of

110 Creek. Flower blue, sometimes purple.
540 17 August, 1846. About 118 miles west of Indepen-
dence, creek bottom. 3-15 September, 1847. Ford of Arkansas River
to Cottonwood Creek, banks and creek bottoms.

CAMPANULACEAE
541 22 September, 1847. Woodland about 6 miles north of
the upper ferry of Kansas River. Flower blue.
542 22 September, 1847. Bottom land of Kansas River.
543 16 June—28 July, 1847. Santa Fe Creek, 9 miles above
Santa Fe, sunny side of mountains between rocks.

ERICACEAE
544 16-17 June, 1847. Santa Fe Creek, overshaded margin
of the creek.
545 28 July, 1847. Santa Fe, ravines high up on the moun-
tains.
546 26 April-2 May, 1847. About 6 miles east of Santa Fe,

creek bottom. Shady declivities of the mountains. Evergreen. Wild

turkies eat the berries as soon as the snow that covered them is
melted away.

AQUIFOLIACEAE
547 25 May, 1847. Rio del Norte, steep and rocky bank of
the river. A shrub about 10 feet high.
PRIMULACEAE
548 1 April-15 May, 1847. Santa Fe Creek; gravelly places

of its bed exposed to inundations, moist soil. Petals light rose color
nearly white. Some few specimens I found to have been preserved
under the snow throughout the winter.

) 3-16 June, 1847. Santa Fe Creek, sunny margin, deep
fertile soil. Flower of a very fine purple color before drying.
550 15-18 August, 1847. Bottom land about Moro River.

PLANTAGINACEAE

551 10 May-26 August, 1847. Santa Fe. Dry valley between
the lesser hills, tolerably good soil. Also creek valley, sides of hills
not far from the creek. Also Rio del Norte and a few miles east of
Cold Spring (Cimarron).

AUGUSTUS FENDLER’S COLLECTION LIST 47

552 6 June-25 July, 1847. Santa Fe Creek bottom, near the
creek.
553 17 August, 1846. About 120 miles west of Indepen-
dence.

PEDALIACEAE
554 31 August, 1847. Lower Spring (Cimarron), prairie.

ACANTHACEAE
555 25 August, 1846. Council Grove, gravel of the creek
bed.
596 11 August, 1846. 10 miles south of Fort Leavenworth.
Flower blue.
557 22 September, 1847. Woodland about 6 miles north of
the upper ferry of the Kansas River.

SCROPHULARIACEAE

558 3 June-7 July, 1847. Santa Fe Creek bottom, margins
of the creek, wet soil.
559 27 May-9 June, 1847. Santa Fe. Near and in springs and
rivulets
560 11 May-19 October, 1847. Santa Fe Creek bottom, low
sandy margin of the creek, wet soil.
561 2 September, 1847. Pool of water on the prairies, 40
miles southwest of the ford of Arkansas River.
562 20 August, 1846. Council Grove, creek bottom.
563 26 April-16 June, 1847. Santa Fe Creek bottom, margin

of the water, flat sandy wet banks. Petals blue with violet stripes, the
lower one smaller than the rest and of a lighter blue. Flower fuga-

cious.

564 31 May-8 July, 1847. Santa Fe Creek, margins of the
water, partly immersed. Flower blue, fugacious in a high degree.
965 17 August, 1846. About 25 miles east of Council Grove.
566 24 August, 1846. Council Grove, creek bottom, fertile

soil. Plant 18 feet high.
567 18 September, 1847. Bluff Creek, bottom.

568 14 August, 1846 and 18 September, 1847. Bluff Creek
and Hickory Point, bottom land.
569 ‘ 95-26 August, 1847. Between Rabbit Ear Creek and

Cold Spring (Cimarron). Sandy hillside.

570 20 September, 1847. From 110 Cree
worth; low prairies.

571 3 September, 1847.
of its ford.
572

k to Fort Leaven-
Arkansas bottom, wet places, south

23 August, 1846. Council Grove, low places between

hills. Flower lilac, fugacious.

48 ELIZABETH A. SHAW

573 18 August, 1846. 5 miles east of Council Grove; also 21
September, 1847. Dry prairies a few miles southwest of the upper
ferry of Kansas River.

574 18 August, 1846. 10 miles east of Council Grove. Prai-
rie

575 21 May-26 July, 1847. Santa Fe, dry gravelly hills.

576 4-28 May, 1847. Santa Fe, gravelly hills, dry soil.

577 14 August, 1846. Hickory Point, prairie.

578 10 August, 1847. Between Santa Fe and Pecos, wood-
land.

579 12 August, 1847. Neighbourhood of San Miguel.

580 1-12 July, 1847. Santa Fe Creek, hillside; flower white.
581 22 June-3 July, 1847. Santa Fe Creek bottom. Also: 11

miles above Santa Fe, on sides and ravines of high mountains (where
the moisture is greater than in the plains of Santa Fe).

582 15 May-3 July, 1847. Santa Fe Creek; shady, gently
sloping declivities, foot of mountains. Petals white, stamens 2, nearly
twice the length of the petals.

583 28 July, 1847. Santa Fe Creek, shady sides of moun-
tains, 11 miles above Santa Fe.

584 29 April-15 July, 1847. Santa Fe, gravelly hills and sides
of mountains; Rio del Norte; Poti Creek, and Rayado Creek (between
Bent’s Fort and Santa Fe). 4 October, 1846.

OROBANCHACEAE
585 4 October, 1846. Rayado Creek (between Bent’s Fort
and Santa Fe); gravelly valley.
VERBENACEAE
586 30 April-20 October, 1846-1847. Waste fields and low

valleys in the mountains, Santa Fe. Also: between Rayado and Ocate
Creeks [20 Oct., 1846].

24 May-31 July, 1847. Santa Fe, Creek bottom.
588 22 June—31 July, 1847. Santa Fe, Creek bank.
589 8 May, 1847. Rio del Norte; flat sandy bottom not far
from the river.
see 19 October, 1846. Santa Fe, Creek valley, gravelly moist
soi
591 23 October, 1846. Santa Fe, Creek bottom. 27 August,
1847. Upper spring (Cimarron), bottom.
592 28 August, 1847. 1 September, 1847. Willow Bar, low
sandy places, also: Sand Creek, low places in the prairies.
593 21 September, 1847. 18 miles east of 110 Creek.
594 12 August, 1847. Neighbourhood of San Miguel, prai-
ries.
595 15 August, 1847. Moro River bottom.

AUGUSTUS FENDLER’S COLLECTION LIST 49

596 22 September, 1847. Bottom land of Kansas River.

597 11 June-31 July, 1847. Santa Fe, Creek bottom. In
some specimens the flower white.

598 4 September, 1847. Ford of Arkansas River to Pawnee
Fork, bottom land.

599 17 August, 1846. About 30 miles east of Council Grove,
bottom land.
600 16 August, 1846. About 120 miles west of Indepen-
dence; creek bed, wet places.
601 21 August-6 September, 1847. Rio Colorado to Pawnee
Fork; low places on the prairie.

LABIATAE
602 10 August-10 October, 1847. A few miles east of Santa
Fe to Moro River; mountain valleys and prairies.
603 25 June-15 July, 1847. Santa Fe Creek; foot of moun-
tains (sunny side).
604a 16 June-18 September, 1847. Santa Fe to Bluff Creek,
bottom land.
604b same specimen with white flowers.
605 22 August-27 September, 1846. Council Grove; also:

about 25 miles south of Bent’s Fort.

606 9 June-20 October, 1847. 1846. [i.e., 20 Oct., 1846; 9
June, 1847.] Santa Fe Creek bottom, also sides of hills, between rocks
not far from irrigating ditches.

607 29 September, 1847. [date certainly wrong| Between
Kansas River and Fort Leavenworth, prairies. Flower white.

608 10-16 August, 1847. Pecos to Moro River; woodland in
the mountains, low valleys.

609 1-31 July, 1847. Santa Fe Creek bottom; in moist shady
places

610 14-18 August, 1847. A few miles west of Vegas to Moro
River

611 18 August, 1847. 6 miles east of Moro River, bottom
land.

612 11 August, 1846. 8 miles south of Fort Leavenworth.
613 17 August, 1846. About 25 miles east of Council Grove.
614 23 June—20 October, 1846-47. Santa Fe Creek bottom,

near irrigating ditches 23 June, 1847; Ford of the Arkansas River and
east of Council Grove.

615 15 August, 1846. 117 miles west of Independence, creek
bottom

616 20 August, 1846. Council Grove, creek bottom.

617 14-17 August, 1847. A few miles west of Vegas to Moro

River; woodland and prairies.

50 ELIZABETH A. SHAW

618 10 August-4 October, 1846-47. Pecos. Ojo de Bernal
to Rock Creek [10 Aug., 1847]. Also: Pofi and Rayado Creek (be-
tween Bent’s Fort and Santa Fe), in valleys.

619 13 August, 1846-23 September, 1847. Kansas River bot-
tom.
620 24 May-31 July, 1847. Santa Fe, sunny hillsides facing

the creek, between rocks; also: between rocks near irrigating
ditches—Rio del Norte.
621 3 July-10 August, 1847. Santa Fe Creek valley in the
mountains, sunny mountain sides between rocks; also stone piles near
a field.
BORAGINACEAE

622 21 August, 1847. Between Rio Colorado and Rock Creek.
Low prairies.
623 28 August, 1847. Willow Bar, deep sandy soil.
624 11 August, 1846. 4 miles south of Kansas ferry, low prai-
rie. 23 August, 1847. 7 miles east of Rock Creek.
625 26 April-3 June, 1847. Santa Fe Creek Bottom, foot of
hills at some distance from the water.
626 24 May-28 July, 1847. Santa Fe Creek bottom, moist
fertile soil at the margin of the water.
627 24 June-15 July, 1847. Santa Fe, high up the creek,
sunny side of mountains. In bunches of 114 foot diameter and 2-24%
foot high. Flower short and small.
628 20 May-3 June, 1847. Santa Fe Creek valley, bottom
land and foot of hills. Flower long but narrow.

19 April & 15 May, 1847. Santa Fe, stony side of hills
(sunny side). Flowers long and large.
629b 3 October, 1846. Poni Creek (between Bent’s Fort and
Santa Fe). Bottom prairie, good soil. The root of this plant is said by
the Mexicans to be good (when boiled) to cure sores. Also: 25 August,
1847. Between Rabbit Ear and McNees’s Creeks.

630 18 September, 1847. Bluff Creek, bottom land.
631 25 August, 1847. McNees’s Creek, sandy hillside.
632 29 April-24 May, 1847. Declivities of dry gravelly hills

southwest of Santa Fe. Flower white.

25 June-15 July, 1847. Santa Fe Creek up the moun-
tains; sunny side of mountains between rocks.
63 4 May-9 July, 1847. Santa Fe, valleys between dry and
gravelly hills; also: creek valley close to fields.
635 7-15 July, 1847. Santa Fe, Creek valley near a field.
636 19 May-31 July, 1847. Santa Fe, valleys between the
gravelly hills where they run out onto the plains.

7 18 September, 1847. Bluff Creek, bottom land.

AUGUSTUS FENDLER’S COLLECTION LIST 51

638 7 July, 1847. Santa Fe Creek valley.
639 17-23 August, 1847. Rio Moro to Rock Creek; elevated
prairies.

24 April-30 May, 1847. Santa Fe, valleys between the

gravelly hills south west of Santa Fe.
HYDROPHYLLACEAE
641 3-16 June, 1847. Santa Fe Creek, 8 miles above Santa
Fe; shady damp places on ledges of rocks.
642 3 June-28 July, 1847. Santa Fe Creek bottom, low
rather moist places.
HYDROLEACEAE

643 30 May-21 June, 1847. 5 miles southwest of Santa Fe,
loose tolerably good soil in a valley between the hills. Petals deep
violet.
644 10-14 August, 1847. Santa Fe to Vegas; woodland val-
leys in the mountains.

POLEMONIACEAE
645 18 August, 1847. Low bottom land of Moro River.
646 11 May-25 July, 1847. Santa Fe. Creek valley; on hills,

also foot of hills near irrigating ditches. Flower red.

64 14 August, 1847. 12 miles west of Vegas.

648 4 May-31 May, 1847. Santa Fe; valleys between the
gravelly hills southwest of Santa Fe, where they run out onto the
plains. Petals white anthers of a bluish green color.

649 24 May, 1847. Rio del Norte, low bank between rocks.
Flower blue.

650a 8 May, 1847. Rio del Norte, shady places in ravines,
between rocks.
650b 28 August-1 September, 1847. Between ford of the

Cimarron River and Willow Bar; sandy soil. Also: Sand Creek.

1 17 May, 1847. Santa Fe. Sunny side of a dry and grav-
elly hill. Only found in one place. 7 June, 1847, in fruit.
652 23 August, 1847. 7 miles east of Rock Creek, creek bot-

tom.
653 24 June-28 July, 1847. Santa Fe, 19 miles up the Creek,
high mountain side.
654 19 July, 1847. Santa Fe Creek bottom, fertile soil.
655 16 June-8 July, 1847. Santa Fe, high up the Creek.
Sunny side of steep rocky mountains.

CONVOLVULACEAE
656 11 August, 1846. 8 miles south of Fort Leavenworth;
bottom land.
657 8 September, 1847. Pawnee Fork; steep bank.

658 18 September, 1847. Bluff Creek, bottom land.

52 ELIZABETH A. SHAW

659 9 July, 1847. Santa Fe; waste house gardens and margins
of fields.
659b 6 September, 1846. About 15 miles east of the ford of
a River; level prairies.

21 August, 1847. Between Rio Colorado and Rock Creek;
as Sand Creek. 27 September, 1846. 20 miles south of Bent’s Fort.

2 October, 1846. Between Poni and Rayado Creek, grav-
ally hillside.

12-22 August, 1847. Pecos to Rock Creek.

662 8-10 August, 1847. Santa Fe, cornfields; also: woodland

between Santa Fe and Vegas.

663 10 August, 1847. Between Santa Fe and Pecos. Moun-

tainous woodland.

664 14 June, 1847. Santa Fe; margin of an irrigating ditch

in a waste lying garden.

665 10 August, 1847. Between Santa Fe and Pecos, wood-

land.

666 15 August, 1847. Moro River, bottom land.

667 23 ee 1847. Creek bottom, north of Kansas

River. Flower white, showy.

668 25 May-11 July, 1847. Santa ws at the foot of dry hills

(sunny side). Socially. Flower seldom seen expanded.

668b 10 August, 1847. Between Santa Fe and Pecos.
SOLANACEAE

669 21 June-11 August, 1847. Santa Fe, Creek valley, about

fields and irrigating ditches. Also from Santa Fe to Vegas; woodland,
shady places around trees and shrubs.

670 28 May-11 June, 1847. Santa Fe Creek valley, close to
an irrigating ditch at the foot of a hill.
671 7 June—1 October, 1847-6. Santa Fe; in back streets of

the town and near fields, flat on the ground [7 June, 1847]. Also be-
tween Raton and Bermejo Creeks, level prairies [1 Oct., 1846].

672 6-11 August, 1847. Santa Fe, steep rocky ‘sides of dry
hills. Also from Santa Fe to Vegas, woodland in the mountains.

673 1 July-6 August, 1847. Santa Fe Creek valley, foot of
mountains, sunny side.

674 18 August-29 August, 1847. Rock Creek to a few miles
east of Council Grove; prairies.

675 17 July-13 August, 1847. Santa Fe Creek valley. Sunny
declivities near the creek, loose soil. Also neighbourhood of Ojo de
Bernal.

676 28, 29 August, 1847. A few miles west of Willow Bar
to Middle Spring (Cimarron).

AUGUSTUS FENDLER’S COLLECTION LIST 53

677 20-27 August, 1847. Ocate Creek to Upper Spring
(Cimarron), low prairies.

678 6 June-9 July, 1847. Santa Fe, waste gardenplaces.
679 25 August, 1847. McNees’s Creek, low prairie.

680 14 June-13 July, 1847. Margin of irrigating ditch in a

waste garden. Stamens deep blue. Santa Fe.

681 1 June-30 August, 1847. Santa Fe—Middle Spring
(Cimarron). Ravines and low places in the prairies; also: creek bottom.
Between Poni and Rayado Creeks.

682 31 May-31 July, 1847. Santa Fe, in and near fields. Also
foot of mountains not far from the Creek.

683 11 June-18 July, 1847. Santa Fe Creek valley, between
rocks at the foot of hills.

684 21 June-18 August, 1847. Santa Fe, gently sloping sides
of ravines, loose soil; also in level places (creek bottom). Spreading
on the ground. Flowers compound, consisting of two separate flow-
ers. Also: 6 miles east of Moro River.

685 7 June-10 August, 1847. Santa Fe, waste gardens and
yards; also between Santa Fe and Pecos.

GENTIANACEAE
686 7 June-25 June, 1847. Santa Fe Creek valley, foot of
mountains. Flies seem to be very fond of this plant. 4-5 feet high.
687 28 August-5 September, 1846-7. Prairies a few miles
west of the ford of Cimarron River; also Arkansas River.
688 21 September, 1847. Prairies a few miles southwest of
the upper ferry of Kansas River.
689 21-23 September, 1847. Creek bottom north of Kansas
River, near its upper ferry.

APOCYNACEAE
690 24 June—28 July, 1847. Santa Fe, steep rocky sides of
mountains, facing towards the creek.
9 22 August, 1847. Rock Creek, bottom land.
692 19 August, 1846. Council Grove, bottom land of the
creek.

JASMINACEAE
693 31 May, 1847. Santa Fe, sunny hillsides between rocks,
foot of hills and in valleys. Flower fugacious.

ASCLEPIADACEAE

694 25 June-31 July, 1847. Santa Fe; foot of the mountains
as well as of the lower hills.
695 11 June-1 July, 1847. Santa Fe, near fields.
696 11 June, 1847. Santa Fe, creek valley, sunny side of a

stony hill.

54 ELIZABETH A. SHAW

697 17 August, 1847. 2 miles east of Moro River.

698 15-21 August, 1847. Moro River and Rock Creek.

699 21-25 August, 1847. Between Rio Colorado and Mc-
Nees’s Creek.

700 18 September, 1847. Bluff Creek, bottom land.

701 22 August, 1847. Rock Creek, bottom land.

702 12 August, 1847. Neighbourhood of San Miguel.

703 _ 21 September, 1847. 12 miles southwest of the upper
ferry of Kansas River.

704 12 August, 1846. A few miles south of Kansas River,
prairies.

705 14-25 August, 1847. Between San Miguel and McNees’s
Creek.

706 25 August, 1847. Between Rabbit Ear and McNees’s

Creek; high prairie.

707 24 May, 1847. Rio del Norte; flat sandy river bank, be-
tween stones.

CHENOPODIACEAE

708 16 September, 1846. 20 miles east of Big Sand Creek;
sandy gravelly soil.

709 25 August, 1847-19 September, 1846. 32 miles east of
Bent’s Fort, flat prairies near the bank of the Arkansas River. Also
McNees’s Creek. A shrub 14 to 3% foot high and very much spread-

710 10 August, 1847. Between Santa Fe and Pecos; wood-
land.
711 22 August, 1847. Between Rio Colorado and Rock Creek.

Low parts of the prairie.

712 26 August, 1847. Between McNees’s Creek and Cold
Spring; sandy soil.

7h 10 August, 1847. Between Santa Fe and Pecos.

714 15-17 August, 1847. Moro River; prairie.

715 22 August, 1847. Rock Creek; bottom land.

716 10-11 August, 1847. Between Santa Fe and San Miguel;
woodland.

717 1 September, 1847. Sand Creek; low prairie.

718 17 & 21 August, 1847-46. Council Grove, dry prairie.
Also: a few miles east of Moro River.

719 23 June, 1847. Santa Fe, waste house gardens.

720 17 May-12 July, 1847. Santa Fe Creek valley; foot of
hills where the ground is occasionally irrigated.

721 23 August, 1847. 7 miles east of Rock Creek.

722 9 June—26 July, 1847. About houses, gardens, ete.

723 12 July, 1847. Santa Fe; margin of fields.

AUGUSTUS FENDLER’S COLLECTION LIST 55

724 17 August, 1847. A few miles east of Moro River.

725 23 June—9 July, 1847. Santa Fe; about fields and yards.
AMARANTHACEAE

726 17 September, 1846. About 10 miles east of Big Sand

Creek; high prairie.

726 30 August, 1847. Middle Spring (Cimarron); prairie. The
branches of this plant are not so spreading as in the preceding one.
727 12 August, 1847. Neighbourhood of San Miguel. Leaves

728 10 September, 1846. 5 miles south of the ford of Ar-

14 August, 1847. 16 miles west of Vegas.

730 9 October, 1846. Between San Miguel and Pecos; road-
side.
731 24 May-23 August, 1847. Santa Fe; yards and waste

places. Also: Rock Creek, high prairies.
3 24 June, 1847. About houses and yards, near irrigating
ditches. Santa Fe.

10 July, 1847. About houses and yards, near irrigating
ditches. Santa Fe. Lower side of the leaves marked with red spots.
7 25 May-6 August, 1847. Santa Fe Creek valley. Also:
Rio del Norte, sandy soil, between rocks, flat river bank.

.

35 14 June-1] July, 1847. Santa Fe Creek valley, foot of

hills.

736 19 July, 1847. Santa Fe, near houses and fields.

737 14-21 August, 1846. Hickory Point to Council Grove,

gullies

38 3 September, 1846. Walnut Creek, creek bank.
NYCTAGINAEAE

739 19 May-11 August, 1847. Santa Fe; flat somewhat moist

places not very far from fields and irrigating ditches.

7 10 August—28 September, 1847. Between Santa Fe and
Ojo de Bernal; mountains, woodland. Also: 28 September, 1846. Rio
de los Animos, valley, tolerably good soil. Flowers seldom expanded
(only in the morning).

741 22 August, 1847. Rock Creek, bottom land.

Lae 14 & 15 August, 1847. Vegas to Moro River. Leaves
cordate, involucre of the flower somewhat tomentose. —

J 19 September, 1846. 42 miles east of Bent’s Fort; bank
: the Arkansas River. 17 August, 1847. 2 miles east of Moro River;
elevated :

144 Senha 1846. Fort Leavenworth to Council Grove.
1% 16 June-10 August, 1847. Santa Fe Creek valley, foot
of hills, also near fields between rocks. Involucre of the flower con-

56 ELIZABETH A. SHAW

sisting of 5 segments and containing 3 flower

746 7 July-6 August, 1847. Santa Fe Creek valley near ir-
rigating ditches, amongst the branches of undershrubs. Plants viscose.
Flowers rarely to be seen opened except in the morning. Involucre
resembling a calyx of 5 sepals.

POLYGONACEAE
747 17 August, 1847. 2 miles east of Moro River; foot of
high rocky embankment.
748 8 September, 1847. Pawnee Fork, creek bottom.
749 29 August, 1847. Middle Spring (Cimarron); low places.
50 4 July-8 September, 1847. Santa Fe and Pawnee Fork,
creek bottom.
751 17 September, 1847. Council Grove, creek bottom.
752 22 June-28 July, 1847. Santa Fe, margins of irrigating
ditches

9 July, 1847. Santa Fe Creek bottom.

754 14 June, 1847. Santa Fe, near irrigating ditches.
755 8 September, 1847. Pawnee Fork. Fruit winged.
756 14 August, 1846. Hickory Point, creek bottom. Seeds

much smaller than in the preceding one.
757 22 August, 1847. Rock Creek, overshaded deep ravine.
758 8 May, 1847. Between the Rio del Norte and Santa Fe;
tenia grassy plains.

17 August, 1847. 2 miles east of Moro River, bottom
iad 4-5 feet high.

bey 21 June, 1847. Santa Fe; near the irrigating ditch of a
eld.

761 29 August, 1847. Middle Spring (Cimarron).

762 8 August, 1847. Santa Fe, pasture, wet moory places.
763 7 July-17 August, 1847. Santa Fe Creek valley, up in

the mountains. Also: 2 miles east of Moro River, elevated rocky re-
gion.

764 10 August, 1847. Between Santa Fe and Pecos. Also: 15
July, 1847. Santa Fe Creek valley.

765 25 August, 1847. Between Rabbit Ear and McNees’s
Creeks.

766 16 August, 1847. Moro River; summit of a rocky hill,
between ledges of rocks.

767 21 August, 1847. Between Rio Colorado and Rock Creek.
768 23 September, 1846. 3 miles above Bent’s Fort, Arkan-

sas River, on the rounded, very pebbly and sterile bluffs.
769 15-17 August, 1847. Moro River, rocky elevated region.
Also: 2 October, 1846. Between Poii and Rayado Creeks.
770 15 July, 1847. Santa Fe Creek valley, mountain side

AUGUSTUS FENDLER’S COLLECTION LIST or

between rocks. Leaves more or less persistent through the winter.
771 14 August, 1847. 15 miles west of Vegas.

772 1 September, 1846. Between Walnut and Cow Creeks.
773 28 August, 1847. Willow Bar in sand, socially. 5 to 6
feet high.
ELEAGNACEAE
774 28 July, 1847. Creek valley, 12 miles above Santa Fe,
foot of mountains and in ravines.
ULMACEAE
775, 25 May, 1847. Rio del Norte; somewhat elevated bank

of the river, close to the water. A treelike shrub about 12 feet high,
branches spreading.

EUPHORBIACEAE
776 25 May-13 July, 1847. Rio del Norte; lower part of ra-
vines, sunny side. Also: Santa Fe, sunny steep side of stony hills.
Also: 4 October, 1846. Between Poni and Rayado Creeks.
777 17 August, 1846. A few miles east of Bluff Creek. 18
September, 1847. Bluff Creek, bottom land.
q 14 August, 1846. Hickory Point, creek bottom.
779 21 September, 1847. Creek bottom, 20 miles east of 110

780 29, 31 August, 1846-47. Cow Creek [31 Aug., 1846] and
Middle Spring [29 Aug., 1847] (Cimarron). Dry prairie.

781 30 June-19 October, 1846-47. Santa Fe, lower part of
the town, creek valley, gravelly soil. Also: on steep sides of gullies,
rather stony soil.
182 25 August, 1846. Council Grove, creek bottom.

= 19 June-8 September, 1847. Santa Fe, in and near
fields. Pawnee Fork, steep banks of the river.

4 17 August, 1846. 120 miles west of Independence to
Waggon Mound (New Mexico); in gullies and low places of the prai-
rie.

#85 16 & 17 September, 1847. 18 miles west of Council
Grove to Fort Leavenworth.

6 2 May-28 July, 1847. Santa Fe Creek valley; foot and
Sunny sides of high mountains.

dl 8 ss 1847. Pawnee Fork. Steep banks of the

creek

a 31 August, 1846. Between Cow Creek and little Arkan-
Sas River. :
» 10-16 August, 1847. Between Santa Fe and Moro River.

Around the stand of shrubs, mountainous country.
oy 24 aise "1847. Rio del Norte, low sandy river bank, be-

tween stones. Leaves somewhat serrate, especially towards their apexes.

58 ELIZABETH A. SHAW

791 6-12 August, 1847. Santa Fe to San Miguel.

792 23 August, 1846. Council Grove; hills, ledges of rocks,
in crevices.

793 25 August, 1847. McNees’s Creek, sandy hillside.

794 28 August & 1 September, 1847. Ford of the Cimarron

to Sand Creek; low prairies.

795 25 May-3 October, 1846-7. Rio del Norte [25 May,
1847], sandy soil, low river bank between rocks. Also: Santa Fe and
Poni Creek [3 Oct., 1846].

796 30 June-10 July, 1847. Santa Fe; valleys between the
dry gravelly hills. Branches pilose.

797 6 June-18 August, 1847. Santa Fe, foot of hills.

798 8 September, 1847. Pawnee Fork; steep bank of the
creek, loose soil.
799 8 September, 1847. Pawnee Fork; prairie and steep bank
of the creek.
800 4 May-18 July, 1847. Santa Fe, sides of hills and moun-
tains
801 21 June, 1847. Santa Fe, dry hillsides.
802 14 August, 1846. Hickory Point, dry bed of a creek.
802b 1 September, 1847. Sand Creek, prairie.
803 24-25 May, 1847. Rio del Norte.
804 7 June, 1847. Santa Fe. Differs from the preceding one
in its seeds.

CUPULIFERAE
805 8 June, 1847. Santa Fe, eastern mountains. Shrub oak,

8 feet high.
806 8 June, 1847. Santa Fe, eastern mountains. A shrub
about 8 or 9 feet high.

807

6 June-1 November, 1846-47. Santa Fe, mountains. A
shrub about 9 feet high.
808

0 8 June, 1847. Santa Fe, mountains. A shrub 10 feet
igh.

809 8 June-19 July, 1847. Santa Fe Creek valley; banks of
the creek and foot of mountains. A shrub 8—12 feet high.

810 6 June, 1847. Santa Fe, sides of mountains. Shrub 8-10
feet high.

BETULACEAE
811

16-25 June, 1847. Santa Fe, upper part of the creek
valley, margin of the water. A shrub 15 to 20 feet high. 30 March-1
April in flower.

SALICACEAE
812 28 April-15 May, 1847. Santa Fe Creek bottom, exposed
to the inundations of the creek. Shrub 10-14 feet high. In drying the

AUGUSTUS FENDLER’S COLLECTION LIST 59

twigs leave a greasy mark on the paper. 30 March-1 April, 1847, in

flower.

813 12-20 May, 1847. Santa Fe Creek bottom; margin of the

creek and near irrigating ditches.

814 25 May, 1847. Rio del Norte; foot of steep mountain

banks, amongst rocks near the water.

815 11-20 May, 1847. Santa Fe, margin of the creek. A tree-

like shrub about 15 feet high. I could find no male flowers on this

shrub.

816 30 April-2 May, 1847. Santa Fe Creek, bottom bank of

the creek. Treelike shrub about 20 feet high.

817 12 May, 1847. Santa Fe, margin of the creek. The buds

of this tree are coated with a very sticky substance, which resembles

bees-wax, is transparent and in drying between paper is taken up by

the latter and makes the same transparent and sticky also. April in
ower.

818 Santa Fe. Close to the creek. Trees about 40 feet high;

lower branchlets long, slender and pending; trees of a fine growth in

the neighbourhood of a house. I could find no male flowers to this

tree. 10 May in flower.

819 Santa Fe Creek bottom, 4 miles above Santa Fe. A tree

with spreading branches; the bark is a shining white color, given to

it by a coat of a very white powdery substance which easily comes

off, adheres to clothes etc., with great tenacity. 12 April, 1847, in

Ower.
URTICACEAE
= 24 May, 1847. Rio del Norte, rocky bank of the river.
821 15 August, 1847. Moro River, bottom land. :
= 18 August, 1846. 15 miles east of Council Grove, cree
bottom, rich soil. ‘
823 15 August, 1846. 100 miles west of Independence, cree

bottom, fertile soil

~ 10 August, 1846. 8 miles south of Fort pale soba
‘ie 15 July, 1847. Santa Fe Creek, 6 miles above Santa Fe,
foot of mountains. :

= 15 August, 1847. Moro River, bottom land.

- 21 August, 1846. Council Grove, creek bottom. 15 miles
south of Fort Leavenworth. 11 August, 1846. Ravine.

CONIFERAE
ag 1 November, 31 January, 4 March, 1847. Santa Fe. In

ee higher part of the mountains about 5 or 6 miles east oe
Especially on the northern declivity of a sharp oe. rl eae

ey are to be found in great numbers with but very few _ aa
between them, while the southern declivity of that same 1 g

60 ELIZABETH A. SHAW

is occupied by a long-leaved pine no. 831, to the exclusion of almost
every other kind. The pistillate flower branches are to be found on
the top part of the tree, the staminate ones on the lower branches.
Bark of the younger trees smooth and of a white color, on old trees
only the summit and the branches are smooth and whitish. The ter-
minal buds of the branches furnished during the winter with a thick
and transparent coat of a resinous substance. A kind of balsam or
turpentine is elaborated in little cavities under the surface of the bark
around the larger branches and the stem of the tree by which the
surface is raised into scattered nodules. This turpentine is clear and
of a very light color and has an agreeable odor. Trees 60—80 feet
high, some of a very handsome pyramidical growth.

829 1 November, 23 February, 1847. Santa Fe. Eastern
mountains, upper regions as well as along the higher part of the creek
valley. I saw several trees of this kind from which the bark had been
peeled, probably for the use of tanning leather. The trunks of some
of the larger trees of this kind measure 7 feet in circumference, 5
feet above the ground.

830 30 March, 1847. Hills, eastern mountains, Santa Fe. A
shrublike tree, which attains a greater height towards the summit of
the mountains, but not above 20 feet. Mostly of a stunted depressed
growth, branching near the ground. Also in great abundance on the
elevated plains and hills west and southwest of Santa Fe. By far the
greater part of the firewood that is brought into Santa Fe by the
Mexicans is of this tree. 20 April in flower.

831 17 February, 2 March, 1847. Santa Fe, eastern mountain
range; low hills as well as high mountains; seems to prefer the drier
north sides of the mountains. Valuable timber trees. Their growth is
above 90 or 100 feet high, straight and some of the trunks are of
great dimensions.

832 8 February, 1847. Santa Fe. Eastern mountain range.
Upper region of high mountains. Fruit suspended, truck 60 to 80 feet
high; branchlets extremely flexible.

8 17 February, 1847. Santa Fe Creek bottom, about 7
miles above Santa Fe. A young, slender and straight tree about 20
feet high; branches arranged in whorls, around the trunk; branches
easily broken with exuding rosin from the broken places. Leaves ar-
ticulated, quadrangular; lower portion persistent to the branchlets. I
found only two trees of this kind.

834 30 March, 1847, in flower. Santa Fe. A shrub (the high-
est from 10 to 15 feet) very common all around Santa Fe. Lower re-
gions of the mountains, but more abundant on the dry gravelly hills
southwest of Santa Fe, and on the plains adjoining them. They are
found also on the summits of high mountains, but far less abundantly.

AUGUSTUS FENDLER’S COLLECTION LIST 61

The branchlets of this shrub dry much quicker than those of the next
following number.
835 30 April, 1847, in flower. Santa Fe, mountain valleys,
in the neighbourhood of creeks. A shrub commonly of 12 to 15 feet
high; some of its lowermost branches (near the ground) often 4 foot
in diameter. A treelike shrub of this kind I saw near the Creek, whose
main trunk measured 8 feet in circumference; the lowermost branches
very much spreading and growing to a height of about 25 feet.

3 June, 1847, in flower. Santa Fe Creek valley, in shady
places at no great distance from the Creek. Low shrubs from | to 244
feet high, branches very tough.

NAJADACEAE
837 21 June-17 July, 1847. Vicinity of Santa Fe, at the mar-
gin of ponds, immerged, leaves floating on the surface.
ALISMACEAE
838 17 August, 1846. About 25 miles east of Council Grove,
marshy ground, creek bed.
839 21-23 June, 1847. Santa Fe, shallow ponds. 17 July, in
fruit.
840 2 September, 1847. Shallow pond-like collections of water
in the prairies, 40 miles west of the ford of Arkansas River.
ORCHIDACEAE
841 14 August-27 September, 1846. Low prairies 9 miles

south of Fort Leavenworth; also about 100° west longitude, not far
from the bed of the Arkansas River. Flowers white.

IRIDACEAE
842 22 June-7 July, 1847. Santa Fe, low grassy places near
irrigating ditches.
SMILACEAE
843 18 June, 1847. Santa Fe, cultivated.
844 17 June, 1847. Santa Fe Creek valley; overshaded bank

of the creek.
845 3-10 June, 1847. Santa Fe, shady bank of the creek,
fertile soil.

3-24 June, 1847. Santa Fe Creek valley, foot of moun-
tains, shady places, northern declivities.

LILIACEAE
847 8 July, 1847. Santa Fe, foot of mountains, creek valley.
848 28 June-1 October, 1846-7. Santa Fe to Rock Creek;

also between Raton and Bermejo Creek (between Bent’s Fort and
Santa Fe). Near creeks and on level prairies, also foot of mountains.

3 June, 1847, in flower. Santa Fe, sunny steep declivi-
ties of mountains, between rocks; ever-green. The rhizomas of this
plant which sometimes are to be found of 3 inches diameter, are used

62 ELIZABETH A. SHAW

by the Mexicans as a substitute for soap. For this purpose, they are
mashed into pieces with a stone, hot water poured onto them, and
the clothes rubbed with them. In drying the leaves I found that from
their bases where they has been cut, some drops of a clear transpar-
ent substance, which was sticky like glue, but had no smell, was ex-
udating. Some of the leaves over three feet long.
850 24 May-21 August in flower. Santa Fe to Rock Creek,
at the foot of mountains, on shady declivities, around shrubs. Also on
dry sterile hills about Bent’s Fort, but more abundantly in the valleys
between these hills. The rhizomas of this plant are likewise used as
a substitute for soap. This plant does not seem to flower every year.
851 17 August, 1847. 2 miles east of Moro River. Elevated
rocky region.

PONTEDERIACEAE
852 2 September, 1847. Pondlike collections of water in the
prairies 40 miles southwest of the ford of Arkansas River. Flowers of
a fine blue colour.

MELANTHACEAE
853 1] August, 1846. 8 miles south of Fort Leavenworth,
high prairies.
854 28 July, 1847. Santa Fe Creek, 12 miles above Santa Fe,

foot of mountains.

JUNCACEAE, CYPERACEAE & GRAMINEAE
855 6 June—26 July, 1847. Santa Fe, wet grassy places of the
creek bottom.
856 20 May-17 August, 1847. Santa Fe and Rio Moro, ele-
vated rocky regions not far from the water.

857 9 June-12 July, 1847. Santa Fe Creek, margin of the
water

858 16 June—28 July, 1847. Santa Fe Creek, margin of the
water

859 3 September, 1847. Ford of the Arkansas River, wet
bottom land.

860 15 May-7 July, 1847. Santa Fe Creek, near the water.
861 3 June, 1847. Santa Fe Creek, bottom land exposed to
inundations.

862

14 August, 1847. 12 miles west of Vegas, elevated rocky
region.

22 August, 1846. Council Grove, shady place underneath
projecting ge8 of rocks, loose fertile soil.
864 28 July, 1847. Santa Fe Creek valley in the mountains.

864b 14 August, 1847. Between San Miguel and Vegas, open
woodland. Flower white.

AUGUSTUS FENDLER’S COLLECTION LIST 63

865 25 June—28 July, 1847. Santa Fe, foot of mountains near
irrigating ditches.

866 14-25 August, 1847. San Miguel—McNees’s Creek, sandy
hillside.

867 27 August, 1846. 30 miles west of Council Grove, low
place.

868 25 July-14 August, 1847. Santa Fe to Vegas, woodland.

869 13 August, 1846. 80 miles west of Independence. 3 Sep-
tember, 1847. Ford of the Arkansas River, wet bottomland.
870 18-26 September, 1847. Bluff Creek, bottom. [N.B. 26
September must be an error; Fendler on that day was near Liberty,
Missouri. ]
871 23 August, 1846. Council Grove, dry parts of the grav-
elly creekbed.

27 May-12 July, 1847. Santa Fe, margin of the creek.

873 26 April, 1847. Santa Fe Creek, low grassy bank.

874 27 May-18 July, 1847. Santa Fe, margin of a pond, wet
soil.

875 21 May-7 June, 1847. Santa Fe, immersed in a pond.
876 3 September, 1847. Ford of the Arkansas River, wet

bottom land.
7 3 September, 1847. Ford of the Arkansas River, low wet
places near the river.

26 April-24 June, 1847. Santa Fe, margin of the creek
and of irrigating ditches.
879 15 May, 1847. Santa Fe, Creek valley, moist places.

880 16 June, 1847. Santa Fe, margin of the creek in the
mountains.

8 21 May-19 July, 1847. Santa Fe, margin of the creek.
882 31 May-16 June, 1847. Santa Fe Creek bank.

883 16 June-7 July, 1847. Santa Fe, margin of irrigating
ditches.

884 15 May-24 June, 1847. Santa Fe Creek bank, near the
water.

885 3 June, 1847. Santa Fe, shady bank of the creek, close
to the water.

886 23 August, 1847. 7 Miles east of Rock Creek.

887 21 May-7 June, 1847. Santa Fe, wet meadows near a
ond.

888 3 June, 1847. Santa Fe Creek valley, not far from the
creek,
889 20 May—6 June, 1847. Santa Fe, margin of irrigating

ditches.

64 ELIZABETH A. SHAW

890 17 April, 1847. Santa Fe Creek, moist places near the
creek.
891 26 April, 1847. Santa Fe, foot of mountains, not far from

irrigating ditches.
892 24-27 April, 1847. Santa Fe, wet meadow near a pond.
893 24 May, 1847. Rio del Norte, margin of the water on
somewhat elevated banks, between rocks.
10 May-—2 August, 1847. Santa Fe, dry loose and level
ground; also sandy soil and gently sloping declivities.
5

89 27 August, 1847. Upper Spring (Cimarron River), bank
of the river.

896 19 July-16 August, 1847. Santa Fe—Moro River, rocky
bluffs.

897 27 August, 1847. Upper Spring (Cimarron River), margin

of the water.
898 27 August, 1847. Upper Spring (Cimarron River), margin
of the water.
899 19 June-10 August, 1947. Santa Fe, elevated places of
the creek bottom; also near the irrigating ditches of fields.

19 August, 1846. Council Grove, bottom land.
901 3-19 July, 1847. Santa Fe, margin of irrigating ditches
at the foot of mountains.
902 22 June-8 August, 1847. Santa Fe Creek bottom.

903 20 May-18 July, 1847. Santa Fe, dry valleys at the foot
of hills, loose soil.
3

903b 31 July, 1847. Santa Fe Creek valley.

904 15 July, 1847. Santa Fe Creek valley, foot of hills.
905 19-31 July, 1847. Santa Fe Creek bottom.

a 21 September, 1847. Bottom land of Kansas River.

8-15 July, 1847. Santa Fe Creek valley near irrigating
ditches.

908 8 June-15 July, 1847. Santa Fe Creek valley at the foot
of hills, not far from the creek bank.
909

22 June-26 July, 1847. Santa Fe Creek valley, foot of

hills.

910 16 June—12 July, 1847. Santa Fe.

911 17 August, 1846. 110 Creek, slopes of the creek bank.
912 21 May, 1847. Santa Fe, in a wet meadow.

913 25 August, 1847. McNees’s Creek, sandy hillside.

914 15 August-18 September, 1847. 100 miles west of In-

dependence to Bluff Creek, fertile bottom land.
5 12 August, 1847. Near San Miguel, open woodland.

916 7 July-16 August, 1847. Santa Fe-Moro River, foot of
rocky hills.

AUGUSTUS FENDLER’S COLLECTION LIST 65

917 28 July, 1847. Santa Fe Creek valley.

918 10-31 July, 1847. Santa Fe, foot of hills at the margin
of fields

919 24 May-16 June, 1847. Rio del Norte-Santa Fe, side of
hills.

920 28 June—7 July, 1847. Santa Fe Creek valley, shady bank

of the creek.
24 June-28 July, 1847. Santa Fe, foot of mountains.

922 7-12 July, 1847. Santa Fe, side of mountains.
923 12-26 July, 1847. Santa Fe, side of mountains.
924 16 June-26 July, 1847. Santa Fe, protected by shrubs,

at the foot of hills near fields. :
28 August, 1847. Between the ford of Cimarron River

and Willow Bar, deep sandy somewhat moist soil. :

9 28 August, 1847. A few miles west of the ford of Cim-

arron River, high sandy prairie.

927 6 June-31 July, 1847. Santa Fe Creek valley, foot of

mountains.

sai 16-22 June, 1847. Santa Fe Creek valley. :

> 3-16 June, 1847. Santa Fe, margin of the creek in the
mountains. ;

929b 3 June-15 June, 1847. Santa Fe, margin of the creek in
the mountians.

= 27 May-12 July, 1847. Santa Fe, wet meadows.

| 27 May-15 July, 1847. Santa Fe, low wet meadows. :
aa 22 April-16 June, 1847. Santa Fe Creek valley, side 0

mountains between rocks, also foot of mountains and steep rocky
banks,

“al 7 June, 1847. Santa Fe, margin of the creek.

~~ 18 September, 1847. Bluff Creek, bottom.

17-20 August, 1847. Ocate Creek to Moro River, ele-

vated rocky places.

ou ‘23 August-26 September, 1847. Council Grove, gravelly

part of the creek. [N.B.: locality and dates do not agree. | ao
10-13 August, 1847. A few miles east of Santa Fe to

neighbourhood of Ojo de Bernal; woodland.

ed 25 May_26 July, 1847. Rio del Norte, steep bank be-

Ween rocks: also Santa Fe, waste gardenland.

7 June—25 July, 1847. Santa Fe, along the mudwalls of

fields,
os 31 May-7 June, 1847. Santa Fe Creek, grassy somewhat
oe parts of its bottom land.

26 July-23 August, 1847. Santa Fe, Rock Creek, dry

Prairie

66 ELIZABETH A. SHAW

941 16 June—7 July, 1847. Santa Fe Creek, foot of hills.
942 15 May, 1847. Santa Fe Creek.

943 30 April-24 June, 1847. Sunny bank of Santa Fe Creek.
944 9 June-29 July, 1847. Santa Fe Creek valley, wet soil

near a spring.
945 1 June—28 August, 1847. Rio del Norte, low sandy bank;
Santa Fe, flat sandy places to the ford of Cimarron River; also be-
tween Bent’s Fort and Santa Fe.
946 17 May-19 July, 1847. Dry sandy plains between Santa
Fe and the Rio del Norte; also Santa Fe, foot of hills near fields.
94 3-31 July, 1847. Santa Fe Creek, valley, foot of hills
near fields.
948 13-23 August, 1847. Ojo de Bernal, Moro River and
Rock Creek.
94 2 August-1 September, 1847. Rather high and dry prai-
ries, Santa Fe to Little Arkansas River, from which region east it is
only found occasionally, small isolated patches along the road as far
as Council Grove.
950 13 August, 1847. Near Ojo de Bernal.
951 29 June-27 August, 1847. Santa Fe, Moro River to
Council Grove.
2 14 August, 1846. A few miles west of Hickory Point, low
wet bottom.
9. 18 September, 1847. Bluff Creek, bottom.
24 September, 1847. 8 miles south of Fort Leavenworth,
along the roadside.
955 24 May-28 August, 1847. Rio del Norte, slopes of some-
what evelated banks, also: a few miles west of the ford of Cimarron
River, prairie.
D6

28 August, 1847. A few miles west of the ford of Cim-
arron River, low prairies.

956b 3 July, 1847. Santa Fe Creek, valley up in the moun-
tains.
957 : 3 July, 1847. Santa Fe Creek, valley not far from the
creek.
958 3 June-19 October, 1846 & 47. Santa Fe Creek valley,

gently sloping declivities, rather dry soil.

959 6 June-15 July, 1847. Santa Fe, margin of the creek.
9 June-10 August, 1847. Santa Fe, foot of dry hills,

somewhat sandy soil.

eh 8 September, 1847. Pawnee Fork, creek bottom, shady

places.

962 26-28 July, 1847. Santa Fe Creek valley, near fields.

963 7 July, 1847. Santa Fe, margin of the creek.

AUGUSTUS FENDLER’S COLLECTION LIST 67

964 24 September, 1847. Near Fort Leavenworth.

965 3 October, 1846. Bottom prairie and banks of Poni
Creek, fertile soil.

966 16 August, 1847. Moro River, hillside between rocks.
967 22 September, 1847. 6 miles north of the upper ferry
of Kansas River.
968 13 August to 1 October, 1846-47. Ojo de Bernal & Rock
Creek [13 Aug., 1847], elevated prairie, in spots where the soil is
loose and fertile; also: between Bermejo and Colorado River [1-Oeb,
1846], in similar localities.

969 1 June-1 October, 1846-47. Santa Fe-Vegas, and Col-
orado River; prairie valleys, good soil.

970 24 June-15 July, 1847. Santa Fe Creek valley near an
irrigating ditch.
971 15 July, 1847. Santa Fe, not far from the creek, at the
foot of mountains.

972 22 June-28 July, 1847. Santa Fe, margin of the creek.

A 30 May-6 August, 1847. Santa Fe, dry hillsides, gravelly
soil.

974 8 September, 1847. Pawnee Fork, creek bottom.

975 24 May, 1847. Rio del Norte, sides of somewhat ele-
vated banks fronting the water, between rocks.

97 1 Sept., 1847. Prairie at Sand Creek.

977 14 Aug., 1847. Between San Miguel and Vegas.

A 19 May-13 July, 1847. Rio del Norte and Santa Fe,
sunny hillsides, between rocks.

he 17 May-6 Aug., 1847. Rio del Norte, sandy elevated

istance from the water; also Santa

places of the river bank, at some d
r protection of shrubs.

Fe, foot of hills, dry loose sandy soil, unde

~ 16 June-10 Aug., 1847. Santa Fe, foot of dry hills.

eds 30 May-13 Aug., 1847. Santa Fe, dry gravelly valley;
also neighbourhood of Ojo de Bernal. ne
me 17 Aug., 1847. 2 miles east of Moro River, foot o igh
rocks,

17-31 July, 1847. Santa Fe, waste fields; Rio del Norte,
Ow sandy river bank.
tg 29 Aug., 1847. Between Willow Bar and Middle Spring

(Cimarron River).
= 15 May-3 June, 1847. S
no great distnce from the creek.

° 19 July, 1847. Santa Fe, grassy m

anta Fe Creek, bottom land at

argin of an irrigating
ditch.

a 12 Aug.—8 Sept., 1
= 16 July—3 Sept., 1847. Santa Fe Cree

7. Pawnee Fork and Pecos, prairies.
soa k valley near fields;

68 ELIZABETH A. SHAW

Willow Bar, sandy wet bottomland; ford of Arkansas River, low bot-
tom lan

989 27 Aug., 1847. Upper Spring (Cimarron River), margin
of the water.
990 21 Sept., 1847. Prairie about 10 miles southwest of the
upper ferry of Kansas River.

991

9 27 Sept., 1847. Bank of Missouri River above Washing-
ton.

992 12 Aug., 1847. 10 miles east of Pecos, low prairie.

993 21 July, 1847. Santa Fe Creek valley, not far from the
creek

994 16 Aug., 1847. Moro River, hillside between rocks.
995 17 Aug., 1846. Creek bottom near Council Grove.

996 26 July-8 Sept., 1847. Santa Fe Creek bottom; Ojo de

Bernal, margin of a pond; ford of the Arkansas River, wet bottom
land; Pawnee Fork, creek bottom.

997 12 Aug., 1847. 10 miles east of Pecos.

998 26 Aug., 1847. A few miles west of the ford of Cimarron
River. :

999 12 Aug., 1847. Ojo de Bernal, low grassy places in the
mountains.

1

3-24 June, 1847. Santa Fe Creek, margin of the creek.
1001 13-16 Aug., 1847. Ojo de Bernal—Moro River, summit
oo a rocky hill.

15 May-12 July, 1847. Santa Fe, margin of the creek.

on 24 Sept., 1847. 8 miles south of Fort Leavenworth, in
ravines along the road side.
1004 17 Aug., 1847. A few miles east of Moro River, deep
standing water.
100 14 Aug., 1846. About 90 miles west of Independence,
under water in a deep large pool.

18 Sept., 1847. Bluff Creek bottom, submerged in the
water of a shallow pool.

1007 & 1008 2 species. 17 Aug., 1847. A few miles east of Moro
River, surface of deep standing water.

009 29 April, 1847. Santa Fe, in a pond-like collection of
water fed by a small spring.

EQUISETACEAE

1010 22 April-22 June, 1847. Santa Fe Creek bottom, near
the creek and irrigating ditches; also Rio del Norte, low bank, margin
of the water.

1011 8-22 May, 1847. Rio del Norte, low wet sandy bank of
the river.

AUGUSTUS FENDLER’S COLLECTION LIST 69

1012 22 June, 1847. Santa Fe, foot of hills, near the creek.
1013 4 March-2 May, 1847. Sunny bank of Santa Fe Creek.
FILICES

1014 3 June, 1847. Santa Fe Creek valley, grassy places.
1015 12 April-11 November, 1846-7. Santa Fe, shady sides

of mountains, near the foot as well as on the summit. In good soil
under protection of rocks and roots, in crevices. Early in the spring,
the leaves show a tendency to roll up as soon as the plant is taken
from the ground. Mostly green throughout the winter.

1016 23 Feb., 1847. Eastern mountains, Santa Fe; sunny de-
clivities in crevices of steep masses of rocks.

1017 8-25 May, 1847. Rio del Norte, deep ravines in the
trap-formation, rachis flexuose, pinnae alternate.

1017b 22 August, 1846. Council Grove, underneath projecting
ledges of rocks, high dry region. rhachis straight, pinnae opposite.
1018 20 Aug., 1846. Council Grove, summit of a hill, deep

crevices in the rock.
101 28 July, 1847. Santa Fe Creek, 12 miles above Santa Fe,
high mountains, shady places in ravines.

1020 5 June, 1847. Santa Fe Creek, bottom, near the creek.
1021 21 Aug., 1847. Council Grove, between rocks, in a ra-
vine.

1022 15 May-24 June, 1847. Santa Fe Creek bottom, shady
places at the foot of rocks, moist fertile soil.

1023 30 March-17 Aug., 1847. Santa Fe Creek, shady bank,

between rocks, good soil. Also 2 miles east of Moro River, green
throughout the winter.

LYCOPODIACEAE
1024 13 Feb.-17 April, 1847. Santa Fe Creek, perpendicular
Side of rocks facing the creek, in the higher regions of the mountain
Tange, as well as on the shady bank of the lower creek. Begins to be
lively green in the month of March.
1025 16 Feb.—April, 1847. Santa Fe,
upper region of the mountians.

between rocks in the

CHARACEAE
1026 29 April, 1847. Santa Fe, in a shallow pond fed by a
Spring
ACKNOWLEDGEMENTS
I wish to bara Mykrantz, Archivist of the Missouri Botanical
Garden Berperecacpener ue ee New York Botanical Garden, respec-
Se ee gs m Asa Gray to George Engel-

tively, for making available to me copies of letters fro

mann, translations of letters from Fendler to Engelmann, copies of letters from Gray

70 ELIZABETH A. SHAW

toa Torrey, and for providing permission to quote from these letters. I would like
also thank Thom 2 of the University of New Hampshire for his interest
N Fendlers pallets for conversations about them.
The collection list and te letters from George Engelmann to Asa Gray are in the
archives of the Gray Herbarium and I am grateful to the Director of the Gray Her-
arium for permission to use these materials.

LITERATURE CITED AND OTHER REFERENCES

BERRY, Kansas Before 1854: A Revised Annals. Part Fifteen, 1846. Kansas
hs povvglen 30: 339-412.
ave W. M. 1885. An era and Some Reminiscences of the Late August
endler. I, Bot. Gaz. 85-290; II, Bot. Gaz. 10: 301-304; III, Bot. Gaz. 10:
ae
Emory, W. H. 1848. Notes of a em Reconnoissance. U.S. 30th Congress, Ist
Sess., Senate Exec. Doc. no. 1-126.
Gray, A. 1843. Notes of Botanical Collections. Amer. J. Arts 45: 225-227.
apinrs 9. Plantae Fendlertanae Novi-Mexicanae. Mem. Amer. Acad., n. ser.,
6.
ra J. 1844. Commerce of the Prairies: or the Journal of a Santa Fe Trader. Henry
Langley, New York. vol. 1, xvi, 320 pp.; vol. 2, viii, pp.
McKee S. D. 1955. Botanical Exploration of Trans- -Mississippi West. 1790-1850.
The Arnold Arboretum, Jamaica Plain, Mass. 1144 pp.
tg C. T. & G. E. Crow. 1980. Type specimens of the Hodgdon Herbarium,
University of New Hampshire. Rhodora 82: 579-597. Published one are pho-
tographs of pages from “Plantae Pondicletine” and from the manuscript collection

RIDDLE, K. 1963. Records and Maps of the Old Santa Fe veg ales and Enlarged
Edition. John K. Riddle, West Palm Beach, Florida.

jibe J. D. 1971. The Santa Fe Trail—A cece bideerabls The Universi-
ty o w Mexico Press, Albuquerque. 271

STANDLEY, P C. 1910. The Type ae ees ‘g ‘ard First Described from New Mex-
ico. Contr. oe mr Herb.

sie oy HL: 1. The road "4 ae cdl Hastings House Publishers, New York,

pp.
THoreavu, H. D. 1862. Walking. eaerg cdo 9: 657-674.
Wane, (ed.) ar The Journals of s Parkman. Harper and Brothers Pub-
ishers, New York, vol. 1, xxv ian ney 2. vii, ais 7
Wisin A. 1848. Memoir of a tour to xico, ‘onetiod with Colonel
no. 26:

Fi Wes expedition. U.S. 30th Congress, lst ra Senate Misc. Doc.

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE)'
REED C. ROLLINS

In many genera of the Cruciferae originally based on species from
the Old World (Arabis, Cardamine, and Draba, for example) there
has never been any doubt about their occurrence natively in the west-
ern hemisphere. Sisymbrium has been similarly treated as an inclusive
genus with species in most parts of the world. But in the latter, treat-
ments have fluctuated enormously. Such early floras dealing with
North American plants as Hooker (1829-30) and Torrey and Gray
(1838-40) interpreted Sisymbrium broadly and attributed many anom-
alous species to the genus. However, since these earlier works there
has been a steady narrowing of the genus. Schulz (1924) was the pri-
mary architect of a reduced Sisymbrium as far as North American
species are concerned. But Greene (1896) and Rydberg (1907) had
already started the process. Schulz recognized that Halimolobos, al-
though initially based on Mexican species, also included others from
farther north [H. diffusa (Gray) Schulz and H. virgata (Nutt.) Schulz]
formerly placed in Sisymbrium. Since then, other species north of
Mexico, H. mollis (Hook.) Roll., H. perplexa (Hend.) Roll. and H.
whitedii (Piper) Roll., were moved from either Sisymbrium or Arabis
to Halimolobos (Rollins, 1943). In another direction, Schulz carved
Romanschulzia and Coelophragma from Sisymbrium. The opposite
trend was initiated by Payson (1922) when he assembled in Sisym-
brium many species that had previously been recognized as belonging
to Thelypodium. In doing so, he placed Schoenocrambe Greene, The-
lypodiopsis Rydb., and Hesperidanthus Rydb. into synonymy under
Sisymbrium.

I have followed Payson in his realignment of Sisymbrium for many
years, but it has been increasingly difficult to do so. The impact of
new and increased knowledge of the North American species involved
has been to emphasize their differences from Sisymbrium rather than
their similarities. It is interesting and telling that Al-Shehbaz (1973)
and Payson, (1923), both studying Thelypodium, came to different
conclusions with regard to many of the species treated in the present
paper. Payson regarded them as belonging to Sisymbrium, Al-Shehbaz
brium. He transferred Thelypodium
(Wats.) Pays.] and Thely-
podium linearifolium (Gray) Wats. [Sisymbrium linearifolium (Gray)
Pays.] to Thelypodiopsis and at
as Rydberg had originally constitute
the eine tke oe Al-Shehbaz with respect to those North American

‘Research supported by National Science Foundation Grant DEB 78-08766.

71

7 REED C. ROLLINS

species referrable to Thelypodiopsis but treated as Sisymbrium by
Payson. I followed Al-Shehbaz in my study of Thelypodiopsis of Mex-
ico (Rollins, 1976) in connection with the preparation of a treatment
of the Cruciferae for Johnston’s Flora of the Chihuahuan Desert. In
doing so, I further narrowed Sisymbrium as far as its involvement with
American species is concerned. In the present study, I have con-
cluded that none of those North American species previously placed
in Sisymbrium should remain there. Sisymbrium is now regarded as
being an Old World genus which, and along with such genera as Bras-
sica, Diplotaxis, Eruca, Raphanus et al., has only alien species in the
North American flora.

Al-Shehbaz’s action (1973) in bringing Sisymbrium linearifolium into
Thelypodiopsis was important because it showed that species had con-
nections with other species of the genus, but he apparently did not
take fully into account its close relationship to Schoenocrambe linifolia
(Nutt.) Greene [Sisymbrium linifolium Nutt.]. These two species can
hardly be placed in separate genera. In this respect I cannot help but
agree with Payson (1923) who, in writing of Sisymbrium linifolium
says, “It seems very doubtful if there is any difference other than one
of degree between the root system of this plant and of Hesperidan-
thus [S. linearifolium]. I should not hesitate to adopt a genus on very
slight morphological grounds if it became evident that it was really
distinct phylogenetically. In this case, however, there seems no reason
to doubt the close relationship between S. linifolium and the related
species [S. linearifolium] and neither is there significant grounds for
maintaining them distinct generically.”

Recent discoveries of new species (Welsh & Atwood 1977; Welsh.
1981) have filled in the gaps between Thelypodiopsis and Schoeno-
crambe to an extent that makes both of these genera pivotal between
Thelypodium and Sisymbrium. The type species of Schoenocrambe,
S. linifolia, has habital phases that resemble S. argillacea Welsh &
Atwood and S$. barnebyi Welsh & Atwood on the one hand and S.
linearifolia on the other. And there are no technical characters that
are sufficiently fundamental to place these species in different genera
from each other.

As in other genera of the Cruciferae, Thelypodiopsis and Schoen-
ocrambe are not well-defined in the sense that all species in each
genus are completely unlike those of any other genus. In fact, The-
lypodiopsis elegans is fairly close to Thelypodium sagittatum and there
are features of Thelypodiopsis ambigua that resemble those of The-
lypodium sagittatum. Our generic concept is based on the premise
that a genus is made up of a constellation of species that are puta-
tively more closely related to each other than to the species of any
other genus. Furthermore, the center of diversity represented by the

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 73

species of such a genus is different from the center of diversity of
any other genus. At the fringe of this diversity there may be species
that come close to the fringe of diversity of another genus, but
taxonomically the submergence of one genus or the other is not re-
quired. The evolutionary connections in such a situation are clearer
than in cases where an evident gap is present between the genera.
The number of instances in the Cruciferae where two or more genera
come together (Rollins, 1939, 1950, 1960) are so numerous that this
is almost the rule rather than the exception. This argues for the rel-
ative “youth” of the family because the evolutionary connections be-
tween the genera are still present whereas in many “ancient” families
great gaps between genera exist. Not only are Thelypodiopsis and
Schoenocrambe genera of the nature mentioned above as far as each
other are concerned, they in turn are closely connected to other gen-
era. This is particularly true of Schoenocrambe where S. linifolia is
very close to Sisymbrium polymorphum, a species of central Asia and
eastern Europe. In fact, S. polymorphum is somewhat out of place
in Sisymbrium and may well be nearer its congeners in Schoeno-
crambe.

Geographically, the species of Schoenocrambe and Thelypodiopsis
occur most abundantly in western United States and northern Mexico.
One species, Schoenocrambe linifolia, barely gets into Canada. The
most northerly known stations are in British Columbia. The most
southerly species is Thelypodiopsis alpina which occurs in Guatemala.
Several species, such as Schoenocrambe linifolia, S. linearifolia, and
Thelypodiopsis elegans, are relatively widespread and abundant. All
three of these species are frequently collected and are well repre-
sented in herbarium collections. Schoenocrambe linearifolia occurs
from Colorado and Utah to Coahuila, Mexico; S. linifolia occurs from
British Columbia to Arizona and New Mexico. Thelypodiopsis elegans
is found in western Colorado, in the northern half of Utah, and in
southwestern Wyoming. Thelypodiopsis purpusii is infrequently col-
lected but it has been collected at disjunct localities over 2 fairly —
range extending from northern Arizona to Coahuila. Otherwise, : e
Species of Thelypodiopsis and Schoenocrambe tend to be local in their
Occurrence and they are infrequently collected.

ly large and showy. In

bright purple, those of T.

elegans ran ; : le. Only three species,

ge from white to light purple ee © :
°crambe linifolia, Thelypodiopsis aurea, and T. divaricata, have yel
‘Ow or yellowish flowers. In all others, the flowers are pu

ish, or white.

rplish, whit-

74 REED C. ROLLINS

THELYPODIOPSIS

Thelypodiopsis Rydberg, Bull. Torr. Bot. Club 34: 432. 1907.

Erect annual or biennial (rarely pean! herbs with one to several stems from the
-_ usually branched; glabrous or if pubescent, trichomes simple; leaves hetero-
lackin

rs
racemes which terminate each branch; s ok glabrous or sparsely pubescent in a few
species; petals narrowly lingulate to eon: or obovate, white to purple (yellow in
T. A p :

KEY TO THE SPECIES
A. Cauline leaves ovate to broadly oblong, strongly auriculate, large auricles clasping
stem.
B. Basal leaves ae eer lobed to pinnatifid, lower cauline leaves at least
pinnatifid. Plants of Arizona to Texas and Mexico
Gc. ue ee beked torulose; etal 8-10 mm. long; lower leaves ~~ gps pe

A NE eg eee woe T: usii.
B. Basa leaves and lower cauline entire or at most dentate (rarely Oe BS pho
one phase of T. elegans).
. Pedicels erect, mostly appressed to rachis; siliques less than bos long;
TNE OF DORON a a 4. oainalor
D. Pedicels divaricately ascending, widely spreading or slightly descending si-
liques 4-10 cm. long; pee annual or biennial (except T. wootonii of Mexi

perenn
E. Siliques stipitate, Souder stipe 3-7 mm. lon
F. Sepals and petals purple; sepals pao te anthes
G. Plants glabrous wonder coarse; styles 1 mm. or ess long; pedicels
at right angles to rachis, slightly curved upward,
cta

(slightly longer in var. erecta) ..................... T. ambigua
G. Plants pilose toward base of st stem, stems slender; swies 2-3 mm igen
pedicels divaricately ascending to widely spreading, 1-2 cm. long o
lower mature siliques ©.04.5..,0 52 13 paris +
F. Mi ca and petals yellow; sepals spreading to somewhat reflexed during
Py aCe ene Pere AU RNC Sire entagaieme ata ap nL euena! ‘ rea.
E. _— wut or with a short thick stipe less than 2 mm. lon od oe
wer cauline leaves oblong to ovate, acuminate to wate feds ob-
tse ee elow ~ middle.
Pedicels less

-5 cm. long, mostly ca. 1 cm. long; plants not
rhizomatous.

J. Inflorescence lax, few- flowered; siliques few and remote; cauline

leaves acuminate, scarcely reduced upward; plants of Guatemala
eee re ee 8. T. alpina.
is ee dense, many-flowered: siliques many on forming a

ense infructe escence: cauline leaves acute, reduced u ards.
K. Petals purplish, light a os or white, sear aty tay inc from

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 75

blade to claw, spatulate, not constricted; sepals purplish, spread-
ing, or if erect not forming an urn-shaped c soles
L. Styles of developed siliques expanded at summit, club-sh
plants sparsely pilose at least at base of stems; ae on.
to somewhat curved, usually not tortuous or torulose .......
Le ta ee ea tn eet ee 4. T. elegans.
L. Styles of developed siliques uniform in diameter, cylindrical;
plants wholly sian or rarely pilose at base of sia asia
tortious and forlilose © ois ass 7. 0
K. Petals yellow, pera constricted between small Stes pats blade
~ broadened membranous aco epals yellowis ra erect,
T. di

t, forming an ee -shape te a i oe toate.
I. Pedicels more than 1.5 cm. Dae mostly ca. 2 cm. long; plants
wootonii.

H. fer apie ra leaves pandurate or obovate, obtuse, sounded at apex;
widest above the mi
M. Pace widely spreading to arched downward; ites rind de-

scending; somewhat arched to nearly straight ....... reuata.
M. Pedicels eee ascending;. siliques lees obser
9. T. shinnersii.

atraleit ois cave ds cece coe ec decor h eee eee em
A. Cauline leaves narrowly oblong without ae or petiolate (weakly auriculate in
some plants of T. vaseyi, auricles not clasping stem).
Cauline leaves petiolate , sharply and eet ae to lobed, very re fruiting

pedicels widely spreading .............:eeeecete recent eee eenes T. byeii.

N. Cauline leave eee at base, entire or rarely weakly dent
O. Pa a s and pedicels ascending; siliques less than 2.5 cm. ie strongly to-
15.27. a

: Bet Sn ieee 4 Saree eh ne rome rae s
O. siues and pedicels strongly reflexed; siliques 4 cm. or more long,
strongly torulose; petals purplish ....-....-.+---+-+++5> 16.7, iciay ene

1. Thelypodiopsis ambigua (S. Wats.) Al-Shehbaz, nee on Thely-
podium ambiguum §. Watson, Proc. Amer. Acad. 290, 1879. Si-
symbrium ambiguum (S. Wats.) Payson, Univ. —. ar Sci. 1: 11,
1922. Thelypodiopsis ambigua (S. Wats.) Al-Shehbaz, Contrib. Gray
Herb. 204: 138, 1973.

Annual or biennial, glabrous ate Hoan glaucous; stem single, coarse, branched
above, up to a meter or more tall; basal leaves — ate, ascending, oblanceolate,
obtuse, coarsely and irregularly dent Re se entire, up t m. long, 1-2 cm. om
i auriculate, ace to lanceolate, up to 1

er often dentate; inflorescences dense with

riage long, bade widely spreading to s
o claw; stamens slightly exserted; Gienents slender; anthers curve ed, 3.4 mm. Se.
ak widely _spreading markedly expanded at summit, usually slightly curved

upward, 6-9 mm. long; siliques widely spreading, s mewhat curved ysesagion to
nearly cera seceke, slightly torulose, 9 cm. long, slightly more than 1 m
diameter, stipitate; stipe m. long; styles ca. 1 mm. long; “— cbsurely 3.
lobed; seeds oblong, not winged, ca. 1.5 mm. long, less than 1 mm. wide. 11
(Rollins and es ag 1971).

KEY TO THE VARIETIES

pr — purple; pedicels spreading at right angles to rachis; genes mee
Re eee et eee hee la. var. ambigua

76 REED C. ROLLINS

Petals light purplish to nearly white; pedicels divaricately ascending; siliques ascending
terete ee ee nt as aN Ste cs lb. var. erecta.

la. T. ambigua (S. Wats.) Al-Shehbaz var. ambigua.

DISTRIBUTION: Arizona in Coconino, Mohave, and Yavapi counties.
It was listed from Washington County, Utah, by Welsh and Reveal
(1977) but the Palmer specimen (no. 27) cited came from Trumbull,
Arizona.

FLOWERING: March-—June.

HOLOTYPE: Long Valley, Mohave County, Arizona, 27 Mar 1858,
Newberry s.n. (GH)).

OTHER SPECIMENS STUDIED: Arizona. Coconino Co.: between Ash Fork and Peach
Springs, 12 May 1931, McKelvey 2170 (cu); 1 mi. SE of Dinosaur, between Peach
Springs and Seligman, 8 May 1967, Rollins 67101 (cH). Mohave Co.: Peach Springs,
June 1884, Lemmon 3246 (Gu, US); same locality 26 May 1884, M. E. Jones 69 (cH);
Arizona Strip District, T36N, R7W, S16, 7 Jun 1979, Coombs & Bundy 2815 (Bry); 1
mi. E of Truxton, 7 May 1967, Rollins 6796 (cH); Trumbull, 60 mi. S$ of St. George,
Utah, 1887, E. Palmer 27 (Gu, ny 5 sheets, vs); 1.6 mi. E of Mt. Trumbull, 25 May
1979, N. H. Holmgren et al. 9169 (cH); ca. 3 mi. N of Mt. Trumbull, 20 May 1973,
Spellenberg et al. 3191 (cu). Yavapai Co.: 30 mi. NW of Wickenburg, 9 Jun 1968,
Barclay & Lockwood 3009 (cn).

lb. T. ambigua var. erecta Rollins, var. nov.

Herba biennis glabra, foliis integris, petalis purpureis vel albidis,
pedicellis adcendentibus vel divaricatis, siliquis erectis.

DISTRIBUTION: southern Utah and northern Arizona.

FLOWERING: April—May.

HOLOTYPE: Utah. Kane Co.: Kanab, 11 May 1941, A. Eastwood &
J. T. Howell 9300 (cu). Isotype (us).

OTHER SPECIMENS STUDIED: Arizona. Mohave Co.: clay hillside, pifion-juniper com-
munity, ca. 18 mi. W of Fredonia, 27 May 1968, Higgins 1370 (GH); chinle formation,
ca. 5 mi. N of Moccasin on Coral Pink Sand Dunes road, 19 May 1972, Atwood &
Higgins 3938 (pry). Utah. Kane Co.: ca. 12 airline mi. NW of Fredonia, Arizona, 4
May 1970, Welsh & Atwood 9706 (sry).

The conspicuous flowers of Thelypodiopsis ambigua are not exactly
regular. Two petals are oriented on the upper side and two on the
lower side of the nearly horizontal flower when it is at full anthesis.
The anthers of the paired stamens protrude slightly from the center
of the flower and are usually exposed to insects entering the flowers
because the petal blades are either widely spreading or somewhat re-
flexed. The dense elongated racemes of deep purple flowers make
T. ambigua a conspicuous member of the juniper woodland flora in
places where it occurs in abundance.

Typical Thelypodiopsis ambigua has a single, rather thick stem that
is densely covered with overlapping thickish oblong leaves. Branching
occurs well above the base. The lower leaves are erect, well-devel-
oped, and tend to persist at least through the early flowering period.
In these respects, this species differs from its close relatives, T. juni-

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) qa

perorum and T. elegans, both of which shed the lowermost leaves by
the time flowering begins. These species usually branch beginning at
the base (this is more consistent in T. elegans than in T. juniperorum).
They have more ovate cauline leaves than is the case with T. ambigua
and these tend to be more remote from each other on the stem.

2. Thelypodiopsis juniperorum (Payson) Rydberg, based on Sisym-
brium juniperorum Payson, Univ. Wyo. Publ. Sci. 1: 12, 1923. The-
lypodiopsis juniperorum (Payson) Rydberg, Fl. Rocky Mts. ed. 2:
1123, 1923. Thelypodiopsis elegans (Jones) Rydb. var. juniperorum
(Pays.) Harrington, Man. Pl. Colorado 280, 1954.

al; 1 d
above base, pilose with simple flattened trichomes at base and at nodes, 3- m.
tall; basal leaves entire to irregularly dentate, oblanceolate, obtuse, narrowed to a
winged petiole, 5-15 cm. long; cauline leaves auriculate, clasping the stem, lower
oblong and obtuse, upper becoming elongated and lax in fruit; inflorescense termi-

. long, : :
exserted, single and paired subequal; anthers erect, curved, 3-4 mm. long; pedicels
widely spreading to divaricately ascending, glabrous to sparsely pilose, 1-2 cm. long;
siliques ascending to nearly erect, stipitate, terete, torulose, glabrous, 5-9 cm. long,
2. n m. :
2-3 mm. long; stigmas shallowly lobed, lobes over replum margin; seeds oblong,
plump, not winged; cotyledons oblique to nearly accumbent.

DISTRIBUTION: west central Colorado.

FLOWERING: May-June.

HOLOTYPE: 6 mi. E of Montrose, Montrose County, Colorado, 15
Jun 1915, E. B. Payson 688 (RM!). Isotype (GH!).

OTHER SPECIMENS STUDIED: Colorado. Gunnison Co.: near Gunnison Canyon, 31 May
1913, Payson 97 (GH). Montrose Co.: 5.4 mi. N of turnoff to Black Canyon, off U.S.
H 1979. R. C. & K. W. Rollins 7994 (cu); 1.3 mi. N of turnoff to

Black C 28 May 1979, R. C. .W. Rollins H); 2.2 mi. on ;
wi a gy dae tg ccs 4 Jul 1969, I. & M. Al-Shehbaz 6902 (cH);

4 mi. E of Cerro Summit, 27 May 1979, R. C. & K. W. Rollins 7975 (GH).

inflorescence is much less dense than in T. ambigua and the presence
of simple flat whitish trichomes toward the base of the stems in T.
juniperorum will consistently separate it from that species.

Evidently Thelypodiopsis juniperorum is more local than most spe-
cies of the genus. As I interpret it, the species includes only those
populations with relatively large purple flowers and with long slender
gynophores supporting the fruits. Thus delimited, some populations
with a mixture of sessile fruits and short stoutish-stiped fruits referred
by some to T. juniperorum fall within the rather polymorphic 7, ele-
gans. This type of material probably misled Harrington (1954) into
believing that T. juniperorum is only a variety of T. elegans.

78 REED C. ROLLINS

3. Thelypodiopsis aurea (Eastwood) Rydberg, based on Thelypodium
aureum Eastwood, Zoe 2: 227, 1891. Thelypodiopsis aurea (Eastwood)
Rydb., Bull. Torr. Bot. Club 34: 432, 1907. Sisymbrium aureum
(Eastwood) Payson, Univ. Wyo. Publ. Sci. 1: 13, 1922.
Biennial or short-lived perennial, glabrous or sparsely pubescent toward base
near

stems with simple trichomes; stems erect, single or branching
above, 2-5 dm

, ca.
wide; petals lingulate or spatulate, tapering gradually to a broad claw or with a slightly
enlarged blade, yellow, 7-10 mm. long: s

and paired stamens subequal; anthers curved when dry; fruiting pedicels divaricately
ascending, straight, 5-10 mm. long; siliques erect to slightly divaricate, slender, nearly
straight, stipitate, 5-7 cm. long, somewhat torulose: stipe slender, 2-5 mm. long;
styles ca. 2 mm. long; stigma two-lobed to nearly entire; immature seeds oblong,
pom, wingless, ca. 2 mm. long, ca. 1 mm. wide. n = 11 (Rollins and Riidenberg,

DISTRIBUTION: southwestern Colorado, northwestern New Mexico
and southeastern Utah

FLOWERING: April—May.

HOLOTYPE: Durango, Colorado, Eastwood (cas, not seen).

OTHER SPECIMENS STUDIED: Colorado. Montezuma Co.: Mancos, Jun 1892, Eastwood
8.n. (GH, NY, Us); SW Colorado, 1875, Brandegee 1120 (cu); 1 mi. E of Utah state line,
23 Apr 1971, Rollins & Stafleu 7148 (cu); hills W of Mancos, 21 May 1943, Ripley
: —— 5352 (Ny); McElmo Creek valley, 20.6 mi. S$ of Cortez, 15 May 1981, R

* x N Fs

Ps ah,
mi. from Colorado line, 23 Apr 1971, Rollins & Stafleu 7146 (GH); road cut, Utah state
route 262, 5-1/2 mi. E of Aneth, 23 Apr 1971, Rollins & Stafleu 7147 (cu); 10 mi.

933, Harrison 5937 (Bry); N end of Montezuma Creek, 6
mi. E of U.S. - 47, 8 May 1969, Welsh et al. 8914 (sry); 15 mi. S of Blanding
on road to Bluff, 13 May 1970, Atwood 2429 (BRY); 6 mi. above Bluff, 29 Apr 1961,
Shumway 61 (BRY); 8 mi. W of Blanding, 19 May 1955, Barneby 12766 (ny).
h it is one of the poorly collected species of Thelypodiopsis,
T. aurea is not uncommon in the area where it occurs as Dr. Frans
Stafleu and I found in 1971 when we collected specimens from four
separated populations in three different states. This was in the four
corners area of Arizona, Colorado, New Mexico, and Utah. There we
observed many populations additional to those we sampled, all in the
valley of the San Juan River. In its habit of growth, T. aurea is some-
what similar to T. ambigua, often having a single stout stem that arises
from a relatively heavy root. So

base but in most, branching occurs above the middle of the stem. It

“
= B
i)
”
ce)
na
po
3
Qu
Ss
co)
a
2
of
a
on
=
*-

some plants persist beyond the second year.
Specimens of Thelypodiopsis aurea are either completely glabrous
or there may be simple spreading trichomes at the base of the stem.

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 79

On a given sheet, both glabrous and sparsely hirsute individuals may
be found, showing that presence or absence of trichomes is not a
characteristic that can be used for taxonomic purposes in this species.
There is also some variation in the petal shape. This ranges from spat-
ulate to broadly oblong.

4. Thelypodiopsis elegans (Jones) Rydberg, based on Thelypodium
elegans M. E. Jones, Zoe 4: 265, 1893. Thelypodiopsis elegans (Jones)
Rydb., Bull. Torr. Bot. Club 34: 432, 1907. Streptanthus wyomin-
gensis A. Nelson, Bull. Torr. Bot. Club 26: 126, 1899. T helypodiopsis
wyomingensis (A. Nels.) Rydb., loc. cit. Thelypodium bakerii Greene,
Pl. Baker. 3: 8, 1901. Thelypodiopsis bakerii (Greene) Rydb., loc. cit.

os, fi e, branched above, sparsely to
ensely pilose toward base or sometimes pubescent throughout, 3-10 d tall; basal
de

9-6 cm. long, 1-1.5 cm.
acute to obtuse, entire or rarely dentate,

usually glabrous, rarely pilose on the lower surface; in
ny in fruit; sepals oblong, purp ish to pale lavender or nearly white, non-saccate,
varicately ascending to erect, 5-7 mm. long; petals spatulate, pale purple to white
unguiculate, 10-14 mm. long; pedicels slender, divaricate, glabrous or less frequently
sparsely pilose, 7-10 mm. long, expanded at summit; i
Al ore 1 mm. or less long, terete, often torulose, stral
varicately ascending to more widely spreading, glabrous
oy densely so, 4-9 cm. long, ca. 1.5 mm. in diameter; styles 1-3 mm. long, often
: vate; stigmas 2-lobed, grooves shallow to prominent; seeds wingless, angular, 0D-
long, ca. 1 mm. wide, less than 1.5 mm. long; radicle exceeding cotyledons; cotyledons
incumbent to obliquely so.
DISTRIBUTION: western Colorado,
FLOWERING: May-June.
HOLOTYPE; Westwater, Grand County, Utah, 6 May 1891, M. E.
Jones s.n. (poM!). Not 7 May 1891 as given in the protologue (cf. Rol-

lins, 1971).

AINE SPECIMENS: Colorado. Delta Co.: 2 mi. W of Delta, 25 May 1938,
ollins 2137 (cu, us). Garfield Co.: near Clenwood Springs, 13 Jun 1957, Rollins 57150
— mi. W of Rifle, May 1938, Rollins 2202 (GH). Gunnison Co.: 3 mi. 0
“Leapehae 23 May 1938, Rollins 2110 (cu). Mesa Co.: © U.S. Interstate 70, 5.7 mi.
sige Utah border, 29 May 1979, R. C. & K- W. Rollins 79102 (cH); 10 mi. N of Mesa,

= lS, pera 2195 (cH). Moffat Co.: Yampa eh a a a

r National M May 1948, Porter 4480 (GH); mi. ayae)

Ma onument, 16 May 1948, Tire Craig, 22 May 1979, Rollins 7937

f Th

Utah, southwestern Wyoming.

m

. y 1948, Harrington 3888 (GH); 4 of , 22 May '
ire Montrose Co.: Cimarron, 6 Jun 1901 Baker 32 (GH; us, isotypes of Thelypodium

N erti); 1.7 mi f Montrose, 28 Ma 1979, Rollins 7952 (G
aturita, 4 May 1914, Payson 272 (cH). Routt Co.: 5 mi. W of Steamboat Springs,
eM 8, Harrington 0. chesne Co.: 3.8 mi. E of Duchesne,
- May 1979, R. C. & K. W. Rollins 79115 (cu). Grand C mi. N of Westwater,
ag 1979, R. C. & K. W. Raollins 79105 (GH); off Deen co if - 3 o
exit, 29 M " W. Rollins 79106 (GH) intah Co.: bluff above
2 m9 RK. C. & K. W. Rollins 79121 (cH);

1 mi. W of Rai 1979, Et al. 1803 (GH); 2.5 mi. SW of
: nbow, 4 1965, N. H. Holmgren et al. (cH); : :
Vernal, 30 May 1979, eh > K. W. Rollins 79118 (GH). Wyoming. Carbon Co.: 16

80 REED C. ROLLINS

mi. N of Rawlins, 21 May 1979, Reed C. and Richard C. Rollins 7927 (cu); 8 mi. N
of Baggs, 22 May 1979, Reed C. & Richard C. Rollins 7933 (GH). Sweetwater Co.:
Green River, 30 May 1897, Nelson 3034 (GH, isotype of Streptanthus wyomingensis);
same locality, 13 Jun 1936, 28 May 1938, Rollins 1177, 2217 (GH); 5 mi. N of Henry’s
Fork, 29 Jun 1951, Rollins & Porter 5140 (cu).

The variation in various features of Thelypodiopsis elegans is frus-
trating because certain facies of the species are recognizable and show
geographical integrity. However, to pin these down with consistent
characters is elusive. Certain trends can be recognized. For example,
there is increased pilosity on the leaves, stems, pedicels, sepals, and
siliques as material from northern Colorado into southern Wyoming
is examined. Some populations from southwestern Wyoming, the type
locality for what has been called Thelypodiopsis wyomingensis, tend
to have torulose siliques and the styles are strongly expanded at the
tip in many plants. But these features are not consistently correlated
even in the same geographical area. Other characters, such as the
length of the gynophore, if it is present, appear to be erratic. In a
single population (Rollins 7933 from 4 miles east of Craig, Colorado),
the siliques of a given plant may be virtually sessile or in different
plants the gynophores may range up to at least 1 mm. long. The
lengths of the siliques vary considerably.

Thelypodiopsis bakerii was founded on plants with relatively shorter
fruits than are present in typical T. elegans but several populations
in the area where the type of T. bakerii was collected have plants
with siliques that range well within the shortest limit of typical T.
elegans. The siliques of plants in the “T. bakerii” area do average
shorter than those in the type area of T. elegans. But this is not suf-
ficiently definitive to provide the basis for the recognition of a sepa-
rate taxon. Over most of the area where T. elegans occurs, the basal
leaves and lower cauline leaves are entire or nearly so. There are
occasional plants with sparsely dentate leaves but this is not the rule.
However, from northern Colorado and southern Wyoming come spec-
imens with lower leaves that are markedly dentate. This feature is
correlated to some extent with increased pilosity. The possibility of
recognizing nomenclaturally an infraspecific taxon based on these cor-
related features was recognized. However, specimens with transitional
features were too numerous to permit clear distinctions to be made.

After studying 13 new collections especially obtained from popula-
tions where field observations were made at the time the material was
taken, I have reluctantly come to agree with Payson (1922). In refer-
ring to this species, he stated that “to separate these forms is at pres-
ent impossible since they seem to show no correlation to one another
nor to geographical areas.” I feel that Payson’s statement is as appro-
priate now as it was when he made it. The 13 collections referred to
were from different localities nearly throughout the range of Thely-

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 8]

podiopsis elegans, including the type areas of ‘ ‘Thelypodium bakerii,”
Thelypodiopsis elegans, and “Streptanthus wyomingensis.

5. Thelypodiopsis vermicularis (Welsh & Reveal) Rollins, comb. nov.,
based on Thelypodium sagittatum (Nutt.) Endl. var. vermicularis Welsh
& Reveal, Great Basin Nat. 37: 357, 1977.

Annual or biennial, glabrous throughout or rarely with a few trichomes near base;
stems one to several from the base, branched above, 2—6 dm. tall; basal leaves few,
quickly shed, broadly oblanceolate, obtuse, entire or with a few small remote teeth,

ti

considerably in » int pay loos ick “ascending [ ae eading Hae aor oblong,
white to purplish or greenish, non-saccate, 4.5-5.5 mm. oc 1.2-1.5 mm. wide; pet-
als ane white, tapering pradually from blade to broad claw, 9-11 mm. long, 3-4

wide; stamens erect; anthers straight, 2-3 mm. long; pedicels widely spreading
to “divans: ascending, slender, 5—8(-10) mm. long, abruptly expanded at — si-
iques erect or ascending, terete, torulose, tortuose, 2-4 cm. long, 1.2-1.5
iameter, sessile or with a short gynophore up to 1.5 mm. lon ng; styles clini, ce -
mm. sis stigmas equalling but not ee styles in diameter; seeds narrowly ob-
ong, wingless, p ee ca. 2 mm. long, ca. 1 mm. wide; cotyledons acitabent ie
Gecbuing cotyledon

DISTRIBUTION: Utah and eastern Nevada.

FLOWERING: April-June.

HOLOTYPE: Arapien shale formation, greasewood community, ca. 4
mi. SE of Sigurd, Sievier Co., Utah, 29 May 1972, Welsh & Atwood
11718 (BRY!).

OTHER SPECIMENS STUDIED: Nevada. White a Co.: 3.9 km. N of U.S. Hwy. 50 on
road to Long ae ns Jun 1979, N. H. & P. K. Holmgren 9373 (GH); along road to
Duckwater, 50. 3 mi. from U.S. Hwy. 50, 26 May 1979, Williams 79- - " (BRY, GH).
Utah. Box Elder Co. = rouse Creek Valley on the Etna Road, south of Etna, 27 May
1973, sie 1355 (sn) Grouse Creek Rd., 11.5 mi. N of Utah Hwy. 30, 27 May
1971, or Holmgren 4862 (GH). Iron Co.: Escalante Desert near Lund, 8
May 196. Ripley : glia 4373 (GH). Juab Co.: Dugway Valley NW of Delta, 29
May 1981, Tiehm 6516 (cu); E of Fish Springs Range on a Sarcobatus and Atriplex
flat near Cane Springs, 9 May 1968, Reveal & Thomas 984 (BRY, GH). Millard Co.: 6

mi. W of Hi oeiey: 9 May 1935, Harrison “6301 (Bry); Wah Wah Valley, 27 May 1981,
Tiehm 6513 (cH); Black Rock Desert, 1 mi. N of Borden so os = oe Pos

S of Black oil Station, 4 May 1968, olmgren

Sng ect aa es ae wy. 6, 9 May 1979, Atwood 7142
(BRY). Salt Lake ea. Salt Lake, June ’ 1869, Watson 108 (cn); alkali area on way to Salt
Lake, 12 Jun 1933, Eastwood & Howell 376 (GH). Sanpete Co.: ca. 2 mi. E of Aurora,
5 Jun 1979, R. C. & K. W. Rollins 79161 (cu); 1 mi. W of Gunnison, 23 Apr 1979,
Neese & Williams 7140 (sry); South Valley allotment, T20S, R1W, $20, 25 May 1979,
Greenwood s.n. (BRY); 15 mi. SSW of Levan, 24 Apr 1979, hang et ate (BRY).
Sevi ide of ee un 1938, apap! s.n. (BRY). Tooele :
fon 36. ea : se = potich 6 Jun 1 ; 970, N. H. & P. K. Holmgren 4178 (cu). ae
Co.: E side of Coyote Pass, 1.2 mi. W of Utah Hwy. 68, 15 May 1968, Weston
ec T. el d

Thelypodiopsis vermicularis is very closely related to T. elegans an
a number of the specimens of it in herbaria have been so determined
in the past. The complex variability of T. elegans obscured and made
difficult the recognition of the consistent pattern of differences that

82 REED C. ROLLINS

characterized T. vermicularis. Only after a number of specimens had
been accumulated was it possible to be assured that such a pattern
existed. Some of the oldest specimens were determined as Thelypo-
dium sagittatum and as indicated above, var. vermicularis was de-
scribed under that species. That S. vermicularis could have been con-
sidered to be varietal in a species of another genus emphasizes the
closeness of the genera Thelypodium and Thelypodiopsis.

The vermiform siliques and cylindrical styles are the most distinc-
tive features of Thelypodiopis vermicularis, but the dense capitate in-
florescences ultimately elongating into relatively long infructescences
with numerous siliques tend to distinguish this species from T. elegans
which has fewer flowers and fruits. In most of the specimens of T.
vermicularis seen, the cauline leaves are broader, shorter, and more
obtuse than those of T. elegans. However, there are enough excep-
tions to make leaf-shape unreliable as a distinguishing feature.

Geographically, Thelypodiopsis vermicularis occurs west of the area
occupied by T. elegans. It is found mostly in the western half of Utah.
I have seen only two collections from eastern Nevada.

6. Thelypodiopsis divaricata (Roll.) Welsh & Reveal, based on Cau-

lanthus divaricatus Rollins, Contrib. Gray Herb. 201: 8, 1971. The-

lypodiopsis divaricata (Roll.) Welsh & Reveal, Great Basin Nat. 37:
1977.

>

o
g
Pa
os
$0
8
a
®
Qu.
i)
3 5
<
oO
°
5
ces
ae
<
go.
3
7
oO
+e
S
7)
ce
<<
6
=|
>
o
7)
io)
g
5
—
9
im
iv)
°o
5
+
o
ot
o
a
fo
3
Qu

u-
e, elongating in fruit; sepals ae
ish, erect, forming an urn-shape calyx, wie pubescent, 3.5-5 mm
ellow with a whitish claw, 7-10 mm long; claw bro: yearn ca. 2 mm. wide; blade
oblong, folded and crisped, 3-5 mm. long, ca. 1 mm. wide; ovary sparsely pubescent;
fruiting pedicels divaricately ascending to more widely s sh eading, net sparsely
pubescent to glabrous, slender, 7-12 m m bone siliques terete, straight, nearly erect
” divaricately ascending or more wide ly spreading, sparsely sole eit oe glabrous,
cm. Gas 1-1.5 mm. in diameter, i or with a oa ophore up to 1 mm.
sited styles subclavate, 1-2 mm. long; stigmas slightly bilobed Arg the lobes over the
replum amyaee eae oblong, wingless, 1.5-2 mm. lon g, ca. 1 mm. wide; cotyledons
incumbent 1 (Rollins oe Riidenberg, 1977).
Hot AC, eastern Utah: Carbon, Emery, Garfield, Grand, San
Juan, and Wayne counties.
FLOWERING: April—June.

HOLOTYPE: Utah, San Juan County, about 75 miles west of Blanding
and 10 miles east of Hite, T34S, R14E, 16 May 1961, Arthur Cron-
quist 9033 (GH)).

sebseeeat SPECIMENS: Utah. Emery Co.: 10 mi. E of Huntin on, 30 Apr
Higgins eal 1256 (cn); Hrateinogy 2 Jun 1944, as ed irs 4735 ic Pe
U.S. os 48 mi. SE of Price, 29 May 1979, R. C. & K. W. Rollins 79108 es

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 83

off U.S. Hwy. 6, 28.8 mi. SE of Price, 29 May 1979, R. C. & K. W. Rollins 7911
(cu); ca. 3 mi. ESE of Castledale, 20 May 1979, Welsh & Neese 18379 (BRY); 13 mi.
NNW from Green River, 29 Jun 1977, Welsh & Taylor 15217 (Bry); W base of As-
sembly Hall Peak, San Rafael Swell, 10 June 1979, J. d> M. Harris 395 (Bry); San Rafael
Swell, 6.5 mi. SE of Ferron, 20 May 1979, N. H. Holmgren et al. 9103 (cu). Grand
Westwater. 6 May 1891, M. E. Jones s.n. (GH); 3 mi. E of Caineville, 5 May

—

un , M. E. Jones s.n. (GH);
9.9 mi. E of Green River, 29 May 1979, R. C. & K. W. Rollins 79107 (cu). Garfield
Co.: SE of Mt. Hillers, junction between Starr Spring road and Utah Hwy. 276, 3
May 1976, Neese 1802 (Bry); Cave Flat road, W side of Henry Mts., 15 May 1970,
Welsh 9815 (Bry); crossing, Bull Frog Creek, 5 mi. W of turnoff from Utah Hwy. 276,
29 Apr 1977, Neese ¢ White 2736 (BRY). Wayne Co.: 5 mi. W of Hanksville, 3 May
1973, Albee 1286 (cu); 1 mi. S of junction Utah Hwy. 24 and Notom Road, 7 Apr
1968, Atwood 1255 (cu); ca. 29 mi. SW of Hanksville, 29 Apr 1961, 8934
(cu); summit of North Cainville Mesa, 18 May 1976, Welsh et al. 13098 (BRY); Flint
Trail, Maze Overlook road, 19 Apr 1979, Welsh 9600 (BRY).

Field work during the spring of 1979 gave me an opportunity to
study populations of Thelypodiopsis divaricata. Near Green River,
Utah, we found a population mostly in fruit but with one flowering
individual having sterile siliques that appeared to be a hybrid between
T. divaricata and T. elegans. Both species are known to occur in this
area although we did not find a truly mixed population of the two.
The sterile plant had flowers with petals somewhat similar to those
of T. elegans while the sepals were erect and formed an urn-like calyx
similar to that of T. divaricata. The putative hybrid is ims a
by R. C. & K. W. Rollins 79107a (GH). =

Although Thelypodiopsis divaricata has a caulanthoid” flower, i.e.,
an urn-shaped connivent calyx and petals with a broader pela
blade with a stricture between the blade and claw and a crispe
blade, other characters are so similar to T. elegans that it seems best
to follow Welsh and Reveal (1977) in removing this species from Cau-
lanthus. Indeed, Jones made mixed collections of these two species
and for some time specimens of T. divaricata were identified as 7.
elegans (Rollins, 1971). The evidence cited above, suggesting inter-
specific hybridization, further supports the placement of this species
in Thelypodiopsis.

7. Thelypodiopsis wootonii (Robins. ) Rollins, based on Sisymbrium

wootonii Robinson, Bot. Gaz. 30: 59 1900. Thelypodiopsis wootonu

(Robins.) Rollins, Contrib. Gray Herb. 206: 12, 1976. . =
Biennial or perennial, glabrous throughout, branched, up to 6 dm. tall; stems leafy,

oa rosette leaves formed a season before flowering occurs not

ODIO;

; oa
(2-)3.5-4.5 mm. wide: paired stamens erect, single stamens curved upward a -~
filaments slightly oetaund a terete; anthers 1.2-1.5 mm. long; stigma bilobed wi

84 REED C. ROLLINS
lobes over replum margin; fruiting pedicels slender, divaricately ascending to more
d, 1.5-2 cm

orulose, narrowed below and above, nerved from base to apex, 3-5 cm. , ca.
; i n 0.5-1 mm. long; stigmas persistent, distinctly lobed,
sharply expanded over the styles; seeds plump, broadly oblong, 1.2-1.5 mm. long, ca.
mm. in diameter; cotyledons incumbent.

KEY TO THE VARIETIES

Petals obovate, 8-10 mm. long, 3.5-4.5 mm. wide .............. 7a.var. wootonii.
Petals spatulate, ca. 6 mm. long, ca. 2 mm. wide ............ 7b. var. parviflora.

7a. T. wootonii (Robins.) Rollins var. wootonii.

DISTRIBUTION: known only from the type collection.

FLOWERING: July—August.

HOLOTYPE: Mexico, state of Chihuahua, 16 km. southeast of Colonia
Garcia, Sierra Madre, 7600 ft., 27 Jul 1899, C. H. T. Townsend ¢> C.
M. Barber 176 (cu!). Isotypes (GH!, US).

7b. T. wootonii var. parviflora Rollins, var. nov.

Herba perennis, floribus parvis, sepalis albis ca. 4 mm. longis, petalis spathulatis
albis ca. 6 mm. longis, siliquis teretibus, 3-5 cm. longis.

DISTRIBUTION: Mpio. de Bocoyna, Chihuahua, Mexico.

FLOWERING: August-November.

HOLOTYPE: Mexico, state of Chihuahua, Gonogochic, E of Creel,
Mpio. de Bocoyna, open flat of pine-oak forest, ca. 7,400 ft., 16 Oct.
1975, Robert A. Bye, Jr. 7128 (cu).

OTHER SPECIMENS STUDIED: Mexico. Chihuahua. Mpio. de Bocoyna: W of Gonogochic,
13 Aug 1977, Bye 7994 (cH): E of Gonogochic, 12 Aug 1977, Bye 7935 (GH); same
locality, different dates, Bye 4287, 4775, 5479 (cn).

The difference in flower size between the collections of Dr. Bye

, but at the same time it serves to point up the dif-
ferences between the two presently known populations without ob-
scuring the fact that they are undoubtedly very closely related.

8. Thelypodiopsis alpina (Standl. & Steyerm. )

manschulzia alpina Standley & Steyermark, Fie
377, 1946. Thelypodiopsis alpina (

Rollins, based on Ro-
Idiana, Bot. 24, pt. 4:
Standl. & Steyerm.) Rollins, Con-

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 85

trib. Gray Herb. 206: 12-13, 1976. Sisymbrium standleyi Rollins,
Rhodora 58: 156, 1956, based on Romanschulzia alpina, not Sisym-
brium alpinum Fourn., Recherch. Crucif. 131, 1865.

Annual or possibly biennial, glabrous throughout or with a few simple trichomes on
the margins of the lower petiole; stems one or few, simple to sparingly branched, 3-7
dm. tall; strictly basal leaves absent; lowest cauline leaves petiolate, narrowly lanceo-
late; cauline leaves scarcely reduced upward, remote, glaucous, lanceolate, acuminate,
auriculate, clasping the stem, remotely denticulate, 3-7 cm. long, 6-12 mm. wide;
inflorescences lax, few-flowered; sepals erect, oblong, purplish, scarious-margined,
3-3.5 mm. long; petals purplish, lingulate, narrowed near insertion, 5-6 mm. long;

m. or less long; fruiting pedicels divaricately ascending,
slender, 7-12 mm. long; siliques erect, nearly terete to slightly compressed contrary
to septum, slightly torulose, subsessile, 4—5.5 cm. long, wide; styles 1

. or less long; stigma larger in diameter than style, the lobes ap-
parently over the valves; immature seeds oblong, wingle . long.

DISTRIBUTION: Guatemala, Departments of Huehuetenango and
Quezaltenango.

FLOWERING: July—August.

HOLOTYPE: Guatemala, Dept. of Huehuetenango, between Tojquid
and Caxin bluff, summit of Sierra de Los Cuchumatanes, 6 Aug. 1942,
Julian A. Steyermark 50144 (F!).

OTHER SPECIMENS STUDIED: Guatemala. Huehuetenango: near Tunima, Sierra de los
Cuchumatanes, 6 Jul 1942, Steyermark 48923 (F). Quezaltenango: Volcdn Santa Maria,
27 Jul 1934, Skutch 864 (cH).

This species is imperfectly known. Only three collections have been
available for study. None of these have fully mature siliques. The
abrupt transition from the lowermost two or three leaves that are
petiolate to the sessile and fully auriculate leaves only three or four
leaves up from the base is very unusual. On the specimens seen there
is no suggestion of a basal rosette. Usually in the Cruciferae, if there
are dimorphic leaves, the petiolate type is characteristic of the basal
rosette or are at least clustered at the base of the stem where the
internodes between the leaves are extremely short. In the case of
Thelypodiopsis alpina, where all of the leaves are truly cauline, the
internodes are one to several centimeters long even in the lowermost
leaves which are petiolate. The petiole is very slender. The third or
fourth leaf up the stem is abruptly sessile and auriculate. The situa-
tion in T. shinnersii is not far different from this except that the pet-
ioles are winged with small auricles at the base and the lower leaves

are broadly obovate.

9. Thelypodiopsis shinnersii (M. C. Johnst.) Rollins, based on Sis-
abe: seman M. C. Johnston, Southw. Nat. 2: 129, 1957, which
in turn was based on Thelypodium vaseyi Coulter, Contrib. U.S. Nat.
Herb. 1: 30, 1890, not Sisymbrium vaseyi S. Watson ex Robinson in
Gray, Synop. Fl. N. Amer. 1: 138, 1895. Thelypodiopsis shinnersii
(M. C. Johnston) Rollins, Contrib. Gray Herb. 206: 13, 1976.

Ca. 1, :
aintly bilobed,
ss; ca, F.2

86 REED C. ROLLINS

glaucous, obovate, 5-10 cm. long, up to 3 cm. wide, with a prominent broad central
i t

i . long; te
pedicels divaricately ascending to more widely spreading, straight, 6-12 mm. long;
siliques terete, divaricately ascending, straight to slightly in-curved, subsessile, 4-6
em. long, c . wide, not torulose; styles less than 1 mm. long; stigmas nearly
entire, slightly larger in diameter than style; seeds plump, oblong, wingless, less than
1 mm. long; cotyledons incumbent or obliquely so.

DISTRIBUTION: Mexico, State of Tamaulipas, and the lower Rio
Grande Valley of Texas, USA.

FLOWERING: July—April.

HOLOTYPE: near Rio Grande City, Texas, 1889, G. C. Neally 188
(us!). Tracing of holotype in (GH)).

OTHER SPECIMENS STUDIED: Mexico. Tamaulipas: region of Rancho Las Yucas, ca. 40

NW of Ald

cality, 16 Oct 1957, Dressler 2427 (GH); 9 mi. S of Ciudad Victoria, 9 Feb 1961,
Pegi :

(GH, TEX-LL); near San Benito, 20 Apr 1959, Rollins & Correll 5954 (cu, TEX-LL); along
Arroyo Colorado, Harlingen brushlands, 13 Mar 1964, Correll 28953 (GH).

Recent acquisitions from Mexico and Cameron County, Texas, have
helped to clarify this species which was known from only two or three
collections for many years. The flowers are very small and although
the infructescences are considerably elongated, reaching three deci-
meters or more in some plants, the dense little cluster of flowers is
always at the very top of the inflorescence. This is because the floral
parts are shed very quickly after anthesis and the developing young
siliques just below the pollinated flowers are devoid of the remnants
of any flowers.

The nomenclatural situation is fairly complicated but without am-
biguities. This has been discussed previously (Rollins, 1960)

10. Thelypodiopsis byeii Rollins, sp. nov.

Annual; stems single, erect, branched beginning near base or above, 2-5 dm. tall,
lower hirsute with simple spreading trichomes, becomin
fe

: est broadly obovate
to nearly orbicular, shallowly crenate repan y entire, middle and upper
leaves lobed to sharply incised and irregularly dentate, middle | ves largest, 4-10
cm. ; wide, ovate to oadly oblong, apex rounded or tuse, the upper
acute, petioles slender; inflorescences becoming a Oose raceme as development pro-
ceeds; ring pedicels slender, divaricately ascending, glabrous, longest up to 1 cm.
long; buds purplish; sepals oblong, glabrous, purplish, 3.5—4 m. long,

rous, ; ca. 2 mm.
wide, inner pair non-saccate, outer pair slightly saccate, petals purplish to nearly

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 87

white, obovate to broadly spatulate, gradually narrowing to point of insertion, 7-9
mm. long, 4—5 mm. wide; stamens erect, included, filaments slender, straight, anthers

or with a shor re less t m
long; stigmas slightly lobed, lobes over the replum; seeds numerous,
mp wi i

ulibus erectis, ramosis, 2-5 dm. altis; inferne pilosis; superne gla-
bris; foliis petiolatis, inferne sparse pilosis, repandis vel lobatis, obtusis vel acuminatis,
4-10 em. longis, 2-4 cm. latis; pedicellis anguste ca. 1 cm. longis patentibus; sepalis
oblongis, purpureis, glabris, 3.5—4.5 mm. longis, ca. 2 mm. latis; petalis purpureis vel
albis, obovatis vel late spathulatis, 7-9 mm. longis, 4-5 mm. latis; infructescentiis laxis;
siliquis immaturis, teretibus, patentibus, 3-5 cm. longis.

DISTRIBUTION: western Chihuahua, Mexico.

FLOWERING: February—April.

HOLOTYPE: in the Gray Herbarium, collected along Arroyo Wimivo
(= Arroyo Samachique) between Wimivo and Rio Batopilas on N side
of Barranca de Batopilas, Sierra Madre Occidentale, Mpio. Batopilas,
Chihuahua, Mexico, 28 Feb 1973, Robert A. Bye 3433.

OTHER SPECIMENS STUDIED: Mexico. Mpio. Batopilas: Sierra Madre Occidentale, Bar-
ranca de Batopilas, in moist shaded rock area with large oulders, near seepage area,
above burro trail along Arroyo Wimivo, 28 Feb 1973, Bye 3436 (GH); vicinity of Wim-
ivo, Feb 1973, Bye 9904 (cu); between La Bufa and Bacosiachi, Mar 1973, Bye 9902
(GH); same locality, Feb 1973, Bye 9903 (GH).

Thelypodiopsis byeii is unlike any other species of the genus in that
it has cauline leaves that are petiolate and very thin. These typically
have broad, lobed to dentate blades and very slender petioles. In tex-
ture of the leaves and slenderness of the siliques it simulates T. shin-
nersii to some extent. But that species has entire auriculate leaves and
siliques that are erect rather than widely spreading as in T. byeii. In
habit and general appearance, T. byeii resembles Dryopetalon runci-
natum Gray var. laxiflorum Roll. but the petals are absolutely entire
which precludes its being placed in the genus Dryopetalon. If the
petal character were to be ignored and the species placed in Dry-
opetalon, it would still be an undescribed species. However, if it were
placed in that genus, the entire genus Thelypodiopsis would have to
disappear into Dryopetalon, it being the older of the two generic
names. There is no justification for such an action.

Thelypodiopsis byeii grows in open areas, or in the shade of rocks
and boulders, in a short tree and thorn forest at about 1,000 meters
elevation. According to Dr. Bye, for whom the species is named, the
plants are cooked and eaten by the native Tarahumara people who

populate the area where it grows.

11. Thelypodiopsis arcuata (Roll.) Rollins, based on Sisymbrium ar-
cuatum Rollins, Rhodora 62: 58, 1960. T helypodiopsis arcuata Rollins,
Contrib. Gray Herb. 206: 14, 1976.

88 REED G ROLLINS

Annual, glabrous throughout; stems siNg|e from base, weak, usually branched at each
node, occasionally simple, 5-10 dm. tall; eaves all cauline, sessile, auriculate, entire,
sometimes with minute teeth, lower nearly pandurate, rounded above, gradually ovate
€ath, greenish above, 2-8 cm. long, 1-4 cm.
“gating into a lax elongated infructescence;

purplish, narrowly oblong, non-saccate, 3-4
: > spatulate with a slender claw, 5-6 mm. long,
ca. 2 mm. wide; filaments 3-4 mm. long. nearly equal on all six stamens; fruiting

e

wide; inflorescences few-flowered, elo
flower parts quickly shed; sepals erect,
mm. long, ca. 1 mm. wide; petals whi
pedicels widely spreading to gently recu
~ pendulous, slightly arched, 7-10 cm. long,
= styles 1-1.5 mm. long; stigmas unexpanded;
aL
ong.
vo Leén and San Luis Potosi,

exico.

FLOWERING: July—October.

HOLOTYPE: 1-2 miles SW of Paplillo. Nuevo pe wee ee
1958, D. S. Correll & 1. M. Johnston 19941 (cH).

OTHER SPECIMENS STUDIED: Mexico. Sa . ie
Punta Huerta, 87 Oct 1984. Reeddivakt Rows poe Sierra de Alverez, cerca del

Only the above ~~ cited collections of Thelypodiopsis arcuata are
known to ara and neither of these has mature seeds. Evidently the
plants of this species are quite Weak-stemmed and, according to the
collectors, tend to sprawl. In the type collection, the lowermost leaves

specimen from San Luis Potosi

> where new growth has started a
broken-off stem, the lowest le zs

fat @'Cuata are wing-petioled and probably

ies to those . s ee Whe influrescenos ‘of this epvies be:

flower ae a ig ; a f peers in the early shedding of
resulung INfructescence is

in the most general features. is very different except

yy aiid i alieg (Brandegee) Rollins, based on Thelypo-
cs Ped eps: randegee, Zoe 5, 939 1906. Sisymbrium purpusii
(Brandeg.) O. E. Schulz, Pflanzénr, §6 (IV, 105) 58, 1924. Thelypo-
diopsis purpusii (Brandeg.) Rolly. Candy Gia Henk. 206: 14

& Standley, Contrib. U.S. Nat.
‘ low mountains west of San Antonio,
oad ae Wooton 3847 (US, not seen; apparently
Stand.) 6 ¢ i. Hs e P- 141)]. Sisymbrium vernale (Woot. &
Se Ee ee, Oe ee ee Sisymbrium kearneyi Rollins
Leafl. West. Bot. 7: 15, 1953. :

Annual, ir,
. Pree tinagaaloe hig coon fruiting’ from base, branched, often purplish and
leaves oblanceolate in outline to oblong, Laeacl§ is Die Pps eane hao
ee ; e, with a stron
renee hh nid Tea nearly rosulate or with evident esau

? ? * wide; cauline leaves sessile, auriculate except

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 89

west one or two in some plants, reduced upward and becoming lanceolate with

cence lax, remaining few-flowered as the fruiting axis elongates; sepals oblong, non-
saccate, purplish or greenish, scarious-margined, 3.5-5 mm. long; petals white to light
lavender, spatulate narrowing to a slender base, 4-6 mm. long; paired stamens ene
g to

exserted; fruiting pedicels divaricately ascending to nearly erect, straight, 1-1

long; sere terete, narrowly linear, divaricate to ascending, sessile to Cpe 3-6
cm. long, ca. 1 mm. wide; styles ca. 1 mm. long; seeds oblong, wingless, plump, less
than 1 mm. gle cotyledons incumbent

DISTRIBUTION: Grand Canyon ‘of Arizona to New Mexico, Texas,
south to Coahuila, Mexico.

FLOWERING: February—May.

HOLOTYPE: Sierra de Parras, Coahuila, Mexico, March 1905, C. A.
Purpus 1392 (uc!). Isotype (GH).

THER SPECIMENS STUDIED: Mexico. Coahuila: Sierra Pata Galena, Feb 1905, Purpus
1329A (cH). USA. Arizona. Mohave Co.: along trail into $.B. Canyon, Grand Canyon
National Park, 8 May 1952, McClintock 481a caves of Sisymbrium GS sie
same locality and date, McClintock 53-481 (G (cu). New Mexico. Dona Ana Co.
end of the San Andres Mts., on NE ge of Black ner - May pee Spllenber
& Todsen 5497 (GH, NMC). Luna Co.: N side of T as Mts., abo
up the North Sister, S of Deming, . Apr 1973, Spelenberg "3002 (aay Otero Ge.
about 3.5 mi. due E of U.S. Hwy. 70 through Alamogordo, in Marble Canyon, 11
May 1980, R. & M. ieee 5501 (GH). Socorro she Sevilleta Grant, west slope

of Los Pinos Mts., directly E of Nunn-Burris Ranch he adquarters, 2 May ite Spel-
lenberg & Ward 5485 (cu, NMC). Texas. Hudspeth Co.: canyon, S en of Hueco Mts.,
18 Mar 1979, Worthington 4012 (GH); same tacality: 12 May 1979, Worthington 7474
(cu). Presidio Co.: McCormick Ranch, near Fresno Canyon, 1 Mar 1959, Johnston
& Warnock 3734, 3735, 3736 (GH); head of Fresno Canyon, Big Bend Ranch, 20 Apr
1961, Rollins ¢ Correll 61174 (cH).

The unfolding of the nature and distribution of Thelypodiopsis pur-
pusii has taken many decades. Only the original two collections of
Purpus from Mexico were known to me when Sisymbrium kearneyi
was described from the Grand Canyon of Arizona in 1953, and these
were the only specimens for comparison. At that time, the Grand
Canyon specimens were thought to be sufficiently distinct from the
Coahuila material of Purpus to represent a distinct taxon. Then came
specimens from Fresno Canyon of western Texas in 1959, later ma-
terial from southern New Mexico (1973), and more recently (1979)
specimens from the Hueco Mountains of extreme northwestern Texas.
In addition, specimens from the general area where Thelypodium ver-
nale was originally collected have permitted us to interpret the =P
plication of that name. As indicated above, the holotype of T. vernale
known isotypes so that the
ithout anyone knowing for sure to what

New Mexico, do fit the protologue of this name and they in turn are
very similar to the collections from the
These populations represent a minor deviant from the other materi

of Thelypodiopsis purpusii in that a definite rosulate cluster of leaves

90 REED C. ROLLINS

is developed and there is a sharp distinction between these leaves and
the strictly cauline leaves. In other populations, such as those of the
Grand Canyon, there is a weakly developed rosette of basal leaves.
Otherwise, internodes do develop between the first formed leaves and
there is a gradual transition from the lowest petiolate to the sessile
auriculate leaves further up the stem. But these differences are only
a matter of degree. They hardly represent the basis for defining sep-
arately recognizeable taxa.

The present interpretation of Thelypodiopsis purpusii includes what
appear to be disjunct populations, quite far from each other, in six
localities. In all of these, the species occupies similar shady habitats,
either at the base of cliffs, near large boulders, or in the shade of
trees or shrubs. A number of distinctive features characterize plants
of all populations. Aside from the lobed or deeply pinnatifid lowest
leaves, the ascending siliques are very narrow and the seeds are ex-
tremely small in a single row. The anthers of the paired stamens are
exserted and are relatively short, ca. 1 mm. long. Minor trends to-
ward discrete populations are recognizable even though there appears
to be no solid basis for dividing the species into infraspecific taxa.
Given the history of exploration involving S$. purpusii, it is likely that
the future will see at least some of the geographic gaps filled.

13. Thelypodiopsis incisa Rollins, Contrib. Gray Herb. 206: 13. 1976.

Annual or biennial, sparsely pubescent with simple spreading trichomes below, gla-
brous above; stems erect, one to several from base, branched, 4-6 dm. tall; strictly
basal rosette not formed; lower cauline leaves with erect simple tricho

‘ . lo

wi strong central nerve; inflorescences racemose, terminating each branch: sepals

purplish, oblong, 4-5 mm. long, ca. 1-5 mm. wide, outer pair saccate, boat-shaped,
owed above, inner pair flat, blunt at apex; petals lav i

8-10 mm. long, 3-4 mm. wide, not differentiated into blade and claw; stamens

strongly tetradynamous, paired stamens erect, 6-7 mm. long, single

long; divaricate, not expanded toward apex, st

c@) S
scarcely cleft, very slightly lobed over replum m
long, occupying full width of silique, 1.75-2 mm. lon

DISTRIBUTION: known only from the type collection.

FLOWERING: March.

HOLOTYPE: ca. 80 miles north of Saltillo along Highway 57, Coa-
huila, Mexico, 13 Mar 1970, Larry C. Higgins 2717 (wrs!
(ENCB)).

One of the most distinctive features of Thelypodiopsis incisa is the
definite sterile beak narrowing to the apex of the silique. It is nearest

). Isotype

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 91

in its relationship to T. purpusii but differs from that species in having
much larger flowers, torulose instead of plane siliques, a beak tipped
with a prominent style about 2 mm. long instead of no beak and a
style about 1 mm. long as in T. purpusii. The latter also has narrower
nearly terete instead of slightly flattened siliques, as in T. incisa, and
usually white instead of the lavender petals.

14. Thelypodiopsis versicolor (Brandegee) Rollins, based on Thely-
podium versicolor Brandegee, Univ. of Calif. Publ. Bot. 4: 178, 1911.
Sisymbrium versicolor (Brandeg.) O. E. Schulz, Pflanzenr. 86 (IV,
104) 57, 1924. Thelypodiopsis versicolor (Brandeg.) Rollins, Contrib.
Gray Herb. 206: 12, 1976.

B
along petiole margins of basal leaves, otherwise glabrous throughout, glaucous; stems
single or few from the base, branched, slender, 4-12(-15) dm. tall; basal leaves with
a winged petiole, entire or shallowly dentate, often denticulate with minute teeth,
obovate to oblanceolate, obtuse, up to 4 cm. long; cauline leaves auriculate, glabrous,
lower obovate, denticulate with minute teeth middle oblong, entire or denticulate;
owered at the apex

of narrow, greatly elongated infructescences; sepals erect, ob] purplish, ca. 3 mm
long, near e; pe ite, broadly spatulate to obovate, ca. 5 mm. long,
blade abruptly narrowed to a slender claw; stamens included; anthers less th

an 1 mm.
long; fruiting pedicels erect to slightly spreading, often appressed to the rachis, slen-
der, straight, 7-10 mm. long; siliques erect to slightly divergent, terete, Pe ses-
i : ca. 1 mm. in

diameter; styles nearly obsolete to 1 mm. long; seeds oblong, wingless, plump, ca. 1.5
mm. long, ca. 1 mm. in diameter; cotyledons incumbent.

DISTRIBUTION: Coahuila and San Luis Potosi, Mexico.

FLOWERING: June—October.

HOLOTYPE: Sierra de Parras, Coahuila, Mexico, Oct 1910, C. A. Pur-
pus 4978 (uc!). Isotypes (GH!, US).

— SPECIMENS STUDIED: Mexico. meagre .

q : ref >

4941, Seanpord ne mi. SW of Torreon in Sierra de Jimulco,
re ta Apa Henrickson piel ae a on. Ohi
9 ae S org aalipeng saci 3 Jul 1941, Stanford et al. 216 (aRIZ, GH, Ny). San
Luis Potosi: 12 km. al W de Estacion Berrendo, Mpio. de Charcas, 11 Sep 1955,
Rzedowski 6590 (GH, ENCB).

Thelypodiopsis versicolor is distinctive because of the eee
pressed pedicels and strictly erect siliques that are shorter than those
of its nearest relatives. The infructescences are narrower than in most
other species. The cauline leaves are strongly auricled with the au-
ticles fully clasping the stem in most plants. Variation between spec-
imens of the few collections known suggest that future collecting may
show up correlations of characters that are taxonomically aaa
If this happens, more than one taxon may be present in the materi

we have referred to this species.

92, REED C. ROLLINS

15. Thelypodiopsis vaseyi (S. Wats.) Rollins, based on Sisymbrium
vaseyi S. Wats. ex Robinson in Gray, Syn. Fl. N. Amer. 1: 138, 1895.
Sisymbrium watsonii Payson, Univ. Wyo. Publ. Bot. 1: 16, 1922,
based on S. vaseyi S. Wats. Thelypodiopsis vaseyi (S. Wats.) Rollins,
Contrib. Gray Herb. 206: 12, 1976.

Annual, glabrous throughout, somewhat glaucous; stems single, highly branched be-
ginning above base, 4-10 dm. tall; basal rosette not formed; lower cauline leaves nar-
rowly oblong to lanceolate or slightly oblanceolate, sometimes auriculate with small
auricles not fully clasping the stem, entire or rarely denticulate with minute teeth
4-6 cm. long, up to 1.5 cm. wide but usually much narrower, upper leaves cuneate
at base, lanceolate to narrowly linear, acute; inflorescences few-flowered, terminating
each slender branch, not elongating greatly in fruit; infructescences narrow, usually
less than 1 dm. long; i

mm. long; petals white, broadly spatulate to ate, abruptly narrowed from blade
to claw, unguicula ng; filaments relatively stout, white or purplish;
anthers small, m ng; fruiting pedi slender, spreading nearly at

right angles to rachis to slightly ascending, straight, 8-11 mm. long; siliques erect,
subsessile to sessile, tapered both above and below, strongly torulose, 1.5-2.5 cm.

les ca. 0.5 mm. long; stigmas inconspicuously 2-lobed with the lobes over the
replum margin; seeds plump, slightly narrowed at distal end, wingless, oblong, 1.2-1.5
mm. long, less than 1 mm. wide; cotyledons incumbent.

DISTRIBUTION: mountains of central New Mexico.

FLOWERING: July—August.

LECTOHOLOTYPE: Las Vegas (probably in the mountains to the west
of Las Vegas), New Mexico, 1881, G. R. Vasey s.n. (GH)).

OTHER SPECIMENS STUDIED: New Mexico. San Miguel Co.: Windsor’s Ranch, Pecos
River Nat. Forest, 3 Aug 1908, Standley 4741 (cH); Lion Head trail above Cowles, 3
Aug 1934, Goodwin 942 (cu); mts. W of Las Vegas, 1881, Vasey 41 (GH). Otero Co.:
Cloudcroft, 19 Aug 1968, D. S. & H. B. Correll 36129 (GH, TEX-LL); James Canyon,
Sacramento Mts., 5 Aug 1905, Wooton s.n. (NMC); Mescalero Reservation, 21 Jul 1905,
Wooton s.n. (NMC).

The flowers of Thelypodiopsis vaseyi are the smallest in the genus
and both sepals and petals flare during anthesis. The plants are highly
ramified with the numerous branches being much more slender than
the main axis. The habit is very distinctive. The lower cauline leaves
are quickly shed and I have not seen any specimens with all of the
lower leaves present. On the holotype, the middle and below middle
leaves have small auricles but in several other specimens leaves in
this position on the plant are merely sessile and are truncate at the
base without auricles being present. It appears that this is a variable
feature of the species.

The substitution of the name Sisymbrium watsonii for S. vaseyi by
Payson was a mistake as pointed out several years ago (Rollins, 1960).
A detailed discussion of the nomenclatural situation will be found in
that publication.

16. Thelypodiopsis retrofracta (Roll.) Rollins, based on Sisymbrium
retrofractum Rollins, Rhodora 59: 66, 1957. Thelypodiopsis retrofracta
(Roll.) Rollins, Contrib. Gray Herb. 206: 15, 1976.

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 93

Annual or possibly biennial, glabrous throughout, strictly basal leaves not known,
basal rosettes not formed; stems single from base, branched above, 5-8 dm. tall; lower
cauline leaves petiolate-lanceolate to linear-lanceolate, coarsely dentate to somewhat
lobed, 4-6 cm. long; upper cauline leaves cuneate at base, entire to sparsely and
shallowly dentate, nearly linear; inflorescence few-flowered; flowers erect or ascending
on slender ascending pedicels which reflex shortly after anthesis; sepals purplish, ob-
long, 3.5-4 mm. long, ca mm. wi

ong, ca. 1 mm. wide; valves l-nerved nearly full length; styles 1.5-2 mm. long;
mas weakly lobed over the replum margin; seeds plump, slightly longer than broad,
ca. 1.2 mm. long, ca. 1 mm. wide, with a narrow distal wing; cotyledons accumbent.

DISTRIBUTION: known from only two collections, one each from Coa-
huila (?) and Zacatecas, Mexico.

FLOWERING: July—August.

HOLOTYPE: 18 km. west of Concepcion del Oro, Coahuila (?), Mex-
ico, 22 Jul 1941, L. R. Stanford, K. L. Retherford & R. D. Northcraft
570 (cu). Although the printed label reads State of Coahuila, it is
difficult to accept that a point 18 km. W of Concepcion del Oro,
which is in the state of Zacatecas, would be other than in Zacatecas.
More likely the wrong label form was inadvertantly used by the col-
lectors.

OTHER SPECIMENS STUDIED: Mexico. Zaca
Jul 1934, Pennell 17443 (cH).

Fortunately, specimens of the second collection seen of Thelypo-
diopsis retrofracta (Pennell 17443) have mature seeds. It was surpris-
ing to find that these are distally winged and that the cotyledons are
nearly accumbent. Both of these features are characteristic of Arabis
and together with the reflexed siliques make a fairly strong case for
placing the species in that genus. However, the siliques are terete
and the seeds are plump, features not characteristic of Arabis. Until
the species is better known, it is perhaps best to leave it in Thely-
podiopsis where it does not violently disturb the integrity of that ge-
nus. The species is certainly not with its congeners in Sisymbrium,
the genus to which I originally referred it.

tecas: Aranzazu, Sierra Madre Orientale, 19

SCHOENOCRAMBE

Schoenocrambe Greene, Pittonia 3: 124-126. 1896. ae
Hesperidanthus Rydberg, Bull. Torr. Bot. Club 34: 433. ie
Perennials with ligneous caudices and roots, sometimes rhizomatous, old — ra si

present, glabrous or with simple trichomes; stem one to several, usually rigid, ate

i caudex, simple or branched above;

late or oblanceolate, petioled to

i td thickish; flowers in a lax raceme ter-
cuneate at base; entire to somewhat ganas of or with subapical calloses on

nocr
ovate; pedicels slender, widely spreading
erect to widely spreading, sessile or rare

94 REED C. ROLLINS

long; stigmas bifid or conical with lobes over the replum margins; seeds plump, ob-
, wingless; cotyledons incumbent to obliquely incumbent. Type species: Schoen-
ocrambe linifolia (Nutt.) Greene.
KEY TO THE SPECIES
A. Petals not differentiated into blade and claw, gradually tapered from apex toward
base; sepals plane, without a callose just below apex.
B. Petals yellow, without purple veins; filaments more than twice the length of the
anthers; plants branched above, usually rhizomatous ........... 1. S. linifolia.
B. Petals purplish, purple veined; filaments less than twice the length of the an-
thers; plants usually not branched above, not rhizomato
C. Leaves linear, non-petiolate; flowers pale purple ......... 2. S. argillacea.
C. Leaves oblanceolate, petiolate; flowers deep purple ........ 3. S. barnebyi.
A. Petals strongly differentiated into blade and claw, abruptly narrowed from blade
to cl als with a prominent callose just below apex...4. S. linearifolia.
1. Schoenocrambe linifolia (Nutt.) Greene, based on Nasturtium lin-
ifolium Nuttall, Journ. Acad. Nat. Sci. Philad. 7: 12, 1834. Sisym-
brium linifolium (Nutt.) Nuttall in Torrey & Gray, Fl. N. Amer. 1:
91, 1838. Erysimum linifolium (Nutt.) Jones, Proc. Calif. Acad. Sci.
5: 622, 1895. Schoenocrambe linifolia (Nutt.) Greene, Pittonia 3:
124-128, 1896. Nasturtium pumilum Nuttall, Journ. Acad. Nat. Sci.
Philad. 7: 12, 1834. Sisymbrium pygmaeum Nuttall in Torrey & Gray,
Fl. N. Amer. 1: 91, 1838. Schoenocrambe pygmaea (Nutt.) Greene,
Pittonia 3: 128, 1896. Erysimum glaberrimum Hooker & Arnott, Bot.
Beechey’s Voy. 323, 1840. Schoenocrambe pinnata Greene, Pittonia
3: 127, 1896. S. linifolia pinnata (Greene) A. Nelson in Coulter &
Nelson, Manual Bot. Central Rocky Mts., 209, 1909. Sisymbrium lin-
ifolium Nutt. var. pinnata (Greene) O. E. Schulz, Das Pflanzenr. 86,
IV, 105: 104, 1924. Schoenocrambe decumbens Rydberg, Bull. Torr.
Bot. Club 31: 409, 1904. Sisymbrium decumbens (Rydb.) Blankinship,
Mont. Agric. Coll. Sci. Studies Bot. 1: 60, 1905. Sisymbrium linifol-
ium Nutt. var. decumbens (Rydb.) O. E. Schulz, Das Pflanzenr. 86,
IV, 105: 104, 1924.

nnial, rhizomatous, glabrous to sparsely pilose with simple trichomes below,
glabrous above, glaucous: rhi i

ces few to many-
owered, racemose with a bud cluster at apex; sepals ye owish, erect to slightly

g

same diameter as siliques, up to 1 mm. long;
sent plump, slightly over 1 mm. long, ca. 0.5 mm. in diameter, cotyledons incum-
ent.

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 95

DISTRIBUTION: British Columbia and Montana to northern New Mex-
ico west to eastern Nevada. It has been reported from northern Ar-
izona but I have not seen specimens from that state.

FLOWERING: _April-August.

: “sources of the Salmon River, Columbia,” [probably
Idaho] Wyeth (PH!).

REPRESENTATIVE SPECIMENS: Canada. British Columbia: Thompson River Canyon near
Spences Bridge, 29 May 1938, J. W. & E. M. Thompson 29 (Gu); within 5 mi. of
Lillooet, 2 Jul 1916, Macoun s.n. (GH); about 9 mi. SW of Clinton on Loon Lake Road,
11 Jul 1956, Calder et al. peice: (GH H); Fraser River Valley, 19 May “ Macoun 146
(GH); Oliver, 5 May 1960, mish & Vrugtman 60128 (us); 1 mi. — fig , 20
May 1953, Calder ‘ob pies: 7975 (us). USA. Head of the TM eth (holotype

of Sisymbrium pygmaeum Nutt., PH!). Snake country [probably el, “Toimi soe

of Erysimum areata A, Gu!). Colorado. Archuleta Co.:
boles, 15 May 1981, R é& K. W. Rollins 8117 (cn); — Co.: "Ena : Aug
1894, Crandall 65 (GH). ode n Co.: 4 mi. E of Gunnison, 22 May 1938, Rollins

2097 (GH); 2 mi. E of Sapinero, ee, 1957, Beaman & "Erbisch 1186 (GH); Gunnison
Canyon, 7.6 mi. - of Gunnison, 27 May 1979, R. C. & K. W. ha 7968 a . La
Plata Co.: Arboles, Jun 1899, Baker 357 (GH). Montezum ma Co.: just E of Cortez, 25
May 1934, McKelvey 4647 (cH). Montrose Co.: Paradox, 17 Jun 1912, one 98 (GH
Us); Naturita, 4 May 1914, Payson 317 (GH). San sere Co.: 16 m i. N of junction,
Colo. Hwy. 141 and US Hwy. 666, 16 May 1981, R. C. & K. W. Rollins 8130 (cH).
Idaho. Bivins’ o Martin, 5 Jul 1916, Macbride & a 3048 (cu); Picabo, 1 Jul
1916, Macbride ¢ Payson 2973 (cu). Butte Co. : along Hwy. 88, 8 mi. N of its junction
with Hwy. 20/26, 22 May 1967, seis 819 (cu); 1 ods N Midway, 13 Jun 1941,
fo ge se Cassia Co.: near City of Rocks, 11 te N of Idah
2 Jun 1981, d> K. W. Rolling § 81293 (GH). Custer : hills across river from
Challis, 14 ee Po Hitchcock ¢ Muhlick 8977 (GH); sare ide Creek), 30 ju
1911, Nelson ¢ Macbride 1418 (cH). Lemhi be Salmon, 26 Jun 1920, &L
17 alls Co. Falls, 24 Jun 1912, Eat & Macbride
1719 (GH). Montana. Melrose, 6 ie 1895, Roibets 2671 (holotype of Schoenocrambe
decumbens Rydb., Ny! isotype NY!). Deer Lodge Co.: ses nue,
2511 (cH); pipet 11 Jul 1906, ero ad 660 (Us). s & Clar
mi. N of airport, Helena, 25 May 1956, Don Scharff s.n. te H). Wheatland Co.: 13
mi. W of Shawmut, 29 May 1956, Den op Mt - ae evada. Elko Co.: ca. 3/4
mi. bad Pass, Pecou ts. :
sag pee Aug 1868, Wa n 96 oh White Pine Co.: E side
oft ‘re ned Pass, 25 Jun 1979, ash - nt 5287 t
road from Tippett to Pleasant, Kern Mts., 25 pe a iris
iy ce eB lads ge veg bani of Schoenocrambe hago’
reene, us!). Carbon Co.: between dieters Summit an Jelper, 10 May :
0, Maguire 18548 fen. Daggett
McKelvey Sed Gi om Ne ee bie (Ga) G mi. $ of Manila, 16 Aug

eep Creek Canyon, 6 Jun 1932,

1935, pinata 12654 feu Duchesne Co.: near Myton, 13 egret — sored
(GH); 1 mi. up side canyon of fara Canyon, 3 Jun : Nei ce ek Higgi

(GH). Emery Co.: ca. 3 m of San Raphael River bridge, 25 nee Cottam 7202
1316 (cu). Juab Co.: ‘ae Canyon, Deep Creek Mts., 29 Ma bo t 77 mi. E
(CH) Co.: 38 mi. E of Kanab, 19 Apr 1967, Rollins oie (GH); al nent Sag =
of Kanab, 4 M 5, Cronquist 10018 (cH). Ric : —s ee pen ehiers
Jun 1946, Rollins 3069 (cx). San J F ost Maguire S of Ouray, 31
of Bluff, 9 } _ A. H. Holmgren 3162 (ch). Uinta 5 mi. N Thorne’

BE
oO
&
—_
S
—
=
=
e
>
my
5
=
~
=
aA
EE:
~~
5
3
=
°

‘Palmer 20 (G sh
ae a Sem) Washington Co.: Red Creek, ee Pa ay 1936, 2 1371 7 re us); Dry

96 REED C. ROLLINS

Falls near Coulee City, 2 May 1931, Thompson 6153 (GH, Us). Kittitas Co.: side canyon
of Columbia River, Spring, 1978, Richard Old s.n. (cu). Klickitat Co.: rocks, 11 Jun
1881, Suksdorf 3 (GH); near Bingen, 27 Apr 1935, Thompson 11413 (GH, US). Wyoming.
Albany Co.: near Wyo. Hwy. 34, 22 mi. E of Laramie, 24 Jun 1951, Rollins & Porter
5113 (GH, RM); Laramie, 20 Jun 1900, Nelson 7277 (GH, NY, RM, US). Carbon Co.: 1 mi.
S of Baggs, 22 May 1979, Reed C. & Richard C. Rollins 7935 (cH). Fremont Co.: 32
mi. NE of Farson, 25 Jun 1979, R. C. & M. L. Rollins 79322 (cu). Lincoln Co.: 9
mi. N of Kemmerer, 21 Jun 1979, R. C. & K. W. Rollins 79305 (cu). Natrona Co.:
13.5 mi. NE of Muddy Gap, 21 May 1979, Reed C. & Richard C. Rollins 7924 (cu).
Sweetwater Co.: 48 mi. S of Rock Springs, 19 Jun 1981, R. C. & K. W. Rollins 81357
(GH). Uinta Co.: near Lyman, 21 Jun 1937, R. C. Rollins 1773 (GH, us).

As one of the two most widespread species of Schoenocrambe, S.
linifolia is also one of the most polymorphic. In that respect, it rivals
S. linearifolia. The rhizomatous habit of S. linifolia is distinctive but
it is not certain that every population is exclusively made up of plants
with rhizomes. At several different locations, I have attempted to de-
termine the extent to which identifiable rhizomes are present. When-
ever the rhizomes are elongated, they are easily seen as such. How-
ever, many plants have close branching just below the caudex and it
is difficult to tell in the field whether the branches are coming off a
vertical rhizome or whether this is merely an underground branching
system. In any case, there are many plants that are hardly rhizoma-
tous as usually understood. The type of caudex branching is very sim-
ilar to that of the other three species of Schoenocrambe.

Many populations of Schoenocrambe linifolia have at least a few
plants with pinnatifid lower leaves. In a few, pinnatifid lower leaves
predominate but in most populations the leaves are entire. I agree
with Payson (1922) that pinnatifid leaves cannot be made the basis for
a distinct taxon as was done by Greene when he described S. pinnata.

There is a rather wide range of flower size in Schoenocrambe lin-
ifolia. Relatively large-flowered plants occur especially in southwest-
ern Colorado and southern Utah but other characters are very similar
to those of populations in the more northerly areas of the species
range. I was surprised to find a population in central Wyoming where
each plant had six siliques or less when the usual number per fruiting
branch is more than twelve. The siliques of specimens from northern
Washington and southern British Columbia are narrower than in ma-
terial from most of the species range. These types of variations appear
to be characteristic of the species and illustrate the polymorphy men-
tioned above.

2. Schoenocrambe argillacea (Welsh & Atwood) Rollins, comb. nov.,
based on Thelypodiopsis argillacea Welsh & Atwood, Great Basin Nat.
37: 95-96, 1977.

Perennial, wholly glabrous; stems several from a stout, ligneous, el ted all
vertical underground caudex which also supports old broken mk us aga.
unbranched, purplish toward base, greenish above, lax to erect, 1.5-3 dm. tall; leaves

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 97

all cauline, sessile or rarely with a short petiole, overlapping, linear, gerry oe
lanceolate thick, Reriag nisi entire or rarely obscurely dentate, 1-2.5(-3.5

lon sually less than 2 mm. wide (emnstnes up to 6 mm. wide), widest aves
ciate 2 t base; "inflovesdaneen few to about 20-flowered, elongating in fruit; sepals
oblong, aicslul. spreading at anthesis, non-saccate, 5-7 mm. long, mm. wide;

tals pale lavender to whitish, prominently purple-veined, broadly spatulate to lin-

2.5-3 mam. long, filaments of single stamens slightly curved, 1.5-2 mm. long;
ead straight, erect, 2.5-3 mm. long; glandular tissue poorly dev ined. mainly
above 7, insertions; fruiting pedicels widely spreading to divaricately asc te
8-12 mm. long, abruptly expanded at apex; — sessile, git nearly straight to
Saved on outward, slightly torulose, 4-7 cm. lon mm. meter, narrowed
toward apex; styles subclavate, 1-2 mm. long; pei oo lobed “with lobes over

m margins; seeds oblong, Baan mornin? SS, ca. m. long, ca. 1 mm. in di-
ameter; cotyledons obliquely incum
DISTRIBUTION: apparently saderhin to the Uinta Basin of eastern
tah.

FLOWERING: April—Jun

HOLOTYPE: hills west of Willow Creek, on east slope of Big Pack
Mountain, T10S, R20E, Sec. 33, at 5,000 ft. elevation, on Green
River shale, Uintah County, Utah, 11 May 1976, N. D. Atwood 6627
(BRY!). Isotypes (GH!, NY!, US!).

OTHER SPECIMENS STUDIED: Utah. Uintah Co.: 25 mi. due S of Ouray, between Hill
and Willow Creeks, 26 May 1978, Nees 4962, 4972 (pry); Uinta Formation, steep N
facing slope, Big Pack Mountain, 1 mi. W of Willow Creek, 27 Apr 19 978, hee er

; JTC); ri i ck Mountain, on and below red
eed . a nso Mar ee une et al. 3168 (uTC); N side of Big
Pack Mountain, T11S, R20E, Sec. 23, 14 May 1979, Schulz ¢ Schulz 3172 (cH, UTC);
Big Pack Mountain, W of Willow Creek, 16 May 1978, Neese & Peterson 4620 (BRY).

Juvenile leaves of Schoenocrambe argillacea are very much like
mature leaves of S. barnebyi, being oblanceolate to obovate and petio-
late. These leaves are formed on retoos of old stems or caudex
branches that secondarily have been covered with soil. At exposed
places several new branches are produced on these old stems that will
become independent plants after the connection with the neice
plant has been severed. The juvenile leaves are quite different from
the narrowly linear leaves on most mature p
ture irilividlusil with flattened leaves that are somewhat pete
between the usual very narrow nearly terete leaves and the broader
juvenile leaves found on the occasional retonos. The case ay Agee
nizing S. barnebyi as a distinct species is weakened by ps act
there are branches and some plants of S. argillacea with ioe ua
proaching those of S. barneybi in size and shape. ae sha

tween the narrowly linear leaves of S. argillacea and the broa ai
petiolate leaves of S. barnebyi are one of the bases for Sra
species apart. However, a more complete set of material o

98 REED C. ROLLINS

nebyi is needed to be sure whether one taxon or two actually exist.

Schoenocrambe argillacea appears to be restricted to the Uinta and
Green River formations and is local in the Big Pack Mountain area
of the Uinta Basin. Another crucifer, Glaucocarpum suffrutescens is
restricted to the same general area. Schoenocrambe argillacea is dis-
tinctive because of its narrow, usually nearly terete leaves. The closest
species is the recently described S. barnebyi which is similar in most
respects. As compared to S. linearifolia, which is purple-flowered,
usually single-stemmed and much taller, these species are closer to
S. linifolia in habit. But S. linifolia has bright golden yellow flowers
instead of purplish flowers as in S. argillacea and S. barnebyi. As
pointed out above, the buds and young flowers of S. linearifolia al-
ways show a prominent callose or short horn just below the apex of
the inner pair of sepals. There is no such structure on the sepals of
either S. argillacea or S. barnebyi. In these, the filaments are equal
to or shorter than the anthers while in S. linearifolia the filaments are
at least twice as long as the anthers. The stigmas, too, are different.
In S. linearifolia the stigmatic surface extends over the entire apex
and down the sides of the style, remaining larger than the style di-
ameter in mature fruit. Whereas in S. argillacea and S. barnebyi the
stigma is only at the very apex of the style and recedes to an area
smaller than the style apex as the fruit matures.

3. Schoenocrambe barnebyi (Welsh & Atwood) Rollins, comb. nov.,
based on Thelypodiopsis barnebyi Welsh & Atwood, Brittonia 33:
300-301, 1981.

Perennial, glabrous throughout, glaucous; stems several from a ligneous vertical un-
derground caudex which also supports old broke stem-stubs, stems unbranched,

gree ned, oute
saccate, ca. 7-8 mm. long; petals spatulate to lingulate, tapered gradually from outer
blade to point of insertion, not unguiculate, purplish, prominent i .
1 m. in diameter,
not known.

OT PECIMENS STUDIED: Utah. Emery Co.: San Rafael Reef, ca. 18 mi. due NW
of Hanksville, 21 Jul 1980, Harris 1007, 1008 (GH).

This species is quite closely related to Schoenocrambe argillacea
differing mainly in leaf-shape and intensity of flower color. The leaves
of S. barnebyi are definitely petiolate whereas those of S. argillacea
are at most cuneate at base. There may be other character differences
that will show up when specimens with mature seeds and fruits that
have not fully shed are available for examination.

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 99

4. Schoenocrambe linearifolia (Gray) Rollins, comb. nov., based on
Streptanthus linearifolius Gray, Mem. Amer. Acad. 4: 7, 1849. The-
lypodium linearifolium (Gray) S$. Watson, Bot. King’s Exped. 25,
1871. Hesperidanthus linearifolius (Gray) Rydb., Bull. Torr. Bot. Club
34: 434, 1907. Sisymbrium linearifolium (Gray) Payson, Univ. Wyo.
Publ. Bot. 1: 19, 1922. Thelypodiopsis linearifolia (Gray) Al-Shehbaz,
Contrib. Gray Herb. 204: 140, 1973. Streptanthus stenophyllus Rol-
lins, Contrib. Dudley Herb. 3: 175, 1941 [holotype: Sierra de Parras,
Coahuila, Mexico, 5 Sep 1950, Shreve & Tinkham 9886 (cH))].
Perennial, <A e a glaucous; stems one or few from base, erect, stiff,
branched upward, u 15 dm. tall; leaves of first year’s rosette obovate, dentate to
entire, petiolate, ia sae plants mature; lower cauline leaves oblanceolate to nar-

up mm. eC
cately ascending, straight, slender, expanded at apex, 1-2 cm. long; hg ae? terete,
erect to widely spreading, sessile to subsessile, 4-10 em. long, 1-1.5 mm. in nappa
~ Sig 0.5-1.5(-2) mm. long; stigmas ——— ese an seeds plump, angular
mm. long; cotyledons wena incumben = 11, + 20 (Rollins, 1966; Rollins
oad 1977; Rodman

DISTRIBUTION: Colada and Arizona to Texas, southward to San
Luis Potosi and westward to Durango and Sonora.

FLOWERING: July-November.

HOLOTYPE: New Mexico, mountainous regions from Santa Fe to Las
Vegas, July-Aug. 1847, Fendler 24 (cn).

The-

REPRESENTATIVE SPECIMENS: for —e from meg given under the name
lypodiopsis line arifolia, see Contrib. Gray Herb. no. 206, p. 15, 1976. USA. Arizona.
Cochise Co.: 1 m of Rustler’s Park, Chiricahua Mts., 21 Oct 1974, Ro " ors y
74198 (GH); 2.4 mi. 4 own Pinery Canyon from Pinery Campground, Pepsserieo ti : ae :
22 Nov 1974, Rollins et al. 74199 (cH). Coconino Co.: Cape Royal, onage es

ul 1930, Goodman ¢& Hitchcock 1636 (cx); N rim of the Grand Cone, . naeee
Rollins aa pu Gila Co.: Workman Cre ek ae , rat: end of car roa . ene :
a

Cottonwood Cr., ca. 4 mi. above Nat. top SS Sangre de ef aliens
Jul 1936, Rollins 1313 (cu, Us). Las Animas of Mosley, 2 ha Oh ke Me
1818 (cH). Rio Grande Co.: Del Norte, 1 jul 1921, eae et al 4138 ie lec ea
ae ap hap Co; Beaverhead, Datil Forest, 29-30 Jul 1924, bis ait

of canyon 2 end of Organ Mts., pe Sep 1930, Fosberg $3711 (GH). age Co.: Burro

along Ruidoso Cr., Wh “ Mts., 3 Jul 1895, Wooton s.n. (cH). McKinley Co
up and Zuni, 5 Jul 1934, Hodgdon & Rossbach 8 (GH a nig — * San
ye Sep 1942, Kirk Bryan s.n. (GH e mi n 3

E
». (cH). Sante F A 3729 (GH). Socorro
eaten 2304 (cu); C ito, 18 Jun 1897, A. A. & E. G. Heller
oe 1,000 ft. a cane ‘on sees et al. 574 (GH). Texas. Brewster r Co.: 9-point

100 REED C. ROLLINS

Mesa, 60 mi. S of Alpine, 22 Sep 1966, Correll 33817 (GH); Chaney Ranch, ca. 25 mi.
SE of Marathon, 16 Apr 1961, Rollins & Correll 6168 (GH); Boot Spring, Chisos Mts.,
ul nston ivi
the Llano, May 1885, Reverchon 3 (Gu). Culberson Co.: Pine Springs Canyon, Guada-
lupe Mts., 7 Sep 1961, Correll & Johnston 24243 (cu); 6.5 mi. NW of Van Horn,
Beach Mt., 14 Jul 1943, Waterfall 5075 (cu). El Paso Co.: McKelligon Canyon, Frank-
lin Mts., 2 May 1970, D. S. & H. B. Correll 38599 (Gu); 4.5 mi. N of El Paso, 13
May 1946, Cory 52958 (cH). Hudspeth Co.; 12 mi. S of Sierra Blanca, 25 Apr 1961,
Rollins & Correll 61133 (cu); N peaks of Eagle Mts., ca. 7 mi. WSW of Hot Wells,
17 Jul 1943, Waterfall 5154 (cu). Jeff Davis Co.: upper slopes of Sawtooth Mts.,
13 Oct 1952, Correll 15012 (cu); Mt. Locke, Davis Mts., 12 Aug 1934,
Cory 9426 (GH). Pecos Co.: 30.5 mi. S of Ft. Stockton, 5 Sep 1933, Cory 6862 (GH).
Presidio Co.: above Capote Falls, 3 Nov 1966, Correll 34122 (cu); Fresno Canyon,
ca. 50 mi. E of Presidio, 20 Apr 1961, Rollins & Correll 61175 (cu); below San Es-
taban Lake, Marfa, 18 Aug 1940, Hinckeley s.n. (cu).

As far as present records show it, Schoenocrambe linearifolia is the
most widespread species of the genus. Its flowering period, which
stretches from April to November, is unusually long. Part of this is
due to the substantial elevational range the species is able to tolerate
but also the flowering period is very much affected by the large dif-
ferences in climate taking the total geographical range into account.

When Schoenocrambe linearifolia grows in the open, the stems are
erect, stiff, and reed-like. In the shade of small trees or bushes and
on shady cliffs, the stems are much weaker and tend to be less erect.
The plants are usually scattered rather than in dense colonies, most
often in open pine forests or oak woods. The species is taxonomically
distinctive. The flowers are relatively large and reminiscent of Strep-
tanthus where Gray originally described it. Unfortunately, I was mis-
led in the same way when I described Streptanthus stenophyllus as
new. But there is still some possibility of the latter being a species
distinct from S. linearifolia. If this proves to be the case when more
material is available for study, it will certainly belong to Schoeno-
crambe rather than Streptanthus. For the present, the evidence sug-
gests that Streptanthus stenophyllus is to be synonymized under
Schoenocrambe linearifolia.

The nomenclatural history of Scho
only that the placement of thi

elongated, and the prese are often considerably

horn in some individuals)
of sepals. Payson (1922)

gued that these differences are sufficient to justify placing S. linear-

THELYPODIOPSIS AND SCHOENOCRAMBE (CRUCIFERAE) 101

ifolia in a genus of its own. However, the recent discovery of S. ar-
gillacea and S. barnebyi, which are certainly close relatives and which
do not have the stigma and sepal characters mentioned, provide a
connecting link to Schoenocrambe that cannot be overlooked.

An unusual individual plant of Schoenocrambe linearifolia was found
in Pinery Canyon of the Chiricahua Mountains in southern Arizona.
This proved to be a polyploid with much larger petals and other floral
structures than is normal for the species as a whole. It was cited as
an example of a naturally occurring polyploid showing gigas features
(Rollins, 1979) but with these having no known taxonomic significance.
Populations of plants with these features apparently have not become
established.

LITERATURE CITED

AL-SHEHBAZ, I. A. 1973. . oe of the genus Thelypodium (Cruciferae).
Contrib. Gray Herb.

GREENE, E. L. 1896. rics in a Cruci ae I. Pittonia 3: 138.

HakRINGTON, H. D. 1954. Manual of the Plants of Colorado eo a Identification of

F and Flowering Plants Denver, Sage Boo
HOOKER, W. J. 1829-30. Flo Boreals-Asnericwia Vol. 1, pt. 1: 70, Lo
Payson, E. B. 1922. spe of Sisymbrium native to America north of Mexico. Univ.
yo. fun Sci. 1: 1-27. =
23. ‘i monographic study of Thelypodium and its immediate allies.
oi Gard. 325.

Ann. Mo.
ROBINSON, B. L. 1895. In . pe Watson, Synoptical Flora of North America. Vol.
i: 173-178. New York.
Ropman, J. E. 1978. IOPB Chromosome Number Report LXI. Taxon ar: 391.
ROLLins, R. e 1939. The cruciferous genus Physaria. Rhodora 41: 391-4 ee
————-. 1943. Generic revisions in the Ce eek Halimolobos. Ee udley
Herb, 3: 241-265.
————. 1950. Studies on some North American Cruciferae. Contrib. Gray Herb.
171: 42-5
—————.. 1960. Some Sisymbriums (Cruciferae) native to Texas and northeastern
Mexico. Rh dora 62: 55-60.
6 Chisntenas numbers of Cruciferae. Contrib. Gray Herb. 197:
43-65.
a ee 2 re Cruciferae and notes on Arabis and Caulanthus.
Contrib. bs Herb.
Se. 1016 ; Saas on eve n Cruciferae. Contrib. Gray Herb. 206: 3 sacl
1979. Taxonomic salehoesbstion of discontinuities in flower size.
28: 4
980. dass genus Pennellia (Cruciferae) in North America. Contrib. Gray
Herb. 20 5-2
Po AND. LL. ee 1971. Chromosome numbers of Cruciferae II. Con
trib. G
ox Merb 201 1 97 a 7.¢Chromosome numbers of Cruciferae III. Contrib.
Gray Herb. 207: 101-1 :
RYDBERG, P. A. 1907. Stoller on the Rocky Mountain Flora XVIII. Bull. Torrey Bo
Club 34: 417-437.
SCHULZ, O. E. 1924. Engler, Das Pflanzenreich (IV, 105) 86 1-388. a
age urze Ber ae sig neue Gattungen, Sektionen u
Cruciferen, Bot. Jahrb. 66: 9

102 REED C. ROLLINS

eage J. anpD A. Gray. A Flora of North America. Vol. 1: 1-711. 1838-40. New

a S, 1981. New taxa web western plants—in tribute. Brittonia 33: 294

. D. Atwoop. 1977. An essing momrie = Thelypodionis (Bras-
sicaceae) fon the Ce: Basie Utah. t Bas at.
D J. REVEAL. 1977. Utah i espace comet Great Basin

Serer AN
Nat. 37: 279-365.

STUDIES ON ARABIS (CRUCIFERAE) OF WESTERN NORTH
AMERICA II'

REED C. ROLLINS

The present paper continues the results of a restudy of Arabis,
building on my former monograph (Rollins 1941) of the genus for
western North America. New collections of rare and little known taxa
now permit a better assessment of their status in the genus, both as
to relationship with other taxa and as to the nature of the variability
present.

THE ARABIS DEMISSA—A. OXYLOBULA—A. PENDULINA GROUP

At the time of my previous study of Arabis oxylobula Greene and
related taxa, there were only two collections of this species available
for study that could, with confidence, be associated with the name.
Because it was considered rare and possibly endangered, A. oxylobula
was sought at its type locality near Glenwood Springs, Colorado, and
was rediscovered by Barry C. Johnston and Pete Lucas in 1978. Since
then, typical material has been collected at several additional localities
in Colorado. The new material now makes it possible to see some of
the variation present in this species and to better assess its relation-
ships and taxonomic position in the genus. Of particular interest was
whether a fuller knowledge of A. oxylobula would now dictate that
populations previously recognized as A. demissa Greene be incorpo-
rated in a more broadly defined and widespread A. oxylobula. At the
same time, the position of A. pendulina with respect to these taxa has
received attention together with several large collections of new ma-
terial belonging to this alliance.

In the original material of Arabis oxylobula, the erect basal leaves
are nearly glabrous with only a few ciliate trichomes on the slender
petioles. New material from the type area shows that the cilia may
extend fully around the blade as well as on the petioles. Material away
from the type area, for example in northwestern Colorado, northern
Utah and Wyoming, has similar trichomes occurring sparsely on the
narrow leaf-blades as well. These trichomes are large with a broad-
ened base and mostly simple. Occasionally a few forked trichomes are
Present and these are as large as those of the rest of the trichome
compliment. In contrast, the trichomes of A. demissa are much ee
and mostly branched. Those on the leaf-blade are at least forked an
in many cases 3- to 5-branched dendritic trichomes occur there.
Larger simple trichomes sometimes form cilia along the petiole mar-
gins but rarely extend to the blade area of the leaf.

: 7
The first Paper in this series was published in Systematic Botany (1981) 6(1): 55-64.

103

104 REED C. ROLLINS

Typical Arabis oxylobula has acute siliques that are less than 2.5
cm long and they are borne on very slender pedicels that vary from
spreading nearly at right angles to the infructescence rachis to arching
downward. The siliques themselves are at right angles to loosely pen-
dulous. In typical A. demissa, the siliques are obtuse, loosely pen-
dulous and the lower limit of their length is 2.5 cm. They range up
to 4 cm. The style is obsolete in A. demissa, whereas there is a short
but definite style present in A. oxylobula.

Except for the sharp differences in the trichomes, the distinctions
between Arabis demissa and A. oxylobula are rather weak. What tips
the balance in favor of recognizing both of these taxa as species is a
consideration of the populations that occur outside of the type areas.
It has always been uncomfortable for me to recognize as var. languida
and var. russeola plants which possess the same different trichome
types as A. demissa and A. oxylobula. Both varieties occur in the same
areas of Colorado, Wyoming, and Utah, yet they are not in mixed
populations and the distinctions between them are sharp. There is no
evidence of hybridization or introgression. When var. russeola is as-
sociated with A. pendulina and var. languida with A. demissa, as given
below, then the geography is no longer at variance with the general
phytogeographic patterns of the region.

KEY TO THE TAXA

. Basal leaves with predominatly large oe ascicular trichomes, these sometimes
‘restricted to petiole margins, rarely glabrou
B. Siliques acute, less than 2.5 cm long, ee at right _ to ner be loosely
descending; teenie filiform, 3-6 mm long; stems numero xylobula.
B. Siliques obtuse, 2.5-4 cm long, slightly incurved, nee ee par stouter,
eo OU ee a a pendulina.
C. Seeds in two definite rows, narrowly winged at least rian! forked trichomes
Ravi: on leaf-blade surfaces; basal leaves broa wey y peony

CORO RE Ci lel alee ig UN et gael wie we lw acre eel ka

_A aa var. pendulina.
. Seeds in one irregular row, wingless, only simp rae ca on leaf-blade sur-
faces: basal leaves linear to narrowly oblanceolate ...............-0-0ccucees
veeewctewneNeb ets egy bre) en a ose) “t ndulina var. russeola.
A. Basal leaves with much smaller, predominately inl ak mes on the leaf-
lades, sometimes with simple ciliate eee omes on the petioles; pubese

= sparse on basal leaves, never

ere arcuate, widely once - 6 cm long; pedicels at a apa to

SAS RFS RES Re OR cee he wk mb Bing inte lead uel Bint eWeek elute

ence dense

eee 9 ee ae

‘heeen seeds; fas wits ed or wingless ...... missa var Higiens?

Basal leaves oblanceolate; cauline eos vou ba p
wingle Bie icti eine Aaa 5b.

missa var. padoes

STUDIES ON ARABIS (CRUCIFERAE) 105

1. Arabis oxylobula Greene, Pittonia 4: 195. 1900.

This species is presently known from Garfield, Gunnison, and Mon-
trose counties, Colorado. It occurs in rock crevices and tiny cracks of
limestone on ridges, cliff edges, and mesa tops. Its nearest relative
is Arabis pendulina var. russeola which occurs farther north and west
in Colorado, Wyoming, and Utah.

SPECIMENS STUDIED. Colorado. Garfield Co.: Glenwood Springs, 18 Jun 1899, Os-
terhout 1942 (holotype ND; isotype RM; photo of holotype GH); same locality, 6 Jun
1902, Osterhout 2575 (COLO, NY, RM 2 sheets); Glenwood Canyon, rimroc of upper-
most “breaks” in canyon wall, N of head of Hanging Lake Trail, 29 Jun 1978, Johnston
and Lucas 1763 (COLO, RM); very top of exposed breaks, Lea ville Limestone, upper-
Co.: 1/4 mile W of Blue Mesa Dam, Black Canyon, 6 Jun 1979, Ratzloff 1645 (COLO);
same general locality, 17 May 1978, Ratzloff 1329 (coLo). Montrose Co.: Black Canyon
National Monument, 21 May 1962, Weber 11340 (coLo).

2. Arabis pendulina Greene, Leaflets 2: 81. 1910.

2a. Arabis pendulina var. pendulina

Specimens of Arabis pendulina from the type area in southwestern
Nevada are varietally distinct from the material we have from middle
elevations of southern Utah north to Colorado and Wyoming. New
collections from the type locality north of Vernal, Utah, clearly dem-
onstrate that var. russeola, which was described from that area, should
be associated with A. pendulina rather than A. demissa as in my for-
mer treatment (Rollins 1941). It is not necessary to cite specimens of
var. pendulina since a number were previously listed.

2b. Arabis pendulina var. russeola (Roll.), comb. nov., based on A.
demissa Greene var. russeola Roll., Rhodora 43: 398. 1941. A. setulosa
Greene, Leaflets 2: 81. Bo
i this vari
oo 5 anes fas a known from extreme northwestern Utah
as well as from central Wyoming, and from much farther south in
Utah than was previously recognized.
SPECIMENS STUDIED. Colorado. Moffat Co.: terraces above Yampa River, between

1967, Weber s.n. (COLO); summit of Round Top Mt.,
Pheer er aie sepaaes ta n-1 Jul 1948, Wolf and Dever 5192 (COLO); same

locality, 1 Jun 1956, Weber and Welsh 9618 (COLO). Wyoming. bany Co.: Laramie

ety has been much expanded by

east of
1979, R. C. and M. L. Rollin :
tinental Divide, 15 Jun 1978, Johnston and Lucas 1689X (COLO, me ona Co.: 2
Hwy, 220, 21 M :

: é ‘ : i , 12 Jun 1981, R: C. an
e in mineralized chip rock, 4 miles S nee ; Co.: vicinity of Flaming

of ridg
Rollins 81283 (Gu, duplicates to be distributed). Dagget oe
Cone 3 Jun a eins 2272 (GH, RM); same locality, 31 May 1932, Williams 459

106 REED C. ROLLINS

t al. 776 (COLO, GH, RM); between Bryce Canyon and Red Canyon, 6 Jun :
C. and K. W. Rollins 79172 pe 1 mile S of Cannonville, 6 Jun 1979, R. C. and K.
W. Rollins 79171 (cu). Kane Co.: 4 miles south of Cannonville, 12 May 1965, Cron-
quist 10080 (COLO, GH, RM). Millard Co.: windswept rock ridge, near Oak City, 15 Jul
1980, Goodrich 14363 (cH). Uintah Co.: 18 miles N of Vernal, 17 Jun 1937, Rollins
1757 (GH, ose os miles N of Maeser, 21 Jun 1979, Neese and Moore 7783 (RM).
Washington Co.: Hor tie Plateau, Zion National Park, 22 Jun 1965, N. H. Holm-
gren et al. "1986 (G ty

The distinctions between var. pendulina and var. russeola tend to
break down to some extent in the area of overlap in Utah. The south-
western populations, including those in most of Nevada, are clearly
separable from the northeastern populations of Colorado and Wyo-
ming. Some populations of var. russeola take on aspects of A. oxy-
lobula showing a definite relationship in that direction. For example,
plants of the population in Box Elder County, Utah, have on the av-
erage, shorter siliques than most populations of var. russeola and in
this respect, they approach A. oxylobula.

x 3 semase falcatoria Roll., sp. nov.

erba perennis, caespitosa; csnalibuis erectis vel decumbentibus, 1—-2(—3) dm alltis;
Fis radial us integris, linearibus vel lineari-oblanceolatis, hispidulis, acutis, 1-2(-2.5)

cm longis, 2~4(-5) mm latis; foliis caulinis sessilibus, 1-1 “8 cm longis; sepalis pur-
nag non is, spa ntibus vel glabris; petalis albis rpureis,
spathulatis, rectis; pedicellis fructiferis rectis divaricatis, glabris vel sparse pubescen-
tibus, 3—6( longis; siliquis a soa divaricatis, glabris, a acu gis,

Per a pe tose; caudex simple; stems few to several, arising below terminal
chested of erect leaves, simple or branched above, Paitots

1-2(-3) dm tall; outer basal leaves oblanceolate, widely spre
linear to linear-oblanceolate, erect, entire, densely
erect trichomes, 1—2(-2.5) cm lon

ith small auricles; flowers few; ivean erect, non-
saccate, oblong, purplish, sparsely ae oe scarious-margi ned, 2.5-3
m wide —

sed, n
wide, uniseriate; cotyledons
eet aie Wer segs pear td eae rth of Grouse Creek
type on, wikypen a He a and Kathryn W. Rollins 81259 (holo-
eralized chip rook, 4 males $:of Lymn; Ut fou/4@0h oth cee ies me ies
ys ger - Joltos seme artial shade of tt sae same locality, 12 Jun 1981,

GH). Nevada. Elko Co.: 3.2 air miles § of Cass House
I (GH).

The coarse forked to few-branched trichomes on the basal leaves of
Arabis falcatoria are similar to those of A. demissa but the widely

STUDIES ON ARABIS (CRUCIFERAE) 107

spreading, much longer, bow-shaped siliques are radically different
from the pendulous, narrow and straight siliques of that species. The
siliques are closer in shape to A. cusickii or A. sparsiflora var. spar-
siflora than to most other species of the genus. But in A. falcatoria
the caudex is simple, the cauline leaves few and mostly remote, and
the pubescence while coarse, is not hispid as in A. cusickii where the
caudex is branched and the cauline leaves are many and overlapping.
Arabis sparsiflora var. sparsiflora is tall and scarcely caespitose while
A. falcatoria is mostly low-growing and definitely caespitose. Neither
A. sparsiflora nor A. cusickii are closely related to A. falcatoria.
Rather, they belong to another group of species in the genus. In gen-
eral habit, A. falcatoria is closest to A. pendulina var. russeola but
the trichomes are completely unlike in these two taxa and the siliques
of the latter are shorter and pendulous instead of widely spreading
as in A. falcatoria.

There is considerable difference in the size and amount of branch-
ing in different plants of Arabis falcatoria, depending on the site
where they are growing. In open undisturbed marble chip rock, the
plants are mostly 1 dm tall or less, the stems few, unbranched and
slender, and the cauline leaves small and remote. In another area,
where there was disturbance of the chip rock as a result of its being
mined, plants of the disturbed places grew taller, had many more
robust, often branched, stems and cauline leaves that were more am-
ple than in plants from undisturbed sites. Larger plants were also
found in partially shaded crevices of large rocks. In spite of the rather
large range of quantitative differences in plants from the different
sites, the essential distinctive features of A. falcatoria were retained
and the individual plants readily fit within the total spectrum of size
differences.

In the collection from Nevada, Tiehm and Williams 6661, the pu-
bescence is not quite as coarse as in the material from northwestern
Utah. However, in most features, the plants of the two areas are very
similar. Arabis falcatoria has stamens noticeably exceeding the calyx
in length. This is unusual among species of the demissa-oxylobula-
pendulina group. The excerted anther feature in this species deserves
further study in relation to the mechanisms of pollination.

4. Arabis pusilla Rollins, sp. nov. ee,
ibus, erectis vel decumbentibus, 6-12 cm altis; “em

ari-oblanceolatis, 1-1.5 cm longis, 1.5-2 mm l tis,
otis; sepalis oblongis, erectis, nonsaccatis,
+ :

purpureis, ca. 2 mm longis; petalis spathulatis, roseis, 3.5-4. ongis; pedicellis
: . 3

Herba perennis; caulibus filiform
radicalibus erectis, linearibus vel line

glabris, 1-1.5 cm longis, ca. 2 mm ant semini

ca. 1 mm latis; cotylendonibus accumbentibus.
Perennial with ic sian: thickened, mostly nonbranching caudex; stems ~~ gel

slender, slightly decumbent toward base, 6-12 cm tall; basal leaves erect, line

108 REED C. ROLLINS

linear-oblanceolate, petiolate, entire, acute to acuminate, 1-1.5 cm long, 1.5-2 mm
wide; sparsely pubescent with erect 2-3 branched trichomes, rarely ciliate on the

simple or forked trichomes; cauline leaves 3-5, remote, sessile, nonauriculate, 4-8 mm
long; sepals oblong, nonsaccate, erect, purplish, scarious-margined, glabrous or with
w trichomes, 2—2.5 mm long; petals spatulate, erect, light lavender, 3.5—-4.5 mm

~

long; fruiting pedicels widely spreading, straight, glabrous, 2-3 mm long; siliques
widely spreading to slightly ascending, glabrous, acuminate, nearly straight but with
slightly undulating margins, 1-1.5 cm long, ca. 2 mm wide; st solete;
seeds oblong, slightly compressed, wingless or occasionally with a slight distal margin,
ca. 2 mm long, ca. 1 mm wide; radical pointed, slightly exceeding cotyledons; coty-
ledons accumbent.

TYPE: Wyoming, Fremont County, in cracks and crevices of huge metamorphosed
rocks, off Wyoming State Highway 28, 39 miles SW of Lander, 20 Jun 1981, Reed
C. and Kathryn W. Rollins 81366 (holotype: GH; isotypes: to be distributed).

In general appearance, Arabis pusilla is more like A. oxylobula than
any other species. The very slender stems, few cauline leaves and
widely spreading siliques are features they share in common. But the
seeds of A. oxylobula are winged all around and are in a single row
in each silique whereas the seeds of A. pusilla are wingless and are
disposed in a double row. The siliques of A. pusilla are nearly twice
as wide as those of A. oxylobula and the margins are slightly wavy,
not straight as in the latter. Most importantly, the trichomes of the
two species are vastly different. Those of A. pusilla are small, mostly
forked or 3-branched and only a few along the petiole margins are
simple. In contrast, the trichomes of A. oxylobula are large, ascicular
and simple; often the plants are completely glabrous. Arabis oxylobula
is restricted to west central Colorado, where it overlaps to some ex-
tent with A. demissa var. demissa and is wholly separated geograph-
ically from A. pusilla.

Both Arabis pendulina var. russeola and A. demissa var. languida
grow in west central Wyoming, but they are comprised of more ro-
oe erat with much longer and narrower siliques than A. pus-
: a. es have pendulous siliques and arched pedicels while those of

- pusilla are at right angles to the infructescence rachis to slightly
ascending.

- pig ar yb sen Greene, Pl. Baker. 3: 8. 1901. A. rugocarpa Os-
vee out, . orr. Bot. Club 31: 357. 1904. A. aprica Osterhout ex

elson in Coulter and Nelson, New Man. Bot. Rky. Mts. 228. 1909.

5a. Arabis demissa var. demissa

STUDIES ON ARABIS (CRUCIFERAE) 109

but rarely extending somewhat up the leaf-blade, and smaller, more
abundant trichomes on the leaf-blades and margins. In addition to
those specimens previously cited (Rollins 1941), the following are rep-
resentative.

Colorado. Garfield Co.: flat limestone ridge overlooking Colorado River to the E
and Glenwood Canyon to the S$, 29 Jun 1978, Johnston and Lucas 1756 (COLO). Gun-
nison Co.: 1.6 mi. E of U.S. Hwy. 50, 17.2 mi. W of Gunnison, 27 May 1979, R.
C. and K. W. Rollins 7974 (cu); near Woods Gulch road, E of Gunnison, 19 Jun 1979,
Ratzloff 1648 (coLo); 1 mi. SW of Powderhorn store, 22 Jun 1979, Ratzloff s.n. (COLO).
Hinsdale Co.: 3 mi. S of Lake City, 26 Jun 1957, Beaman and Erbisch 1117 (coo,
cH). Lake Co.: Twin Lakes, 1903, Juday B108 (coLo). Ouray Co.: Ouray, E. Bethel
s.n. (RM). Pitkin Co.: Aspen, E. Bethel s.n. (RM). Saguache Co.: E of Cochetopa Creek,
20 Jun 1979, Ratzloff s.n. (COLO).

5b. Arabis demissa var. languida Rollins, Rhodora 43: 388. 1941.

Although definitely caespitose, as in var. demissa, var. languida is
taller, coarser, and with more spreading siliques. The upper cauline
leaves of var. languida have auricles while in var. demissa all cauline
leaves are sessile and without auricles. Some populations of var. lan-
guida, in unusually favorable sites, come close to attaining the size
of the most depauperate populations of A. fendleri. However, the
habitats of the two species are quite different. Arabis fendleri tends
to occur on slopes, cliffs, and rocky places in woodland areas while
A. demissa var. languida is found in open, often windswept ridges,
calcareous slopes and benches. Some previously uncited collections of
var. languida are as follows.

Colorado. Gunnison Co.: dike just west of and parallel to “the Dike,” S ridge of
Ruby Peak, 16 Jul 1980, B. C. Johnston 2832 (COLO, GH). Moffat Co.: Douglas Moun-
tain, 4.5 air km S$ of Graystone, 24 May 1978, Peterson 492 (COLO). Wyoming. Albany
Co.: ca. 2-1/2 mi. SE of Laramie, TI5N, R73W, S13, 23 May 1980, Dorn 2631 (RM);
8 mi. E of Laramie, 10 May 1936, M. Ownbey 1019 (GH). Sweetwater Co.: Aspen
Mt., TI7N, R104W, S17, 3 Jun 1980, Dorn 3440 (rm); Lichvar 2758 (RM). Utah. Uintah
Co.: 10 mi. W of Vernal, 11 Jun 1965, Despain 59 (nM). Duchesne ca. 3.5 mi.
WNW of Duchesne, 16 Jun 1980, Neese and Welsh 8900 (cH). Emery Co.: ca. 9 mi.
E of Huntington, 30 Apr 1968, Higgins and Reveal 1249 (GH).

ARABIS PLATYSPERMA AND RELATED SPECIES

The previously presumed distributional range of Arabis platysperma
for which there has been vouchered evidence, centers in the
Sierra Nevada extending north into southern Oregon and southward
to the San Jacinto Mountains of southern California. In Nevada, it has
been known from the Mount Rose area, which is part of the Sierra
Nevada system. Recent collections extend this species to the White
Mountains of Mono County, California, and Esmeralda County, Ne-
vada. The White Mountain material, although usually with narrower
siliques, is referrable to var. platysperma, not var. howellii — oe
curs at higher elevations in the Sierra Nevada and northward. The

110 REED, © ROLLINS

Nevada Seapriciy were foung by Ann Pinzl of the Nevada State
Museum in Carson City. She pa, also discovered a new species of
Arabis, related to a platysper. "ma, growing in the same area. Both
species grow mostly above 10,409 feet in a bowl on the northeastern
side of Boundary Peak. The ne, species is named as follows.
seni pinzlae Rollins, sp. nov.
erba Sia, caulibus tenuibus, : a esha ti Fg
eae ae pubescentibus ee 3-8 cm longis; foliis radicalibus lineari
perne glabra ain seaulis oblon rpu
thulatis, erectis, purpureis; Godot
pubescentibus, 3-7 mm longis; siliquis
ris, 2-4 cm lon 2-3 mm latis;

bl
: velo Seakeviiccs any ee stigmatibus sessilibus; seminibus late o ongis

n mm latis, cotyledonibus accumben-
Perennial; stems simple, slender, nS or few from a simple or branched caudex, 3-8

em tall, densely pubescent below Pal ely pubescent to glabrous above; basal leaves

erect, greyish, tufted, linear- lanceolate to narrower, tapering to an acute ‘tip, thickish,

pel inrolled, densely ache : rae Wei
—2 mm wide; cauline leave

eaoen pubescent, upper seahaaceak em

wide; sepals non-saccate, oblong, purpl”

wc ephw se oblong, non- sdoriculate, lower

: e, scarious-margined, ca. 3 mm long, ca. 1 mm
wide; petals p rple-tipped, whitish below patulate, erect fy pedicels
erect to divaricately ascending, straight pi oly nisbenoeait © gabrous, 355 ’ g
siliques erect to ascending, linear sles ven marg acuminate, glabrous, purplish
at maturity, 2-4 cm lo —

ins,
seeds in a single row filling width of | Ne valves obscurely veined, 1-nerved near —
a oblong to nearly orbicular, ca ‘ilique, widely winged all around, compressed,
; wings varying in width, narr rowe! . 2.5 mm long, ca. 2 mm wide not including
na fc cotyledons; cotyledor, . ron the sides than distally, up to 0.5 mm wide;
TYPE: Nevada, Esmeralda County, Pg accumbent.
northeast side of Bou ndary Peak, Weer center of bowl, A eee 11,200 ver

Mountains, 12 July 1981, Ann Pinzl 444
pring Cray Herbert, Stipes: pe be distributed cour iho Nevada State Me
Boundary Peak, 10,400 re aa “Toro, smeralda Co.: White Mts., northeast side of
of Bouk Peak. onethind of the «: Ann Pinzl 2397, J. Pinzl (cu); bowl on NE side
1981, pee T. Austin an. Hpk aad down the middle, approx. 11,300 ft., 20 Jul
21 Jul 1981, Ann Pinzl 4448 (cu). al bowl, NE side of Boundary Peak, 10, 800 tt,

In habit, Arabis pinzlae is m
tysperma, although all three s
three erect or ascending siliqu
either of the other species. Thy, tek
flattened and more narrowly
A. pygmaea. But the most fun
zlae as a distinct species is in
chomes on the leaves of A.

bre like A. pygmaea Roll. than A. pla-
hecies are closely related. The one to

of A. pinzlae are narrower than in
ds of A. pinzlae are smaller, less
nged than in either A. platysperma or
amental difference that marks A. pin-
Pein nature of the trichomes. The tri-

STUDIES ON ARABIS (CRUCIFERAE) 111

is dense and abundant, extending from basal leaves to the cauline
leaves, upper stems and even occurring sparsely on the pedicels and
sepals in A. pinzlae whereas at least the upper parts of the plants in
the other two species are glabrous. Typically, the caudex of A. pinzlae
is branched below the ground surface and the elongated tap root ap-
pears to be adapted to moving soil. Remnants of old basal leaves are
often present on the caudex branches that show evidence of having
been buried beneath the soil.

Arabis platysperma, A. pygmaea, and A. pinzlae form a distinctive
group in Arabis. They share features such as erect to divaricately as-
cending siliques, relatively widely winged seeds, broad siliques, and
small sessile non-auriculate cauline leaves. The nearest relative of
these species is A. suffrutescens but that species has reflexed siliques.

At the time of my former treatment, Arabis pygmaea was known
from only two collections. Since then, it has been collected several
times but always in a relatively restricted area of the southern Sierra
Nevada, and so far, not outside of Tulare County, California. I have
seen only three flowers of this species. The sepals are erect, nonsac-
cate, broadly oblong and sparsely pubescent. The petals are white,
erect, more lingulate than spatulate, ca. 4 mm long, and less than
twice as long as the sepals. There are minor differences between some
of the five collections at my disposal. The importance of these dif-
ferences cannot be assessed without a fuller sampling of the species
as a whole.

SPECIES RELATED TO ARABIS COBRENSIS

Aside from the loose branching habit, narrow entire basal leaves
densely pubescent with fine dendritic trichomes characteristic of Arabis
cobrensis M. E. Jones, the nearly orbicular substantially winged seeds
are a distinctive feature of this species. Although somewhat related
to A. puberula Nutt., A. cobrensis has, in the past, stood alone as a
relatively uniform, easily distinguished species. Material from the
Bodie Hills of eastern Mono County, California, collected in 1945
(Rollins 1946), might have cast doubts on the singularity in Arabis of
A. cobrensis had the appropriate connection been made at that time.
But the specimens then referred to as possible hybrids between A.
sparsiflora var. subvillosa and A. fernaldiana var. stylosa were so
different from A. cobrensis that the chance of their being a related
species, not hybrids at all, did not surface. New material of still a
third taxon of this alliance from Elko and Washoe counties, Nevada,
has sparked the necessary study to clarify the situation. Many new
collections of A. fernaldiana var. stylosa (Rollins 1981) are available
and that taxon is much better understood than previously. I do not

112 REED C. ROLLINS

now believe var. stylosa is present in eastern California as previously
indicated (Rollins 1946). It appears to be common but restricted, as
far as our present information shows it, to the mountains of central
and northern Nevada.

Thus delimited, var. stylosa cannot logically be considered parental
to putative hybrids in the Bodie Hills area of Mono County, Califor-
nia, as was presumed to be the case earlier. Furthermore, an exten-
sive sampling of Arabis populations of the Bodie Hills by Tim Messick
has shown that what we name A. bodiensis below is consistent in its
characters and is present on many appropriate sites throughout the
area. This adds up to the necessity of recognizing the Bodie Hills
material as an undescribed species.

The following key gives the readily seen distinctions between Arabis
cobrensis and the two related new species, A. bodiensis and A. fal-
cifructa.

KEY TO THE SPECIES

wing at least 0.25 mm wide

a
QO
5.
3 oO
oe
2
n
oO

1. Arabis cobrensis M. E. Jones.

The known geographic range of this species has changed but little
since my earlier treatment. It has been found in eastern California
(Rollins 1946) but otherwise no significant extensions of range have
been reported. It is surprising that it has not been found in Utah.
Evidently the sweep of occurrence is northward from Mono County,
California, across Nevada to eastern Oregon, then eastward through
southern Idaho to south central Wyoming.

2. Arabis falcifructa Rollins, sp. nov.

Herba =— eeraes caudicibus ramosis; caulibus simplicibus vel sparse ra-
mosis, erectis, 2—4. tis, inferne dense pubescentibus. su if
centibus vel glabratis; foliis ra cot wl Wine ee
latis, canescentibus, 1.5-2.5 cm longis, 2-3
sessilibus, auriculatis, 8-15 mm longis, 1.5-3 mm la
catis, ca. 3 mm longis; petalis albo-purpureis, anguste spathulatis, 5-7 mm longis,
pedicellis fructiferis patentibus vel recurvatis, sparse pubescentibus, 7-10 mm longis;
scent - ha antl patentibus, 4-6 cm longis, ca. 1.5 mm latis: seminibus
oblongis vel late oblongis, anguste alatis vel exalatis, ca. 1.5 is: 00 i
a mm longis; cotyledonibus

STUDIES ON ARABIS (CRUCIFERAE) 113

Perennial with numerous stems and many branched caudices; stems erect, numer-
us, mostly simple, occasionally branched above, densely pubescent below, sparsely
pubescent to glabrate above, 2—4.5 dm tall; basal leaves numerous, erect, entire, linear
to linear-oblanceolate, acuminate, densely pubescent with minute dendritically branched
trichomes, 1.5—2.5 cm long, 2-3.5 mm wide; cauline leaves oblong, acute, sessile,
auriculate, few and remote, 8-15 mm long, 1.5-3 mm wide; sepals erect, oblong,

nonsaccate, sparsely , scarious-margi above, m ; petals erect,
narrowly spatulate, not differentiated into blade and claw, pale purplish, 5-7 long;
stamens excerted above calyx; fruiting pedicels widely arched downward, slender, pu-
bescent, sparsely so or glabrate, 7-10 mm long, slightly swollen toward apex; siliques
widely ar downward, gently curved inward, acuminate, l-nerved be middle,
glabrous, 4-6 cm long; ca. 1.5 mm wide; valves not constricted between seeds, plain;
styles ca. 1 mm long; seeds oblong to broadly oblong, narrowly winged mainly distally
1 :

or the wings obsolete, ca. 1.5 mm long, ca. 1.2 mm wide; cotyledons accumbent.
TyPE: Nevada, Elko County, in crevices of rocks, sagebrush area on slope of a high
ridge, near U.S. Hwy. 93 between Thousand Springs and Jackpot, 37 miles south of
Jackpot, 18 Jun 1979, Reed C. and Kathryn W. Rollins 79267 (holotype: GH; isotypes:
distributed as A. cobrensis).
OTHER COLLECTIONS STUDIED. Nevada. Elko Co.: 1 mile N of San Jacinto, 30 May
1945, Ripley and Barneby 6465 (cu). Lander ee oo Mts., ca. 18 mi. and 315
J

in the case of A. cobrensis vs. A. falcifructa, this can be decisive in
correlation with other characters in pointing to taxa that should be
recognized. The siliques of A. falcifructa are acuminate and tipped
with a style nearly 1 mm long while the siliques of A. cobrensis are
blunt or at most acute and there is either no style at all or if one is
present, it is so short that it can scarcely be seen without considerable
magnification.

3. Arabis bodiensis Rollins, sp. nov.
Herba perennis, multicaulis; caulibus erectis,

sparse pubescentibus vel glabratis, 1.5-3.5 dm ss

tegris, lineari-oblanceolatis, canescentibus, 13 cm longis, cs ig

ili a longis; ped
sessilibus, oblongis, auriculatis, pebensonhs. © 2 Bye sphgge aie paistagereee

atis: seminibus anguste-alatis, oblongis vel

is, C
i s.
obicularibus, ca. 1.5 mm longis; cotyledonibus accumbentibu
ial; caudex usually a many branches; stems several to numerous, erect,
sely pubescent below, s arsely pu-

114 REED C. ROLLINS

of minute dendritically branched trichomes, usually with a few large simple or forked

trichomes near base of petiole, entire, linear to linear-oblanceolate, acute to acumi-

or nearly so, expanded toward apex, 3-6 mm long; siliques widely spreading to slightly
ascending, gently curved to nearly straight, glabrous, faintly 1-nerved below, acute at
pex, 3-5 cm long, ca. 2 mm wide; seeds broadly oblong to nearly orbicular, narrowly
winged, ca. 1.5 mm long; cotyledons accumbent.

TYPE: California, Mono County, in loose soil of old mine, 2 miles northwest of Ma-
sonic, 5.9 miles from Bridgeport-Sweetwater highway, 3 Aug 1945, Ira L. Wiggins an
Reed C. Rollins 536 (holotype: GH; isotype: Ds).

OTHER SPECIMENS STUDIED. California. Mono Co.: small hill adjacent to Masonic Peak
and on Masonic Peak, 4 Aug 1945, Wiggins and Rollins 553 (ps, GH); 2 mi. NW of
Masonic, 5.9 mi. from Bridgeport-Sweetwater hwy., 3 Aug 1945, Wiggins and Rollins
537 (Ds, GH); NE slope of Masonic Peak, ca. 8 mi. NE of Bridgeport-Sweetwater hwy.
4 Aug 1945, Wiggins and Rollins 559 (Ds, Gx); top of cliff across Clearwater Creek

summit of Biedeman, 18 Jun 1979, Messick 515 (Hsc); 5.1 mi. S of Bo e, 20 Jun
1980, Messick 992 (usc); between Ne rk Hill and Masonic Mt., 25 Jun 1980, Mes
sic ear W summit of Masonic Mt., 24 Jul 1981, Messick 1726 (HSC)

Nevada. Mineral Co.: B
1945, Alexander and Kellogg 4449 (cu).

Although Arabis bodiensis is most closely related to A. cobrensis on
the basis of similarity of pubescence, basal leaves and the remoteness
of the cauline leaves, it also has some features in common with A.

ACKNOWLEDGEMENTS. The research upon which this paper is based was supported by
National Scie

LITERATURE CITED

Rotiins, R. C. 1941. A monographic study of Arabis in western North America. Rho-
dora 43: 289-325, 347-411, 425-481,

—————. 1946. Some new or noteworthy North American Cruciferae II. Contrib.
Dudley Herb. 3: 366 '

—————._ 1981. Studies on Arabis (Cruciferae) of western North America. Syst.

Bot. 6: 55-64.