WiM^m

\ovtf ^ - 0\ \&x*

s>

N.C. DOCUMENTS

CLEAPIMGHOUSe

AUG 2 6 1998

r»if. immt 8F NORTH CAROLINA

MUltH

number 25

july 1998

EDITORIAL STAFF

Richard A. Lancia, Editor

Suzanne A. Fischer, Assistant Editor
Stephen D. Busack, Managing Editor

Brimleyana, the Zoological Journal of the North Carolina State Muse-
um of Natural Sciences, appears twice yearly in consecutively numbered issues.
Subject matter focuses on systematics, evolution, zoogeography, ecology, behav-
ior, and paleozoology in the southeastern United States. Papers stress the results
of original empirical field studies, but synthesizing reviews and papers of signif-
icant historical interest to southeastern zoology are also included. Brief com-
munications are accepted.

All manuscripts are peer reviewed by specialists in the Southeast and
elsewhere; final acceptability is determined by the Editor. Address manuscripts
and related correspondence to Editor, Brimleyana, North Carolina State Museum
of Natural Sciences, P.O. Box 29555, Raleigh, NC 27626. Information for con-
tributors appears in the inside back cover.

Address correspondence pertaining to subscriptions, back issues, and
exchanges to Brimleyana Secretary, North Carolina State Museum of Natural
Sciences, P.O. Box 29555, Raleigh, NC 27626.

In citations, please use the full name - Brimleyana.

North Carolina State Museum of Natural Sciences

Betsy Bennett, Director

North Carolina Department of Environment,

Health, and Natural Resources

James B. Hunt, Jr., Governor
Jonathon B. Howes, Secretary

CODN BRIMD 7
ISSN 0193-4406

IN MEMORIAM

Dr. Joshua Laerm, Professor at the University of Georgia and Curator of
Zoological Collections at the University of Georgia Museum of Natural History, died
28 September 1997. He was bom and raised in Pennsylvania and received an under-
graduate degree from Pennsylvania State University and graduate degrees from the
University of Illinois. He joined the faculty at the University of Georgia in 1976.

Dr. Laerm published numerous works in systematics, mammalogy, and nat-
ural history. He was particularly interested in rare and threatened or endangered mam-
mals and contributed significantly to understanding their natural history and distribu-
tion in the Southern Appalachian Mountains. His enthusiasm for science, his prolific
contributions, and his eagerness to help colleagues will be deeply missed.

Turtles (Reptilia: Testudines) Of The Ardis Local Fauna Late
Pleistocene (Rancholabrean) Of South Carolina

Curtis C Bentley and James L. Knight

South Carolina State Museum, 301 Gervais Street,

P.O. Box 100107

Columbia, South Carolina 29202-3107

ABSTRACT- The Ardis local fauna (late Pleistocene) was collected from
a group of interconnecting sediment-filled solution cavities, located in
the Giant Cement Quarry near Harleyville, Dorchester County, South
Carolina. Fossil material from the lowermost levels and the extreme
upper layer of the deposit have been radiocarbon dated at 18,940 ± 760
and 18,530 ± 725 y.b.p., respectively. These dates are considered con-
temporaneous within present resolution. Approximately ninety verte-
brate taxa were collected from the site. Fourteen were species of turtles,
including eight not previously reported from the Pleistocene of South
Carolina. Among these is the southeasternmost occurrence of Emy-
doidea blandingii. This record, in conjunction with other vertebrate fos-
sils from the site, suggests a north-south dispersal route of species along
the Atlantic Coastal Plain during interglacial-glacial transitions. Geo-
graphically isolated eastern and western populations of Emydoidea
blandingii may have existed during the maximum advance of the Lau-
rentide ice sheet. Unusually complete fossils of large box turtles recov-
ered from the site corroborate the previously suggested synonymy of the
extinct Terrapene Carolina putnami with T c. major. The fossil turtle
community of the Ardis local fauna has no modern analogue. Like the
Ardis mammals, it comprises a "disharmonious" fauna which suggests
that, during the height of the Wisconsinan glaciation, the region experi-
enced a more equable climate than that of today .

The Ardis local fauna has yielded approximately 90 species of late
Pleistocene fossil vertebrates from the Coastal Plain of South Carolina, includ-
ing a substantial mammalian (Bentley et al. 1994), avian, reptilian, amphibian,
and fish faunas currently under study. We present data on the Ardis turtle col-
lection (Appendix I), the largest Rancholabrean fauna reported from the state.
Dobie and Jackson (1979) and Roth and Laerm (1980) reported fossils of late
Pleistocene age from Edisto Island, the only other Pleistocene fossil turtle fauna
from South Carolina described to date. Among the Edisto Island fauna were ten

Curtis C. Bentley and James L. Knight

taxa of turtles, including Gopherus sp. and Malaclemys terrapin which were not
recovered from the Ardis local fauna and possibly three species of Pseudemys, P.
floridana and/or P. concinna, and P nelsoni.

The Ardis local fauna was discovered in a large open-pit mine, operat-
ed by the Giant Cement Company, located 5 km NNE of Harley ville, Dorchester
County, South Carolina (33° 14'N, 80° 26'W). Quarry operations exposed San-
tee Limestone (middle Eocene) and the clay-rich Harleyville Formation (late
Eocene) which underlie Plio-Pleistocene surficial deposits (Ward et al. 1979,
Harris and Zullo 1991). Locally, groundwater differentially dissolved the Santee
Limestone in its upper portions, so that many solution cavities contacted and
penetrated the overlying Harleyville Formation and thereby opened several of
the cavities to the Pleistocene surface (Bentley et al, 1994). The radiocarbon
dates of the Ardis material place the time of deposition at or near the height of
the Wisconsinan glaciation (Bowen 1988, Tushingham and Peltier 1993). Fur-
ther discussion on the geology, dating methods of the Ardis fossil material, pre-
vious fossil collections from the quarry, fossil collection procedures, and a local-
ity map are available in prior publications (Bentley and Knight 1993, Bentley et
al. 1994).

TAPHONOMY

At least part of the fossil assemblage collected from inside the solution
cavities at the Ardis site appears to represent an obrution deposit, the very rapid
burial of intact organisms (Brett 1990) in which many of the specimens exhibit
incipient decay. Surface openings leading to the cavities varied from a gentle
downward slope to a vertical shaft, generally allowing the Pleistocene fauna ease
of ingress and egress. This permitted animals to enter the cavities in three dif-
ferent ways: (1) "walk-in" taxa, which may have used the site for
estivation/hibernation or as denning sites and hunting grounds, for example
muskrats, mink, and woodrats (Bentley et al. 1994); (2) "wash-in" taxa from the
surface, either alive or dead, which applies most readily to large animals known
only from isolated remains e.g., Mammut sp., Bison sp., Equus sp. (Bentley et al.
1994), that would have been unable to enter the cavities during life; and (3) "fall-
in" taxa which fell into exposed verticle shafts, fossil accumulations resulting
from this type of natural trap are well documented (e.g., Webb 1974).

Because of the interconnecting "tunnel-like" nature of the cavities, a
single episodic event could produce differing water velocities within the cavities
and different rates of deposition. Seasonal flooding, depending on the intensity,
may have simultaneously smothered living animals within the cavities and
buried or reworked those that had died just prior to, or in a preceding, deposi-
tional event. Consequently, specimens incompletely or shallowly buried during
an event with a low sedimentation rate (low energy) could be completely or par-
tially exhumed and reburied by a succeeding event. This resulted in the preser-

Turtles

vation of specimens in various orientations to the bedding planes (Fig. 1), vari-
ous degrees of disarticulation, and the occasional mixing of individual elements.
An articulated Emydoidea blandingii specimen preserved in its life position with
axial skeletal elements inside, indicates little or no decay prior to its final burial
(Fig. 2). This preservation suggests the turtle was buried quickly in a high ener-
gy, high sedimentation environment (Brett and Speyer 1990), resulting in burial
deep enough to avoid reworking during subsequent episodic events. Several
articulated turtles were collected with limbs and skulls preserved within the
shells in various orientations to the bedding plane. A high energy hydrological
environment before or shortly after death would likely explain the various orien-
tations observed in well preserved specimens. Retention and preservation of
limb elements, cervical and caudal vertebrae, and skulls inside the shell may
reflect a withdrawal by the turtles in response to a catastrophic event.

Fig. 1 Emydoidea blandingii, only the carapace (.547) was preserved, ventral
side up (side view), among clay clasts from the surrounding Harleyville Forma-
tion. This illustrates the hydrodynamic effect upon some specimens prior to final
burial.

Curtis C. Bentley and James L. Knight

Fig. 2 Complete E. blandingii (.546) in situ, with axial skeleton preserved inside
the shell, indicating a "withdrawal response" and final burial prior to any signif-
icant decay.

MATERIALS AND METHODS

Morphological terminology used in this paper is taken from Carr
(1952), Ernst and Barbour (1989), Holman (1967, 1977, 1985), Holman and
Grady (1987), and Preston (1979). Taxonomy follows Conant and Collins
(1991).

Morphological comparisons of Recent skeletons to fossil material were
made against available specimens in the Florida Museum of Natural History and
the South Carolina State Museum collections. Additionally, specimens from The
University of Michigan Museum of Zoology of Emydoidea blandingii, UMMZ
155047-155054, Clemmys guttata, UMMZ 51235, 51236, 51240-51242,
159219, 155001, 155002, and Clemmys muhlenbergii, UMMZ 77140 and,
130840 were studied.

Most of the specimens in the South Carolina State Museum collections
are deposited under the base number of S.C. 94.10. and for brevity, are refer-
enced in the text only by the digits following this base number. Specimens
accessioned separately are designated by the institutional prefix of SCSM. Fos-
sil specimens deposited in the National Museum of Natural History and the
Florida Museum of Natural History are designated by USNM and UF, respec-
tively.

Turtles

SYSTEMATIC PALEONTOLOGY

Testudines
Kinosternidae
Material: 1 right xiphiplastron (.25); 16 peripherals (.26-.44); 2 humeri (.45-. 46);
2 partial jaw rami (.755-. 756).

Remarks: These fossil elements could only be identified with confidence to fam-
ily.

Kinosternon subrubrum - Eastern Mud turtle (Lacepede, 1788)

Material: 3 nuchals (.11 -.13); 2 right, 3 left hyoplastra (.6-. 10); 5 left hypoplas-
ia (.1-.5).

Characters used for identification: The hyoplastron of Kinosternon subrubrum
can be separated from other North American Kinosternon and Sternotherus
because the axillary notch is narrower, and from Kinosternon baurii because the
axillary notch is wider and shallower (Holman 1985). K. subrubrum hyo- and
hypoplasia differ from Sternotherus odoratus in that the elements are shorter lat-
erally than medially in S. odoratus (Preston 1979). Characters used to identify
nuchal material are discussed by Holman (1975). In addition, nuchals of K. sub-
rubrum can be distinguished from nuchals of S. odoratus because the anterior lip
of the nuchal, viewed anteriorly, is nearly straight in K. subrubrum. Nuchals of
S. odoratus, viewed anteriorly, have a decided arc.

Remarks: The eastern mud turtle inhabits a variety of shallow slow to non-mov-
ing bodies of water with a soft substrate, such as swamps, ponds, marshes, wet
meadows, and lagoons (Ernst and Barbour 1989). Kinosternon subrubrum today
ranges from southern Massachusetts and Pennsylvania along the Atlantic coast,
to the tip of Florida and west into Texas and Oklahoma (Conant and Collins
1991). K. subrubrum is common in the area of the Ardis site today and may be
sympatric with K. baurii (Lamb and Lovich 1990).

This is the first report of this species from the fossil record of South
Carolina. Dobie and Jackson (1979) and Roth and Laerm (1980) both reported
the same single pygal bone from Edisto Island as "Kinosternon sp."

Sternotherus odoratus - Common Musk turtle (Latreille, in Sonnini and

Latreille, 1802)

8 Curtis C. Bentley and James L. Knight

Material: 3 nuchals (.22-.24); 2 right hyoplastra (.14-. 15); 1 right, 5 left
hypoplasia (.16-. 21).

Characters used for identification: Identification is based on characters provid-
ed in discussion for Kinosternon subrubrum. The hyo-hypoplastron of S. minor
can be separated from the same elements in S. odoratus because the area that
forms the bridge between the plastron and the carapace is dorsally compressed
or flattened in S. minor, and not raised as in S. odoratus. The nuchals compare
most favorably to S. odoratus, following the characters used above, and addi-
tionally exhibit a strong dorsal medial keel that is generally lacking in K. sub-
rubrum.

Remarks: The common musk turtle inhabits areas very similar to that of K. sub-
rubrum, preferring slow to non-moving bodies of water with a soft bottom. It
has also been collected from fast moving, gravel bottomed, streams (Ernst and
Barbour 1989). S. odoratus occurs from southern Maine and Canada southward
through Florida and as far west as Kansas and central Texas (Conant and Collins
1991), and occurs in the area of the Ardis site today.

Chelydridae
Chelydra serpentina - Snapping turtle (Linnaeus, 1758)

Material: 2 right parietals (.50, .53); 1 left postorbital (.51); 1 right prefrontal
(.52); 1 left quadratojugal (.54); 1 partial left mandible (.55); 2 right mandibles
C56-.57); 6 cervical vertebrae (.102-. 107); 1 humerus (.63); 2 radii (.69-.70); 1
right scapulo-acromial process (.64); 1 right partial acromial process (.65); 5
femora (.58-.62); 3 ilia (.66-.68); 1 caudal vertebra (.108); 1 nuchal (.95); 1 right
1st peripheral (.71); 16 unassigned peripherals (.72-.83)(2 USNM)(2 UF); 1 left
1st costal (.84); 24 partial costals (.85-.92)(8 USNM)(8 UF); 2 associated costals
C93-.94); 2 neurals (.96-.97); 2 epiplastra (.100-. 101); 2 hypoplasia (.98-.99).

Characters used for identification: Chelydra shell material is very distinctive and
easily separated from other turtles including Macroclemys. Preston (1979) pro-
vides characters that allow the identification of fragmentary material. All listed
fossil elements compare favorably to Recent skeletal materials.

Axial and appendicular skeleton - The large size and diagnostic orna-
mentation of the Chelydra skull roof elements distinguish them from all other
turtles. Macroclemys lacks the rugose cranial ornamentation of Chelydra. A
pectoral girdle was assigned to this species based on the 90° angle between the
scapula and acromial process and on the heavily striated distal ends (Holman
1966). Femora and humeri could not be separated from Recent material of C.
serpentina, and are more robust than other genera of fresh water turtles (Holman
1964) except Macroclemys, which is generally considerably larger.

Turtles

Remarks: Snapping turtles are generally found in freshwater to brackish water
habitats with soft muddy substrate (Ernst and Barbour 1989) from eastern Cana-
da through the United States east of the Rockies south through Mexico and into
Ecuador. C. serpentina occurs today in the Ardis area.

Dobie and Jackson (1979), first reported fossil material of C. serpenti-
na from Edisto Island with additional material reported by Roth and Laerm
(1980).

Macroclemys temminckii - Alligator snapping turtle (Gray, 1855)

Material: 1 partial right parietal (.109).

Characters used for identification: The fossil parietal is identical to Recent spec-
imens of this turtle, differing from C. serpentina in that the dorsal surface is
smooth, i.e. without any of the prominent ornamentation consistently found in C.
serpentina (Fig. 3). The parietal of M. temminckii is generally more robust and
is longer with respect to width than specimens of C. serpentina of comparable
sizes. This was the only fossil element of this species collected from the site.
Because this represents a significant range extension, assignment to this species
was made only after exhaustive comparisons to the fossil and Recent collections
at the Florida Museum of Natural History and the South Carolina State Museum
negated all other possibilities.

Remarks: This is the largest freshwater turtle in North America and possibly the
heaviest in the world (Ernst and Barbour 1989). The alligator snapping turtle
often can be found in the deep waters of lakes, ponds, rivers and bayous that con-
tain abundant aquatic vegetation and muddy bottoms. This turtle is highly aquat-
ic and ranges westward from northern Florida into Texas along the Gulf Coast
and thence northward up the Mississippi Valley into Illinois, Iowa and Kansas
(Ernst and Barbour 1989).

This is the first fossil or Recent evidence of Macroclemys from South
Carolina and the Atlantic Coastal Plain.

Emydidae

Emydinae

Chrysemys picta - Painted turtle (Schneider, 1783)

Material: An individual specimen consisting of a complete carapace (missing the
3rd right marginal) and plastron (.110); 7 cervical vertebrae (.110.1 -.110.7); 1
ulna (.110.16); 1 radius (.110.17); 4 phalanges (.110.18-.110.21); 1 ungual
(.110.22); 2 partial scapulo-acromial processes (.11 0.11 -.11 0.1 2); 2 coracoids

10

Curtis C. Bentley and James L. Knight

Fig. 3 Right parietals of Macroclemys temminckii and Chelydra serpentina. A)
Recent M. temminckii. B) Ardis fossil. C) Recent C. serpentina.

3 cm

(.1 10.13-.1 10.14); 2 dorsal vertebrae (.1 10.9-. 1 10.10); complete pelvic girdle
(.110.15); 1 caudal vertebra (.110.8). An individual specimen consisting of a
nearly complete carapace and plastron (.111). An individual specimen consist-
ing of a complete carapace (missing left 8-9th peripherals and 6th neural) and
plastron (.112); partial skull and mandible (.112.1); partial hyoid process
(.11 2.2); 7 cervical vertebrae (.112.20-. 11 2.26); 2 humeri (.11 2.5-. 11 2.6); 2 ulnae
(.112.7-.112.8); 2 radii (.112.9-. 112. 10); 2 scapulo-acromial processes (.112.15-
.112.16); 2 coracoids (.112.17-.112.18); 2 femora (.112.3-.112.4); 2 tibiae
(.112.11-.112.12); 2 fibulae (.112. 13-. 112. 14); 7 metapodial elements (.112.42-
.112.48); 31 phalanges (.112.49-. 112-80); 13 unguals (.112.81-.112.91); 2 sacral
ribs (.112.92-. 112.93); partial pelvic girdle (.112.19); 3 dorsal vertebrae (.112.27-
.112.29); 12 caudal vertebrae (.112.30-. 11 2.41). An individual specimen con-
sisting of a nearly complete plastron and attached 4-7th right peripherals (.113);
skull (including inner ear ossicles) and mandible (.113.1); partial hyoid appara-
tus (.113.2); 4 cervical vertebrae (.113-.9-.113.12); 2 humeri (.113.3-.113.4); 1
scapulo-acromial process (.113.7); 2 partial femora (.11 3.5-. 11 3.6); 1 tibia
(.113.8); 2 metapodial elements (. 1 1 3. 15-. 1 13. 16); 15 phalanges (.113.17-
.113.31); 5 unguals (1 13.32-. 1 13.36); 2 caudal vertebrae (. 1 13. 13-. 1 13. 14).
Specimen with partial carapace (.114); 3 cervical vertebrae (.114.3-. 114.6); 1

Turtles

11

humerus (.114.1); 1 radius (.114.2); 1 phalange (.114.7). 2 mixed specimens
both with partial fragmented carapaces and plastra (.115); 3 cervical vertebrae
(.115.5-. 115.6); 2 humeri (.115.1 -.115.2); 2 scapulo-acromial processes (.115.3-
.113.4); 2 coracoids (. 1 15.8-. 1 15.9); 2 tibiae (.115.10-.115.il); 1 ilium (.115.12);
1 phalange (.115.13). Four individual specimens with fragmented carapace and
plastron (.116-. 119). Two mixed specimens with badly fragmented carapaces
and plastra (.120).

Isolated elements: 16 nuchals (.123-.132)(3 USNM)(3 UF); (SCSM
91.170.1) partial plastron; 10 right and 2 left epiplastra (.133-. 144); 2 entoplas-
tra (.169-.170); 10 left and 3 right hyoplastra (.145-.151)(3 USNM)(3 UF); 1
right hypoplastron and xiphiplastron (.122); 5 left and 7 right hypoplasia (.152-
.157)(3 USNM)(3 UF); 6 left and 9 right xiphiplastra (.158-.168)(2 USNM)(2
UF); 8 right and 6 left 2nd costals (.227-.240); 3 right and 1 left 3rd costals (.241-
.244); 6 right and 3 left 4th costals (.245-.253); 7 right and 8 left 5th costals
C254-.268); 4 right and 3 left 6th costal (.269-.275); 4 right and 1 left 7th costal
(.276-.280); 56 peripherals (.171-.226); 1 mandible (.761); 2 right and 1 left
mandibular rami (.758-.760); 10 humeri (.289-.298); 8 femora (.281-.288).

Characters used for identification: Identification of complete shells was based
on nuchal characters (Bentley and Knight 1993), and the alignment of the verte-
bral and pleural sulci (Ernst and Barbour 1989) (Fig. 4).

Hyoplastron - The humeral sulcus does not cross dorsally over the plas-
tral scute overlap area, as in Clemmys. Terrapene and Emydoidea have hinged
plastra. Elements of comparable size can be separated from Deirochelys reticu-
laria as the dorsal scute overlap area in C. picta is wider and more sharply curved

Fig. 4 Fossil carapace of Chrysemys picta (.110) from the Ardis local fauna.

1 2 Curtis C. Bentley and James L. Knight

distally, and the articulating surface of the epiplastron and hyoplastron between
the entoplastron and the outside edge is wider in C. picta. Trachemys and
Pseudemys are larger and more robust than C picta as adults. Young specimens
of Trachemys and Pseudemys can be separated by a less pronounced or inflated
scute overlap area and signs of incomplete ossification.

Hypoplastron - The inguinal sulcus runs diagonally to the peripherals
and into the inguinal notch in Chrysemys but is parallel to the peripherals in
Clemmys. This element can be distinguished from Deirochelys as it is less elon-
gate with respect to width in C picta, and the scute overlap area is wider. It can
also be separated from Terrapene Carolina and Emydoidea by the lack of a hinge.
Adult Trachemys and Pseudemys differ in being larger and more robust than
adults of C. picta. Young specimens of Trachemys, comparable in size to adult
C. picta and completely ossified, can be separated from C. picta by a larger
bridge with respect to the hypoplastron proper and a greatly reduced or absent
epidermal attachment scar.

Entoplastron - The humero-pectoral sulcus does not cross the entoplas-
tron as in eastern species of Clemmys and in Terrapene Carolina. It can be tenta-
tively separated from Pseudemys, Trachemys, and Emydoidea by overall size, as
specimens of the preceding genera tend to exhibit incomplete ossification when
of comparable size to adult C. picta. However, size alone is not a reliable char-
acter for this element. We were unable to separate this element from that of
Deirochelys, so entoplastra are only tentatively assigned to C. picta.

Epiplastron - This element differs from other species (except Trache-
mys) in that the anterior edge exhibits a degree of serration. This element often
is serrated in specimens of Trachemys, but the size of these specimens allows
easy separation from C. picta. Young specimens of Trachemys generally lack
this serration and have a poorly developed scute overlap area in comparison to
C. picta of comparable size.

Xiphiplastron - This element can be separated from Clemmys, Ter-
rapene, and Emydoidea by the scute overlap area, which in those genera is more
pronounced than in C. picta. Further, Clemmys muhlenbergii has a posterior
edge tapered to a point, while in Clemmys insculpta the element is longer with
respect to width, with a pronounced notch where the anal sulcus wraps over the
edge, a condition minimal or lacking in C. picta. In C picta the scute overlap of
this element is wider and more pronounced on the posterior edge than on any
examined specimens of Deirochelys. This element in Trachemys is generally
more robust in adult specimens than in C. picta, and in young specimens of com-
parable sizes the scute overlap area is much less pronounced than that of C picta.

2nd Costal - This element differs from C guttata in being approxi-
mately 30% wider with respect to length, and the junction between the 2nd ver-
tebral sulcus and the 1st and 2nd pleural sulcus is located generally more distal-
ly than in C. guttata. This element differs from C. muhlenbergii by having a

Turtles 1 3

greater curvature, and the above mentioned junction point forms a distinct "T"
shape in C. muhlenbergii and dips into a general "U" or "V" shape in C. picta.
The 2nd costal of C. picta differs from C. insculpta by being completely smooth
and lacking any "tortoise-like" bulges, by being more distally flared, and by hav-
ing greater curvature of the element. It can be separated from Trachemys and
Pseudemys by a lack of sculpting and smaller size, and from Deirochelys by lack
of sculpting, a more proximal rib attachment (Jackson 1978), and the proximal
tapering of the element.

3rd Costal - Differs from Clemmys, Terrapene, Trachemys, Pseudemys,
Emydoidea, and Deirochelys in that it lacks the vertebral sulcus between the 2nd
and 3rd vertebral scutes. Additionally, in Deirochelys the rib attachment is sig-
nificantly more distal than in C picta. This 3rd element can be separated from
the 5th costal in C. picta in that it lacks the posteriorly directed curvature of the
5th costal.

4th Costal - Can be separated from all other emydid turtles in that it
exhibits alignment of the sulci between the 2nd and 3rd vertebral and pleural
scutes. This sulcal alignment occurs only in the subspecies C picta picta
(Conant and Collins 1991).

5th Costal - Can be separated from Clemmys, Trachemys, Deirochelys,
Pseudemys, Emydoidea, and Terrapene by characters given for the 3rd costal.

6th Costal - Differs from Clemmys, Trachemys, Deirochelys, Pseude-
mys, Emydoidea, and Terrapene in that it lacks the sulcus of the 3rd and 4th ver-
tebral scutes.

Peripherals, femora, humeri, and mandibular rami - These elements are
tentatively assigned to this species as they compare most favorably to Recent and
fossil material of C. picta.

Remarks: The painted turtle has a wide distribution, occurring from southern
Canada south into Mexico and across the entire continental United States (Ernst
and Barbour 1989). Lakes, ponds, and streams are typical habitats of the paint-
ed turtle. Slow to non-moving, shallow aquatic environments with soft bottoms
are favored.

The completeness of many of the painted turtles recovered from the
Ardis site are strong indicators of an obrution deposit. Chrysemys picta occurs
in the Piedmont and mountains of South Carolina today but does not inhabit the
Ardis site or any other part of the Coastal Plain. This is the first fossil record
from South Carolina.

Clemmys guttata - Spotted turtle (Schneider, 1792)

Material: An individual specimen consisting of a nearly complete shell (SCSM
93.90.1) missing only the right hypoplastron and xiphiplastron, figured in Bent-

1 4 Curtis C. Bentley and James L. Knight

ley and Knight (1993), and the following associated cranial and postcranial ele-
ments; articulated skull fragment (prefrontal, frontal, postorbital, parietal, and
supraoccipital), right maxilla, both quadrates, both opisthotics, basisphenoid, and
articulated lower jaw, 1st cervical vertebra, 2 humeri, 1 ulna, 2 coracoids, 1 ischi-
um, partial pubis and ilium, 1 sacral rib, 1 fibula, 3 phalanges, and vertebrae
fragments. An individual specimen consisting of a partial carapace and plastron
missing only the left hypoplastron and xiphiplastron (.299). Isolated skull frag-
ment (parietal, supraoccipital, both maxillae, basisphenoid, 1 quadrate, 1 postor-
bital, 1 partial squamosal) (.299.1). A single sub-adult individual with partial
plastron and 2 peripherals (.12 1.1 -.12 1.2).

Isolated elements: 17 nuchals (SCSM 93.90.2-.8)(5 USNM)(5 UF); 3
right and 2 left 2nd costals (.334-.338); 6 right and 5 left 3rd costals (.339-349);

6 left and 6 right 4th costals (.350-.361); 3 right and 2 left 5th costals (.362-366);

7 right and 4 left 6th costals (367-377); 1 left 7th costal (378); 34 peripherals
(300-333); 5 right and 3 left epiplastra (379-386); 3 entoplastra (.417-.419); 5
right and 13 left hyoplastra (387-399,.432)(2 USNM)(2 UF); 3 left hypoplasia
(.400-.402); 6 right and 10 left xiphiplastra (.403-.416)(l USNM)(1 UF); 1
mandible (.757); 3 humeri (.426-.428); 6 femora (.420-.425).

Characters used for identification: Identification of the two most complete spec-
imens is discussed by Bentley and Knight (1993).

Nuchals - See Bentley and Knight (1993), for characters used to distin-
guish this from other possible identifications.

Epiplastron - Differs from C. picta in that the scute overlap area is more
robust in C. guttata, and the anterior edge is not serrated as is common with C.
picta. It can be separated from C. muhlenbergii because the bulbous area where
the gular sulcus wraps onto the scute overlap is usually considerably wider medi-
ally to laterally in specimens of C. guttata, whereas the scute overlap portion that
is posterior to the bulbous area is narrower in C. muhlenbergii than C. guttata.
The length of the scute overlap posterior to the gular sulcus is longer than that of
C. picta. Also, the epidermal attachment scar in C. muhlenbergii is deeply
incised and tends to undercut the scute overlap area. The epiplastron of C. gut-
tata is rarely incised to this extent. The epiplastra of C. insculpta differ from C.
guttata in that the bulbous area where the gular sulcus crosses the dorsal surface
is only slightly or not at all bulbous in specimens of similar size. The epiplas-
tron of Deirochelys, Trachemys and Pseudemys that fall within the size range of
C. guttata have less pronounced scute overlap area compared to that of C. gutta-
ta. E. blandingii specimens of comparable size show sub-adult traits (incomplete
ossification) and have a less pronounced scute overlap area. Terrapene epiplas-
tra differ in that the area posterior to the scute overlap is concave, forming a
depression posteromedially to the gular sulcus, generally absent in C. guttata.

Turtles 1 5

Hyoplastron - Differs from Terrapene and Emydoidea in that C. gutta-
ta lacks the hinge components. C. muhlenbergii differs slightly from C. guttata
in that the scute overlap is narrower in C. muhlenbergii. Specimens of Trache-
mys and Pseudemys that fall within the size range of C. guttata have scute over-
laps that are greatly reduced in comparison to C. guttata and the elements exhib-
it incomplete ossification. Deirochelys and Chrysemys picta can be separated
from Clemmys guttata because the distance between the entoplastron and
hypoplastron is ca. 40% greater in adults of the former two genera. Also, in C.
picta, the humeral sulcus does not cross dorsally over the scute overlap area. C.
insculpta exhibits incomplete ossification when elements fall within the size
range of C. guttata.

Hypoplastron - This element can be separated from other genera of
emydid turtles by the lack of hinge components (separating it from Terrapene
and Emydoidea), or by its posterior width being greater than its length (separates
it from Trachemys and Pseudemys). Holman (1977) gives characters used to sep-
arate this element from C. muhlenbergii and C. insculpta.

Xiphiplastron - Separation of this element from C. picta is listed under
that species. It can be separated from C. muhlenbergii and C. insculpta in that the
posterior edge is generally squared off rather than tapering to a point, as in the
other two species. However, some specimens of C. guttata do exhibit a pointed
condition. These still can be separated from C. muhlenbergii because the area
where the abdominal muscle attaches to the xiphiplastron is more pronounced.
C. insculpta can also be separated from C. guttata because, in the area where the
anal sulcus crosses onto the scute overlap area, the xiphiplastron is deeply
notched, being greatly reduced or lacking in C. guttata.

Entoplastron - In C. guttata the humero-pectoral sulcus crosses the
entoplastron within the anterior half of that element. In C. muhlenbergii the
humero-pectoral sulcus may cross the entoplastron at its posterior extremity, but
typically it does not cross the entoplastron at all (Bentley and Knight 1993).
These fossil entoplastra are tentatively assigned to C. guttata, and not C. insculp-
ta, because this element is generally more robust in C. insculpta and the humero-
pectoral sulcus crosses the entoplastron more posteriorly in C. insculpta than in
C. guttata.

Costal - Characters used to differentiate costal elements from C. picta
are described in that section. Costals of Clemmys guttata differ from C. muh-
lenbergii in having substantially more curvature. The dorsal surface of costals in
C. guttata is smooth, lacking any exterior bulges or sculpturing common to adult
C. muhlenbergii, C. insculpta, Deirochelys, Trachemys, and Pseudemys. The
costals of Emydoidea, Trachemys, and Pseudemys exhibit immature traits when
they are within the size range of C. guttata. Terrapene has deeply incised sulci,
and its elements tend to be more acutely angled proximally and exhibit "bul-
bous" sculpturing.

1 6 Curtis C. Bentley and James L. Knight

Peripherals, femora, humeri, and mandible - These elements are tenta-
tively referred to this species because they compare most closely to Recent and
fossil C guttata.

Remarks: The spotted turtle ranges from northern Illinois into Ohio and Ontario,
east to Maine and New York, and south along the Atlantic Coastal Plain into
northern Florida (Conant and Collins 1991). Clemmys guttata occurs most com-
monly in bogs or marshy pastures, but it also can be found in woodland streams.
It favors habitats with soft substrates. C. guttata is frequently found away from
water, but even so it is the least terrestrial of the three eastern species of Clem-
mys (Ernst and Barbour 1989). The fossil spotted turtle remains from the Ardis
local fauna represent the oldest known material of this species (Bentley and
Knight 1993), and the first fossil record for the eastern United States. Holman
(1990) reports a right epiplastron from a 6,000-year-old fauna near Lansing,
Michigan. Interestingly, Ernst and Barbour (1989) noted a relationship between
this turtle and the burrows of muskrats, which the turtles apparently use for esti-
vation and hibernation sites. Fossil muskrats were the most common mammals
found at the Ardis site (Bentley et al. 1994) and are believed to have used the
solution cavities as burrow sites. This may help to explain the abundance of this
turtle at the Ardis site.

Clemmys muhlenbergii - Bog turtle (Schoepff, 1801)

Material: An individual consisting of a partial carapace (nuchal, 1st and 2nd left
costals and peripherals, 2 peripherals, numerous shell fragments) and plastron
(both epiplastra, and partial hyoplastron) (.429).

Isolated elements: 2 nuchals (.430-.431) ; 2 right epiplastra (.433-.434).

Characters used for identification: C. muhlenbergii fossils were distinguished
from other emydid turtles based on characters listed in previous sections, along
with additional nuchal and sulci characters given by Bentley and Knight (1993)
(Fig.5).

Remarks: The soft bottoms and slow moving waters of swamps, bogs and
marshes are typical aquatic habitats of the bog turtle (Ernst and Barbour 1989),
but this turtle can also be found on land. Clemmys muhlenbergii has a patchy
distribution in the Northeast, and ranges as far south as northern Georgia and
extreme northwestern South Carolina. This disjunct spatial pattern has been
interpreted as suggesting a larger former range (Smith 1957). The fossil evi-
dence from the Ardis site suggests that the species' range extended at least 250
km farther southward during the late Pleistocene.

Turtles 1 7

This is the second report of fossil material of C. muhlenbergii (Holman
1977) and represents the first fossil record from the eastern United States south
of Allegany County, Maryland. This is the first sympatric occurrence of C. muh-
lenbergii and C. guttata in the fossil record.

Fig. 5 Partial fossil carapace of Clemmys muhlenbergii (.429). from the Ardis
local fauna.

0 3 cm
1 1 1 1

Terrapene Carolina major- Gulf Coast Box turtle (Agassiz, 1857)

Material: An individual male specimen consisting of a partial carapace (SCSM
91.165.1) lacking it's anterior edge (Fig. 6), complete plastron (SCSM 91.165.2),
partial skull (SCSM 91.165.19), 8 cervical vertebrae (SCSM 91. 165. 10-. 17), 1
humerus (SCSM 91.165.5), both scapulo-acromial processes (SCSM 91.165.6-
.7), both coracoids (SCSM 91.165.8-.9), complete pelvic girdle (SCSM
91.165.3), 1 femur (SCSM 91.165.4), 1 sacral rib (SCSM 91.165.18). An indi-
vidual female specimen consisting of a complete carapace (SCSM 91.163.1) and
plastron (SCSM 91.163.2), 2 cervical vertebrae (SCSM 91.163.10-.il), both
scapulo-acromial processes (SCSM 91. 163. 4-. 5), both coracoids (SCSM
91.163.8-.9), both femora (SCSM 91.163.6-.7), and complete pelvic girdle
(SCSM 91.163.3). An individual female specimen consisitng of a complete
carapace and plastron (SCSM 9 1.1 64.1 -.2), 2 cervical vertebrae (SCSM
91.164.9-.10), 1 humerus (SCSM 91.164.8), 1 scapulo-acromial process (SCSM
91.164.4), both coracoids (SCSM 91.164.5-.6), 1 femur (SCSM 91.164.7), com-
plete pelvic girdle (SCSM 91.164.3), and 1 caudal vertebra (SCSM 91.164. 11).
An individual male specimen with complete carapace (SCSM 91. 166. 1) and pos-
terior half of plastron from hinge (SCSM 91.166.2). An individual female spec-

1 8 Curtis C. Bentley and James L. Knight

imen consisting of a partial carapace and one half of posterior plastron from the
bridge (SCSM 91.168.1). An individual female specimen consisting of a com-
plete carapace and plastron (.435-.435.1). A individual partial carapace (SCSM
91. 167.1).

Isolated elements: 8 nuchals (.459-.466); 2 right 1st costals (.471 -.472);
1 fused left 7th and 8th costal (.473); 5 costals (.474-.478); 3 left 5th peripherals
(.479-.481); 2 fused peripherals (1 USNM) (1 UF); 32 peripherals (.510-.541)(2
USNM)(2 UF); 2 complete, 10 partial anterior plastral lobes (.467-.469)(6
USNM)(3 UF); 1 right epiplastron (.470); 2 entoplastra (.508-.509); 2 complete,
9 partial posterior plastral lobes (.507)(5 USNM)(5 UF); 32 large shell fragments
(.510-.541); 1 partial skull (.436); 1 right maxilla (.437); 3 postorbitals (.438-
.440); 4 mandibles (.441-.444); 3 cervical vertebrae (.456-.458); 4 humeri (.445-
.448); 4 femora (.449-.451); 4 ilia (.452-.455).

Characters used for identification: The more complete specimens are easily sep-
arated from other emydid turtles based on their hinged plastra and overall mor-
phology. Emydoidea differs from the box turtle by its smooth, unkeeled carapace
and tends to be anteriorly constricted (Holman 1985).

Plastron - The hinged plastral elements prevent confusion with any
other emydid of North America except E. blandingii. The anterior end of this
attachment area between the carapace and plastron differs from E. blandingii, as
the carapace and plastron of Terrapene have a heavily sutured interlocking pro-
trusion and pocket respectively. In Emydoidea this area lacks the sutured "ball
and socket" mechanism and instead has a pronounced lateral flare generally
located on the 5th marginal. The large size and robust nature of the fossil ele-
ments suggest an affinity to this subspecies.

Peripherals - These elements are distinguished by an upwardly curved
anterior edge, forming, in some specimens, a "gutter-like" effect.

Humeri, femora, and ilia - Humeri and femora were separated from
other emydid turtles based on comparison to Recent specimens and characters
provided by Holman (1967, 1975). Ilia of T. Carolina have a distinctive
"boomerang-shape" and are straighter in other species (Holman 1977). Addi-
tionally, these fossil elements were identical to those retrieved from within the
shells of the more complete fossil T. c. major collected from the Ardis site.

Remarks: This is the largest of the North American box turtles and today ranges
from the coast of eastern Texas eastward along the Gulf coast into the Florida
panhandle (Carr 1952). The Gulf Coast box turtle is commonly found in marsh-
es, palmetto-pine forests, and upland hammocks. It enters water with a frequen-
cy similar to T. c. Carolina (Carr 1952, Conant and Collins 1991).

Turtles 1 9

Fig. 6 Partial fossil carapace of Terrapene Carolina major (SCSM 91.165.1)
which contained the skull (SCSM 91.165.19) shown in figure 6d.

Large fossil box turtle remains have generally been referred to as T. c. putnami
or T. c. putnami x major (Milstead 1969). T. c. putnami differs from T. c. major
only in size, attaining lengths upwards of 300 mm (Auffenberg 1967, Milstead
1969). The largest Recent specimen of T. c. major on record has a carapace
length of 216 mm (Conant and Collins 1991). Auffenberg (1967) reported a
large box turtle (233 mm), with skull, from Haile 8A, stating that it was very sim-
ilar to T. c. major. Blaney (1971) placed T. c. putnami in synomyny with T. c.
major, based on the shared characters of the two subspecies, and stated that size
alone was not a justifiable reason to recognize a subspecies. The fossil box tur-
tles from the Ardis site exhibit all the characters Milstead (1969) used to distin-
guish T. c. putnami. Furthermore, the five specimens had carapace lengths of
190.0 mm to 260.0 mm. Shell and axial elements of fossil box turtles from the
Ardis site could not be consistently distinguished from Recent specimens of T. c.
major except by size. Milstead (1969) stated that populations of T. c. Carolina in
Massachusetts and Michigan, on the northwestern edge of the subspecies range,
appear to have strong morphological affinities to T. c. major having average cara-
pace lengths of 140 mm and 139 mm respectively. Milstead suggested that this
relationship may be due to a pre-Wisconsin influence of T. c. putnami or the
influence of T. c. triunguis. An isolated, fused posterior plastral lobe (82.26 mm)
collected from the Ardis site was estimated to belong to a specimen with a cara-
pace length of about 140-145 mm. This plastral lobe might indicate the presence

20 Curtis C. Bentley and James L. Knight

of box turtles at or near the size of the northwestern populations discussed by
Milstead (1969). Additionally, several significantly smaller isolated peripherals
were collected from the Ardis site, but these peripherals are largely unfused and
may represent juveniles. The association of the smaller fused plastral lobe with
significantly larger specimens suggests that during the height of the Wisconsin
glaciation there may have been intergradation between local and northerly dis-
placed populations of T. c. Carolina, and populations of T. c. major, T. c. triun-
guis, and T. c. baud radiating from the south. The Ardis population, being pre-
dominantly large box turtles, suggests that T. c. major traits (very large size,
elongated shells, and upward curvature of peripherals) were more predominant
during this time period.

The Ardis site produced two partial fossil skulls; SCSM 91.165.19 asso-
ciated with a carapace of 245-250 mm. in length, and (.436), an isolated partial
skull comparable in size to SCSM 91.165.19 (Fig. 6 and Fig. 7). We failed to
distinguish any consistent differences between our skulls and Recent specimens
except for size and an exaggerated upward curvature of the supraoccipital crest
in specimen SCSM 91.165.19. This extreme curvature is considered an anom-
aly due to its complete absence in other fossil specimens; however, it is noted in
Recent specimens but is less developed. The supraoccipital is thought to be one
of the most variable cranial elements for box turtle systematics (W. Auffenberg,
University of Florida, personal communication).

The parietals of most Recent specimens examined of Terrapene Caroli-
na had parietals that were inflated anteriorly, with a reduction in the tabled, dor-
sal surface of this element. Although extremely preliminary, we suggest a cor-
relation between size and the degree of parietal inflation in T. c. major. Anteri-
or inflation of the parietals is greatly reduced to absent among the largest speci-
mens. In other subspecies of T. Carolina, the inflation of the parietals remains
fairly constant. The two skulls from the Ardis site do not exhibit anterior infla-
tion of the parietals. The morphological similarities between the fossil box tur-
tles of the Ardis site and the Recent and fossil specimens examined from muse-
um collections (Fig. 7) suggests that the greatest affinity of the Ardis specimens
is to T. c. major. They support the synonymy of T. c. putnami with T. c. major
(Blaney 1971). Affinities noted by Milstead (1969) between northwestern pop-
ulations and T. c. major may be a result of the proposed intergradation between
box turtles during the height of the Wisconsin glaciation. The synonymy of T. c.
major with T. c. putnami also suggests that T. c. major may have been capable of
obtaining a considerably larger size than that observed in living specimens.

Although we believe a systematic revision of the genus Terrapene is
needed, it exceeds the bounds of this faunal review. One of the goals of this dis-
cussion, however, is to emphasize the need for such a revision, based both on
"standard" characters and on osteological characters of all fossil and extant
forms.

Turtles

21

Fig. 7 Terrapene Carolina major skulls. A) Recent (UF 18963). B) Haile 8A (UF
3148). C) Ardis fossil (SCSM 91.165.19). D) Ardis fossil (.436).

3 cm

3 cm

B

3 cm

D

22 Curtis C. Bentley and James L. Knight

Deirochelys reticularia - Chicken Turtle (Agassiz, 1857)

Material: 2 peripherals (.542-.543).

Characters used for identification: Identification is based upon the distinctive
"spike-like" pattern of the dorsal sculpturing on these elements (Jackson 1964,
Holman 1978).

Remarks: This turtle today has a continuous range from Texas and Oklahoma
through the Gulf states and along the southern half of the Atlantic Coastal Plain
states into North Carolina, with isolated populations in southeastern Virginia
(Conant and Collins 1991). Still-water habitats, such as ponds, swamps, marsh-
es, and temporary pools, are commonly occupied by chicken turtles, which
reportedly do not favor moving water (Ernst and Barbour 1989).

This is the first fossil record of Deirochelys from South Carolina. The
species most extensive fossil record and probable origin is in Florida (Jackson 1978).

Emydoidea blandingii - Blanding's Turtle (Holbrook, 1838)

Material: An individual specimen consisting of a complete carapace, anterior
plastral lobe (.544), left lower jaw (.544.1), partial hyoid (.544.2), 1 partial
humerus (.544.4), 1 partial scapulo-acromial process (.544.3), 1 femur (.544.5),
3 partial dorsal vertebrae (.544.8-.544. 10), 1 ilium (.544.6), 1 pubo-pectineal
process (.544.7), 1 sacral rib (.544.11). An individual specimen consisting of a
complete carapace, 1 phalange, 2 partial vertebrae (UF). An individual specimen
consisting of a nearly complete carapace, anterior plastral lobe, 4 partial dorsal
vertebrae, 3 caudal vertebrae (USNM). An individual specimen consisting of a
complete carapace, posterior plastral lobe (.545), 5 cervical vertebrae (.545.1-
.545.5), 2 humeri (.545.21-.545.22), 1 ulna (.545.31), 2 scapulo-acromial
processes (.545.27- .545.28), 2 coracoids (.545.25-.545.26), 1 dorsal vertebra
(.545.6), 2 femora (.545.23-.545.24), 1 fibula (.545.29), 1 tibia (.545.30), 1 com-
plete pelvic girdle (.545.20), 5 phalanges (.545.32-.545.35), 1 ungual (.545.36),
2 sacral ribs (.545.37-.545. 38), 13 caudal vertebrae (.545.7-.545. 19). An indi-
vidual specimen consisting of a complete carapace and plastron, partial skull (2
maxillae, 1 quadrate, basioccipital-condyle, basisphenoid, frontal-postorbital-
parietal skull fragment) (.546), partial hyoid apparatus (.546.2), 6 cervical verte-
brae C546.3-.546.8), 1 humerus (.546.20); 2 ulnae (.546.25-.546.26), 1 radius
(.546.27), 2 scapulo-acromial processes (.546.28-.546.29), 2 coracoids (546.30-
.546.31), 1 femur (.546.21), 1 tibia (.546.22), 2 fibulae (.546.23-.546.24), com-
plete pelvic girdle (.546.19), 10 phalanges (.546.32-.546.41), 1 ungual (.546.42),
2 sacral ribs (.546.43), 10 caudal vertebrae (.546.9-.546. 18). An individual spec-
imen consisting of a complete carapace (.547). An individual specimen with the

Turtles 2 3

anterior one half of the carapace (.548). An individual specimen with the ante-
rior two-thirds of the carapace (.549). An individual juvenile specimen consist-
ing of a partial carapace (nuchal, 1-2,4,8 neurals, pygal, 2-5, 7-11 left peripher-
als, 1-4 right peripherals, 3-8 right costals, 3rd left costal) and anterior lobe of
plastron missing left epiplastron (.550), left xiphiplastron (.550.1), 1 dorsal ver-
tebra (.550.2).

Isolated elements: 1 nuchal (.588); associated nuchal and 1st left and
right costal with 1st left peripheral (USNM), 4 associated costals and single
peripheral (.566); 1 left and 1 right 1st costal (.567-.568); 1 left 3rd costal (with
rodent gnaw marks) (.569); 10 costals (.556-.565); 6 neurals (.570-.575); 3 right
1st peripherals (.576-.578); 1 left 5th peripheral (sub-adult) (.580); 1 left 6th
peripheral (.579); 1 partial plastron (.607); 2 anterior plastral lobes (2 UF), 1 par-
tial anterior plastral lobe (USNM), 2 pairs of associated epiplastra (.589-. 590); 4
left and 2 right epiplastra (3 USNM)(3 UF), 7 entoplastra (.581-.587); 4 left and
6 right hyoplastra (.601-.602)(4 USNM)(4 UF); 4 posterior plastral lobes
(.591)(1 USNM)(2 UF); 1 associated right hypoplastron and xiphiplastron
(.592); 7 right and 8 left hypoplastra (.593-.600)(3 USNM)(4 UF); 3 right and 3
left xiphiplastra (.603-.606)(l USNM)(1 UF); 2 partial skulls (.551-.552); 1 left
postorbital (.553); 2 left auditory bullae and quadrate (.554-. 555); 5 cervical ver-
tebrae (.614-.618); 3 humeri (.608-.610); 3 femora (.611-.613); 1 partial pelvic
girdle (.619).

Characters used for identification: It is possible to distinguish the more complete
specimens of Emydoidea blandingii from Deirochelys reticularia on the basis of
the hinged plastral elements and the lack of carapacial sculpturing in the former
(Jackson 1978). Characters that distinguish this species from Terrapene Carolina
are given under that account. Isolated specimens can be distinguished from other
emydid turtles based on characters mentioned in other sections of our paper and
the following:

Epiplastron - Emydoidea epiplastron can be contrasted to Terrapene
epiplastron in several ways. This element in Emydoidea differs from Terrapene
by the presence of a depression located on the dorsal surface and medially to the
anterior scute overlap area, which is less pronounced in Emydoidea. When com-
pared to specimens of Terrapene of comparable size, this element is less robust
and somewhat dorso-ventrally compressed. However, there is some difficulty in
distinguishing large specimens of T Carolina major from E. blandingii. This ele-
ment differs from Trachemys in that it is more elongated and thinner in E.
blandingii.

Xiphiplastron - This element is most easily confused with Clemmys.
Preston and McCoy (1971), suggested that the xiphiplastron of Clemmys is wider
with respect to its length. Preston (1979) discusses additional characters used to
identify this element in Emydoidea.

24

Curtis C. Bentley and James L. Knight

Neurals - These thin, very broad, smooth elements are distinctive
among emydid turtles.

Axial and appendicular skeleton - Although cranial material of this
species is distinctive, the generalized nature of the postcranial material makes the
assignment to E. blandingii tentative.

Remarks: The recent distribution of E. blandingii is limited to southern Ontario
and the Great Lakes region, with scattered populations occuring westward into
northeastern Nebraska and south into northeastern Missouri, and eastward into
New York and Massachusetts on the Atlantic coast (Conant and Collins 1991).
The nearest fossil records of Emydoidea blandingii to the Ardis site are from
Catalpa Creek, Mississippi (Jackson and Kaye 1975), and at New Trout Cave,
West Virginia (Holman and Grady 1987). Both records are late Pleistocene.

The well preserved fossil material from the Ardis site (Fig. 8) is the first
report of this species on the Atlantic Coastal Plain and is a range extension of
about 1,200 km from its present continuous distribution and nearly 525 km south
of the nearest reported fossil locality at New Trout Cave.

Habitats frequented by E. blandingii are generally in shallow, lentic
waters with soft substrate, such as ponds, streams, marshes and sloughs (Ernst
and Barbour 1989).

Fig. 8 Fossil Emydoidea blandingii carapace (.547).

5 cm

Turtles 2 5

Trachemys scripta - Slider Turtle (Schoepf, 1792)

Material: SCSM 91.169.1 a nearly complete carapace and plastron, partial right
lower jaw, 1 scapulo-acromial process, 1 coracoid, 1 partial vertebra, partial
pelvic girdle, 1 fibula. An individual specimen consisting of a partial carapace
and plastron (.620).

Isolated elements; 3 nuchals (.621 -.623); 23 partial costals (.624-
.640)(3 USNM)(3 UF); 15 peripherals (.641-.655); 1 suprapygal (.656); 10 neu-
rals C657-.666).

Characters used for identification: This turtle is distinguished from other turtles
by characters given by Holman (1985). Additionally, it has a diagnostic carapa-
cial ornamentation.

Remarks: The slider has a nearly continuous distribution from Illinois southward
into Texas and New Mexico and thence eastward across northern Florida north
along the Atlantic coast into Virginia, with populations in Mexico and Maryland
(Conant and Collins 1991). Trachemys scripta can be found in most freshwater
habitats, but seems to prefer slow to non-moving water with a soft substrate
(Ernst and Barbour 1989).

Today Trachemys scripta is common around the Ardis site and we have
observed more than 10 within 100 m of the excavation site.

Pseudemys sp.-Cooters (Gray, 1855)

Material: 1 right 1st peripheral (.667); associated 9- 10th right peripherals (.668);
2 peripherals (.669-.670).

Characters used for identification: These compare most favorably to species in
this genus based on the thin, elongated sloping nature of the elements. We were
unable to identify any diagnostic characters on these elements that could be used
make a species placement. The only genus with which these may be easily con-
fused is Trachemys. The fossil elements lack the sculpturing found in Trache-
mys and are significantly thinner and more elongated. The fossil peripherals also
have a straight distal margin which contrasts with the notched margin in Trache-
mys.

26 Curtis C. Bentley and James L. Knight

Remarks: Pseudemys is a common genus of the Southeast, found in various
aquatic systems (Conant and Collins 1991). The two species found in South Car-
olina today are P. floridana and P. concinna.

Testudinidae
Hesperotestudo crassiscutata -Giant Tortoise (Williams, 1950)

Material: 9 carapacial and plastral fragments (.671-.677)(1 USNM)(1 UF); 1
right xiphiplastron (.678); 1 vertebra (.685); 1 ungual (.686); 6 osteoderms (.679-
.684).

Characters used for identification: The fragmented shell elements were assigned
to this species rather than H. incisa on the basis of their extremely large size and
robustness (40.0 mm thick). The large osteoderms, phalange, and vertebra could
not be distinguished from H. crassiscutata in the Florida Museum of Natural
History.

Remarks: Bramble (1971), Preston (1979), and Meylan (1995) place all North
American non-Gopherus tortoises into the genus Hesterotestudo, and that prac-
tice is followed here. Dobie and Jackson (1979), provided the first report of
(Geochelone) H. crassiscutata in South Carolina from the late Pleistocene of
Edisto Island.

Trionychidae
Apalone sp.- Softshell turtle (Rafinesque, 1832)

Material: 1 partial nuchal (.47); 1 costal distal end (.48); 1 partial neural (.49).

Characters used for identification: These fossils are easily assigned to this genus,
based on the relatively thin shell elements with characteristic pitting of the dor-
sal surfaces, general morphology, and the geographical distribution of Triony-
chidae. However, based on these elements, we were unable to identify a species
with any certainty.

Remarks: Two species of the genus Apalone now occur in South Carolina, A.
ferox and A. spinifera. Both species inhabit various aquatic environments with
muddy or sandy bottoms in deep or shallow water (Ernst and Barbour 1989).
Apalone ferox occurs throughout Florida and in the southern portions of Alaba-
ma, Georgia, and South Carolina. A. spinifera is restricted in Florida to rivers in
the extreme northeastern and northwestern portions, and ranges no farther north
along the Atlantic coast than North Carolina. It ranges westward into Colorado
and north into Minnesota, with isolated populations in Montana, California, and

Turtles 2 7

Mexico (Conant and Collins 1991). Only A. spinifera inhabits the area of the
Ardis site today. Dobie and Jackson (1979) reported "Trionyx sp." from Edisto
Island, South Carolina. Meylan (1987) has shown the correct name for North
American softshells to be Apalone.

DISCUSSION AND PALEOECOLOGY

The turtle assemblage of the Ardis local fauna provides important new
data for the late Pleistocene of the southeastern United States. In particular, it
documents a shift in the spatial patterns of several turtle species during the late
Pleistocene. The turtle fauna is unique in its geographical and temporal setting,
and it contains the first sympatric fossil occurrences of several taxa, e.g., Clem-
mys muhlenbergii, C. guttata, Emydoidea blandingii, Macroclemys temminckii.

All of the fossil turtles collected from the site, except Hesperotestudo
crassiscutata and Terrapene Carolina major, are primarily aquatic and are com-
monly found in, or require, still or slow moving water with a soft substrate and
aquatic vegetation (Ernst and Barbour 1989). Additional evidence of a nearby
body of water included the presence of Alligator mississippiensis and elements
of the fish fauna which are currently under study. This agrees with the habitat
suggested by Bentley et al. (1994) based on the Ardis mammal fauna that indi-
cates an ecotone between a mixed forest of conifers, hardwoods and meadows,
and a permanent body of water such as a river or stream which may have given
way to a bog or marsh. Portions of the mammalian fauna and avian material
from the Ardis site further suggest the presence of a nearby large body of water
such as a lake or pond. This association is based on the life histories and habitat
requirements of many of the extant species represented in the Ardis fauna.

The Ardis turtle fauna consists of thirteen extant and one extinct
species, with five taxa considered extralimital. Three of the five have northern
affinities: Emydoidea blandingii, Clemmys muhlenbergii, and Chrysemys picta.
Terrapene Carolina major has a strong southern affinity. Macroclemys has a pri-
marily Gulf coast distribution but extends as far north up the Mississippi Valley
as Iowa (Pritchard 1989). Although Hesperotestudo was widespread in North
America by the Miocene, Hibbard (1960) suggested that the presence of Hes-
perotestudo crassiscutata in a fauna indicated a mild climate with frost-free win-
ters. The sympatric occurrence of species that are apparently ecologically
incompatible today constitutes a "disharmonious fauna" (sensu Lundelius et al.
1983), which has been interpreted by many authors (Hibbard 1960, Holman
1980, Lindelius et al. 1983) as indicating a more equable climate (reduced sea-
sonal temperature gradients) than that experienced in the region today. The
Ardis local fauna reflects such a disharmonious biota, which clearly has no mod-
ern analogue. The turtle fauna corroborates conclusions made on the basis of the
Ardis mammal fauna (Bentley et al. 1994), which also suggests that a more

28 Curtis C. Bentley and James L. Knight

equable climate than today prevailed near or during the maximum advance of the
Laurentide ice sheet in the southeastern United States.

Based on the distribution of terrestrial vertebrates, Smith (1957) stated
that the northeastern biota of North America was displaced southward during the
Wisconsinan glacial maximum. Smith also suggested that during a "Climatic
Optimum," southern counterparts dispersed northward. During the height of the
Wisconsinan glaciation, E. blandingii would have been extirpated from its pre-
sent-day northerly distribution (Mickelson et al. 1983, Conant and Collins
1991). The southern boundary of the Laurentide ice sheet, while abutted against
the Appalachian Plateau (Mickelson et al. 1983), could potentially have forced
the range of E. blandingii to be split, with one population occurring along the
Mississippi Valley and another along the Atlantic Coastal Plain. This may have
produced geographically isolated eastern and western populations of Emydoidea
blandingii during the late Pleistocene.

In addition to Emydoidea blandingii, Spermophilus tridecemlineatus
(thirteen-lined ground squirrel) was also collected from the Ardis site (Bentley et
al. 1994). Based on Recent and fossil distributions (Kurten and Anderson 1980),
the present northeastern distribution of the thirteen-lined ground squirrel also
may have been displaced southward along the Atlantic Coastal Plain by the
advancing Laurentide ice sheet.

As with the Gulf Coast Corridor, depressed sea levels during the Pleis-
tocene glacial stage exposed much or all of the Atlantic continental shelf (Bloom
1983), thereby widening the Atlantic Coastal Plain. This may have facilitated
the dispersal of glacially-displaced species southward along the coast. The
newly emergent land area provided expanded habitats for species to utilize
(Blaney 1971), and a more equable climate may have allowed for the range
extention of both northern and southern species into these new areas. Northern
populations of T. c. Carolina in Massachusetts and Michigan that show affinity to
T. c. major, discussed by Milstead (1969), may be relicts left over from a Pleis-
tocene interval of extensive intergradation between the northerly displaced sub-
species and radiating southern subspecies.

Bleakney (1958) suggested that the Recent population of E. blandingii
in Nova Scotia survived glaciation in an "Atlantic Coastal Plain refuge" and then
dispersed northward up the coast into Canada and Maine. Preston and McCoy
(1971), suggested that "colonization of the Atlantic Coastal Plain from the Great
Lakes region, along a 'steppe corridor' (Schmidt 1938) through the Mohawk
Valley" was a more plausible hypothesis. Preston and McCoy (1971) also stat-
ed that the study of specimens from the eastern limits may provide an answer to
the possibility of a "minor Atlantic Coast refuge for Emydoidea during ice
advances." The presence of numerous E. blandingii at the Ardis site, as well as
fossil material from New Trout Cave in West Virginia (Holman and Grady 1987),
suggest that the Recent extreme northeastern populations may be products of a

Turtles 2 9

re-invasion of the northeast by a substantial Atlantic Coastal Plain stock that had
spread at least 900 km southward during the glacial advance of the late Pleis-
tocene. Evidence from the Ardis local fauna indicates significant shifts longitu-
dinally in the spatial distribution of E. blandingii along the Atlantic Coastal
Plain during the late Pleistocene.

ACKNOWLEDGMENTS - We wish to express our gratitude to the staff of the
Giant Cement Plant, Harleyville, South Carolina, for their generous cooperation,
and in particular to Burt Ardis, for whom the fauna is named. Many thanks go
to all the field volunteers: Vance McCollum, Linda Eberle, Craig and Alice
Healy, Derwin Hudson, Ray Ogilvie, Lee Hudson, Tom Reeves, Martha Bentley,
and Karin Knight. Robert Weems, U.S. Geological Survey, and Peter Meylan,
Eckerd College, Petersburg, Florida, provided helpful discussion and review of
this manuscript. We also wish to express our appreciation to Overton Ganong of
the South Carolina State Museum; David Webb, Gary Morgan, and David Auth,
of the Florida Museum of Natural History; Dennis Hermann, of Zoo Atlanta; and
Greg Schneider, Museum of Zoology, The University of Michigan, and Justin
Cougdon, Savannah River Ecology Labratory, Aiken, South Carolina, for allow-
ing us access to their collections. Thanks go to Mike and Debi Trinkley, the
Chicora Foundation, Columbia, South Carolina, for the use of their field equip-
ment, and to Mike Runyon of Lander College, Greenwood, South Carolina, for
donating fossil material for C14 dating. Darby Erd provided the illustrations.

LITERATURE CITED

Auffenberg, W. 1967. Further notes on fossil box turtles of Florida. Copeia
2:319-325.

Bentley, C. C, and J. L. Knight. 1993. The oldest spotted turtle: Clemmys gut-
tata (Testudines:Emydidae), from the late Pleistocene (Rancholabrean)
Ardis local fauna, Dorchester County, South Carolina. South Carolina Geol-
ogy 36:59-63.

Bentley, C. C, J. L. Knight, and M. A. Knoll. 1994. The mammals of the Ardis
local fauna (late Pleistocene), Harleyville, South Carolina. Brimleyana 21:1-
34.

Blaney, M. R. 1971. An annotated check list and biogeographic analysis of the
insular herpetofauna of the Apalachicola region, Florida. Herpetologica
27:406-430.

Bleakney, J. S. 1958. A zoogeographical study of the amphibians and reptiles of
eastern Canada. Bulletin of the National Museum of Canada 155: 1-1 19.

Bloom, A. L. 1983. Sea level and coastal morphology of the United States
through the late Wisconsin glacial maximum. Pages 215-229 in Late-Qua-
ternary environments of the United States (H.E. Wright, Jr., editor). Uni-
versity of Minnesota Press, Minneapolis. Volume 1.

3 0 Curtis C. Bentley and James L. Knight

Bowen, D. Q. 1988. A stratigraphic framework for multidisplinary work. Perg-
amon Press, Oxford, New York.

Bramble, D. M. 1971. Functional morphology, evolution, and paleoecology of
gopher tortoises. Ph.D. Thesis, University of California, Berkeley. Library
Photographic Service: Abstract.

Brett, C. E. 1990. Obrution deposits. Pages 239-243 in Paleobiology: a syn-
thesis. (E. G. Briggs and P. S. Crowther editors). Blackwell Scientific Pub-
lications, Oxford, England.

Brett, C. E., and S. E. Speyer. 1990. Taphofacies. Pages 258-263 in Paleobi-
ology: a synthesis. (E. G. Briggs and PS. Crowther, editors). Blackwell Sci-
entific Publications, Oxford, England.

Carr, A. 1952. Handbook of turtles. The turtles of the United States, Canada,
and Baja California. Cornell University Press, Ithaca, New York.

Conant, R., and J. T. Collins. 1991. A field guide to reptiles and amphibians.
Eastern and central North America. Hought Mifflin Company, Boston,
Massachusetts.

Dobie, J. L., and D. R. Jackson. 1979. First fossil record for the Diamondback
Terrapin, Malaclemys terrapin (Emydidae), and comments on the fossil
record of Chrysemys nelsoni (Emydidae). Herpetologica 35:139-145.

Ernst, C. H., and R. W. Barbour. 1989. Turtles of the world. Smithsonian Insti-
tution Press, Washington, D.C.

Harris, W. B., and V. A. Zullo. 1991. Eocene and Oligocene stratigraphy of the
outer Coastal Plain. Pages 251-262 in The geology of the Carolinas (J.W.
Horton, Jr. and V. A. Zullo, editors). University of Tennessee Press,
Knoxville.

Hibbard, C. W. 1960. Pliocene and Pleistocene climates in North America.
Annual Report of the Michigan Academy of Science, Arts, and Letters 62:5-
30.

Holman, J. A. 1964. Pleistocene amphibians and reptiles from Texas. Herpeto-
logica 20:73-83.

Holman, J. A. 1966. Some Pleistocene turtles from Illinois. Transactions Illi-
nois Academy Science 59:214-216.

Holman, J. A. 1967. A Pleistocene herpetofauna from Ladds, Georgia. Bulletin
of the Georgia Academy of Science 25: 154-166.

Holman, J. A. 1975. Herpetofauna of the Wakeeney local fauna (lower Pliocene:
Clarendonian) of Trego County, Kansas. Pages 49-66 in Studies on Ceno-
zoic paleontology and statigraphy in honor of Claude W. Hibbard (G.R.
Smith, and N. E. Friedland, editors). Papers on Paleontology Volume 3, Uni-
versity of Michigan, Museum of Paleontology, Ann Arbor.

Holman, J. A. 1977. The Pleistocene (Kansan) herpetofauna of Cumberland
Cave, Maryland. Annals of the Carnegie Museum of Natural History
46(11): 157-172.

Turtles 3 1

Holman, J. A. 1978. The late Pleistocene herpetofauna of Devil's Den sinkhole,
Levy County, Florida. Herpetologica 34:228-237.

Holman, J. A. 1980. Paleoclimatic implications of Pleistocene herpeto faunas of
eastern and central North America. Transactions of the Nebraska Academy
of Sciences VIII: 13 1-140.

Holman, J. A. 1985. Herpetofauna of Ladds Quarry. National Geographic
Research l(3):423-436.

Holman, J. A. 1990. Vertebrates from the Harper site and rapid climatic warm-
ing in the Mid-Holocene Michigan. Michigan Academician 22:205-217.

Holman, J. A., and F. Grady. 1987. Herpetofauna of New Trout Cave. National
Geographic Research 3(3):305-317.

Jackson, C. G., Jr. 1964. The status of Deirochelys floridana Hay with com-
ments on the fossil history of the genus. Tulane Studies in Geology
2:(3)103-106.

Jackson, C. G., and J. M. Kaye. 1975. The occurrence of Blanding's turtle, Emy-
doidea blandingii, in the late Pleistocene of Mississippi (Testudines: Testu-
dinidae) Herpetologica 30:417-419.

Jackson, D. R. 1978. Evolution and fossil record of the chicken turtle
Deirochelys, with a re-evaluation of the genus. Tulane Studies in Zoology
and Botany 20(1 -2):35-55.

Kurten, B., and E. Anderson. 1980. Pleistocene mammals of North America.
Columbia University Press, New York, New York.

Lamb, T, and J. Lovich. 1990. Morphometric validation of the striped mud tur-
tle (Kinosternon baurii) in the Carolinas and Virginia. Copeia 3:613-318.

Lundelius, L. E., Jr., R. W. Graham, E. Anderson, J. Guilday, J. A. Holman, D.
W. Steadman, and S. D. Webb. 1983. Terrestrial vertebrate faunas. Pages
311-353 in Late-Quaternary environments of the United States (H. E.
Wright, Jr., editor). Volume 1, University of Minnesota Press, Minneapolis.

Meylan, P. A. 1987. The phylogenetic relationships of soft-shelled turtles (Fam-
ily Trionychidae). Bulletin of the American Museum of Natural History
186(1):1-101.

Meylan, P. A. 1995. Pleistocene amphibians and reptiles from the Leisey Shell
pit, Hillsbourough County, Florida. Bulletin of the Florida Museum of Nat-
ural History 37 (9):273-297.

Mickelson, D. M., L. Clayton, D. S. Fullerton, and H. W. Borns, Jr. 1983. The
late Wisconsin glacial record of the Laurentide ice sheet in the United States.
Pages 3 1 1-353 in Late Quaternary environments of the United States. (H. E.
Wright, Jr., editor). Volume 1, University of Minnesota Press, Minneapolis.

Milstead, W. W. 1969. Studies on the evolution of box Turtles (Genus Ter-
rapene). Bulletin of the Florida State Museum 14(1): 1-1 13.

32 Curtis C. Bentley and James L. Knight

Preston, E. R. 1979. Late Pleistocene cold-blooded vertebrate faunas from the
mid-continental United States. 1. Reptilia: Testudines, Crocodilia. Universi-
ty of Michigan, Museum of Paleontology. Papers on Paleontology 19:1-53.

Preston, R. E., and C. J. McCoy. 1971. The status of Emys twentei Taylor (Rep-
tilia: Testudinidae) based on new fossil records from Kansas and Oklahoma.
Journal of Herpetology 5(l-2):23-30.

Pritchard, P. C. H. 1989. The alligator snapping turtle: biology and conservation.
Milwaukee Public Museum, Milwaukee, Wisconsin.

Roth, J. A., and J. Laerm. 1980. A late Pleistocene vertebrate assemblage from
Edisto Island, South Carolina. Brimleyana 3:1-29.

Schmidt, K. P. 1938. Herpetological evidence for the postglacial eastward
extension of the steppe in North America. Ecology 19:396-407.

Smith, W. P. 1957. An analysis of Post-Wisconsin biogeography of the Prairie
Peninsula region based on distributional phenomena among terrestrial ver-
tebrate populations. Ecology 38:205-218.

Tushingham, M. A., and W. R. Peltier. 1993. Implications of the radiocarbon
timescale for ice-sheet chronology and sea-level change. Quaternary
Research 39:125-129.

Ward, L. W., B. W Blackwelder, G. S. Gohn, and R. Z. Poore. 1979. Strati-
graphic revision of Eocene, Oligocene and Lower Miocene Formations of
South Carolina. South Carolina Geological Survey Publication. Geological
Notes 23(l):2-32.

Webb, S. D. 1974. Introduction. 3-4 in Pleistocene mammals of Florida (S. D.
Webb, editor). University of Florida Presses, Gainesville.

Received 3 January 1996
Accepted 13 August 1996

Turtles 3 3

APPENDIX I

Taxa and Minimum Number of Individuals Present

Taxa Minimum number of individuals

Kinosternon subrubrum* 5

Sternotherus odoratus* 4

Chelydra serpentina 3

Macroclemys temminckii* 1

Clemmys guttata 19

Clemmys muhlenbergii* 3

Chrysemys picta* 25

Deirochelys reticularia* 1

Trachemys scripta 4

Pseudemys sp. 1

Terrapene Carolina major 12

Emydoidea blandingii* 16

Hesperotestudo crassiscutata 1

Apalone sp. 1

Number of turtle species = 14

Number of individuals = 96

* = first fossil report from South Carolina

Observations of Freshwater Jellyfish, Craspedacusta sowerbyi
Lankester (Trachylina: Petasidae), in a West Virginia Reservoir

Ted R. Angradi

U.S. Department of Agriculture, Forest Service

Timber and Watershed Laboratory,

Parsons, West Virginia 26287

ABSTRACT. — A swarm of medusae of the freshwater jellyfish
Craspedacusta sowerbyi was observed in a cove of a West Virginia
reservoir in August and September, 1995. Medusae were abundant

(>1000/iTr) but extremely localized. Distribution of medusae in the
cove did not appear to be linked to water chemistry. Size of medusae
ranged from 6-21 mm in diameter and increased significantly with dis-
tance from the center of abundance, suggesting that the localized distri-
bution of medusae resulted from dispersion rather than from environ-
mentally-induced aggregation. Measurements of mean diameter of
medusae on separate dates indicated a growth rate of about 0.2 mm/d,
and a medusa life cycle of approximately 102 days.

The freshwater jellyfish Craspedacusta sowerbyi Lankester 1880 is an
exotic species first observed (as medusae) in the United States in 1908 (Kramp
1950, Pennak 1989). Native to the Yang-tse River system in China (Kramp
1950), C sowerbyi has been reported from many localities worldwide between

45° north and 45° south latitude (Acker and Muscat 1976, Pennak 1989). True
freshwater jellyfishes are few, limited to about a dozen species worldwide
(Hutchinson 1967, Pennak 1989).

Craspedacusta sowerbyi has been reported from 3 1 states and the Dis-
trict of Columbia; it has not been reported from northern New England, the
Northern Rocky Mountains, or the Northern Great Plains (DeVries 1992). In
West Virginia there are records of C. sowerbyi from Barbour, Fayette, Mercer,
Monogalia, Wayne, and Wood counties (Reese 1940, Lytle 1962, Koryak and
Stafford 1981, and D. Tarter, Marshall University, personal communication).

Craspedacusta sowerbyi has two life stages, a free-swimming medusa
(10-20 mm diameter), and a sessile hydroid polyp (1 mm long, Acker and Mus-
cat 1976). Lytle (1959) reviewed the developmental biology of the species.
Medusae of C. sowerbyi appear sporadically in lentic and even less frequently in
lotic ecosystems in the United States (Acker and Muscat 1976, Beckett and
Turanchik 1980, DeVries 1992). Usually a swarm of medusae appears in sum-

34

Jellyfish 35

mer where it has never before been observed or where it has not been observed
for many years (Slobodkin and Bossert 1991). Lentic systems in which the
medusae have been observed include reservoirs, natural lakes, ponds, quarries,
ornamental pools, and aquaria.

Specific environmental factors associated with the formation of
medusae from polyps via asexual reproduction (budding) are poorly understood.
Factors suggested include increasing water temperature (McClary 1959)
increased alkalinity (Koryak and Clancy 1981, McCullough et al. 1981),
increasing dissolved CO2 (Acker and Muscat 1976), decreasing stream flow

(Brussock et al. 1985), changing reservoir levels (Deacon and Haskell 1967) and
increasing supply of zooplankton (Lytle 1959), on which the medusae prey.

Dispersal of C. sowerbyi among water bodies probably occurs via
polyps attached to aquatic plants or waterfowl, or in tanks used to transport fish
(Byers 1945, Bushnell and Porter 1967, Howmiller and Ludwig 1970). The
polyps can survive in moving water (Hutchinson 1967), so once the polyps enter
a river system, the medusae may eventually appear in downstream reservoirs
(e.g., Yeager 1987).

Because the medusae occur unpredictably and the polyps are micro-
scopic and easily overlooked, the complete geographic distribution and ecology
of C. sowerbyi are not well known. Field studies have been mostly descriptive
(e.g., Garman 1916, Deevy and Brooks 1943, Dexter et al. 1949, Chadwick and
Houston 1953, Bushnell and Porter 1967, Koryak and Clancy 1981, McCullough
et al. 1981, Dodds and Hall 1984). Deacon and Haskell (1967) examined diel
activity patterns of medusae at Lake Mead, Nevada. Dodson and Cooper (1983)
examined trophic relationships of the medusae in the laboratory. Acker and
Muscat (1976) and DeVries (1992) reviewed the literature on the ecology of C.
sowerbyi.

The purpose of this paper is to describe a swarm of C. sowerbyi
medusae I observed at Stonewall Jackson Lake, Lewis County, West Virginia, in
August and September 1995. My initial observations of the medusae at
Stonewall Jackson Lake suggested that the size distribution of medusae varied
with distance from the main concentration of medusae (swarm). I hypothesized
that the distribution of medusae resulted from dispersion from the apparent pop-
ulation center at the swarm, and I predicted that medusae collected away from
the main swarm location would be larger than medusae within the swarm
because more distant medusae would have had more time to grow. The null
hypothesis that medusae collected from all locations have the same size class dis-
tribution implies that the dense concentration of medusae at the swarm location
results primarily from aggregation due to water chemistry, temperature, food,
current, wind, or some other factor rather than from dispersion. I collected and
measured specimens to test this hypothesis. I also compared the mean size of

36 Ted R. Angradi

specimens collected on separate dates to calculate an approximate growth rate
for the medusae.

METHODS

I observed the medusae at Wolf Fork (80°28'W, 38°59'N), a cove
formed by a flooded tributary to Skin Creek which forms the east arm of
Stonewall Jackson Lake (Fig. 1). Wolf Fork is about 1.5-km long and 30-150-m
wide and has abundant flooded timber. The cove is well sheltered from winds
and is a no-wake boating zone. A culvert connects Wolf Fork and the stream
draining the upper watershed. Stonewall Jackson Lake is a 1,070-ha Army
Corps of Engineers reservoir filled in 1986. The main West Fork River arm of
the reservoir is a tributary of the Tygart River in the Ohio River drainage.

During my initial visit to Wolf Fork (16 August 1995) I estimated the
density (number m3) of medusae using two methods. Where the medusae were
abundant I used a 20-L plastic bucket. From a small boat I slipped the bucket
into the water and withdrew it with minimal turbulence. I poured the bucket con-
tents through a fine sieve and transferred the medusae to a tray for enumeration.
Where the medusa were scarce, I estimated the density visually. Both methods
are biased toward the upper 0.5 m of water surface because I could not see or
sample medusa at greater depths. On the first visit to Wolf Fork I collected water
samples and recorded water temperature and dissolved oxygen at several loca-
tions in the cove. Water samples were analyzed at the U.S. Forest Service Tim-
ber and Watershed Laboratory, Parsons, West Virginia.

Fig. 1 Map of Stonewall Jackson Lake showing the swarm location at Wolf
Fork. Inset map shows location of the reservoir in West Virginia.

Jellyfish 37

On subsequent visits to Wolf Fork (16 August and 12 September 1995)
I collected medusae with an aquarium dip net from a small boat. I measured bell
diameter of live specimens under a dissecting microscope by placing a ruler
under a clear plastic petri plate containing a few medusae and a small amount of
water. I judged measurement error to be ±1.0 mm.

RESULTS
On 16 August 1995, I observed a dense swarm of medusae near the
head of Wolf Fork. Medusae decreased greatly in abundance with distance from
the swarm. Density of medusae in three bucket samples was 1.2, 1.9, and 4.8
medusa/L. This is approximately equivalent to 1,000-5,000 medusae/m3 in the
upper 0.5 m of the water column within an area of about 25 m2. At 100-200 m
from the swarm (toward the main channel of the reservoir) there were 10-50
medusae/m3; at 300-400 m medusae were scarce (<1 medusa/m3). I did not
observe medusae in lower Wolf Fork, the main channel, or in other coves of Skin
Creek, although I did not make an exhaustive search. My conversations with
anglers, reservoir managers, and local fish biologists suggest that this is the first
record of C. sowerbyi at Stonewall Jackson Lake. The origin of C. sowerbyi in
the drainage is unknown.

Table 1. Selected water quality measurements for Wolf Fork, 16 August 1995.
Site distances are from the swarm toward the main channel. All values for sam-
ples collected at the water surface. DO=dissolved oxygen.

Site

Temp

DO

PH

Conductivity

Alkalinity

so4

Ca

Medusa

(m)

C

(mg/L)

nSlcm

mg/L
(CaCO,)

(mg/L)

(mg/L)

abundance

(#/m<)

-100

30.0

7.7

7.0

98.3

19.1

21.9

12.1

<1

0

30.0

8.1

7.0

100.5

18. 2

21.6

12.8

1000-5000

100

30.5

8.3

7.1

99.8

17.8

22.0

12.5

10-50

200

30.5

8.3

7.2

99.6

17.5

23.7

12.6

<1

300

30.5

8.3

7.2

98.1

17.2

22.6

12.7

<1

Water chemistry did not differ appreciably among sites within Wolf
Fork (Table 1). Dissolved oxygen decreased with depth to 5 mg/L at 2.1 m and
1.8 mg/L at 2.7 m. Depth of the channel in Wolf Fork was 2.5-3.5 m. Flow from
the Wolf Fork watershed (the stream) into Wolf Fork (the cove) could not be
detected at the culvert.

On 18 August 1995, medusae collected ranged in size from 6 to 20 mm
(Fig. 2). Mean (SE) size and size class distribution of medusae collected from
the swarm location were different from medusae collected at three other stations
in Wolf Fork. Swarm medusae were 13.3 (0.2) mm in diameter; medusae from
outside the swarm were 15.2 (0.3) to 16 (0.4) mm in diameter depending on col

38

Ted R. Angradi

Fig. 2 Size distribution of C. sowerbyi medusae collected 0, 100, 200, and 350+
m from a dense concentration (swarm) of medusae at Wolf Fork, Stonewall Jack-
son Lake, West Virginia on 18 August 1995. Figure legends give distance from
the main swarm location, mean (SE) medusa diameter and sample size.

T3
0)

T5

o
o

<D
CO
C/>

13

■D
<D

E

CD
.a

ou -

200 m

E

3

x=15.8(0.3)

C

20

n=110

16
o

10

..■Jill.

0

■■iililllli.

30

20

10

350+ m

x=16.0(0.4)

n=52

4 6 8 10 12 14 16 18 20 22

Medusa diameter (mm)

Jellyfish

39

lection site, a significant difference (Kruskal-Wallis ANOVA on Ranks; H =
75.1, P < 0.01). Among sites outside the swarm location, there was also a trend
of increasing size with distance from the swarm; mean medusa diameter was
largest 350 m from the swarm, the location at which medusae were first observed
when entering Wolf Fork from the reservoir. Within the swarm there was a
bimodal size distribution of medusae concentrated in 9 - 13 mm and 15-18 mm
size classes, with more medusae in the smaller size classes. Away from the
swarm most medusae were in the larger size classes, but a semblance of the
bimodal pattern was apparent (Fig. 2).

Fig. 3 Size distribution of C. sowerbyi medusae collected 0 and 100 m from a
dense concentration (swarm) of medusa at Wolf Fork, Stonewall Jackson Lake,
West Virginia on 18 August and 12 September 1995. Otherwise as for Fig. 2.

18 August 1995

4 6 8 10 12 14 16 18 20 22 4 6 8 10 12 14 16 18 20 22

Medusa diameter (mm)

On September 12, I observed fewer medusae than on previous visits to
Wolf Fork. Water temperature was 25 C, and reservoir level had dropped about
1 m since the previous visit. The sky was overcast and light rain was falling,
whereas the sun shone on previous visits. Medusae were most abundant near the
18 August swarm site; at 100 m they were scarce, and were not observed else-
where. Mean diameter of medusae was the same at both locations (Fig. 3),
although more medusae from the 100 m site were in the largest size classes (20-
21 mm). Between August 18 and September 12, mean diameter of medusae at
the swarm location increased slightly more than 4 mm (Fig. 3). This is equiva-

40 Ted R. Angradi

lent to an average growth rate of about 0.2 mm/d. Assuming the medusa stage
originates at about 0.5 mm and attains a maximum size of about 21 mm (Pennak
1989), this rate of growth indicates a life cycle for the medusa of about 102 days.
The apparently slower growth of medusae away from the main swarm (Fig. 3)
may be attributed to disproportionate mortality of the full-growth medusae.

DISCUSSION

I observed great variation in abundance of medusae over a scale of a
few meters that is not readily explained by available water chemistry data. My
finding of larger individuals away from the main concentration of medusae is cir-
cumstantial evidence that spatial variation in abundance of the medusae results
from dispersion from a highly localized swarm location.

If the polyps are lotic (Hutchinson 1967), then the distribution of the
medusae in Wolf Fork results from events and conditions since delivery of imma-
ture medusae from upstream earlier in summer. During most of the year, flow
from the stream would induce some current at the head of Wolf Fork, and the
location and localization of the swarm at Wolf Fork may have resulted from
flow-related concentration (e.g., in an eddy) of polyps or small medusae origi-
nally exported from the stream.

Several authors have also found a highly localized distribution of
medusae. Dodson and Cooper (1983) found medusae only in one small sheltered
cove of a Wisconsin Lake. Deacon and Haskell (1967) reported medusae from
sheltered coves at Lake Mead, Nevada, and not from the open waters of the reser-
voir. Garman's (1916) description of the location of a dense swarm in a narrow
flooded tributary of an impounded reach of the Kentucky River is similar to Wolf
Fork and to other published accounts (e.g., Lytle 1962).

Few authors have explicitly commented on factors accounting for spa-
tial variation in medusae abundance within a reservoir or lake. Acker and Mus-
cat (1976) noticed an apparent effect of light on the distribution of medusae
which they attributed to a direct light effect or to an indirect effect of light on
food concentration. Other factors such as wind (Deevy and Brooks 1943), and
ebbulition (Koryak and Stafford 1981) have been proposed. I noticed no appar-
ent light effect at Wolf Fork, and the cove is well protected from wind and boat-
wake disturbance; these factors are unsatisfactory for explaining the large local
variation in abundance of medusae at Wolf Fork. My observations suggest
instead that the distribution of medusae may result from the pattern of dispersion
from a location that is determined by the habitat requirements of the polyps in
the reservoir or in tributary streams.

My observation of a bi-modal size distribution of medusae (Fig. 2) sug-
gest the possibility of more than one period of medusa formation at Wolf Creek,
perhaps associated with variation in water temperature, water chemistry, or
streamflow during early summer. McClary (1959) reported medusa budding

Jellyfish 41

only within a narrow temperature range, 26 C to 33 C. However, I found very
few small (<8 mm) medusae in August while sampling within this temperature
range (Fig. 2) suggesting that formation of medusae was not ongoing and most
likely occurred during two periods in early summer.

ACKNOWLEDGMENTS

Cliff Brown of the West Virginia Division of Natural Resources pro-
vided access to Wolf Fork. Emmett Fox conducted water quality analysis. Neal
Auvil brought the jellyfish to my attention. Linda Plaugher assisted with the fig-
ures.

LITERATURE CITED

Acker, T S., and A. M. Muscat. 1976. The ecology of Craspedacusta sowerbii
Lankester, a freshwater hydrozoan. The American Midland Naturalist
95:323-336.

Beckett, D. C, and E. J. Turanchik. 1980. Occurrence of the freshwater jelly-
fish Craspedacusta sowerbyi Lankester in the Ohio River. Ohio Journal of
Science 80:95- 96.

Brussock, P. P., L. D. Willis, and A. V. Brown. 1985. Flow reduction may
explain sporadic occurrence of Craspedacusta sowerbyi (Trachylina)
medusae. Proceedings of the Arkansas Academy of Science 39: 120.

Bushnell, J. H., and T W. Porter. 1967. The occurrence, habitat, and prey of
Craspedacusta sowerbyi (particularly polyp stage) in Michigan. Transac-
tions of the American Microscopy Society 86:22-27.

Byers, C. F. 1945. The fresh-water jellyfish in Florida. Proceedings of the Flori-
da Academy of Sciences 7:173-180.

Chadwick, C. S., and H. Houston. 1953. A "bloom" of fresh water medusae
Craspedacusta ryderi (Potts) in Kentucky Lake, Tennessee. Journal of the
Tennessee Academy 28:36-37.

Deacon, J. E., and W. L. Haskell. 1967. Observations on the ecology of the
freshwater jellyfish in Lake Mead, Nevada. The American Midland Natu-
ralist 78:155-166.

Deevey, E. S., and J. L. Brooks. 1943. Craspedacusta in open water, Lake
Quassapaug, Connecticut. Ecology 24:266-267.

DeVries, D. R. 1992. The freshwater jellyfish Craspedacusta sowerbyi: a sum-
mary of its life history, ecology and distribution. Journal of Freshwater
Ecology 7:7-16.

Dexter, R. W., T. C. Surrarrer, and C. W. Davis. 1949. Some recent records of
the freshwater jellyfish Craspedacusta sowerbii from Ohio and Pennsylva-
nia. The Ohio Journal of Science 49:235-241.

Dodds, R. B., and K. D. Hall. 1984. Environmental and physiological studies
of the freshwater jellyfish Craspedacusta sowerbyi. Bios 55:75-83.

42 Ted R. Angradi

Dodson, S. L, and S. D. Cooper. 1983. Trophic relationships of the freshwater
jellyfish Craspedecusta sowerbyi Lankester 1880. Limnology and
Oceanography 28:345-351.

Garman, H. 1916. The sudden appearance of great numbers of fresh-water
medusa in a Kentucky Creek. Science 44:858-860.

Howmiller, R. P., and G. M. Ludwig. 1970. A record of Craspedacusta sower-
byi in Wisconsin. Transactions of the Wisconsin Academy of Sciences, Arts
and Letters 60:181-182.

Hutchinson, G. E. 1967. A treatise on limnology: Vol. II. Introduction to lake
biology and the limnoplankton. John Wiley and Sons, Inc., New York, New
York.

Koryak, M., and P. J. Clancy. 1981. Craspedacusta sowerbyi Lankester (Hydro-
zoa) medusoid generations in two western Pennsylvania impoundments.
Proceedings of the Pennsylvania Academy of Science 55:43-44.

Koryak, M., and L. Stafford. 1981. Bubbles and jellyfish: ebullition and
Craspedacusta sowerbyi Lankester (Hydrozoa) in Tygart Lake. Unpub-
lished Report. U.S. Army Corps of Engineers, Pittsburgh District, Pitts-
burgh, Pennsylvania.

Kramp, P. L. 1950. Freshwater medusae in China. Proceedings of the Zoolog-
ical Society of London 120:165-184.

Lytle C. F. 1959. Studies on the developmental biology of Craspedacusta.
Ph.D. Thesis, Indiana University, Bloomington.

Lytle, C. F. 1962. Craspedacusta in southeastern United States. Tulane Stud-
ies in Zoology 9:309-314.

McClary, A. 1959. The effect of temperature on growth and reproduction in
Craspedacusta sowerbii. Ecology 40:158-162.

McCullough, J. D., M. F. Taylor, and J. L. Jones. 1981. The occurrence of the
freshwater- medusa Craspedacusta sowerbyi Lankester in Nacogdoches
Reservoir, Texas and associate physical-chemical conditions. Texas Jour-
nal of Science 33:17-23.

Pennak, R. W. 1989. Freshwater invertebrates of the United States. Protozoa to
mollusca. Third edition. John Wiley and Sons, New York, New York.

Reese, A. M. 1940. Craspedacusta again. American Naturalist 74:180.

Slobodkin, L. B., and P. E. Bossert. 1991. The freshwater Cnidaria — or Coe-
lenterates. Pages 125-143 in Ecology and classification of North American
freshwater invertebrates (J. H. Thorp and A. P. Covich, editors). Academic
Press, Inc., San Diego, California.

Yeager, B. L. 1987. Drainagewide occurrence of the freshwater jellyfish,
Craspedacusta sowerbyi, Lankester 1880, in the Tennessee River system.
Brimleyana 13:94-98.

Distribution and Status of Selected Fishes in North Carolina,
With a New State Record

Fred C. Rohde

North Carolina Division of Marine Fisheries

127 Cardinal Drive Extension

Wilmington, North Carolina 28405-3845

Mary L. Moser
Center for Marine Science Research

7205 Wrightsville Avenue
Wilmington, North Carolina 28403

and

Rudolf G. Arndt

Faculty of Natural Sciences and Mathematics

The Richard Stockton College of New Jersey

Pomona, New Jersey 08240

ABSTRACT - We made status surveys of 22 state-listed fishes in the
French Broad River, Nolichucky River, and Dan River systems from
1991-1995 from 105 collections at 99 sites made by us, augmented by
data from 146 collections made by others, personal communications,
and the literature. We believe state-listed endangered Polyodon spathu-
la and Percina sclera have been extirpated from the state whereas
Exoglossum maxillingua, Thoburnia hamiltoni, Noturus flavus, N.
gilberti, and Percina burtoni are either secure or rare. All six threatened
species surveyed were collected. Of these, Lampetra appendix,
Cyprinella monacha, Cottus carolinae, and Aplodinotus grunniens
probably warrant being elevated to state-endangered status, whereas
Luxilus chrysocephalus and Percina caprodes appear to be secure.
Nine species of special concern were surveyed. Noturus eleutherus and
Etheostoma simoterum are presumed to be extirpated from the state and
Acipenser fulvescens, Hiodon tergisus, and Carpiodes carpio, if not
extirpated, probably do not now have reproducing populations in North
Carolina. Etheostoma vulneratum is restricted to the Little Tennessee
River system and Percina squamata occurs in low numbers in the
French Broad, Hiawassee, Little Tennessee, and Nolichucky river sys-
tems. Both warrant consideration for elevation to threatened status.

43

44 Fred C. Rohde, Mary L. Moser and Rudolf G. Arndt

Scartomyzon ariommus and Etheostoma podostemone have apparent
healthy populations in the Dan River system. The first record of
Ichthyomyzon bdellium from North Carolina is presented.

The North Carolina Wildlife Resources Commission lists 9 species of
fishes in this state as endangered, 1 1 as threatened, and 30 as of special concern
(Article 25, Chapter 113 of General Statutes of the State of North Carolina, 1987,
amended 1991). Since 1988 we have surveyed some of these species, and to date
have reported on the distribution and status of the sandhills chub, Semotilus lum-
bee, and the pinewoods darter, Etheostoma mariae (Rohde and Arndt 1991), and
of the sharphead darter, E. acuticeps (Rohde and Arndt 1994). In this paper we
add new, and summarize existing, data on the distribution and status of 22 of the
25 state-listed fishes that occur in the North Carolina portions of the French
Broad and Nolichucky river systems (Tennessee River drainage) and in the Dan
River system (Roanoke River drainage). The former are located in the Blue
Ridge Physiographic Province and the Dan River headwaters are located in
Appalachian Mountain remnants in north central North Carolina in the Piedmont
Physiographic Province. We also provide records on a species of fish new to
North Carolina.

Three endangered, five threatened, and nine of special concern fish
species (34% of the state total) occur in the French Broad River system; two
endangered, two threatened, and one species of special concern (10% of the
total) occur in the Nolichucky River system; and three endangered and two of
special concern fishes (10% of the total) occur in the Dan River system (Table
1).

SURVEY AREAS

The French Broad River originates in North Carolina and runs some
166 river kilometers (rkm) to where it enters Tennessee and drains approximate-
ly 4,163 km2 of North Carolina (Fig. 1). River elevation over this reach (Fig. 2)
drops from 640 m to 378 m; river gradient from its headwaters to Asheville is 2.6
m/km and from Asheville to the Tennessee border is 5.2 m/km (Richardson et al.
1963). Redmon Dam near Marshall in Madison County, North Carolina, pre-
vents upstream movement of fishes, and six species that might be expected to
occur farther upstream are known only from below the dam (Menhinick 1986).

The Nolichucky River and its three major tributaries (Fig. 2), the Cane,
North Toe, and South Toe rivers, drain an area of about 1,666 km2 (Crowell
1965). The Nolichucky River enters Tennessee at an elevation of 539 m (Crow-
ell 1965), and joins the French Broad River at Douglas Reservoir in Jefferson
County, Tennessee (Fig. 1).

Distribution of Fishes

45

Table 1. List (1991) of endangered, threatened, and special concern fishes found
in the mountainous portions of North Carolina (from Article 25, Chapter 1 13 of
General Statutes of the State of North Carolina, 1991).

Scientific Name

Common Name

River System Occurrence

ENDANGERED

Polyodon spathula
Exoglossum maxillingua
Scartomyzon hamiltoni
Noturus flavus
Noturus gilberti
Percina burtoni

Percina sclera

THREATENED
Lampetra appendix
Cyprinella monacha

Hybopsis rubrifrons
Luxilus chrysocephalus
Etheo stoma acuticeps
Percina caprodes

Aplodinotus grunniens
Cottus carolinae

Paddlefish
Cutlips minnow
Rustyside sucker
Stonecat

Orangefin madtom
Blotchside logperch

Dusky darter

Amer. brook lamprey
Spotfin chub

Rosyface chub
Striped shiner
Sharphead darter
Logperch

Freshwater drum
Banded sculpin

French Broad

Dan

Dan

Nolichucky

Dan

French Broad

Nolichucky

French Broad

French Broad
French Broad
Little Tennessee
Savannah
Nolichucky
Nolichucky
French Broad
New

French Broad
French Broad

SPECIAL CONCERN

Acipenser fulvescens Lake sturgeon

Hiodon tergisus Mooneye

Clinostomus funduloides Rosyside dace
Notropis lutipinnis Yellowfin shiner

Phenacobius teretulus
Carpiodes carpio
Scartomyzon ariommus
Noturus eleutherus
Etheostoma inscriptum
Etheostoma jessiae
Etheostoma podostemone
Etheostoma simoterum

Kanawha minnow
River carpsucker
Bigeye jumprock
Mountain madtom
Turquoise darter
Blueside darter
Riverweed darter
Snubnose darter

French Broad

French Broad

Little Tennessee

Little Tennessee

Savannah

New

French Broad

Dan

French Broad

Savannah

French Broad

Dan

French Broad

46

Fred C. Rohde, Mary L. Moser and Rudolf G. Arndt

Table 1. Continued.

Etheostoma vulneratum Wounded darter

Percina macrocephala
Percina oxyrhyncha
Percina squamata

Longhead darter
Sharpnose darter
Olive darter

French Broad
Little Tennessee
French Broad
New

French Broad
Hiwassee
Little Tennessee
Nolichucky

Fig. 1. Upper Tennessee River drainage, North Carolina and Tennessee.

C A R O UNA _

£.-~£o'u Na"~"

TENNESSEE /

G E O R *G I A

.-sovnH

The Dan River is the major southern tributary to the Roanoke River. It
originates on the Blue Ridge uplands in south central Virginia and, after a course
of 59 rkm, enters North Carolina in northwest Stokes County; it crosses the state
line five more times before joining the Roanoke River at Kerr Reservoir in Hal-
ifax County, Virginia (Fig. 3). The 140 rkm North Carolina portion drains
approximately 4,410 km2 of the state. River elevation drops from 366 m at the
point where it first enters North Carolina to 140 m where it first exits North Car-
olina in northeast Rockingham County (from U.S. Geological Survey 7.5 minute
series topographic maps).

Distribution of Fishes

47

Fig. 2. French Broad and Nolichucky river systems, North Carolina.

Fig. 3. Dan River system with sites (dots) we sampled from July 1992 to May
1995. Some dots overlap.

Patrick County %
TOWNESOAM ^

N

0 7.5 15
kilometers

PINNACLES \
POWER 4
PLANT \

JVV* TALBOTTOAM

. /LITTLE OAN RIVER T\

Homy County

VIRGINIA ~"S—

NORTH CAROUNA

Surry County

i

C DAN RIVER

J MAYO RIVER

-A

^S M

MMSON \r~r- "~^""

^

^

StotosCounty V*

Rockingham County

48

Fred C. Rohde, Mary L. Moser and Rudolf G. Arndt

All three rivers are generally bordered by forest, although some land is
pasture. River substrate ranges from bedrock to boulders to cobble to silt.

Fig. 4. French Broad and Nolichucky river systems with sites (dots) we sampled
from July 1991 to April 1995. Some dots overlap.

TENNESSEE

/

<^ p

\ -^r

/

Nolichucky River

_•'"

N^^

i "-vt L.

y

"\

/**"

French Broad River -*>Jfc=-;

Cane Rfver jj.

\ North Toerfiiver

vt\

'w J

/'jm

-^ — ^-r ) L

A South Toe River

1

N

) J

1

1 1

NORTH CAROLINA

\f

0

16 32
KILOMETERS

METHODS AND MATERIALS

We sampled 99 sites in 105 collections: 32 sites in the lower reach of
the French Broad River and some of its tributaries from May 1994 to May 1995,
and at 2 sites in the middle reach of the river in September 1993 (Fig. 4); 39 sites
in the Nolichucky River system between July 1991 and November 1994 (Fig. 4);
and 15 sites in the Dan River and in 2 of its tributaries 25 times from July 1992
to May 1995 (Fig. 3). We sampled most sites only once, although four sites each
in the French Broad and Nolichucky rivers were each sampled twice. We also
sampled 1 1 sites in the Virginia portion of the upper Dan River from November
1993 to October 1994 (Fig. 3). All site locations and dates are available from the
senior author. In addition, we include data from 146 collections made by others,
other personal communications, and from the literature.

We sampled primarily with a backpack electroshocker and seine, using
the technique described by Jenkins and Burkhead (1975). Each site was elec-
trofished for 45- 190 minutes, i.e., until we believed that sampling had been com-

Distribution of Fishes 4 9

prehensive. However, in the French Broad and in the Nolichucky rivers, we sam-
pled 12 sites in 1991 and 2 in 1994 only by seine (3.05 m x 1.2 m, 0.64 cm mesh),
and we sampled 2 sites in the former river in 1994 only by a 25 m, 14 cm
stretched mesh monofilament gill net and a 50 m, 5.1 cm stretched mesh
monofilament gill net. The nets were deployed overnight and fished on consec-
utive days for a total of six net days. In the Dan River in 1992 and in 1994, we
sampled two sites in each year only by seine (size as above). At five sites in the
North and South Toe rivers, in addition to sampling with electroshocker and
seine, we also surveyed fishes underwater by snorkeling. In all sampling efforts,
the known preferred habitat for each species was sampled most intensively.

In addition to the fishes taken, data on stream depth, width, and sub-
strate type; current; air and water temperatures; pH; and dissolved oxygen con-
centration were often recorded at a site, and we include these data when avail-
able. Fishes were preserved in 10% formalin upon capture for subsequent exam-
ination. Fish measurements when available are given; TL is total length and SL
is standard length. We deposited preserved specimens in the North Carolina
State Museum of Natural Sciences in Raleigh. Scientific and common names of
fishes used herein follow Mayden et al. (1992), except for Cyprinella monacha
which follows Jenkins and Burkhead (1994).

We include figures that show all of our known fish capture localities in
North Carolina. In our species accounts we occasionally include records of fish-
es taken in portions of adjacent states in an effort to make the accounts more
accurate and complete.

Positive results are gratifying, but, as usual, negative results are not nec-
essarily conclusive. When fish populations in rivers and large creeks decrease
strongly, it becomes virtually impossible to differentiate between occurrence at a
low level and extirpation (Etnier 1994).

RESULTS AND DISCUSSION
TENNESSEE RIVER DRAINAGE
ENDANGERED SPECIES

Paddlefish, Polyodon spathula (Walbaum)

The paddlefish once occurred throughout the Mississippi River and its
larger tributaries, but its distribution has decreased coincidental with river chan-
nelization, damming, and overfishing (Burr 1980). Cope (1870) maintained that
it migrated up the French Broad River as far as Asheville in Buncombe County,
North Carolina. Fishermen in this state reported that it had been caught in the
lower reaches of the French Broad River as recently as 1983, but none of these
reports has been substantiated by specimens (E. Menhinick, personal communi-
cation, 1994). We sampled with large-mesh gill nets in the French Broad River
downriver of Hot Springs and in the river at the mouth of Big Laurel Creek, both
Madison County, North Carolina, on 14 and 15 May and 15 and 16 August, 1994,

50

Fred C. Rohde, Mary L. Moser and Rudolf G. Arndt

respectively. Swift currents limited our efficiency. We caught no paddlefish.
Local fishermen in Madison County whom we questioned in 1994 told us that
they had never seen a paddlefish from North Carolina nor heard of one caught
there. We presume that the paddlefish has been extirpated from North Carolina.

Fig. 5. Distribution of the stonecat, Noturus flavus, (circle) and the blotchside
logperch, Percina burtoni, (star) in the French Broad and Nolichucky river sys-
tems, North Carolina. An open circle overlaps two historical sites where the
stonecat was not taken in this survey. Specific historical sites for the blotchside
logperch in Cane Creek and the Swannanoa River are not known and are plotted
as circles with a question mark.

Stonecat, Noturus flavus Rafinesque

The stonecat is distributed through portions of the Mississippi River
basin, the Great Lakes, the Ohio River basin, and the St. Lawrence, Mohawk, and
Hudson River systems (Rohde 1980). In North Carolina it is documented from
only three sites in the Cane River, where it was collected on 1 8 June (one speci-
men) and 26 June (six), 1984, and 15 September 1985 (two) (Menhinick 1986).
Despite our efforts to collect it at all three sites, we took two adults (pho-
tographed and released) only at the downstream-most site on 4 September 1993
(Fig. 5).

Distribution of Fishes 5 -j

We record it here for the first time from the Ivy River, a tributary to the
French Broad River, from upstream of Marshall in Madison County, North Car-
olina, where we took three adults (90-99 mm SL) on 14 August 1994 (Fig. 5).
One adult (86 mm TL) and one specimen (released, not measured) were collect-
ed in the Little Tennessee River at Needmore in Swain County, North Carolina,
by Tennessee Valley Authority (TVA) personnel on 20 June 1994 (E. Scott, per-
sonal communication, 1994) (Fig. 1). This site had been sampled five times by
TVA biologists during the period 1988-1993 and once by us in 1993. The dis-
covery of the species there in 1994 was unexpected. Preferred habitat was grav-
el riffles. Current at the Ivy River site was 0.31 m/sec, water temperature 20.6
C, pH 8.0, and dissolved oxygen concentration 8.0 ppm. Its status of endangered
in North Carolina is warranted.

Blotchside logperch, Percina burtoni Fowler

The blotchside logperch occurs in disjunct populations in the Ten-
nessee River drainage from westcentral Tennessee to southwestern Virginia
(Page and Burr 1991); where it occurs it is localized and rare (Etnier 1994). In
North Carolina it was taken at one site in Cane Creek (Henderson County) in
1902, at one site in the Swannanoa River (Buncombe County) in 1934, and at
two sites in the South Toe River (Yancey County) in 1975 and 1977 (Menhinick
1986) (Fig. 5). We collected two adults (120 mm SL, one released) at a new
locality in the South Toe River near its confluence with the North Toe River in
September 1993. Since we noted that this fish can readily avoid electroshockers
and seines, we observed 2 to 6 adults in each of 4 visits by snorkeling at the two
upstream historic sites in the South Toe River in July and September 1993 and
August 1995 (Fig. 5). Preferred habitat was in pools below riffles. Menhinick
(1986) presumed P. burtoni to have been extirpated from the Swannanoa River
and from Cane Creek since he did not obtain it there in 14 collections nor did the
North Carolina Division of Environmental Management in 5 collecions (V
Schneider, personal communication, 1994). We did not collect it in either stream
in two collections made there in 1993, and we concur with Menhinick (1986). Its
continued presence in North Carolina is tenuous.
Dusky darter, Percina sciera (Swain)

The dusky darter occurs from the Wabash River drainage in Indiana
south and west to the Guadalupe River drainage in Texas and east to the Tombig-
bee-Black Warrior river system in Alabama (Page 1980). In North Carolina it is
known only from Spring Creek, Madison County, where U.S. Forest Service per-
sonnel collected one specimen in 1966 and one in 1969 (Auburn University Col-
lection 3442), although the specific sites are now not known (M. Seehorn, per-
sonal communication, 1994). We made 13 collections at 7 sites with suitable
habitat in Spring Creek over a distance of 27.3 rkm in 1994 and 1995. We did
not collect it. Apparently never widespread or common in North Carolina, we

52

Fred C. Rohde, Mary L. Moser and Rudolf G. Arndt

consider the dusky darter to have been extirpated from this state by unknown
causes. It appears to be a relatively tolerant species in other portions of its range.

THREATENED SPECIES

American brook lamprey, Lampetra appendix (DeKay)

The American brook lamprey is widely distributed in the St. Lawrence
and Mississippi river basins from New York to northern Arkansas, and on
Atlantic Slope drainages from southern Quebec south to the Roanoke River
drainage in Virginia (Page and Burr 1991). In North Carolina it was known from
only one site in the downstream reach of Spring Creek, Madison County, at a
point where a railroad trestle crosses this creek, where 26 individuals were taken
in 1980 and 1 in 1983 (Menhinick 1986). We took one adult some 200 m down-
stream of the above site (and about 50 m upstream of the confluence of Spring
Creek with the French Broad River, Madison County) on 22 April 1995 (Fig. 6),
and two adults on the same day, also in Spring Creek at a point about 0.9 rkm
above this confluence. They were males and measured 144, 147, and 157 mm
TL. All were taken in gravel riffles where the current was 0.45 m/sec. We col-
lected one ammocoetes of 152 mm TL in a sandy-bottomed pool about 50 m
downstream of the last-mentioned Spring Creek site on 14 August 1994. The pH
here was 6.9, and the dissolved oxygen concentration was 8.4 ppm. This species
appears to be restricted to this creek in North Carolina. Its status of threatened
in North Carolina appears to be conservative.

Fig. 6. Distribution of the American brook lamprey, Lampetra appendix, (circle)
and the striped shiner, Luxilus chrysocephalus, (star) in the French Broad and
Nolichucky river systems, North Carolina. Some symbols overlap sites.

NORTH CAROLINA

Distribution of Fishes 5 3

Spotfin chub, Cyprinella monacha (Cope)

The spotfin chub is endemic to the Tennessee River drainage in disjunct
populations from southwestern Virginia to northwestern Alabama and in the Buf-
falo River in central Tennessee (Jenkins and Burkhead 1984, Etnier and Starnes
1994). It has apparently been extirpated from Alabama and Georgia (Etnier and
Starnes 1994). In North Carolina it is restricted to a 16.8 rkm section of the Lit-
tle Tennessee River in Macon and Swain counties (Alderman 1987). We col-
lected and released 27 spotfin chub in this river at Needmore in Swain County
(Fig. 1) on 25 September 1993. Three previous North Carolina records, two
from the French Broad River system (1888) and one from the Tuckaseegee River
(1940), apparently represent now extirpated populations (Menhinick 1986). We
did not take it in 36 collections in the French Broad River system, and we con-
cur with Menhinick (1986) that it has been extirpated from this drainage. Its sta-
tus of threatened in North Carolina appears to be conservative.

Striped shiner, Luxilus chrysocephalus Rafinesque

The striped shiner is common in the southern Great Lakes basin from
western New York and southeastern Wisconsin south through much of the Mis-
sissippi River basin almost to the Gulf of Mexico (Page and Burr 1991). A dis-
junct population was discovered in the Cane River, Yancey County, North Car-
olina in 1980 by E. Menhinick, and soon thereafter it was known from five sites
in the Cane River system (Menhinick 1986). We found it in four sites in 14.5
rkm of the middle reach of the Cane River and in two tributaries, Bald and Indi-
an creeks (Fig. 6), in 1994. Numbers taken per our collections ranged from 2-
23. Specimens ranged from 29-107 mm TL. Tennessee Valley Authority biolo-
gists took 61 individuals in one collection in the Cane River in 1992 (E. Scott,
personal communication, 1994) (Fig. 6). Preferred habitat was pools and runs.
Current ranged from 0.39-0.57 m/sec, pH 7.0-7.5, and dissolved oxygen concen-
tration 6.8-8.4 ppm. Its status of threatened in North Carolina is warranted.

Banded sculpin, Cottus carolinae (Gill)

The banded sculpin inhabits mountainous areas of the Mississippi River
basin from West Virginia west to Kansas and from the Ozark Mountains south-
east to southern Alabama (Page and Burr 1991). In North Carolina it was report-
ed only from Big Laurel and Spring creeks, Madison County (Robins 1954).
Menhinick (1986) later reported it as restricted to the main stream of the French
Broad River in North Carolina near the Tennessee line and absent from the two
creeks. E. Menhinick (personal communication, 1994) took two adults and eight
juveniles with rotenone in the downstream-most 100 m of Shut-in Creek, Madi-
son County, North Carolina in July 1994 (Fig. 7). We did not collect it at two
upstream-sites in this creek in 1994. However, we did take 58 specimens in two
collections made on 14 May and 19 July 1994 throughout the lower 300 m of

54

Fred C. Rohde, Mary L. Moser and Rudolf G. Arndt

Paint Creek, Greene County, Tennessee. This creek enters the French Broad
River some 120 m downstream of the North Carolina/Tennessee line (Fig. 7). Its
status as threatened appears to be conservative.

Fig. 7. Distribution of the banded sculpin, Cottus carolinae, (circle) and the
freshwater drum, Aplodinotus grunniens, (star) in the French Broad River sys-
tem, North Carolina. A circle with a question mark indicates an undefined his-
torical site of the banded sculpin. Some symbols overlap sites.

Logperch, Percina caprodes (Rafinesque)

The logperch occurs from central Canada and the upper Mississippi
River and adjacent drainages south to the Gulf of Mexico, and on Atlantic Slope
drainages from the Hudson River south to portions of the Chesapeake Bay
drainage (Rohde et al. 1994). In North Carolina it is known from four sites in
the French Broad River between Redmon Dam and the Tennessee state line
(Harned 1979) (Fig. 8), four specimens were collected below Redmon Dam in
1986 and 1987 (Birchfield et al. 1987) (Fig. 8), and from one site in the New
River, Allegheny County (Menhinick 1986). We collected eight adults (88-132
mm SL) at four sites in the downstream reaches of the French Broad River and
at two sites in the downstream portion of Spring Creek, Madison County on 13
May; 19, 22 July; and 5 November 1994 (Fig. 8). One specimen was taken in

Distribution of Fishes

55

Fig. 8. Distribution of the logperch, Percina caprodes, (circle) and the olive
darter, Percina squamata, (star) in the French Broad and Nolichucky river sys-
tems, North Carolina. Some symbols overlap sites.

TENNESSEE

/

cd^ ^

^W^

/'

N— I

WQJj/^hi ir*ir-iy Diuar

\^£

.s-~

S

-\

f**

French Broad River -*~^~,

~}\

Cane River j

1 North Toeffiiver

vV

*w \

r"J\

-*/ — ^-r ) F

l/_

) r"

\

/* South Toe River

1

N

1 J

NORTH CAROLINA

\f

0

1 6 32
KILOMETERS

Spring Creek 7.1 rkm upstream from its mouth in July 1994 (S. Bryan, personal
communication, 1994) (Fig. 8). Preferred habitat in the river was runs with large
boulders. Current at our Spring Creek site was 0.58 m/sec, pH 7.1-7.6, and dis-
solved oxygen concentration 7.1-10.5 ppm. Its status of threatened is warranted.

Freshwater drum, Aplodinotus grunniens Rafinesque

The freshwater drum occurs throughout the Mississippi River basin
from southern Canada and the Great Lakes to western Texas and western Flori-
da (Rohde et al. 1994). Prior to this survey, it was known in North Carolina from
six sites in the lower reaches of the French Broad River downstream of Redmon
Dam, Madison County (Harned 1979) (Fig. 7). We collected one large specimen
(305 mm TL) in a pool in Spring Creek, at a point 1 rkm upstream of its conflu-
ence with the French Broad River, on 22 July 1994, and E. Menhinick (personal
communication, 1994) took one specimen in the same month in Spring Creek at
this confluence (Fig. 7). Its status of threatened is warranted due to the lack of
juveniles in collections.

5 6 Fred C. Rohde, Mary L. Moser and Rudolf G. Arndt

SPECIES OF SPECIAL CONCERN

Lake sturgeon, Acipenser fulvescens Rafinesque

The lake sturgeon is usually found over shoals in lakes and large rivers
in central Canada and Hudson Bay and St. Lawrence River drainages, and in
much of the Mississippi River drainage south to northeastern Louisiana (Page
and Burr 1991). Eight specimens, presumably of this species, were taken from
the French Broad River near Hot Springs in Madison County, North Carolina in
1945 (Brimley 1946). An occasional lake sturgeon is still reported from Douglas
Reservoir in Jefferson County, Tennessee, but these are unsubstantiated records
(Etnier and Starnes 1994). We set 2 large-mesh gill nets of 25 and 50 m total
length in the French Broad River downstream of Hot Springs in mid-May 1995
and in the river at the mouth of Big Laurel Creek in mid- August but failed to col-
lect sturgeon. Swift current limited sampling location possibilities at the former
site and reduced gear efficiency. Local North Carolina state fishery biologists
have no reported sightings (J. Borawa, personal communication, 1994). Men-
hinick (1986) considers the lake sturgeon to have been extirpated from North
Carolina, and we concur.

Mooneye, Hiodon tergisus Lesueur

The mooneye is found in central and southern Canada and in much of
the Mississippi River basin from the Great Lakes south to the Gulf of Mexico
(Page and Burr 1991). It historically occurred in the upper reaches of the French
Broad River near Bowman's Bluff, Henderson County, North Carolina in 1902
(Smith 1907), but it is now known only from Redmon Dam to the Tennessee
state line (Menhinick 1986) based on several mooneye obtained from fishermen
in the French Broad River just above the confluence with Big Laurel Creek by
Harned (1979). We did not take it in this river in our electroshocker or gill net
collections. Its status of special concern in North Carolina appears to be conser-
vative.

River carpsucker, Carpiodes carpio (Rafinesque)

The river carpsucker occurs throughout the Mississippi River basin
from Montana to Pennsylvania and south to the Gulf of Mexico (Lee and Plata-
nia 1980). There is one North Carolina 1947 record from the French Broad River
near Hot Springs in Madison County (Menhinick 1986). It was also captured in
the same river in Tennessee 41 rkm downstream of the North Carolina state line
in 1979 (Harned 1979), but we failed to collect it in this study. Its status of spe-
cial concern appears to be conservative.

Mountain madtom, Noturus eleutherus Jordan

The mountain madtom occurs in disjunct populations from northwest-
ern Pennsylvania south and west through the Ohio River basin to the Red and

Distribution of Fishes 57

Ouachita river drainages in Oklahoma and Arkansas (Page and Burr 1991). The
only verified North Carolina specimens are from Spring Creek, Madison Coun-
ty and were collected in 1889 (Taylor 1969). It was also collected at two sites in
the French Broad River just upstream of Douglas Reservoir, Cocke County, Ten-
nessee (32 km downriver of North Carolina) during a 1979 TVA survey (Harned
1979). We did not collect it at any of the 34 sites we surveyed in the lower reach-
es of the French Broad River system. We concur with Menhinick (1986) that it
has been extirpated from North Carolina.

Snubnose darter, Etheostoma simoterum (Cope)

The snubnose darter is abundant in the Tennessee River drainage from
southwestern Virginia to northern Alabama (Rohde et al. 1994). The only puta-
tive extant specimen from North Carolina was reported by Cope (1870) and is
now in the United States National Museum, but it is unclear from Cope's records
whether its provenance is North Carolina or Tennessee (Menhinick 1986). Men-
hinick (1986) reported two unverified records from Laurel and Spring creeks,
Madison County, North Carolina. We collected no snubnose darter, nor did Men-
hinick (1986). M. Hopey, who made 13 collections for a general survey of the
streams in this area for the Western North Carolina Alliance in 1992, did not col-
lect it (M. Kelly, personal communication, 1994). We consider the past or pre-
sent occurrence of this darter in North Carolina to be highly doubtful.

Wounded darter, Etheostoma vulneratum (Cope)

The wounded darter is restricted to the upper Tennessee River drainage
from Virginia to Georgia (Rohde et al. 1994). It is abundant in the Little Ten-
nessee River in North Carolina (F. Rohde, personal observations). Although the
type locality is Spring Creek, Madison County, North Carolina (Cope 1870),
none has been reported from the French Broad River system in North Carolina
since then, including our survey. Harned (1979) collected one specimen in the
French Broad River in Tennessee at a point approximately 23 km downstream of
the North Carolina state line. We conclude that it has been extirpated from the
French Broad River system in North Carolina. Its status of special concern in
North Carolina appears conservative.

Olive darter, Percina squamata (Gilbert and Swain)

The olive darter is confined to the Rockcastle and Big South Fork rivers
in the Cumberland River drainage in Kentucky and Tennessee and to the upper
Tennessee River drainage (Rohde et al. 1994). There are five records from the
lower reaches of the French Broad River system (three in the main river and two
in Spring Creek) in North Carolina, and four records from the Nolichucky River
system (three in Cane River and two in North Toe River); it also occurs in the
Little Tennessee and upper Hiwassee rivers in the state (Menhinick 1991). We

58

Fred C. Rohde, Mary L. Moser and Rudolf G. Arndt

did not collect it in the French Broad River system, but we collected one juve-
nile (38 mm SL) in the Cane River on 19 July 1993, and three adults (104 mm
SL, two released) in the South Toe River on 5 September 1993 (Fig. 8). Pre-
ferred habitat is around large boulders in fast riffles. Its status of special concern
in North Carolina appears to be conservative.

NEW STATE RECORD

Ohio lamprey Ichthyomyzon bdellium (Jordan)

The Ohio lamprey occurs in disjunct populations in the Ohio River
basin, where it is uncommon (Rohde and Lanteigne-Courchene 1980). Because
of its presence in nearby Tennessee, Menhinick et al. (1974) listed its occurrence
in North Carolina as probable. However, there were no records from North Car-
olina until we took one male and three females from the mouth of Spring Creek,
Madison County, on 14 May 1994 (Fig. 9). Each was adult (220-260 mm TL),
had a well-developed digestive tract, and each female was gravid. We took
another three males (243-246 mm TL), one female (244 mm TL), and two juve-
niles (not ammocoetes, nor mature adults) (152, 153 mm TL) here on 22 April
1995, as well as two adult males (239, 248 mm TL) 1 rkm further upstream in
this creek on the same day. We took three females (235-240 mm TL) in nearby
Paint Creek, Greene County, Tennessee on 14 May 1995. All our specimens
were taken over rocky riffles with a current from 0.45-0.78 m/sec.

Fig. 9. Distribution of the Ohio lamprey, Ichthyomyzon bdellium, in the French
Broad River system, North Carolina.

TENNESSEE

Nolichucky River ^ -.

Distribution of Fishes

59

DAN RIVER SYSTEM
ENDANGERED

Cutlips minnow, Exoglossum maxillingua (Lesueur)

The cutlips minnow occurs on the Atlantic slope from the St. Lawrence
River and eastern Lake Ontario drainages south to the upper Dan River in North
Carolina (Gilbert and Lee 1980). Menhinick (1986) reported it from one site on
the Dan River in Stokes County, North Carolina, within 1.6 rkm downstream of
the Virginia state line. We found it in the North Carolina portion of this river at
four sites from the Virginia line downstream to NC Route 704 and at six Virginia
sites upstream to the Pinnacles Power Plant, over a total distance of 43 rkm (Fig.
10). Numbers (39) taken in our collections ranged from 1-6 (mean 3.9), and their
length ranged from 69-133 mm SL; most specimens were adults. This species
preferred fast-flowing runs or pools, near large rocks or boulders over sand and
gravel. Current where it was collected was 0.54-0.75 m/sec; water temperature
7.7 C (November)-22 C (July); pH 6.8-7.6; and dissolved oxygen concentration
10.6-11.8 ppm (both November). The species appears to be secure in its limited
distribution in North Carolina.

Fig. 10. Distribution of the cutlips minnow, Exoglossum maxillingua, in the Dan
River system, North Carolina and Virginia.

60

Fred C. Rohde, Mary L. Moser and Rudolf G. Arndt

Rustyside sucker, Thoburnia hamiltoni Raney and Lachner

The rustyside sucker is endemic to the upper Dan River system in North
Carolina and Virginia (Jenkins and Burkhead 1994). In North Carolina it is
known only from the 1 .4 rkm downstream-most portion of the Little Dan River
in Stokes County. Here Menhinick (1986) collected four specimens at a point
some 400 m downriver of the Virginia line in 1985 (Fig. 11). We made three col-
lections in the Little Dan River, from its confluence with the Dan River upstream
to the North Carolina/Virginia line, and took one adult (144 mm SL) in a run with
gravel and rubble substrate on 21 December 1992. In Virginia we took three
adults (1 18-142 mm SL) in the Dan River at one site located 365 m, and at anoth-
er site 914 m, downriver of the Pinnacles Power Plant on 28 November 1993
(Fig. 11). Both sites were deep and fast rock-strewn riffles, current velocity 0.62-
0.75 m/sec, water temperature 8.3 C, pH 6.8, and dissolved oxygen concentra-
tion 10.0 ppm. The distribution of the rustyside sucker in North Carolina is
extremely limited, and its continued existence there is precarious.

Fig. 11. Distribution of the rustyside sucker, Thoburnia hamiltoni, in the Dan
River system, North Carolina and Virginia.

Distribution of Fishes

61

Fig. 12. Distribution of the orangefin madtom, Noturus gilberti, in the Dan River
system. An open circle indicates an historical site where the species was not
taken in this survey.

Patrick County

TOWNES DAM

PINNACLES
POWER
PLANT

Surry County

Orangefin madtom, Noturus gilberti Jordan and Evermann

The orangefin madtom is a Roanoke River drainage endemic (Jenkins
and Burkhead 1994). Menhinick (1986) reported it from five North Carolina
localities in the Dan River from the Virginia line downstream to Danbury, all dis-
covered after 1968. Simonson and Neves (1986) found it at four Dan River sites
in 45.2 rkm in North Carolina during a 1985 survey; including three of the Men-
hinick (1986) sites. We collected two adults (photographed and released) at 2
sites (of 16 sampled) in the Dan River in North Carolina 1.6 and 7.3 rkm below
the Virginia line on 16 and 17 July 1992 (Fig. 12). Personnel from the North
Carolina Wildlife Resources Commission took two individuals (93 and 96 mm
TL) at an intermediate site on 5 October 1990 (A. Braswell, North Carolina State
Museum of Natural Sciences, personal communication, 1992) (Fig. 12). We did
not take it in three collections made at the downstream-most historical locality in
the Dan River near Danbury in Stokes County, North Carolina, nor in five col-

6 2 Fred C. Rohde, Mary L. Moser and Rudolf G. Amdt

lections made at three other historic localities upstream of this site in this river
and state (Fig. 12). We took the first specimens known from the Little Dan River
in North Carolina, at a point 1.1 rkm upstream of the confluence with the Dan
River: one adult (released) on 21 December 1992, and one subadult (68 mm SL)
on 21 August 1993 (Fig. 12). We also took three adults and one subadult (44-75
mm SL) at one locality in the Dan River in Virginia on 10 June 1994 (Fig. 12).
Simonson and Neves (1986) took it at four Dan River sites in 12.4 rkm in Vir-
ginia. All fishes were taken in riffles with a gravel/rubble substrate; pH was 7.6-
8.0.

Our capture of the two specimens in the Little Dan River was surpris-
ing, especially since Jenkins and Burkhead (1994) presumed that it had disap-
peared from this river. Since we did not take it at the four historical downstream-
most sites in the Dan River in North Carolina (Fig. 12), we suspect that it may
now be absent there. Its status of endangered in North Carolina is warranted.

SPECIAL CONCERN

Bigeye jumprock, Scartomyzon ariommus (Robins and Raney)

The bigeye jumprock is endemic to the upper and middle portions of the
Roanoke River drainage in North Carolina and Virginia (Jenkins and Burkhead
1994). We found it at eight sites in a 41 rkm section of the Dan River, Stokes
and Rockingham counties, North Carolina, between July 1992 and May 1995
and in the Mayo River, a tributary of the Dan River, 1.1 rkm below the Virginia
state line on 22 August 1993 (Fig. 13). We collected 32 adults (106-170 mm SL,
mean 154.1 mm) and 3 juveniles (52-91 mm SL, mean 75.3 mm); collections
ranged from 1-7. Adults were taken in deep runs and heads of pools, usually near
large boulders and rock outcrops, and juveniles in a shallow gravel riffle (one
specimen) and in a sandy-bottomed pool (two); pH at capture sites was 7.4-7.8.
Duke Power Company personnel (unpublished data) took two individuals by
electroshocker at the Dan River Steam Station in Eden in Rockingham County,
North Carolina, in August 1990 (Fig. 13). Within its limited distribution in North
Carolina, this difficult-to-collect species is apparently secure.

Riverweed darter, Etheostoma podostemone Jordan and Jenkins

The riverweed darter is endemic to the upper Roanoke River drainage
in North Carolina and Virginia (Jenkins and Burkhead 1994). It is widely dis-
tributed in the Dan River and its tributaries (Menhinick 1991). We found it in a
124 rkm section of the main Dan River in North Carolina between July 1992 and
May 1995, as well as in the Little Dan River, Mayo River, and Virginia portion
of the upper Dan River (Fig. 14). We collected 434 specimens from 21-61 mm
SL, ranging 2-62 (mean 21.7) per collection. It was common in shallow riffles
with a gravel/cobble substrate; pH was 7.4-8. 1. Within its limited distribution in
North Carolina, the species is apparently secure.

Distribution of Fishes

63

SUMMARY AND CONCLUSIONS

We conducted surveys for 22 North Carolina fishes currently consid-
ered to be endangered, threatened, and of special concern and known from the
French Broad River (14 species), Nolichucky River (5 species), and Dan River
system (5 species). We made 105 collections over the period July 1991 -May
1995 inclusive, and augmented these collections with data from the literature and
from personal communications.

Fig. 13. Distribution of the bigeye jumprock, Scartomyzon ariommus, in the
Dan River system, North Carolina.

64

Fred C. Rohde, Mary L. Moser and Rudolf G. Arndt

We did not collect two of the seven species considered to be endan-
gered, the paddlefish and dusky darter, and believe that they have been extirpat-
ed from the state. The cutlips minnow in North Carolina is restricted to the Dan
River, where we found it in 2 1 rkm rather than the 1 .6 rkm in which it was pre-
viously known; this population at present appears to be secure. The rustyside
sucker in North Carolina is confined to the downstream reaches of the Little Dan
River, where it is extremely rare. We discovered that the North Carolina distri-
bution of the stonecat is significantly larger than the small portion of the Cane
River from which it was previously known, and to also include the Ivy River and
the Little Tennessee River. Nevertheless, its status of endangered is warranted.
The orangefin madtom in North Carolina is restricted to the Dan and to the Lit-
tle Dan rivers; its distribution there appears to have decreased. The blotchside
logperch in North Carolina has been extirpated from historical sites in the French
Broad River system, and it is today restricted to 24 rkm of the South Toe River;
its continued existence there appears to be tenuous.

Fig. 14. Distribution of the riverweed darter, Etheostoma podostemone, in the
Dan River system, North Carolina and Virginia.

Records of all six North Carolina species considered threatened were
collected in our survey. The American brook lamprey in North Carolina is
known only from the downstream portion of Spring Creek; consequently, we

Distribution of Fishes 6 5

believe that its status in North Carolina should be increased to endangered. The
spotfin chub has apparently been extirpated from at least Alabama and Georgia,
as well as from the French Broad River system and the Tuckaseegee River in
North Carolina, and it remains only in the Little Tennessee River; its status in
North Carolina should probably be elevated to endangered. A viable population
of the striped shiner is present in the Cane River. Its status of threatened in North
Carolina is appropriate. Although we did not collect the banded sculpin, E. Men-
hinick took 10 specimens in North Carolina in a creek tributary to the lower
French Broad River in 1994. Its status more accurately may be described as
endangered. The logperch was found at several sites in the French Broad River
and in Spring Creek. Its North Carolina status of threatened is warranted. Two
freshwater drum were taken in Spring Creek. We doubt that a viable population
of this species occurs in North Carolina, and it appears to be endangered.

Of the nine listed species of special concern that we surveyed, the big-
eye jumprock and riverweed darter have apparently healthy populations in the
limited Dan River system area in which they occur in North Carolina. The
wounded darter in North Carolina apparently has been extirpated from the
French Broad River system, although it is still abundant in the Little Tennessee
River. Its status in North Carolina is more accurately described as threatened.
The olive darter apparently still occurs in very low numbers in the French Broad
River system, as well as in the Cane, North Toe, and South Toe, but its current
status appears too conservative. We believe that the mountain madtom and the
snubnose darter have been extirpated from North Carolina. We did not collect
the lake sturgeon, mooneye, or the river carpsucker, nor could we obtain any
recent records. The occasional specimens of these larger species, if they still
occur, would be exceedingly difficult to catch. If they have not yet been extir-
pated from North Carolina, they almost certainly do not today have reproducing
populations there.

We report the first North Carolina specimens of the Ohio lamprey, from
Spring Creek, Madison County.

ACKNOWLEDGMENTS - We thank for assistance in field work those students
too numerous to mention here individually from The Richard Stockton College
of New Jersey (TRSCNJ) and the University of North Carolina at Wilmington,
and especially H. Jackson and T. Ward from the latter; this work benefitted great-
ly from their help. E. Menhinick, University of North Carolina-Charlotte, gen-
erously provided collection data from the French Broad River system. A.
Braswell, North Carolina State Museum of Natural Sciences, answered questions
on specimens in his care. We are indebted to all individuals, named and
unnamed in this paper, who kindly responded to our requests and provided infor-
mation. We are indebted to B. Bragin and to V. Schneider of the North Carolina
Division of Environmental Management (now with North Carolina Museum of

6 6 Fred C. Rohde, Mary L. Moser and Rudolf G. Amdt

Natural Sciences), for the loan of backpack electroshockers. The Graphics Pro-
duction Department of TRSCNJ, and L.M. Neveu, helped to prepare the figures.
This research was supported in part by grants from the North Carolina Wildlife
Resources Commission, Nongame and Endangered Wildlife Program, and by
travel funds and vehicles provided by TRSCNJ. These surveys were conducted
under North Carolina Wildlife Resources Scientific Collecting Permit Number
39.

LITERATURE CITED

Alderman, J. M. 1987. Population status and distribution of the spotfin chub,
Hybopsis monacha (Cope), in the Little Tennessee River, North Carolina.
Unpublished manuscript. Final report to the United States Fish and Wildlife
Service.

Birchfield, L. J., W. T. Bryson, C. R. Cofield, K. A. MacPherson, and W. E.
Partin. 1987. Marshall Hydraulic Plant. Fish passage and survival study.
Unpublished manuscript. Carolina Power and Light Company Report.

Brimley, C. S. 1946. The freshwater fishes of North Carolina. Installment 1:
Carolina Tips 9(3): 10-11.

Burr, B. M. 1980. Polyodon spathula (Walbaum), Paddlefish. Pages 45-46 in
Atlas of North American freshwater fishes (D. S. Lee et al.) North Caroli-
na State Museum Natural History, Raleigh.

Cope, E. D. 1870. A partial synopsis of the fishes of the freshwaters of North
Carolina. Proceedings of the American Philosophical Society 11:448-495.

Crowell, T. E. 1965. Survey and classification of the Toe River watershed and
tributaries, North Carolina. Final Report, Job I-U, Project F-14-R, North
Carolina Wildlife Resources Commission, Raleigh.

Etnier, D. A. 1994. Our southeastern fishes — what have we lost and what are
we likely to lose. Proceedings of the Southeastern Fishes Council Number
29:5-9.

Etnier, D. A., and W. C. Starnes. 1994. The fishes of Tennessee. University of
Tennessee Press, Knoxville.

Gilbert, C. R., and D. S. Lee. 1980. Exoglossum maxillingua (LeSueur), Cut-
lips minnow. Page 158 in Atlas of North American freshwater fishes (D. S.
Lee et al.) North Carolina State Museum Natural History, Raleigh.

Harned, W. D. 1979. A qualitative survey of fish and macroinvertebrates of the
French Broad River and selected tributaries, June- August 1977. Technical
Note B35. Tennessee Valley Authority, Norris.

Jenkins, R. E., and N. M. Burkhead. 1975. Recent capture and analysis of the
sharphead darter, Etheostoma acuticeps, an endangered percid fish of the
upper Tennessee River drainage. Copeia 1975(4):73 1-740.

Jenkins, R. E., and N. M. Burkhead. 1984. Description, biology, and distribu-
tion of the spotfin chub, Hybopsis monacha, a threatened cyprinid fish of the

Distribution of Fishes 6 7

Tennessee River drainage. Bulletin of the Alabama Museum Natural Histo-
ry 8:1-30.

Jenkins, R. E., and N. M. Burkhead. 1994. Freshwater fishes of Virginia. Amer-
ican Fisheries Society, Bethesda, Maryland.

Lee, D. S., and S. P. Platania. 1980. Carpiodes carpio (Rafinesque), River carp-
sucker. Page 367 in Atlas of North American freshwater fishes (D. S. Lee
et al.) North Carolina State Museum Natural History, Raleigh.

Mayden, R. L., B. M. Burr, L. M. Page, and R. R. Miller. 1992. The native
freshwater fishes of North America. Pages 827-863 in Systematics, histori-
cal ecology, and North American fishes (Mayden, R. M. editor). Stanford
University Press, Stanford, California.

Menhinick, E. F. 1986. A numerical method for ranking of endangered species
and its application to North Carolina freshwater fishes. Journal of the Elisha
Mitchell Scientific Society 102:54-86.

Menhinick, E. F. 1991. The freshwater fishes of North Carolina. Delmar Com-
pany, Charlotte, North Carolina.

Menhinick, E. F., T. M. Burton, and J. R. Bailey. 1974. An annotated checklist of
the freshwater fishes of North Carolina. Journal of the Elisha Mitchell Sci-
entific Society 90(l):24-50.

Page, L. M. 1980. Percina sciera (Swain), Dusky darter. Page 740 in Atlas of
North American freshwater fishes (D. S. Lee et al.) North Carolina State
Museum Natural History, Raleigh.

Page, L. ML, and B. M. Burr. 1991. A field guide to freshwater fishes of North
America north of Mexico. Houghton Mifflin, Boston.

Richardson, F. R., H. M. Ratledge, and J. B. Messer. 1963. Survey and classi-
fication of the French Broad River and tributaries, North Carolina. Final
Report, Job I-I, Project F-14-R, North Carolina Wildlife Resources Com-
mission, Raleigh.

Robins, C. R. 1954. A taxonomic revision of the Cottus bairdi and Cottus car-
olinae species groups in eastern North America. Ph.d. Thesis, Cornell Uni-
versity, Ithaca, New York.

Rohde, F. C. 1980. Noturus flavus (Rafinesque), Stonecat. Page 455 in Atlas of
North American Freshwater Fishes (D. S. Lee et al.) North Carolina State
Museum Natural History, Raleigh.

Rohde, F. C, and R. G. Arndt. 1991. Distribution and status of the sandhills
chub, Semotilus lumbee, and the pinewoods darter, Etheostoma mariae.
Journal of the Elisha Mitchell Scientific Society 107(2):61-70.

Rohde, EC, and R. G. Arndt. 1994. Distribution and abundance of the
sharphead darter, Etheostoma acuticeps (Percidae), in North Carolina.
Association of Southeastern Biologists Bulletin 41(3): 153-159.

6 8 Fred C. Rohde, Mary L. Moser and Rudolf G. Arndt

Rohde, F. C, and J. Lanteigne-Courchene. 1980. Ichthyomyzon bdellium (Jor-
dan), Ohio lamprey. Page 15 in Atlas of North American freshwater fishes
(D. S. Lee et al.) North Carolina State Museum Natural History, Raleigh.

Rohde, F. C, R. G. Arndt, D. G. Lindquist, and J. F. Parnell. 1994. Freshwater
fishes of the Carolinas, Virginia, Maryland, and Delaware. The University
of North Carolina Press, Chapel Hill.

Simonson, T. D., and R. J. Neves. 1986. A status survey of the orangefin mad-
torn and Roanoke logperch. Unpublished manuscript, Virginia Cooperative
Fish and Wildlife Research Unit, Virginia Tech, Blacksburg.

Smith, H. M. 1907. The fishes of North Carolina. North Carolina Geologic and
Economic Survey, Volume 2, Raleigh, North Carolina.

Taylor, W. R. 1969. A revision of the catfish genus Noturus Rafinesque with an
analysis of higher groups in the Ictaluridae. Bulletin United States Nation-
al Museum 282:1-315.

Received 29 February 1996
Accepted 13 August 1996

Status of the River Frog, Rana heckscheri (Anura: Ranidae), in

North Carolina

Jeffrey C. Beane

North Carolina State Museum of Natural Sciences

P.O. Box 29555

Raleigh, North Carolina 27626-0555

ABSTRACT-The river frog (Rana heckscheri), a large ranid occurring
in aquatic and riparian habitats in the southeastern United States, reach-
es the northern edge of its range in southeastern North Carolina, where
it has been recorded historically from a few scattered localities in the
Lumber and Cape Fear river systems. Currently listed by the state as a
species of Special Concern, R. heckscheri was last documented from
North Carolina in 1975. A survey was undertaken to determine the
frog's status in the state. Considerable field work has failed to yield any
current evidence of its existence in North Carolina, and it appears like-
ly that the species no longer occurs there. Reasons for its apparent dis-
appearance are unknown.

The river frog (Rana heckscheri) is a large ranid occurring in associa-
tion with blackwater river habitats from southern Mississippi to southeastern
North Carolina (Sanders 1984, Conant and Collins 1991). The species is known
in North Carolina from only a few scattered localities in the Lumber and Cape
Fear river systems (Fig. 1). Little has been published on R. heckscheri in North
Carolina. Its occurrence in the state was first suggested by Brimley (1944), who
listed it among the state's fauna with some doubt on the basis of a single speci-
men of a frog found dead at a heron rookery at Battery Island in Brunswick
County on 13 June 1938. His tentative identification was apparently based sole-
ly on the frog's dark ventral coloration, and his brother, H. H. Brimley (1938),
remarked that "it was a noticeably black specimen, with very definite markings
showing on the inside of the thighs, so we brought it back ... to identify . . . but
it turned out to be nothing more than a common bull frog!" DePoe and Funder-
burg (1959) and Simmons and Hardy (1959) discounted that specimen as being
R. heckscheri, probably rightfully so, as the specimen apparently was not
retained and cannot be verified.

DePoe and Funderburg (1959) reported a specimen of R. heckscheri from
Greenfield Lake in New Hanover County, collected 10 May 1948, but that specimen
(Cornell University 5496) was later identified as R. catesbeiana (Sanders 1984).

69

70

Jeffrey C. Beane

Fig. 1 . Historical distribution of Rana heckscheri in North Carolina. Dots rep-
resent localities documented by specimens in curated collections. The triangle
represents an approximate locality supported by a specimen.

\\J\Y 7>

r

Y

The first legitimate records of the river frog in North Carolina were
reported by DePoe and Funderburg (1959) and Simmons and Hardy (1959). The
apparent earliest specimen was collected as a tadpole by a University of North
Carolina student in March, 1957, at an imprecise locality on the Black River in
Sampson County, and kept in an aquarium at the University until it transformed
(DePoe and Funderburg 1959). It was later donated to the North Carolina State
Museum (NCSM 14610) by W. L. Engels. Simmons and Hardy (1959:37)
reported collecting a series of R. heckscheri tadpoles on 21 March and addition-
al specimens on 12 April 1958 (a date of 12 March was also reported later in the
same paper, but appears to be an erroneous reference to 21 March) from "a grav-
el pit pond" near Maxton in Robeson County, "located at the intersection of State
Highway 71 and the Lumbar [sic] River." They further reported that "a local
farmer, obviously familiar with the distinctive tadpoles, has been aware of their
occurrance [sic] in the Maxton ponds for at least fifteen years." It is not known
what became of the specimens Simmons and Hardy collected. No repository was
listed in the paper, and it is possible that the specimens were not saved. A pho-
tograph of one of the tadpoles was included; it appears to be the only published
photograph of a river frog tadpole from North Carolina. A currently existing
small, shallow pond near the west bank of the Lumber River just southwest of
the NC 7 1 bridge is probably the site referred to, but no R. heckscheri have been
found there in recent years. The locality for four specimens in the U.S. Nation-
al Museum (USNM 144367-144370), collected by Hardy 30 May-1 June 1960,
was given as only "Robeson-Scotland Co. line, Air Base," and may or may not

River Frog 7 1

refer to the same site. The air base (now Laurinburg-Maxton Airport) actually
lies in Scotland County ca. 2 air miles WNW of the NC 71 bridge.

On 7 August 1958, C. E. DePoe collected a single juvenile R.
heckscheri (NCSM 7004) at Rhodes Pond, a large cypress lake located 1.5 air
miles NE of Godwin in Cumberland County (DePoe and Funderburg 1959). A
specimen in the American Museum of Natural History (AMNH 22433) is report-
ed as having been collected at Southern Pines in Moore County, with no further
data.

DePoe and Funderburg (1959) reported several additional specimens
found among frogs brought to Carolina Biological Supply Company by com-
mercial collectors, believed to have been collected "either in the Cape Fear or
Pee Dee river drainages in southern North Carolina." Only one of these has been
confirmed as R. heckscheri, a specimen (NCSM 7005) collected ca. mid June
1958 from an undetermined locality.

The remaining records from the state have come from a series of bor-
row pit ponds along the Lumber River near the SR 1433 bridge at the Scotland-
Robeson County line, 5 air miles S of Wagram; and from along the Lumber River
between that site and the NC 7 1 bridge. Voucher specimens from that locality in
the collections of the North Carolina State Museum of Natural Sciences are as
follows:

7 Feb. 1965: Series of 17 larvae seined from borrow pit pond by W. M. Palmer
and J. R.Paul (NCSM 3741).

16 April 1967: Five of six adults taken from borrow pit pond by J. R. Bailey et
al. (NCSM 32080-32084, formerly DU A6819). Bailey (personal field notes)
noted "no significant difference in habit or habitat . . . from bullfrogs taken there
at same time."

23 May 1968: Series of nine adults collected by floating the Lumber River from
8-12 pm "from first half to 2/3 of distance" between the SR 1433 and NC 71
bridges by J. R. Bailey et al. (NCSM 9790 and 32085-32092, formerly DU
A9349). Bailey (personal notes) noted that 40 bullfrogs were also taken along
the same stretch, but that only bullfrogs were taken upstream from the SR 1433
bridge.

6 Feb. 197 1 : Series of larvae seined from borrow pit pond by W. M. Palmer and
D. L. Stephan [NCSM 10058 (5 larvae) and NCSM 26534 (10 larvae)].
21 Oct. 1973: Series of 30 larvae collected from borrow pit pond by A. L.
Braswell and D. L. Stephan (NCSM 12895).

12 July 1975: Adult female taken beside borrow pit pond by A. L. Braswell,
D.L. Stephan, and J. H. Reynolds (NCSM 15659). This is the last known spec-
imen from the state. Its photograph appears in Martof et al. (1980) and in a pop-
ular article by Dopyera (1995).

72 Jeffrey C. Beane

Little else has been published on R. heckscheri in North Carolina, and
little is known about its natural history in the state. Neither eggs nor calling
adults have been reported from the state. Martof et al. (1980) provided a brief
descriptive account of the species in the Carolinas and the aforementioned pho-
tograph. Stephan (1985) and Beane (1993a) wrote popular articles, and Stephan
(1989) provided a brief account of the frog's status in the state. In 1990 it was
granted protection as a species of Special Concern under the North Carolina
Endangered and Threatened Wildlife Law (G.S. 113-331 to 113-337). Beane
(1993b) provided a more detailed summary of its status in the state.

Short accounts of R. heckscheri in other states and general information
on the species may be found in Wright (1924, 1932), Allen (1938), Carr (1940),
Wright and Wright (1949), Mount (1975), Sanders (1984), Behler and King
(1985), Ashton and Ashton (1988), and Conant and Collins (1991). Recordings
of the breeding call are provided by Bogert (1958), Anon. (1982), and Elliott
(1992).

A survey was undertaken to determine the current distribution of the
river frog in North Carolina (if indeed it still occurred in the state), to evaluate
the status of any populations located, to learn more about the biology and habi-
tat requirements of the species, to identify the level of protection it should be
afforded, and to outline any conservation measures that might be justified.

METHODS

Efforts were made to locate all museum specimens and literature
records of the river frog in North Carolina. Field survey work was centered
around the vicinity of these records, as well as other potential sites. Sites inves-
tigated included many areas along the Lumber, South, Black, Northeast Cape
Fear, Cape Fear, Waccamaw, and Lockwood Folly rivers and their larger tribu-
tary streams and swamps.

Field work for this survey was conducted between spring 1987 and fall
1996, and is ongoing; however, most of the work was conducted between April
1992 and September 1993. During 1992-1993 over 1,500 man-hours were
devoted to field work and travel for the project, and a comparable amount of time
was devoted to office work. Over 10,600 miles of travel were logged in that time
period. The area surveyed included portions of Robeson, Scotland, Columbus,
Bladen, Sampson, Cumberland, Pender, Brunswick, Hoke, Duplin, New
Hanover, Moore, Richmond, and Harnett counties, North Carolina; and Horry
County, South Carolina. Beane (1993b) provided a map and list of specific local-
ities visited during 1992-1993 along with dates and survey methods used at each
site.

Survey techniques included navigating rivers and other bodies of water
by canoe or johnboat during the day to search for suitable habitat, adult frogs, or
schools of tadpoles; floating the same areas by night with flashlights and head-

River Frog 7 3

lamps in search of adult frogs; walking or wading potential habitat at night with
lights; visually scanning for the large and conspicuous tadpoles at bridge cross-
ings or other sites with good visibility; seining and dipnetting for tadpoles; slow-
ly driving and walking roads through suitable habitats-particularly bridge cross-
ings—on rainy (and non-rainy) nights; and listening for calling adults at potential
sites by day and night.

Posters depicting a drawing of the river frog's distinctive tadpole were
widely distributed in the southeastern part of the state. Biologists and outdoor
enthusiasts residing in, collecting in, or frequenting areas within the frog's range
were encouraged to report any suspected sightings. Local residents were often
questioned when encountered in the field, and many were shown a large pre-
served tadpole and photographs of adult frogs. Several articles featuring the
river frog project appeared in regional newspapers, and the survey was adver-
tised in several issues of the North Carolina Herpetological Society newsletter.
Several public field trips to search for river frogs were organized through the
North Carolina State Museum, and several public talks on the project were pre-
sented, using slides, photographs, call tapes, preserved specimens, field guides,
and a live adult frog from Florida as educational tools. Participants in the field
work were familiarized with river frog identification.

RESULTS

This survey revealed no current evidence of river frogs anywhere in
North Carolina. All 26 other anuran species known to share the potential range
of the river frog (NCSM files, Conant and Collins 1991) were encountered in the
state during the survey, most of them in relative abundance. The most produc-
tive methods for locating ranids were nocturnal searches with lights, conducted
either by canoe or on foot, and driving roads on rainy nights. All other Rana (R.
catesbeiana, R. clamitans, R. utricularia, R. palustris, R. virgatipes) with similar
habits and utilizing habitats similar to those of the river frog were frequently
encountered. River frogs were encountered with little difficulty in Franklin, Lib-
erty, and Wakulla counties, Florida; Charlton, Clinch, and Ware counties, Geor-
gia; and Hampton, Jasper, and Sumter counties, South Carolina during the time
of the survey.

No reports of river frog encounters in North Carolina were received
during the time of the survey. Only two plausible and previously undocumented
reports of earlier sightings were received, and both may have occurred prior to
the last documented sighting in 1975. J. H. Carter III, an environmental consul-
tant and experienced field biologist with herpetological expertise, reported (per-
sonal communication) having seen what he believed to be an adult river frog in
a large lake on the campus of St. Andrews Presbyterian College in Laurinburg,
Scotland County. (Visits to this site during the day and again at night during a
thunder shower in late June of 1993 yielded no evidence of the species.) David

7 4 Jeffrey C. Beane

Scott of Fair Bluff, an active conservationist and founding member of the Lum-
ber River Basin Committee, also reported (personal communication) having
taken what he believed to be an adult river frog while frog gigging on the Lum-
ber River in the vicinity of Fair Bluff along the Columbus-Robeson County line.
Neither individual could recall the date of the sightings, but both estimated them
to have been in the early to mid-1970s. Unfortunately, neither of these sightings
can be verified because of the similarities between adult river frogs and some
bullfrogs.

Only three responses to the many "wanted" posters distributed were
received, all of them false leads. Most local persons, when shown preserved tad-
poles, had obviously never encountered them before.

DISCUSSION

The results of this survey suggest that the river frog no longer occurs in
North Carolina. However, such a conclusive statement is difficult to make with
absolute confidence. The rather large amount of potential habitat present in the
state, and the limited scope of the current work, make it possible to envision how
populations of this frog could escape detection. Beane (1993b) remarked that if
the species still occurred in the state, it probably deserved Endangered status, but
recommended that it remain Special Concern since its occurrence had not been
verified.

The status and range of the river frog in South Carolina are not well
known. Until recently, the northernmost known populations from that state were
from the vicinity of Poinsett State Park in Sumter County, in the Santee drainage
(Sanders 1984), and the species still appears common at that site (personal obser-
vation). In 1996, R. heckscheri was first documented from the Pee Dee drainage
in South Carolina by Michael E. Dorcas et al. from two sites on the "Woodbury
Tract," a 20,000-acre parcel of land situated at the confluence of the Great Pee
Dee and Little Pee Dee rivers, ca. 16 air miles SSE of Brittons Neck in Marion
County. River frogs were heard calling from two sites on that tract on 6 April
and 18 April 1996. Although no specimens were collected or seen, a recording
was made of three individuals calling on 18 April. The tape was verified by J.
Whitfield Gibbons and is on file at the Savannah River Ecology Laboratory
(Michael E. Dorcas and Katie Distler, personal communication). The species has
yet to be reported from the Waccamaw drainage in either North or South Caroli-
na. It is possible that the lack of records from the northern Coastal Plain of South
Carolina reflect a lack of collecting efforts in that region rather than a genuine
absence, and more field work is needed in that area to determine whether any
currently or previously existing North Carolina populations should be regarded
as peripheral or disjunct.

River Frog 7 5

Much attention has been devoted in recent years to the apparent global
decline in many amphibian populations (Barinaga 1990; Blaustein and Wake
1990; Phillips 1990, 1995; Wyman 1990; Livermore 1992). No single cause
explains all of these widespread and often alarming disappearances, and there is
general agreement that a combination of factors is probably responsible.
Although habitat loss has been associated with the decline of many species, this
does not seem to be the case with the river frog in North Carolina; suitable habi-
tat appears to be plentiful. The fact that oxbow lakes on black water rivers are
rarer in North Carolina (Schafale and Weakley 1990) than in areas further south,
might represent a limit of prime breeding habitat for river frogs. However, the
species breeds in other habitats as well, and places where it has been taken his-
torically in the state do not seem unique in any way that is readily observable.

Reasons for the apparent disappearance of the river frog from sites
where it once occurred are unknown and must remain speculative. Factors lim-
iting the distribution of the species are poorly known. The frog was apparently
never common or widespread in North Carolina, and small, scattered populations
of any species are usually more vulnerable to extinction than are large, wide-
spread ones. Species at the edge of their range likewise tend to be susceptible.
Some possible explanations for the apparent disappearance of R. heckscheri from
North Carolina, all of them speculative, include a number of diverse factors:

1) Frog gigging or spearing still appears to be a popular sport in some parts of
southeastern North Carolina, although many persons encountered during the sur-
vey spoke of its being more widely practiced (as well as more productive) in past
years. The rather unwary adult river frogs (Carr 1940, Wright and Wright 1949,
Mount 1975) probably make easier targets for frog hunters than any other Rana
species. It is conceivable that intensive take by humans in an area could seri-
ously impact or eventually eliminate populations, especially those that were rel-
atively small.

2) Other ranids surely compete with river frogs, both as larvae and adults, and
the niche of the bullfrog in particular seems to overlap that of the river frog rather
broadly (Carr 1940, Wright and Wright 1949). Although the two have been
taken sympatrically at numerous sites, the highly adaptable bullfrog is probably
a better competitor in certain, if not most, situations, and it is possible that a
change in environmental quality or in some particular selective pressure could
offer bullfrogs an advantage leading to river frog extirpation.

3) A general overall decline in environmental quality could be responsible for
the river frog's decline in North Carolina. It was beyond the scope of this survey
to delineate the exact causes or effects of any environmental degradation that
may have occurred over the past several decades. Although precise long-term
data are difficult to obtain, it seems almost certain that some declines in water
quality have occurred in the state's blackwater rivers. Most fishermen and other
local residents encountered during the course of field work seemed of the opin-

7 6 Jeffrey C. Beane

ion that fishing and frogging had declined, and that various species of wildlife
were not as frequently observed as in past years. While frogs were generally
found to be common during this survey, there were occasions when far fewer
were observed than expected. As an example, during four hours of night canoe-
ing on a stretch of the Northeast Cape Fear River in Duplin County on 6 Sep-
tember 1992, a total of only three anurans were observed. Little is known about
the sensitivity of the river frog to environmental changes.

Finally, Pechmann et al. (1991) pointed out the difficulties sometimes
involved in distinguishing true amphibian declines from natural population fluc-
tuations, and Hairston and Wiley (1993) emphasized the value of long-term stud-
ies in determining whether supposed amphibian declines were genuine. While
the apparent absence of R. heckscheri in North Carolina could represent a natur-
al fluctuation, the lack of a single record in more than 22 years suggests other-
wise. Still, conclusive documentation of extinction can be difficult for any
organism, and more field work is needed to determine the river frog's true status
at the northern edge of its range. It is hoped that biologists working in south-
eastern North Carolina will make every effort to collect and report all possible
evidence of R. heckscheri in the state.

ACKNOWLEDGMENTS-- This survey was possible only through the efforts of
many. Funding was provided by a grant from the North Carolina Wildlife
Resources Commission's Nongame and Endangered Wildlife Fund, by the North
Carolina State Museum of Natural Sciences, and by participating individuals. I
especially thank Ann Berry Somers, who gave generously of her time and
resources, assisting with much field work as well as providing equipment, fund-
ing, and invaluable encouragement and friendship. Additional assistance with
field work and the use of various equipment was provided by Jeff Alexander,
Stanley L. Alford, Charles D. and Carolyn L. Beane, Fred Berry, Alvin L.
Braswell, Cheryl K. Cheshire, David G. and Martha R. Cooper, Randy L.
Craven, Robert A. Davis, Darrell L. DeTour, Kelly and Ralph Eanes, John T.
Finnegan, Robert S. Flook, Larry and Roseanne Francis, Beth Gatlin, Kim
Helms, Novella Hopkins, Stephanie J. Horton, Jim and Jane Hunt, Dale and Joe
Jackan, Barbara J. and Raymond W. Johnson, Scott Jones, Samuel Kennedy, Bill
Lillard, William M. Palmer, Robert Pegram, David Pike, Patty Purcell, Bobby F.
Roberts, William H. Rowland, Jr., Vincent P. Schneider, Stephen D. Smith, Chris
Spauer, Ronald W. Sutherland, Thomas J. Thorp, Paul Thrift, Janice M. Weems,
James and Joseph Zawadowski, the North Carolina Division of Environmental
Management, and J. H. Carter III and Associates Environmental Consultants.
David and Donna Scott provided lodging for field crews and assisted in many
other ways. Private access to waterways was permitted by Eugene Shaw,
Tommy Spivey, and the staff of Rhodes Pond Fishing Camp. Tom Squier made
special efforts to generate media publicity. Much support was provided by the

River Frog 7 7

North Carolina Herpetological Society. The late Aubrey M. Shaw helped dis-
tribute posters and allowed the use of his land in Bladen County as a base camp,
but more important was his deep appreciation for the beauty and biodiversity of
southeastern North Carolina, which served, and continues to serve, as an inspi-
ration to me. Earlier drafts of the manuscript benefited from comments by A. L.
Braswell, W. M. Palmer, R. W. Van Devender, and an anonymous reviewer.

LITERATURE CITED

Allen, E. R. 1938. Notes on Wright's bullfrog, Rana heckscheri (Wright) [sic.].
Copeia 1938(1):50.

Anon. 1982. Voices of the night. 12" LP record. Third edition. Library of Nat-
ural Sounds, Laboratory of Ornithology, Cornell University, Ithaca,
New York.

Ashton, R. E., Jr., and P. S. Ashton. 1988. Handbook of reptiles and amphibians
of Florida—part three: The amphibians. Windward Publishing, Inc., Miami,
Florida.

Barinaga, M. 1990. Where have all the froggies gone? Science 247: 1033-1034.

Beane, J. 1993a. The quest for the frog that was. Friends of the North Caroli-
na State Museum of Natural Sciences Newsletter. April 1993:8-9.

Beane, J. C. 1993b. A status survey of the river frog in North Carolina. Final
report. North Carolina Wildlife Resources Commission, Raleigh (unpub-
lished manuscript).

Behler, J. L., and F. W. King. 1985. The Audubon Society field guide to North
American reptiles and amphibians. Alfred A. Knopf, New York, New York.

Blaustein, A. R., and D. B. Wake. 1990. Declining amphibian populations: A
global phenomenon? Trends in Ecology and Evolution 5:203-204.

Bogert, C. M. 1958. Sounds of North American frogs. The biological signifi-
cance of voice in frogs. 19 p., 12" LP record. Folkways Records No.
FX6166, New York, New York.

Brimley, C. S. 1944. Amphibians and reptiles of North Carolina. Reprinted
from Carolina Tips (1939-1943). Carolina Biological Supply Co., Elon Col-
lege, North Carolina.

Brimley, H. H. 1938. The Battery Island heron colony near Southport, N.C. The
Chat 2(5-6):41-43.

Carr, A. F., Jr. 1940. A contribution to the herpetology of Florida. University
of Florida Publications, Biological Science Series 3(1): 1-118.

Conant, R., and J. T. Collins. 1991. A field guide to reptiles and amphibians of
eastern and central North America. Third Edition. Houghton Mifflin Co.,
Boston, Massachusetts.

Depoe, C. E., and J. B. Funderburg. 1959. The river frog, Rana heckscheri
Wright, in North Carolina. Journal of the Elisha Mitchell Scientific Society
75(1):11-12.

7 8 Jeffrey C. Beane

Dopyera, C. 1995. Amphibians on the brink. Wildlife in North Carolina
59(2): 10-15.

Elliott, L. 1992. The calls of frogs and toads. 28 p., 66 minute audiocassette.
NatureSound Studio, Ithaca, New York.

Hairston, N. G., and R. H. Wiley. 1993. No decline in salamander (Amphibia:
Caudata) populations: A twenty-year study in the southern Appalachians.
Brimleyana 18:59-64.

Livermore, B. 1992. Amphibian alarm: Just where have all the frogs gone?
Smithsonian Magazine Oct. 1992:113-120.

Martof, B. S., W M. Palmer, J .R. Bailey, and J. R. Harrison III. 1980. Amphib-
ians and reptiles of the Carolinas and Virginia. The University of North Car-
olina Press, Chapel Hill.

Mount, R. H. 1975. The reptiles and amphibians of Alabama. Auburn Univer-
sity Agricultural Experiment Station, Auburn, Alabama.

Pechmann, J. H. K., D. E. Scott, R. D. Semlitsch, J. P. Caldwell, L. J. Vitt, and J.
W. Gibbons. 1991. Declining amphibian populations: the problem of sep-
arating human impacts from natural fluctuations. Science 253:892-895.

Phillips, K. 1990. Frogs in trouble. International Wildlife 20:4- 1 1 .

Phillips, K. 1995. Tracking the vanishing frogs. Penguin Books, New York,
New York.

Sanders, A. E. 1984. Rana heckscheri Wright. Catalogue of American Amphib-
ians and Reptiles 348.1-348.2.

Schafale, M. P., and A. S. Weakley. 1990. Classification of the natural commu-
nities of North Carolina. Third approximation. North Carolina Natural Her-
itage Program, Raleigh.

Simmons, R., and J. D. Hardy, Jr. 1959. The river-swamp frog, Rana heckscheri
Wright, in North Carolina. Herpetologica 15(1): 36-37.

Stephan, D. 1985. "Herp-of-the-month": Rana heckscheri Wright; river frog.
NC Herps 8(3):3-4.

Stephan, D. 1989. Rana heckscheri (River Frog), page D67 in Scientific coun-
cil report on the conservation of North Carolina amphibians and reptiles (A.
L. Braswell and Committee). North Carolina Wildlife Resources Commis-
sion, Raleigh.

Wright, A. H. 1924. A new bullfrog {Rana heckscheri) from Georgia and Flori-
da. Proceedings of the Biological Society of Washington 37:141-152.

Wright, A. H. 1932. Life histories of the frogs of Okefinokee Swamp, Georgia.
North American Salientia (Anura) Number 2. Macmillan Co., New York,
New York.

Wright, A. H., and A. A. Wright. 1949. Handbook of frogs and toads of the
United States and Canada. Third edition. Comstock Publishing Co., Ithaca,
New York.

River Frog 79

Wyman, R. L. 1990. What's happening to the amphibians? Conservation Biol-
ogy 4:350-352.

Received 20 March 1996
Accepted 13 August 1996

Wood Ducks, Aix sponsa (Anseriformes: Anatidae), and Black-
water Impoundments in Southeastern North Carolina

Craig A. Harper1,
James F. Parnell, and
Eric G. Bolen
Department of Biological Sciences, University of North Carolina at Wilming-
ton, Wilmington, North Carolina 28403

ABSTRACT— Three small (<2.0 ha) newly constructed, wooded
impoundments located on first-order blackwater streams were exam-
ined for suitability as habitat for wood ducks (Aix sponsa). We evalu-
ated brood and roosting cover as well as availability of mast and inver-
tebrate foods. Wood ducks used the new impoundments for nesting,
brood rearing, feeding, and roosting. Feeding was heaviest on
impoundments with dense cover where water oak (Quercus nigra)
acorns were abundant. Brood rearing was restricted to sites with dense
low cover. Roosting activity was highest where live, dense, woody
cover was available. All sites were used by wood ducks throughout
much of the year. Management recommendations include drawdown
prescriptions designed to ensure live, woody shrub cover and continued
mast production. We conclude that construction and active manage-
ment of small blackwater impoundments offer a means of improving
habitat for wood ducks and other wetland wildlife in the Coastal Plain
of North Carolina.

Bottomland wetlands and other sites suitable for wood ducks have
declined across North America (Dugger and Fredrickson 1992). North Carolina
has been no exception, with alteration of more than 50% of the state's palustrine
wetlands (Cashin et al. 1992). Thus, the creation of new wetlands is potential-
ly significant in ameliorating the problem of wetland drainage, and is potential-
ly significant to wood ducks. Three blackwater creeks at Camp Lejeune were
impounded by the Environmental Management Division in October 1990, creat-
ing small (<2.0 ha) wetlands to be managed primarily for wood ducks. Although
much is known about wood ducks elsewhere (Bellrose and Holm 1994), the use-
fulness of creating small impoundments as wood duck habitat is questionable.

Present Address: Department of Forest Resources, Clemson University, Clemson, South Carolina 29634.

80

Wood Ducks 8 1

We monitored environmental parameters, changes in vegetation, and the
response of wood ducks within these blackwater impoundments during the two
years following construction and initial inundation.

STUDY SITES

The study was conducted on three small blackwater impoundments at
Camp Lejeune Marine Base, Onslow County, North Carolina. The impound-
ments were located approximately 5 km east of Stone Bay (on the New River)
and about 8 km northwest of the Atlantic Ocean.

Impoundment 1 (hereafter 1-1) covered 1.2 ha with a mean depth of 1
m. Black gum (Nyssa sylvatica), red maple (Acer rubrum), sweetgum (Liq-
uidambar styraciflua), American holly (Ilex opaca), red bay (Persia borbonia),
sweetbay (Magnolia virginiana), and willow (Salix caroliniana) formed a partial
canopy over the inundation. Fetterbush (Lyonia lucida), wax myrtle (Myrica
cerifera), and titi (Cyrilla racemiflora) furnished a thick undergrowth around and
within 1-1. Duckweeds (Spirodela spp., Wolffia spp., Wolfiella spp., and Lemna
spp.) blanketed much of the water in spring and summer.

Impoundment 2 (1-2) covered 1 .9 ha at a mean depth of 1 .5 m. 1-2 was
heavily wooded by yellow poplar (Liriodendron tulipifera), sourwood (Oxyden-
drum arboreum), water oak (Quercus nigra), American holly, red bay, and sweet-
bay. Other woody vegetation included black gum, sweetgum, loblolly pine, fet-
terbush, wax myrtle, loblolly bay (Gordonia lasianthus), and white oak (Quer-
cus alba). Duckweeds flourished during spring and summer.

Impoundment 3 (1-3) covered 0.36 ha with a mean depth of 1.5 m. Most
of 1-3 was open water, flanked by two wooded coves vegetated with water oak,
loblolly pine, wax myrtle, American holly, sourwood, and red maple. Mats of
bladderwort (Utricularia biflora) spread across the water, and organic matter
supporting herbaceous plants such as yellow-eyed grasses (Xyris spp.) and club
moss (Lycopodium alopecuroides) draped floating logs.

METHODS

Dissolved oxygen, pH, and Secchi disk readings were recorded month-
ly from June 1991 - December 1992 at each impoundment. Vegetation was sam-
pled with a series of 15 m2 circular plots, each selected randomly within a grid at
a ratio of one plot per 0.40 ha. of pond surface. Plots were sampled in August
1991 and again in June and August 1992. All trees within each plot were count-
ed and measured for diameter at breast height (dbh). The number of above- water
stems of shrubs and vines was counted, and percent cover of herbaceous vegeta-
tion was visually estimated to the nearest 10% in a 0.50-m2 quadrat.

Mast was sampled biweekly from September through December 1992
in square baskets of 1-m2 surface area. Ten baskets were placed randomly at each
wetland, with the restriction that they were spaced evenly around each wetland.

8 2 Craig A. Harper, James F. Parnell, and Eric G. Bolen

Five baskets were placed over water at no more than 50 cm of depth, and anoth-
er five were placed in the adjacent uplands within 15 m of the apparent high-
water mark. Mast captured in each basket was identified and weighed in the field
to the nearest 0.01 g. The data obtained at each wetland were log 10 transformed
because of heterogeneity of variances and were tested by species and location
(inside or outside the impoundment) using a one-way ANOVA in the General
Linear Models (SAS 1989) procedure. Transformed means were compared
using the Student-Newman-Keuls test at the alpha = 0.05 level.

Macroinvertebrates were sampled in each vegetation plot monthly dur-
ing April, May, June, and July 1992, using a plankton tow net and a sweep net.
Three water surface samples were obtained with a plankton tow. The water col-
umn was sampled with a sweep net by making semicircular sweeps on the left,
center, and right side of the plot. An attempt was made to sweep across the entire
plot, thus sampling a standardized area. A sweep net also was used to sample the
benthos by scraping the substrate with the net across the left, center, and right
side of the plot. Samples were preserved in 45% isopropyl alcohol and identi-
fied using Thorp and Covich (1991). Specimens were air dried overnight, then
ovendried at 55 C for eight hours, and their biomass was determined with a Met-
tler electronic balance. Data were log 10 transformed because of heterogeneity
of variances and tested for differences in total invertebrate biomass between wet-
lands with a one-way ANOVA in the General Linear Models (SAS 1989) proce-
dure. Transformed means were contrasted using the Student-Newman-Keuls test
at the alpha = 0.05 level.

Three cypress nest boxes designed for wood ducks were erected at each
impoundment upon inundation in October 1990. Each nest box was erected on
a steel post equipped with a predator-proof guard and supplied with wood shav-
ings. Prior to the 1992 nesting season, sufficient numbers of nest boxes were
added to provide one box per 100 m of shoreline at each of the three wetlands,
thereby resulting in eight boxes at 1-1, nine at 1-2, and four at 1-3. The boxes
were inspected biweekly during May, June, and July and monthly August
through April.

Flush counts were conducted biweekly on the impoundments between
September and December 1992. All ducks flushed by an observer walking the
perimeter of each wetland were recorded. Brood surveys were conducted each
week during May, June, and July 1992 from blinds located at sites arbitrarily
selected for their enhanced visibility of shrub-strewn habitat ( Rumble and Flake
1982, Robb and Bookhout 1990). Broods were counted for two-hour periods
beginning one-half hour before sunrise. Species, brood size, age-class (Gollop
and Marshall 1954), and activities were recorded whenever broods were
observed. Evening roost censuses were made biweekly between September and
December 1992 from suitable vantage points adjacent to each wetland. Sampling
effort and number of counts were identical at all impoundments. Flush and roost

Wood Ducks 8 3

count data were analyzed by Chi-square tests to detect differences among wet-
lands.

RESULTS AND DISCUSSION

Dissolved oxygen was low at all three sites, averaging 2.0 - 2.1 ppm
with a range of 0.4 - 6.6 ppm. Acidity was similar among impoundments, aver-
aging 5.1 - 5.4 with a range from 4.1 - 6.1. Secchi disc readings also were sim-
ilar with a range from 0.18 - 0.93 m.

Water control structures in 1-1 and 1-3 remained closed throughout the
two years of this study, while 1-2 was partly drained half-way through the second
year. Density of living upland vegetation within each impoundment diminished
with time after flooding, whereas the density of aquatic plants increased (Table
1). Most woody vegetation at 1-1 remained vigorous during the first year of
inundation, but after two years more than 55% had died. Duckweeds and blad-
derwort first appeared during the summer of 1991 and covered approximately 50
% of the water's surface by the following summer. Aquatic emergent vegetation,
including cattail (Typha latifolia), common rush (Juncus effusus), arrowhead
(Sagitarria latifolia), and swamp loosestrife (Decodon verticillatus), was pio-
neering along the perimeter of 1-1 by the second summer of inundation (Table 1).
About 75% of woody vegetation at 1-2 was dead by the end of the first growing
season; 90% was dead a year later. Duckweeds and bladderwort quickly colo-
nized 1-2 and covered more than 40% of the surface one year after inundation
(Table 1). Downed timber and dying, woody, emergent vegetation at 1-2 provid-
ed conditions generally favored by wood ducks and their broods in the southern
United States (see Cottrell et al. 1990). However, favorable conditions declined
after two years of inundation, as woody cover died and disappeared rapidly cre-
ating a more open impoundment.

A lack of inundated woody vegetation at 1-3 resulted in an open pond,
except for one wooded cove and vegetation bordering the stream that fed the
impoundment. Some perimeter vegetation was inundated when the water-control
structure was closed. Plant mortality in 1-3 thus occurred primarily in the wood-
ed coves, which covered only 0.14 ha. Aquatic vegetation present at 1-3 con-
sisted primarily of bladderwort and duckweeds. Cattail was growing along the
perimeter of 1-3 by the end of the second growing season, and floating organic
muck supported thick patches of swamp loosestrife. Swamp loosestrife, giant
cane (Arundinaria gigantea), and fetterbush within and along the wooded coves
offered cover for wood ducks as they foraged on water oak acorns.

Mast production varied greatly among impoundments. In the uplands
adjacent to the impoundments, more mast was collected at 1-1 (158.7 g, of which
southern red oak acorns contributed 148.5 g) than 1-2 (52.2 g) or 1-3 (68.7 g).
Mast collected in baskets within the impoundments probably better represents
availability in terms of food for wood ducks. However, wood ducks were

84

Craig A. Harper, James F. Parnell, and Eric G. Bolen

o

o (N

>.ON

.-s o\

c« ^

C —

1) </5

•a s

.S 3
<* <

CM

§ s

E

5 -.

o

u&

^

&

.» «*

'vs

*J 3

c

S 2?

a)
Q

H<

oo

o

o

(N|

o

ci

vd

ON

o

r«S

CI — -h

<N — <

<N — i

< ^

On

ON

(X

o\

ON

Os

On

</5

t/J

3

3

4J

60

00

C

3

3

3

<

<

&

On On

On <n On

O

3,

S

I 1

•B S f-S

;s -s

observed feeding upon water oak acorns both at the water's edge and as far as 5
m above the shoreline of the wooded coves at 1-3. Within the coves of 1-3, water
oaks yielded significantly (p < 0.05) more acorns (>28.0 g/m2) than at 1-1 or 1-2
and attracted large numbers of wood ducks on a regular basis.

Acorn yields adjacent to the impoundments were similar to those in
Missouri, where the upper range of acorn production in an upland forest varied
from 4.0 to 25.5 g/m2 during a 2-year period (Dalke 1953). Elsewhere in Mis-

Wood Ducks 8 5

souri, pin oaks (Quercus palustris) within green-tree reservoirs produced an
average of 14.2 g/m2 (McQuilken and Musbach 1977).

Invertebrates are important waterfowl foods because of their protein
and mineral contents, which are essential for egg production in hens and tissue
growth of chicks (Krapu 1974, Swanson et al. 1974, Drobney and Fredrickson
1979, Drobney 1990). Invertebrate biomass within the impoundments ranged
from 0.06 - 0.20 gm/m2 (Table 2). Insects in the order Odonata (primarily Libel-
lulidae) and Heteroptera (primarily Notonectidae) were the most abundant inver-
tebrates. Most other groups Were much less abundant (Table 2).

Table 2. Macroinvertebrate biomassa (in grams) in three blackwater impound-
ments at Camp Lejeune in April, May, June, and July 1992.

Taxab Pond

1-1 1-2 1-3

Chelicerata

Araneae 0.01 0.05 0.01

Crustacea

Amphipoda 0.02

Decapoda 2.22 1.34

Insecta

Coleoptera

Diptera

Heteroptera

Megaloptera

Odonata
Total for all plots
Average/ plot
Number Plots (15m2)

a Includes adults, nymphs, pupae, and larvae.

b For further taxonomic breakdown, see Harper (1993).

c tr = trace.

While investigating invertebrate productivity within a blackwater river
in south Georgia, Benke et al. (1984) recorded 20-50 times more standing stock
biomass on snags than in sandy habitat and 5-10 times more than in muddy habi-
tat. They also found increased species richness on snag habitats as compared to
other benthic habitats and concluded that production on snags appeared to be
limited by available substrate. Thus, blackwater impoundments with significant

0.27

0.33

0.07

if

0.09

ti"

4.91

1.34

0.36

0.02

0.45

2.10

8.39

0.07

7.33

12.87

1.85

2.44

3.22

0.93

3

4

2

8 6 Craig A. Harper, James F. Parnell, and Eric G. Bolen

amounts of flooded vegetation and downed timber (such as 1-2 in our study)
seem to provide habitat for a larger standing stock and increased production over
riverine sites (e.g., non-impounded blackwater streams). High diversity and bio-
mass of invertebrates make an important contribution to nesting and brooding
wood ducks.

Wood ducks utilized the impoundments as nesting and brooding sites
during both years of the study. One wood duck nest was located in 1991 (nine
boxes were available). In 1992, two wood duck nests were among 15 bird nests
found in 21 boxes available on the impoundments. Other birds using the nest
boxes included eastern bluebirds (Sialia sialis), Carolina wrens (Thryothorus
ludovicianus), great-crested flycatchers (Myiarchus crinitus), and prothonotary
warblers (Protonotaria citrea).

Broods were sighted 34 times on 1-1 during the survey period (May -
June 1992). Because no eggs hatched in nest boxes at 1-1, these broods either
emigrated from other wetlands or hatched from nests in natural cavities nearby.
The thick cover of fetterbush, titi, wax myrtle, honeysuckle (Lonicerajaponica),
and blackberry {Rubus spp.) around and within 1-1, coupled with an abundance
of invertebrates, made this impoundment an attractive brood rearing site. Molt-
ing adults also were observed within 1-1 during brood surveys. Broods and
adults were observed using floating logs in 1-1 as loafing sites.

Brood habitat also was favorable at 1-2, where broods were sighted on
35 occasions. The largest of these consisted of seven ducklings, but because this
brood was observed before any clutches hatched in nest boxes located at 1-2, the
ducklings must have come from another site or a natural cavity in the local area.
Movement of broods should not be unexpected in light of the mobility
McGilvrey (1969) recorded for wood duck broods on impoundments elsewhere.

Thick cover at both 1-1 and 1-2 precluded precise censuses of broods;
however, two broods were commonly observed concurrently on each impound-
ment, thus providing a conservative estimate of use. This compares favorably
with the average of three broods per sampling date observed on impoundments
of 0.71 to 2.70 ha in South Dakota, where visibility was much greater (see Rum-
ble and Flake 1982). Similarly, Belanger and Couture (1988) recorded the great-
est density of dabbling duck broods (2.0 broods/ha) on man-made ponds con-
taining >30% cover and 0.35 g of invertebrate biomass per m2, similar to condi-
tions at 1-1 and 1-2.

Broods were not observed at 1-3 during the survey, and the 10 ducklings
that hatched from a nest box there in 1991 left the impoundment shortly after-
ward. The relative lack of invertebrates and plant cover, along with the small
size of 1-3, likely contributed to the absence of broods (Sousa and Farmer 1983,
Drobney and Fredrickson 1979, Haramis 1990). Continued use of these
impoundments as brood cover will depend on the schedule of flooding and
drawdown. Brooding, molting, and roosting wood ducks depend on thick cover.

Wood Ducks 8 7

Unless managed to maintain living woody cover, impoundments will, in time,
become more open, much like beaver (Castor canadensis) ponds.

Significantly (p < 0.05) fewer wood ducks were flushed from 1-2 (n =
38) than 1-1 (n = 82) and I-3(n=71). Numbers of wood ducks using the impound-
ments increased dramatically during autumn. Numbers recorded during flush
counts were highest in November: 1-1 (n = 50), 1-2 (n = 20), and 1-3 (n = 45).
Decreased density of vegetation within 1-2 probably accounted for fewer ducks
flushed. Mortality of woody vegetation was highest at 1-2 in 1992. By fall 1992,
almost 90% of the standing vegetation within 1-2 was dead, whereas only about
60% had died in 1-1 and within the coves of 1-3 (Table 1). Apparently, the avail-
ability of acorns in the two wooded coves at 1-3 attracted wood ducks to these
specific sites, which were the only locations on 1-3 where wood ducks were
flushed (i.e., no ducks were flushed from open water).

Wood ducks used I- 1 and 1-2 regularly for roosting during autumn and
winter. Roosting activity increased from a low average of seven wood ducks per
night on 1-1 and 10 wood ducks per night on 1-2 in September to a seasonal high
of 140 per night on 1-1 in November. Dense, low, evergreen shrub cover within
1-1 similar to conditions described by Parr et al. (1979), provided attractive roost-
ing habitat. Differences in cover apparently influenced the degree to which each
site served as a roosting area, because significantly (p < 0.05) more wood ducks
roosted under the live vegetation in I- 1 as opposed to the dead vegetation in 1-2.
While only 10 wood ducks per day were flushed from 1-2 in November, many
more (average 65/night) sought roosting cover at night. Because very few acorns
were available within 1-1 or 1-2, we believe wood ducks flushed during the day
were loafing and not necessarily actively feeding. The wooded coves of 1-3 were
presumably too narrow for use as a secure roosting area. Wood ducks avoided
roosting on 1-3, which was mostly open water, thereby underscoring the value of
impoundments with thick living emergent woody vegetation for loafing and for
roosting.

MANAGEMENT CONSIDERATIONS AND CONCLUSIONS
The presence of myriad, first-order blackwater rivulets and streams in
eastern North Carolina offers opportunities for creating many small impound-
ments suitable as habitat for wood ducks and many other species. Within a year
of construction, small wooded impoundments were used readily by wood ducks
for nesting, rearing broods, molting, roosting, and feeding. Depending upon
physiographic features of each site (e.g., presence of live, low, woody cover; size
of impoundment; topography; presence of preferred mast trees; etc.), use by
wood ducks varied among the impoundments, yet certain biological needs were
met at each.

Management goals for waterfowl on small impoundments are best
achieved where water levels can be manipulated (Johnsgard 1956). Additional-

8 8 Craig A. Harper, James F. Parnell, and Eric G. Bolen

ly, water-level management enhances biodiversity in wetland environments
(Fredrickson and Taylor 1982). Maximum benefits are achieved when water-
level management follows a schedule of drawdowns and flooding designed to
maintain suitable communities of food and cover plants ( Fredrickson and Tay-
lor 1982, Fredrickson and Batema (no date), Fredrickson 1991). When manag-
ing small impoundments for wood ducks, we recommend initiating a flooding
and drawdown schedule that allows maintenance of living, emergent woody veg-
etation at a density of > 3 - 5 stems/m2 where possible. Drawdowns and flood-
ing should be completed slowly (i.e., approximately 3 weeks), thereby maxi-
mizing food availability (e.g., invertebrates and mast) within the wood duck's
foraging niche (<20 cm water depth) and allowing time for the establishment of
moist-soil vegetation. Drawdowns should be timed so that they coincide with
the spring migration of wood ducks. When managing for broods, drawdowns
should not be completed until August, when most wood duck broods have
fledged. Upon drawdown, areas inundated for brood utilization during the grow-
ing season should not be flooded again for 2 - 3 years to allow complete aeration
of soil in order to perpetuate live woody cover. Reflooding in the fall should be
timed so that at least 85% of the impoundment is inundated for the peak migra-
tion period of wood ducks (for our study, the first week of November. (See Harp-
er 1993 for additional details.)

Water-level management enables managers to provide needed
resources for wood ducks year-round. Obviously, this would be very difficult
with only one wetland; however, with a complex of small, managed impound-
ments, the task is more feasible. With a schedule where an impoundment pro-
vides flooded cover only every 2 to 3 years, it is essential to have several
impoundments that are diverse and can be inundated on a rotation, thus meeting
the various needs of wood ducks each year. Three ponds, with one flooded each
year and with the other two recovering, appear to represent a minimal manage-
ment unit.

Although it is usually possible for landowners to construct small
impoundments on small streams for wildlife enhancement, the regulatory office
of the local United States Army Corps of Engineers should always be contacted
prior to initiating work to determine the effect of regulations relative to section
404 of the Clean Water Act.

ACKNOWLEDGMENTS -We are indebted to Charles Peterson and his staff at
Camp Lejeune for their financial and logistical support. The counsel of Martin
Posey, Department of Biological Sciences at the University of North Carolina at
Wilmington and Hoke Hill, Department of Experimental Statistics, Clemson
University, was most helpful. The University of North Carolina at Wilmington
provided basic support.

Wood Ducks 8 9

LITERATURE CITED
Belanger, L., and R. Couture. 1988. Use of man-made ponds by dabbling duck
broods. The Journal of Wildlife Management 52:718-723.

Bellrose, F. C, and D. J. Holm. 1994. Ecology and management of the wood
duck. Stackpole Books, Mechanicsburg, Pennsylvania.

Benke, A. C, T C. Van Arsdall, Jr., D. M. Gillespie, and F. K. Parrish. 1984.
Invertebrate productivity in a subtropical blackwater river: the importance
of habitat and life history. Ecological Monographs 54:25-63.

Cashin, G. E., J. R. Dorney, and D. J. Richardson. 1992. Wetland alteration
trends on the North Carolina coastal plain. Wetlands 12:63-71.

Cottrell, S. D., H. H. Prince, and P. I. Padding. 1990. Nest success, duckling sur-
vival, and brood habitat selection of wood ducks in a Tennessee riverine sys-
tem. Pages 191-197 in Proceedings of the 1988 North American wood duck
Symposium, (L. H. Fredrickson, G. V. Burger, S. D. Havera, D. A. Graber,
R. E. Kirby, and T. S. Taylor, editors.). St. Louis, Missouri.

Dalke, P. D. 1953. Yields of seeds and mast in second growth hardwood forest,
southcentral Missouri. The Journal of Wildlife Management 17:378-380.

Drobney, R. D. 1990. Nutritional ecology of breeding wood ducks; a sympo-
sium for wetland management. Pages 77-82 in Proceedings 1988 North
American wood duck Symposium, (L. H. Fredrickson, G. V. Burger, S. P.
Havera, D. A. Graber, R. E. Kirby, and T. S. Taylor, editors.). St. Louis,
Missouri.

Drobney, R. D., and L. H. Fredrickson. 1979. Food selection by wood ducks in
relation to breeding status. The Journal of Wildlife Management 43:109-120.

Dugger, K. M., and L. H. Fredrickson. 1992. Life history and habitat needs of
the wood duck. Leaflet 13.1.6. Waterfowl management handbook. United
States Fish and Wildlife Service, Washington, D.C.

Fredrickson, L. H. 1991. Strategies for water level manipulations in moist-soil
systems. Leaflet 13.4.6. Waterfowl management handbook. United States
Fish and Wildlife Service, Washington, D.C.

Fredrickson, L. H., and D. L. Batema. (No date). Greentree reservoir manage-
ment. University of Missouri-Columbia, Puxico, Missouri.

Fredrickson, L. H., and T S. Taylor. 1982. Management of seasonally flooded
impoundments for wildlife. Resource Publication 148. United States Fish
and Wildlife Service, Washington, D.C.

Gollop, J. G., and W. H. Marshall. 1954. A guide for aging duck broods in the
field. Mississippi Flyway Council Technical Section (unpublished mimeo.).

Haramis, G. M. 1990. Breeding ecology of the wood duck: a review. Pages 45-
60 in Proceedings of the 1988 North American wood duck Symposium, (L.
H. Fredrickson, G. V. Burger, S. P. Havera, D. A. Graber, R. E. Kirby, and
T. S. Taylor, editors.). St. Louis, Missouri

9 0 Craig A. Harper, James F. Parnell, and Eric G. Bolen

Haramis, G. M, and D. Q. Thompson. 1984. Density-production characteristics
of box-nesting wood ducks in a northern greentree impoundment. The Jour-
nal of Wildlife Management 49:429-436.

Harper, C. A. 1993. Small wooded blackwater impoundments: their role as
wood duck habitat. M.S. Thesis, University of North Carolina at Wilming-
ton.

Johnsgard, P. A. 1956. Effects of water fluctuation and vegetation change on
bird populations, particularly waterfowl. Ecology 37:689-701.

Krapu, G. L. 1974. Foods of breeding pintails in North Dakota. The Journal of
Wildlife Management 38:408-417.

McGilvrey, F. B. 1969. Survival in wood duck broods. The Journal of Wildlife
Management 33:73-76.

McQuilkin, R. A., and R. A. Musbach. 1977. Pin oak acorn production on
greentree reservoirs in southeastern Missouri. The Journal of Wildlife Man-
agement 41:218-225.

Parr, D. E., M. D. Scott, and D. D. Kennedy. 1979. Autumn movements and
habitat use by wood ducks in southern Illinois. The Journal of Wildlife Man-
agement 43:102-108.

Robb, J. R., and T. A. Bookhout. 1990. Female and juvenile wood duck survival
and movements in Indiana. Pages 179-184 in Proceedings 1988 North
American wood duck Symposium, (L. H. Fredrickson, G. V. Burger, S. P.
Havera, D. A. Graber, R. E. Kirby, and T. S. Taylor, editors. ). St. Louis,
Missouri.

Rumble, M. A., and L. D. Flake. 1982. A comparison of two waterfowl brood
survey techniques. Journal of Wildlife Management 46:1048-1053.

SAS. 1989. SAS/STAT User's Guide, Version 6, Fourth Ed., Vol. 1. SAS Insti-
tute, Inc., Cary, North Carolina.

Sousa, P. J., and A. H. Farmer. 1983. Habitat suitability index models: wood
duck. United States Fish and Wildlife Service FWS/OBS-82/10.43.

Swanson, G. A., M. I. Meyer, and J. R. Serie. 1974. Feeding ecology of breed-
ing blue-winged teal. The Journal of Wildlife Management 38:396-407.

Thorp, J. H., and A. P. Covich. 1991. Ecology and classification of North Amer-
ican freshwater invertebrates. Academic Press, Inc., San Diego, California.

Received 10 April 1996
Accepted 1 August 1996

Relationship of Mass to Girth in Raccoons, Procyon lotor (Mam-
malia: Procyonidae), from West Tennessee

Troy A. Ladine1

The University of Memphis, Edward J. Meeman Biological Station

and Ecology and Organismal Biology, Campus Box 526080,

Memphis, Tennessee 38152-6080

ABSTRACT--Mass to girth ratio of 103 captured, adult (>14 months)
raccoons (Procyon lotor) was analyzed to assess allometric relation-
ships. Use of this ratio as a measure of physical condition was investi-
gated with linear regression and a comparison of the ratio by year, sea-
son, sex, and age. There was no statistically significant difference
among years (F = 0.17; P = 0.8473) or between seasons (F = 1.13; P =
0.2916). The ratio differed significantly among ages and was smallest
for raccoons 14 to 38 months in age (F = 12.31; P = 0.0001). Males
exhibited a significantly larger ratio than females (F = 25.03; P =
0.0001). There was no difference found within all groupings in paral-
lelism or coincidence based on the regression equations. The regression
equation for mass to girth for all individuals was Mass = 0.02916 *
(Girth) - 5.5904 (r2 = 0.8044; P = 0.0001). Although the allometric rela-
tionship between the mass and girth of raccoons was significant, this
relationship does not appear to be indicative of the physical condition
of the animal.

Estimating fitness of individuals in a population requires direct count of
the reproductive success of the females or some other measure of physical con-
dition such as antler beam diameter of white-tailed deer (Odocoileus virgini-
anus)(Ua\\s 1984), or kidney fat index of white-tailed deer and raccoons (Procy-
on lotor)( Johnson 1970, Glenn and Clark 1990). When studying medium-sized
nocturnal mammals, it is difficult to accurately determine the number of off-
spring produced per female in live animals because typical measures of physical
condition (such as placental scar counts, body fat indices, kidney fat indices)
involve sacrificing the individual (Johnson 1970, Sanderson and Nalbandov
1973, Glenn and Clark 1990). In long-term population and community ecology
studies, sacrificing individuals is not a suitable method. Thus, a nondestructive
method of quantitatively reporting physical condition is needed.

1 Present Address: Division of Mathematics and Science, Bethel College, McKenzie, Tennessee 38201.

91

9 2 Troy A. Ladine

The inherent relationship known for mass and length (Thompson 1961,
Schmidt-Nielsen 1984), suggests that a relationship between mass and girth
could be an estimable parameter to investigate as a measure of physical condi-
tion. The girth of an animal could increase disproportionately to overall mass if
there is a great amount of fat in relation to the skeletal and muscular tissue. This
would indicate an improvement in the relative condition of an individual. Con-
versely, a decrease in girth measurement would likely be noticed before a large
decrease in mass due to the loss of fat around the thoracic region, the first fat to
be lost during a shortage of food (Stuewer 1943). Bissonette and Csech (1938)
found smaller litters were produced by malnourished female raccoons, suggest-
ing a possible correlation of mass to girth with reproductive fitness in raccoons.

No indirect method of estimating physical condition that avoids sacri-
ficing individuals has been described for the raccoon. The objectives of this
study were to investigate the allometric relationship between mass and girth in
raccoons and to evaluate the suitability of this relationship as an indirect method
of quantifying physical condition in this species.

STUDY AREA AND METHODS

The study was conducted on the 252 ha Edward J. Meeman Biological
Station (MBS) located ca. 20 km north of Memphis, Tennessee, 35°20'N latitude
and 90°01'W longitude. Tree species are described as western mixed mesophyt-
ic forest (Braun 1950, Miller and Neiswender 1987). Five permanent ponds, one
temporary pond and permanent and intermittent streams provided water sources
to MBS. No trap site was located at a distance greater than 450 m from perma-
nent water. Topography varies from gentle (<3%) to steep (>45%) slopes. Cli-
matic conditions varied greatly throughout the three years of the study. Rainfall
at MBS from April to September was 77.4 cm in 1991, 50.9 cm in 1992, and 48.9
cm in 1993, and from October to March was 71.2 cm in 1991, 52.7 cm in 1992,
and 26.2 cm for October and November 1993. Mean temperature did not vary
greatly from the 30-year mean temperature at MBS. Mean seasonal temperature
was 16.5 C in spring, 26.6 C in summer, 17.3 C in autumn, and 6.0 C in winter.
There was an extended period equal to 30 days of >32 C daytime temperature
from July to August 1993 that did not occur during the previous two years.
Detailed analyses of the habitat and climatic patterns at MBS can be found in
Ladine (1995).

Beginning 3 February 1991 and running through 30 November 1993, a
5 X 10 trapping grid was established with folding Tomahawk live traps (Toma-
hawk Live Trap Co.; Tomahawk, Wisconsin) placed 150 m apart (grid size =112
ha). Traps were open four nights per week from 3 February 1991 to 31 January
1992, after which they were open three nights per week. Traps were baited with
canned cat food.

Mass to Girth 9 3

Captured raccoons were anesthetized with a 1 : 1 mixture of ketamine
hydrochloride (Ketaset; Bristol Laboratories, Syracuse, New York) and acepro-
mazine maleate (PromAce; Ayerst Laboratories, New York, New York), based on
estimated body mass, then tagged in both ears with No. 3 Monel ear tags (Nation-
al Band and Tag Co., Newport, Kentucky). Data collected from captured rac-
coons included mass, girth, and age. Raccoons were weighed at the site of cap-
ture to the nearest 0. 1 kg with a spring loaded scale. Girth measurements were
taken at the posterior end of the sternum immediately anterior to the xiphoid
process with a flexible tape pulled taught against the raccoon. Age was deter-
mined by tooth wear (Grau et al. 1970). Age classes were: 0-14 months (Age
I), 14-38 months (Age II), 38-56 months (Age III), 56-84 months (Age IV), and
>84 months (Age V).

Only adult raccoons, age II and older, were included in the analysis to
avoid confounding relationships due to different patterns in growth for younger
animals (Johnson 1970). To maintain independence of observations, only the
first capture of an individual was used in data analysis, except for analysis of
individual variation. Date of capture was categorized into two seasons, summer
(April through September) and winter (October through March). Data were ana-
lyzed by year, season, age, and sex.

All data were analyzed using Statistical Analysis Systems (SAS Insti-
tute 1989). Comparison of the mass to girth ratio was analyzed with General
Linear Models (PROC GLM) using Tukey's multiple comparison procedure to
test among classes. The relationship between mass and girth was analyzed using
linear regression (PROC REG). Because of the possibility that the allometric
relationship may differ among age classes, years, or between sexes and seasons,
parallelism and coincidence of the regression equation of the allometric relation-
ship between mass and girth were tested within each grouping. Parallelism tests
for the equality of the slope of the equation for more than one population. Coin-
cidence tests for the equality of the intercept of the y-axis of the equation.
Dummy variables were created to test for parallelism and coincidence for the
three years of the study, each season, the four adult ages, and sex.

To investigate the stability of the allometric relationship, the linearity of
the mass to girth ratio was examined with PROC REG for individuals with >5
captures. This analysis was used to determine if the allometric relationship
exhibited temporal variation among individuals.

RESULTS

One hundred and five raccoons (61 males, 42 females) were measured.
There was no difference among years (F = 0. 17; P = 0.8473), or between seasons
(F = 1.13; P = 0.2916). Differences in the mass to girth ratio were found among
age classes (F = 12.31; P < 0.0001) and between sexes (F = 25.03; P < 0.0001).
Age class II raccoons exhibited a smaller ratio than did the three older age class-

94

Troy A. Ladine

es (Table 1). The ratio was larger in males than females. There was no two-way
interaction for SEX X AGE (F = 1.08; P = 0.3622), SEASON X AGE (F = 0.97;
P = 0.4118) and SEASON X SEX (F = 3.36; P = 0.0709).

Table 1 . Mass to girth ratio for year, season, sex, and age for captured raccoons
(Procyon lotor). See text for explanation of age classes and definition of seasons.

Grouping
Year

Mean

Season

Sex

1991
1992
1993

Summer
Winter

Males
Females

Age class

III
IV

V

40

1.27:1.00

31

1.25:1.00

32

1.24:1.00

44

1.14:1.00

59

1.16:1.00

61

1.27:1.00*

42

1.05:1.00*

37

1.01:1.00*

38

1.32:1.00

22

1.22:1.00

6

1.23:1.00

* Means are different within grouping (P < 0.05).

There was no difference between sexes in either parallelism (F =
0.00002; P = 0.9964) or coincidence (F = 0.0407; P = 0.9601). There was no dif-
ference among ages for either parallelism (F = 0.0105; P = 0.9985) or coinci-
dence (F = 0.0277; P = 0.9998), between seasons for parallelism (F = 0.0365; P
= 0.8489) or coincidence (F = 0.0407; P = 0.9601), and among years for paral-
lelism (F = 0.0173; P = 0.9829) or coincidence (F = 0.0216; P = 0.9957).

Because there was no difference in parallelism or coincidence regard-
less of grouping, all data were analyzed in a single regression equation (Fig. 1).
The regression equation, was significantly linear (Mass = 0.02916 * (Girth) -
5.5904; r2 = 0.8044; P = 0.0001) and indicates a strong positive relationship
between mass and girth. The lack of difference for either parallelism or coinci-
dence indicate that the equation is useful across all age and sex classes regard-
less of season or year.

Mass to Girth

95

Thirteen raccoons were captured 5 or more times for analysis of indi-
vidual variation in the mass to girth ratio. The stability of the allometric rela-
tionship between girth and mass exhibited substantial variability in correlation
and linearity (Table 2). Only three raccoons showed a linear correlation between
girth and mass during their capture histories. One female (4160M) exhibited a
significant linear correlation without a strong fit of the data to the line indicating
little relationship between mass and girth for this individual.

Fig. 1 . Linear relationship between mass and girth of raccoons captured in west-
ern Tennessee from February 1991 to November 1993.

O)

9.0
8.0
7.0
6.0
5.0 h

co 4.0

3.0
2.0
1.0

0.0

200

/= 0.02916 (X)- 5.5904
r2 = 0.8044
P= 0.0001
/V=103

250

300 350

Girth (mm)

400

450

DISCUSSION

Several studies have shown a relationship between length and mass
when length is converted to a volumetric measure (see Johnson 1970, Dunn and
Chapman 1983, Vogel 1979, Schmidt-Nielsen 1984). The mass of an animal is
believed to be approximated by volume, thus, the reason comparing mass to the
cubed length of and animal. Because girth is a circumgeal measurement, a clos-
er aproximation to mass may be achieved. The relationship between mass and
girth has been shown for phocid seals (Hofman 1975, Castellini and Kooyman
1990) and dromedary camels (Camelus dromedarius) (Schroter et al. 1992).

Due to the manner in which most subcutaneous fat is accumulated in
raccoons (Stuewer 1943), girth should vary in relation to mass. Regression
analysis indicated that this occurred. Additionally, the mass to girth ratio was

9 6 Troy A. Ladine

linearly constant regardless of year, season, and sex class, with slight differences
among age classes. The lower ratio for age II raccoons might indicate that these
individuals are not able to accumulate the fat reserves of older animals resulting
from potential lower dominance status. The potential for dominance hierarchies
in raccoons has been shown experimentally (Barash 1974).

Table 2. Correlation of mass and girth of 13 individual raccoons (Procyon lotor).
See text for explanation of age class.

Raccoon

2760M
4022M
4145M
5208M
4080M
4083M
4090M
4100M
4117M
4160M
4186M
4188M
4190M

a Adjusted R2 was used due to the small sample sizes for each individual.

Sanderson (1950) noted that the strain of pregnancy and lactation have
a marked effect on the mass of female raccoons. Thus, adult females would be
expected to have lower fat reserves than adult males due to the stress of rearing
young. Furthermore, the observed difference between the males and females
might be due to potential physiological differences between the sexes, a poten-
tial social dominance of the males (see Barash 1974), or a combination of the
two.

The seasonal variation observed in individuals could be due to season-
al mass changes. Raccoons are known to exhibit seasonal fluctuations in mass
(Stuewer 1943, Mech et al. 1968, Moore and Kennedy 1985). As indicated pre-
viously, the mass to girth ratio in raccoons is a linearly constant relationship.
Consequently, one would expect to see girth fluctuate with mass.

Class

Sex

Number of
Captures

Adjusted R2

P

V

M

5

-0.1134

0.4923

III

M

8

0.7185

0.0049

II

M

10

0.9030

0.0001

II

M

5

-0.2968

0.7901

II

F

5

0.1993

0.3172

III

F

8

0.3829

0.0601

II

F

6

0.8742

0.0127

II

F

10

-0.0667

0.5269

IV

F

6

-0.0259

0.4130

IV

F

9

0.5973

0.0089

II

F

7

0.2152

0.1648

II

F

9

0.2329

0.1065

II

F

6

-0.0819

0.4746

Mass to Girth 9 7

The assumption for using a mass to girth ratio as a measure of physical
condition was: the greater the amount of fat the better the physical condition of
a raccoon. This assumption is based on the manner in which subcutaneous fat is
accumulated in raccoons (see Stuewer 1943). Body fat and kidney fat indices
have shown some correlation of the amount of fat to physical condition for the
species (Johnson 1970, Dunn and Chapman 1983).

My findings indicate that the mass to girth ratio was not indicative of
physical condition in raccoons for several reasons. No difference in slope or
intercept existed between any groupings for the equation relating mass to girth.
Although a difference existed among age classes and between sexes, this differ-
ence can be explained by reasons other than physical condition. Further evi-
dence of the shortcoming of the mass to girth ratio as an indicator of physical
condition was the lack of any consistent relationship between the two variables
when the relationship was tested for each individual. All individuals appeared to
be in good condition when captured, and differences in the ratio probably only
reflect individual variation and might not be indicative of physical condition.

Thus, I concluded that the mass to girth ratio was not a good indicator
of physical condition. Nevertheless, because of need for determining physical
condition during long-term population and community ecology studies, there
remains a necessity to determine physical condition through nonintrusive meth-
ods.

ACKNOWLEDGMENTS - I thank Michael L. Kennedy and Timothy M. Stew-
ard for their aid in collecting data, and Mark E. Ritke for supplying rainfall data
for 1991 and 1992. I thank John R. Ouellette and the anonymous reviewers for
their constructive criticism.

LITERATURE CITED

Barash, D. P. 1974. Neighbor recognition in two "solitary" carnivores: the rac-
coon (Procyon lotor) and the red fox (Vulpesfulva). Science 185:794-796.

Bissonette, T. H., and A. G. Csech. 1938. Sexual photoperiodicity of rac-
coons on low protein diet and second litters in the same breeding season.
Journal of Mammalogy 19:342-348.

Braun, E. L. 1950. Deciduous forests of eastern North America. The Blakiston
Company, Philadelphia, Pennsylvania.

Castellini, M. A., and G. L. Kooyman. 1990. Length, girth and mass relation-
ships in weddell seals (Leptonychotes weddellii). Marine Mammal Science
6:75-77.

Dunn, J. P., and J. A. Chapman. 1983. Reproduction, physiological responses,
age structure, and food habits of raccoon in Maryland, USA. Zeitschrift fur
Saugetierkuande 48:161-175.

9 8 Troy A. Ladine

Glenn, T. C, and W. R. Clark. 1990. Correlation of raccoon pelt qualities with
age, sex, and physical condition. Journal of the Iowa Academy of Sciences
97:33-35.

Grau, G. A., G. C. Sanderson, and J. P. Rogers. 1970. Age determination of rac-
coons. The Journal of Wildlife Management 34:364-372.

Halls, L. K. 1984. White-tailed deer: ecology and management. Stackpole
Books, Harrisburg, Pennsylvania.

Johnson, A. S. 1970. Biology of the raccoon {Procyon lotor varius) Nelson
and Goldman in Alabama. Auburn University, Agricultural Experiment Sta-
tion Bulletin Number 402.

Ladine, T. A. 1995. Ecology of co-occurring populations of Virginia opossums
{Didelphis virginiana) and raccoons {Procyon lotor). Ph.d. Thesis. Univer-
sity of Memphis, Tennessee.

Mech, L. D., D. M. Barnes, and J. R. Tester. 1968. Seasonal mass changes, mor-
tality, and population structure of raccoons in Minnesota. Journal of Mam-
malogy 49:63-73.
Miller, N. A., and J. B. Neiswender. 1987. Plant communities of the third
Chickasaw loess bluff and Mississippi River alluvial plain, Shelby County,
Tennessee. Journal of the Tennessee Academy of Science 92: 1-6.

Moore, D. W., and M. L. Kennedy. 1985. Mass changes and population struc-
ture of raccoons in western Tennessee. The Journal of Wildlife Management
49:906-908.

SAS Institute, Inc. 1989. SAS/STAT User's Guide, Version 6, 4th edition, Cary,
North Carolina.

Sanderson, G. C. 1950. Methods of measuring productivity in raccoons. The
Journal of Wildlife Management 14:389-402.

Sanderson, G. C, and V Nalbandov. 1973. The reproductive cycle of the rac-
coon in Illinois. Illinois Natural History Survey Bulletin 31:29-85.

Schmidt-Nielsen, K. 1984. Scaling: why is animal size so important? Cam-
bridge University Press, New York, New York.

Schroter, R. C, D. H. O'Neill, M. R. Achaaban, and A. Ouhsine. 1992. The pre-
diction of body weight of the dromedary camel using girth at the hump as an
index. Indian Veterinary Journal 69:925-930.

Stuewer, E. W. 1943. Raccoons: their habits and management in Michigan.
Ecological Monographs 13:203-257.

Thompson, D. W. 1961. On growth and form. Cambridge University Press,
New York, New York.

Vogel, S. 1979. Life's processes: the physical world of animals and plants.
Princeton University Press, Princeton, New Jersey.

Received 23 August 1996
Accepted 9 September 1996

A Multiscale Approach to Capture Patterns and
Habitat Correlations of Peromyscus leucopus (Rodentia, Muridae)

Troy A. Ladine1 and Angela Ladine1
The University of Memphis, Edward J. Meeman Biological Station and Ecology
and Organismal Biology, Campus Box 526080, Memphis, Tennessee 38152-6080

ABSTRACT— Capture patterns (presence/absence) of Peromyscus leu-
copus were examined in relation to 1 2 selected habitat variables at three
spatial scales. Trapping was conducted on a 14 X 14 trapping grid
established at the Edward J. Meeman Biological Station in southwest-
ern Tennessee. Density of the population was estimated at 18.5 mice
per hectare. Twelve habitat variables were collected in three circular
plots (1 m2, 5 m2, 10 m2) centered on 60 trap sites (30 trap sites where
captures of P. leucopus occurred, 30 randomly selected sites where no
captures occurred). There was a significant difference among spatial
scales for six habitat variables. We observed no discernable patterns
through principal components analysis for any scale. However, the cen-
troid of the cluster of traps in principal component space shifted from
negative to positive as scale increased. Sites where captures occurred
and those where no captures occurred were not significantly different at
the 1-m2 scale for any habitat variables. Capture occasions differed sig-
nificantly for stems 10-15-cm diameter and logs 10-15 cm at the 5-m2
and 10-m2 spatial scales, respectively. Our study emphasizes the need
for including multiscale assessments of habitat use. Scales might best
be selected by assessing the habitat of the study site and the behavior of
the species being studied.

The concept of scale, while not a new concept in other disciplines, has
only recently been investigated in ecology (Wiens 1989). Levin (1991) stated
that because there is an absence of any correct scale at which to investigate a
population, a multiscale approach should be taken. Thus, investigations of
species relating habitat use to capture success could be affected by the selected
scale. Studies relating habitat use to capture success have generally selected a
single scale in which to measure the habitat. This scale of habitat assessment is
usually based on amount of time spent for amount of data return. Thus, the scale

Present Address: Division of Mathematics and Science, Bethel College, McKenzie, Tennessee 38201 .

99

100 Troy A. Ladine and Angela Ladine

selected for habitat assessment might not be representative for the species being inves-
tigated or may affect the results of the study (see Levin 1991, Schneider 1994).

The need for a multiscale approach has been demonstrated in several
studies of species interactions. Depending on the scale selected, species of
marine birds were or were not associated with their prey species (Woodby 1984,
Schneider and Piat 1986). Least flycatchers (Empidonax minimus) and redstarts
(Setophaga ruticilla) had a negative association at small scales and a positive
association at larger scales (Sherry and Holmes 1988). Furthermore, behavior of
an animal can be affected by the spatial scales at which prey are distributed
(Boyd 1996).

Similarly, the association between habitat around a live trap and capture
of a selected species lends itself to a multiscale approach. However, capture-
recapture studies rarely, if ever, use multiple scales to assess correlations
between captures and habitat use. Using capture success is warranted for stud-
ies investigating habitat correlations because densities within a habitat can be
influenced by factors (e.g. intra- and interspecific interactions) that place subor-
dinates into suboptimal habitats (van Home 1983). Also, factors such as curios-
ity of a new object (e.g. a trap) in an area may influence captures (Lackey et al
1985). However, an animal must be present in a habitat for a capture to occur;
thus, must use the habitat in some way.

The objective of our study was to investigate the association between
captures of Peromyscus leucopus and selected habitat variables at three spatial
scales centered on location of live traps. Although there is a large amount of lit-
erature on P. leucopus (see Lackey et al. 1985), to our knowledge, no study has
been conducted relating spatial scale to the association between capture success
of P. leucopus and selected habitat variables.

Peromyscus leucopus is an excellent organism to use in multiscale
analyses of correlations between captures and habitat. The species is well stud-
ied throughout its range, and habitat affinities are well documented (see Lackey
et al. 1985). Because P leucopus is a small mammal, a multiscale study design
can be done at small scales, and fine grained changes in habitat are more likely
to be exhibited. Previous investigations of habitat affinities of P. leucopus (see
Lackey et al. 1985) indicate loosely defined associations. However, these loose-
ly defined associations may become more clearly defined with a different or
more meaningful choice of scales.

STUDY AREA AND METHODS
The study was conducted at the 252-ha Edward J. Meeman Biological
Station (hereafter referred to as the station) located ca. 20 km north of Memphis,
Tennessee, (35°20' N, 90°01* W) on the third Chickasaw loess bluff. The station
is surrounded on three sides by private lands and on the fourth by the Shelby For-
est Wildlife Management Area.

Multiscale 1 o 1

Habitat has been described as a western mixed mesophytic forest
(Braun 1950, Miller and Neiswender 1987). Dominant canopy plants are sweet
gum {Liquidambar stryaciflua), tulip poplar (Liriodendron tulipifera), elms
(Ulmus spp.), oaks (Quercus spp.), and hickories (Carya spp.). There is an
extensive network of grape (Vitis spp.) and poison ivy (Toxicodendron radicans)
vines throughout the canopy. The understory is dominated by spicebush (Lin-
dera benzoin). Dominate ground cover species are Osmorhiza sp., Smilacina
racemosa, Toxicodendron radicans, Urtica sp., various woodland grass species,
and seedlings of the dominant canopy and understory species. A detailed analy-
sis of the habitat on the station can be found in Ladine (1995).

A 14 X 14 trapping grid was established using folding Sherman live
traps (H. B. Sherman Traps, Inc.; Tallahassee, Florida) spaced ca. 10-m apart.
Trapping was conducted from 28 January 1995 through 9 February 1995. Traps
were baited with oatmeal, left open during the day, and checked at sunrise. Esti-
mation of population size was made using the Schnabel method (Krebs 1989).

Location of the trapping grid was entirely within a mature stand of oak,
sweet gum, and tulip poplar trees. The selected location has been shown to be
homogenous on the macrohabitat scale (Ladine 1995). Placing the grid in this
location avoided potential confoundment during statistical analyses posed by
placing traps in differing macrohabitats.

Trap sites were classified according to the occurrence of captures of P.
leucopus. Trap sites with at least one capture were classified as capture sites.
Other sites were classified as no-capture sites. To strengthen the multivariate
analyses and remove the possibility of nonorthoganal functions and components
(Tabachnick and Fidell 1989), thirty randomly selected no-capture sites were
designated for habitat association analyses.

Twelve selected habitat variables (Table 1) were measured at each cap-
ture site and at each no-capture sites. All selected habitat variables were mea-
sured at each of three spatial scales (1 m2, 5 m2, and 10 m2) in circular plots cen-
tered on each trap. These scales were selected following Noon (1981) who sug-
gested that a more homogeneous habitat be sampled more finely than a hetero-
geneous habitat in order to detect the inherent heterogeneity. Thus, because of
the apparent homogeneity of the habitat within the trapping grid (Ladine 1995),
these selected scales were used.

All statistical analysis were conducted using Statistical Analysis Sys-
tems (SAS Institute 1989). Habitat variables for capture and no-capture sites
were compared at each scale with a Kruskal-Wallis test of Chi-square approxi-
mation. Selected habitat variables between scales were compared with a
Kruskal-Wallis test of Chi-square approximation to test for differences among
selected scales. To control for group-wide Type I error, all multiple pairwise
comparisons were made using a sequential Bonferroni adjustment (Rice 1989)
with initial « = .05.

102

Troy A. Ladine and Angela Ladine

Table 1 . Description of selected habitat variables measured at 30 sites with cap-
tures of Peromyscus leucopus occurred and 3 1 sites with no captures of P. leu-
copus for a study in western Tennessee.

Habitat variables Description

COVER0 Percent green vegetation at ground level

COVER 1 Percent green vegetation at 1 m height

COVER2 Percent green vegetation at 2 m height

STEMS0-5 Number of vertical woody stems with diameter <5 cm

STEMS5-10 Number of vertical woody stems with diameter 5-10 cm

STEMS 10- 15 Number of vertical woody stems with diameter 10-15 cm

STEMS>15 Number of vertical woody stems with diameter >15 cm

LOGS0-5 Number of horizontal woody stems on the ground with

diameter of <5 cm
LOGS5-10m Number of horizontal woody stems on the ground with

diameter 5-10 cm
LOGS 10- 15 Number of horizontal woody stems on the ground with

diameter 10-15 cm
LOGS>15 Number of logs at ground level with diameter >15 cm

LITTER Mean of seven leaf litter depths taken for each scale

The existence of potential patterns at each scale was examined with
principal components analysis. Discriminate function analysis was used to fur-
ther examine the difference between sites with captures and sites where no cap-
tures occurred. Initial discriminating variables were selected with stepwise selec-
tion discriminate analysis and an initial entry level of significance of P = 0.15.
Variables were removed or added to check the selection of variables from the
stepwise selection procedure for improvement of the discriminating capabilities
of the variables. No addition or subtraction improved the classification for any
scale.

RESULTS
Thirty-one Peromyscus leucopus were captured 55 times at 30 trap
sites. Population size was estimated at 32 mice (range = 26 - 38) with a mean
density of 18.5 mice per hectare. Other species, Tamias striatus (n = 1), Mari-
na carolinensis (n = 2), Glaucomys volans, (n = 7) were captured at eight addi-
tional sites. No P. leucopus were captured at trap sites where captures of other
species occurred.

Multiscale

103

■3

00

sS

•a

X

i

€ 2

«3

o

W5

■s

P

<**

.5

o

x

c

T3

_o

§

a.

'C

Q

o

<o

-H

T3

£|

M

"8 e

" fi

X) 6

vor-voom>ncsor^oo-Hfn

f^Ttr^OOOOOOOf^^

<5<5c5<5<5<od>d>d>d>c>c>

ISt

00 OO N rf

■S a

r^ — i — 'U:o\<N\Or«">oooo<N©

+) +i +i +i -h +i +i

+1 -H

(N

cncJ — TtodoSocDO

■«t cs «n oo o o
■H -H -H -H -H +1

(Nl O

o 2 o o o

SO v£> 0O SO

d d r-" o d c>

+1 -H -H -H -H ""

m h Nn oo -h *
«- — • rf -<3- O — • "1

ON

o o o

ONn't -t h- oo © r- r- r~

CNOVOCStNtNCNvqiOcncS^

■<*— loiv-idodr-'ddd —
■h-h-h-h-h-h-h-h-h-h-h*

JU

JO

"fi

>

o

-^

CM

«_,

9$

OS

^

2

£>

S

&

•s

n»

^

^

«

p o o

X

o o u

^ 2 "7 «o

*? ^ d *^

O "O *n A
§ § § §

SSSSOOOOH

■fi

o u-i i; a ps

Co co 03 to ^

bKK

104 Troy A. Ladine and Angela Ladine

Significant differences among the selected scales were found for
STEMS<5, STEMS5-10, STEMS>15, LOGS<5, LOGS5-10, and LOGS10-15
(Table 2). Except for LOGS<5, the 1-m2 and 5-m2, scales did not differ signifi-
cantly for selected variables. The 10-m2 scale was significantly different from
both the 1-m2 and 5-m2 scales for all variables exhibiting significant differences
between the three scales.

No significant difference was observed between capture and no-capture
sites for selected habitat variables at the 1-m2 scale (Table 3). Except for
STEMS 10- 15, no significant difference was found between capture and no-cap-
ture sites for selected habitat variables at the 5-m2 scale (Table 4). At the 10-m2
scale, a significant difference was found between capture and no-capture sites for
LOGS10-15 (Table 5).

Sites with no captures tended to be centered in the cluster of sites on
graphs of principal components for all scales. Outliers from sites where no cap-
tures occurred were only observed at the 1-m2 spatial scale. Percent variation
accounted for by the first three principal components for each scale was 42.0%
at 1 m2, 39.5% at 5 m2, and 45.3% at 10 m2 . Variables loading on each of the
first three principal components varied for each scale. For the 1-m2 scale, all

Table 3. Selected habitat variables (x ± SD) and Kruskal-Wallis test (%2 approx-
imation and probability values) for differences between sites where Peromyscus
leucopus were captured and randomly selected sites where no captures occurred
for the 1-m2 scale. (See Table 1 for description of habitat variables. See text for
explanation of capture and no-capture sites.)

Habitat variable

Capture

No-Capture

t

P

COVER0

2.50 ± 4.84

2.29 ±

3.68

0.02

0.8841

COVER 1

0.13 ± 0.73

0.29 ±

1.19

0.36

0.5507

COVER3

0.07 ± 0.37

0.74 ±

3.51

0.36

0.5507

STEMS0-5

4.63 ± 5.33

4.59 ±

5.57

0.46

0.4995

STEMS5-10

0.03 ± 0.18

0.09 ±

0.29

0.99

0.3210

STEMS 10- 15

0.13 ± 0.35

0.03 ±

0.17

2.04

0.1536

STEMS>15

0.03± 0.18

0.06 ±

0.24

0.31

0.5766

LOGS0-5

8.13 ± 8.49

5.32 ±

6.51

2.06

0.1512

LOGS5-10

0.17 ± 0.46

0.29 ±

0.52

2.04

0.1532

LOGS 10- 15

0.13 ± 0.51

0.03 ±

0.17

0.43

0.5128

LOGS>15

0.13 ± 0.35

0.03 ±

0.17

2.04

0.1536

LITTER

49.22 ± 13.52

44.45 ±

11.28

2.10

0.1471

Multiscale

105

Table 4. Selected habitat variables (x ± SD) and Kruskal-Wallis test (%2 approx-
imation and probability values) for differences between sites where Peromyscus
leucopus were captured and randomly selected sites where no captures occurred
for the 5-m2 scale. (See Table 1 for description of habitat variables. See text for
explanation of capture and no-capture sites.)

Habitat variable

Capture

No capture

t

P

COVERO

2.20 ± 4.63

2.85 ± 3.82

2.02

0.1558

COVER 1

0.17 ± 0.65

0.94 ± 3.79

0.22

0.6383

COVER2

0.33 ± 1.09

1.18 ± 6.86

1.02

0.3132

STEMSO-5

8.60 ± 9.35

8.09 ± 9.17

0.32

0.5716

STEMS5-10

0.17 ± 0.38

0.09 ± 0.29

0.64

0.4227

STEMS 10- 15

0.23 ± 0.43

0.06 ± 0.24

5.32

0.0211

STEMS>15

0.07 ± 0.25

0.15 ± 0.36

1.32

0.2503

LOGS0-5

12.47 ± 14.37

10.38 ± 11.18

0.02

0.8793

LOGS5-10

0.27 ± 0.58

0.56 ± 1.39

3.26

0.0709

LOGS 10- 15

0.13 ± 0.51

0.03 ± 0.17

0.43

0.5128

LOGS>15

0.17 ± 0.46

0.06 ± 0.24

0.85

0.3565

LITTER

48.80 ± 12.37

49.53 ± 11.34

0.06

0.8118

Table 5. Selected habitat variables (x ± SD) and Kruskal-Wallis test (x2 approx-
imation and probability values) for differences between sites where Peromyscus
leucopus were captured and randomly selected sites where no captures occurred
for the 10-m2 scale. (See Table 1 for description of habitat variables. See text
for explanation of capture and no-capture sites.)

Habitat variable

Capture

No capture

t

P

COVERO

2.53 ± 4.14

3.47 ± 4.86

1.62

0.2033

COVER 1

0.47 ± 1.48

1.29 ± 4.93

0.13

0.7152

COVER2

0.60 ± 2.06

1.41 ± 7.72

0.24

0.6275

STEMS0-5

15.43 ± 15.34

15.15 ± 16.54

0.91

0.3395

STEMS5-10

0.33 ± 0.61

0.26 ± 0.57

0.12

0.7298

STEMS 10- 15

0.30 ± 0.47

0.18 ± 0.39

1.63

0.2015

STEMS>15

0.17 ± 0.38

0.27 ± 0.51

0.83

0.3611

LOGSO-5

20.63 ±25.15

16.38 ± 17.18

0.02

0.8851

LOGS5-10

0.67 ± 1.09

0.71 ± 1.06

0.28

0.5937

LOGS 10- 15

0.50 ± 0.90

0.09 ± 0.29

5.97

0.0144

LOGS >15 cm

0.23 ± 0.50

0.12 ± 0.33

0.62

0.4318

LITTER

52.26 ± 13.04

50.65 ± 13.05

0.12

0.7292

106 Troy A. Ladine and Angela Ladine

percent cover measurements correlated positively along the first principal com-
ponent. Vertical stems correlated along the second principal component with the
two smaller stem categories correlating positively and the larger stem categories
correlating negatively. Logs were correlated to the third principal component
with all but LOGS>15 correlated positively. For the 5-m2 scale, the first princi-
pal component was correlated positively to all cover measurements and
STEMSO-5, and negatively to STEMS 10- 15. The second principal component
was correlated positively to LOGSO-5 and LOGS5-10. The third principal com-
ponent was correlated positively to STEMS5-10 and STEMS>15 and LITTER,
and correlated negatively to LOGS 10- 15 and LOGS>15. For the 10-m2 scale, the
first principal component was correlated positively to STEMS 10- 15, LOGSO-5,
and LITTER, and correlated negatively to COVER0 and COVER 1. The second
principal component was correlated positively to LOGS5-10, LOGS10-15, and
LOGS>15. The third principal component was correlated positively to COVER 1
and STEMS5-10, and correlated negatively to STEMSO-5, and STEMS>15.

Correct classification of sites with captures was poor for all scales: 1 m
~ 46.7%; 5 m~ 56.7%; 10 m ~ 40.0%. Classification of sites where no captures
occurred was better at all three scales: 1 m — 70.0%; 5 m — 98.3%; 10 m -
76.7%. Variables selected for discriminating between capture and no-capture
sites were different for each scale. LITTER, and LOGS>15 were selected for the
1-m2 scale. STEMS10-15, 2 m COVER, and LOGS>15 cm were selected at the
5-m2 scale. LOGS10-15, 2 m COVER, STEMSO-5, and LITTER were selected
at the 10-m2 scale.

DISCUSSION

Members of the genus Peromyscus exhibit habitat generality, at least on
a local scale, and often occur across a broad range of habitats within a small geo-
graphic area (Kirkland 1976, Batzli 1977, Sullivan 1979, Van Home 1981,
Martell 1983, Adler et al. 1984). There are conflicting reports concerning rela-
tionship between density of P. leucopus and habitat type (Klein 1960, Stickel and
Warbach 1960, Getz 1961, Bongiorno and Pearson 1964, Kaufman and Fleharty
1974). Density in our study was within the reported ranges for the species (see
Lackey et al. 1985). Findings of our study at the 5-m2 scale, in concurrence with
Kaufman and Fleharty (1974), suggest a relationship between number of stems
10-15 cm and captures of P. leucopus. However, this relationship was not
observed at the 10-m2 scale. A relationship between logs 10-15 cm and captures
of P. leucopus was observed at the 10-m2 scale. These finding are similar to those
of Getz (1961). Although relationships were observed between two of the select-
ed habitat variables and captures of P. leucopus, lack of significant relationships
between other variables for all scales appears to reflect the habitat generality of
the species.

Multiscale 107

Lack of readily discernable patterns between captures and selected
habitat variables could be a reflection of the variables we selected and lack of dif-
ferences between spatial scales for some variables. There are at least three
potential reasons for the lack of a readily discernable pattern between capture
and no-capture sites. First, low statistical power might have resulted in the lack
of differences observed in our study. However, because of the large amount of
variation observed for all means and the finding of significant differences for the
larger scales, the lack of differences is most likely not due to low statistical
power. Second, captures of P. leucopus are often related to factors other than
habitat. The species is known to respond to new objects placed within a famil-
iar area (Lackey et al. 1985), and densities have been shown to correlate to food
distribution (Getz 1961). Third, the spatial scales selected for study might have
been of an incorrect size for ascertaining capture patterns. However, the lack of
patterns in our study does not necessarily indicate scale is not important in asso-
ciating captures of P. leucopus with habitat, only that a different scale may be
warranted for future studies.

Our study shows at least two potential means by which selection of
scale could influence results of a study warranting investigation of potential pat-
terns at each scale. First, significant differences found for some variables sug-
gest difference at the 10-m2 scale, but not at smaller scales. Second, the differ-
ences in the loadings of variables on the principal components axes, differences
in variables selected for use in discriminant analysis, and the decrease in outliers
as scale increases suggest differences between all of the scales. These differ-
ences between scales may potentially have a large effect on the conclusions (see
Schneider and Piat 1986, Woodby 1984). For example, our study had inconse-
quential findings at the 1-m2 scale; but significant results at the larger scales.

Our data show that, even when using the small scales that we selected,
differences in habitat affinities for capture can occur around the same trap site.
Because our data were collected in a homogenous habitat over a short period of
time, the differences observed in our study concerning correlations between
habitat and capture can be attributed to the different scales. While a species may
appear to be a habitat generalist with an affinity toward a variety of habitats (e.g.
P. leucopus), studies incorporating a multiscale approach may indicate a narrow-
er range of optimal habitat affinities. Thus, studies assessing habitat use should
incorporate analyses at multiple scales. It appears this may be achieved by the
incorporation of at least three scales of assessment allowing for comparison at
different scales in the same habitat.

More study is needed in the selection of scale to be measured. Selec-
tion of scale is difficult to evaluate due to differences in habitat at each study site.
However, we feel choice of scale should be selected based on at least the fol-
lowing factors. Of primary concern should be the habitat in which the study is
conducted. More homogeneous habitats may require a larger number of scales

108 Troy A. Ladine and Angela Ladine

to detect observable differences. Additionally, the behavior of the species being
studied must be addressed in selecting the size and number of scales to be
assessed. For example, species with large ranges will require large scales to
account for greater movement of individuals of these species.

ACKNOWLEDGEMENTS - We thank Michael L. Kennedy for the use of the
Sherman traps and the Edward J. Meeman Biological Station for the use of facil-
ities. We thank the reviewers for their constructive comments on a previous ver-
sion of this manuscript.

LITERATURE CITED

Adler, G. H., L. M. Reich, and R. H. Tamarin. 1984. Characteristics of white-
footed mice in woodland and grassland. Acta Theriologica 29:57-62.

Batzli, G. O. 1977. Population dynamics of the white-footed mouse in flood-
plain and upland forest. American Midland Naturalist 97:18-32.

Bongiorno, S. F., and P. G. Pearson. 1964. Orientation of Peromyscus in rela-
tion to chronic gamma radiation and vegetation. American Midland Natu-
ralist 72:82-92

Boyd, I. L. 1996. Temporal scales of foraging in a marine predator. Ecology
77:426-434.

Braun, E. L. 1950. Deciduous forests of eastern North America. The Blakiston
Company, Philadelphia, Pennsylvania.

Getz, L. L. 1961. Notes on the local distribution of Peromyscus leucopus and
Zapus hudsonicus. American Midland Naturalist 65:486-500.

Kaufman, D. W, and E. D. Fleharty. 1974. Habitat selection by nine species of
rodents in north-central Kansas. Southwestern Naturalist 18:443-452.

Kirkland, G. L., Jr. 1976. Small mammals of a mine waste situation in the cen-
tral Adirondacks, New York: A case of opportunism by Peromyscus man-
iculatus. American Midland Naturalist 95: 103-1 10.

Klein, H. G. 1960. Habitat relationships of Peromyscus leucopus noveboracen-
sis and P. maniculatus gracilis in central New York. Ecological Mono-
graphs 30:387-407.

Krebs, C. J. 1989. Ecology methodology. Harper Collins Publishers, New York,
New York.

Lackey, J. A., D. G. Huckaby, and B. G. Ormiston. 1985. Peromyscus leucopus.
Mammalian species. Number 247.

Ladine, T. A. 1995. Ecology of co-occurring populations of Virginia opossums
(Didelphis virginiana) and raccoons (Procyon lotor). Ph.D. Thesis. Uni-
versity of Memphis, Tennessee.

Levin, S. A. 1991. The problem of pattern and scale in ecology. Ecology
73:1943-1967.

Multiscale 109

Martell, A. M. 1983. Demography of southern red-backed voles (Clethriono-
mys gapped) and deer mice {Peromyscus maniculatus) after logging in
north-central Ontario. Canadian Journal of Zoology 61:958-969.

Miller, N. A., and J. B. Neiswender. 1987. Plant communities of the third
Chickasaw loess bluff and Mississippi River alluvial plain, Shelby County,
Tennessee. Journal of the Tennessee Academy of Science 92: 1-6.

Noon, B. R. 1981. Techniques for sampling avian habitats. Pages 42-52 in The
use of multivariate statistics in studies of wildlife habitat (D. E. Capen, edi-
tor). United States Forest Service General Technical Report RM-87.

Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225.

SAS Institute, Inc. 1989. SAS/STAT User's Guide, Version 6, fourth edition,
Cary, North Carolina.

Schneider, D. C, and J. F. Piat, 1986. Scale-dependent correlation of seabirds
with schooling fish in a coastal system. Marine Ecology — Progress Series
32:237-246.

Schneider, D. C. 1994. Quantitative ecology: Spatial and temporal scaling.
Academic Press, New York, New York.

Sherry, T. W., and R. T. Holmes. 1988. Habitat selection by breeding American
Redstarts in response to a dominant competitor, the Least Flycatcher. The
Auk 105:350-364.

Stickel, L. F., and O. Warbach. 1960. Small-mammal populations of a Maryland
woodlot, 1949-1954. Ecology 41:269-286.

Sullivan, T. P. 1979. Demography of populations of deer mice in coastal forest
and clear-cut (logged) habitats. Canadian Journal of Zoology 57:1636-
1648.

Tabachnick, B. G. and L. S. Fidell. 1989. Using multivariate statistics. Harper
Collins Publishers, Inc., New York, New York.

van Home, B. 1981. Demography of Peromyscus maniculatus populations in
serai stages of coastal coniferous forest in southeast Alaska. Canadian Jour-
nal of Zoology 59:1045-1061.

van Home, B. 1983.- Density as a misleading indicator of habitat quality. Jour-
nal of Wildlife Management, 47:893-901.

Wiens, J. A. 1989. Spatial scaling in ecology. Functional Ecology 3:385-397

Woodby, D. A. 1984. The April distribution of murres and prey patches in the
southeastern Bering Sea. Limnology and Oceanography 29:181-188.

Received 23 August 1996
Accepted 25 February 1997

Methods Used To Survey Shrews (Insectivora: Soricidae) And
The Importance Of Forest-Floor Structure

Timothy S. McCay

Institute of Ecology and Department of Statistics, University of Georgia,

Athens, Georgia 30602

Joshua Laerm
Museum of Natural History, University of Georgia, Athens, Georgia 30602

M. Alex Menzel

Daniel B. Warnell School of Forest Resources, University of Georgia,

Athens, Georgia 30602

and

William M. Ford

Westvaco Corporation, Timberlands Division, Box 577, Rupert,

West Virginia 25984

ABSTRACT— We examined shrew (Insectivora: Soricidae) capture
rates using selective (best-site) transects, linear transects, and drift-
fence arrays to better understand how pitfall trap arrangement might
affect our perception of shrew assemblages in the southern Appalachi-
an mountains. Also, we studied the use of microhabitat structure
(coarse woody or rocky debris) by shrews to determine how microhab-
itat selection might affect capture probabilities. The distributions of
shrew captures were similar at selective and linear transects, but differ-
ent between either transect type and the drift-fence arrays (P < 0.05).
Differences in the effectiveness of trap arrangements were apparently
related to microhabitat use. We found a gradient of selection for habi-
tat structure among Sorexfumeus, S. cinereus, and Blarina brevicauda,
although relationships were weak. Captures of S. fumeus were most
closely associated with the abundance of and distance to woody or
rocky debris, and those of B. brevicauda were independent of these
microhabitat factors. Caution should be used when comparing the
results of surveys using different pitfall trap arrangements.

110

Survey of Shrews 1 1 1

Ecologists studying small mammals often attempt to accurately depict
the structure of small mammal assemblages from trapping data. This effort is
complicated by differences in size and microhabitat use among species, which
can affect species- and trap type-specific probabilities of capture. Some types of
sampling, notably mark-and-recapture (Otis et al. 1978), may be used to estimate
capture probability and avoid this as a confounding factor. However, survey of
shrew (Soricidae) assemblages using live-trapping methods is made problematic
by high rates of trap mortality, and removal sampling is commonly employed
using pitfall traps (Kirkland and Sheppard 1994). Therefore, particular care must
be taken to minimize biases associated with sampling shrew communities.

Many studies have examined differences among types of traps used to
sample shrews (e.g., Williams and Braun 1983). However, there is little infor-
mation regarding biases introduced through the arrangement of traps. Despite
recent efforts to promote standardized methods (Handley and Varn 1994, Kirk-
land and Sheppard 1994), many different pitfall-trap arrangements have been
used to survey shrews (Kalko and Handley 1993). Because trap arrangements,
like trap types, vary in their effectiveness at catching certain species (Bury and
Corn 1987, Mitchell et al. 1993), the assessment of shrew community structure
could be affected by trap arrangement.

Pitfall trapping designs often take advantage of patterns of microhabitat
use, such as drifting behaviors often observed when small mammals encounter
an obstruction (Brillhart and Kaufman 1991). Because these behaviors may vary
among species, methods that rely on drifting could selectively under- or over-
represent certain species in samples. Two methods that take advantage of drift-
ing behavior are transects of traps placed along natural habitat structures, such as
fallen logs or exposed rock (selective transects), and drift-fence arrays, which
use artificial obstructions to direct small mammal movement.

To assess how perception of a shrew assemblage might vary with trap-
ping design, we concurrently sampled shrews with selective transects, linear
transects, and drift-fence arrays in the southern Appalachians. To gain insight
into behaviors that might affect capture success with these trapping techniques,
we also examined microhabitat (coarse woody or rocky debris) use by shrews.

METHODS
STUDY AREA

We conducted our study at the Coweeta Hydrologic Laboratory
(35o03'N,83o25'W), located in the Nantahala Mountain Range of Macon Coun-
ty, North Carolina (Swank and Crossley 1988). Elevation at our study plots
ranged from 792 to 1,524 m above sea level. Study plots were restricted to plant
communities typical of cove hardwood and northern hardwood forests (Wharton
1977). Cove hardwood forests were characterized by the dominance of yellow
poplar (Liriodendron tulipifera), yellow buckeye (Aesculus octandra), black

112 Timothy S. McCay, Joshua Laerm, M. Alex Menzel

and William M. Ford

cherry {Prunus serotina), and birch (Betula spp.) in the canopy, sparse woody
vegetation below the canopy, and lush herbaceous vegetation. Northern hard-
wood forests were dominated by black oak (Quercus velutina), northern red oak
(Q. rubra), yellow birch (B. luted), and black cherry in the canopy. Rhododen-
dron {Rhododendron maximum) and mountain laurel (Kalmia latifolia) were
common shrubs, and the composition and density of herbaceous vegetation was
variable.

COMPARISON OF TRAPPING METHODS

At each of 12 plots we established one selective transect and one linear
transect of pitfall traps in July 1994. Both transects consisted of 20 traps placed
at 5-m intervals, were approximately parallel, and were separated by 50 m. Pit-
falls in selective transects were placed along logs, rocks, and stumps where our
previous experience had indicated that chances for shrew capture might be good.
Traps in linear transects were placed without regard to microhabitat conditions.
Pitfalls were tapered plastic cups (11-cm lip diameter and 14-cm depth) partial-
ly filled with preservative and set flush with the ground surface.

In August 1995, we constructed a series of five Y-shaped drift-fence
arrays at each of four plots randomly chosen from among the 12 original plots.
Each array consisted of three, 3-m "arms" of 36-cm-wide aluminum flashing
radiating from the center of the array. Arms were set at 120° angles, and flash-
ing was buried to 3 cm to prevent mammals from burrowing under the fences
(Handley and Varn 1994, Kirkland and Sheppard 1994). Nine pitfall traps were
set in association with each array, such that three were placed in the middle, and
two at the ends of each of the three arms. The five arrays were set in a line
approximately parallel to, and 50 m from, the previously established transects at
these plots. Individual arrays were spaced 25 m apart, so that the length of the
array series was equal to the length of the transects (100 m).

We operated the two types of transects at 12 plots from 9 to 23 July
1994 for a total of 3,360 trapnights (TN) per method. We operated all three
methods at four plots from 4 to 1 1 August, and again from 1 8 November to 2
December 1995. Trapping effort was equal at the two types of transects (2,240
TN), but greater at the arrays (4,040 TN). Because pitfalls associated with an
array are interdependent, it is not meaningful to compare sampling effort
between transects and arrays. Thus, we used methods of analysis that were not
influenced by differences in sampling effort. All specimens were identified to
species and accessioned into the collections of the University of Georgia Muse-
um of Natural History.

The distributions of capture frequencies using each survey method were
compared using likelihood-ratio tests of independence (Agresti 1990). Rejection
of the null hypothesis of independence indicated that the methods produced dif-
ferent distributions of capture frequencies, and thus different perceptions of

Survey of Shrews 113

shrew assemblage structure. We also partitioned the data table involving all
three methods into several independent, four-fold (2-by-2) tables (Lancaster
1949) to better determine patterns of dependence. For example, capture rates of
Sorex spp. (both species combined) and Blarina brevicauda were compared
between transects (both types combined) and arrays. For each four-fold table,
we calculated the corresponding odds ratio and tested the hypothesis that the
odds ratio was equal to unity (Agresti 1990).

MICROHABITAT ANALYSES

In July 1994, we measured several microhabitat variables surrounding
each of the 240 pitfall traps of the linear-transects. Because these traps were
placed without regard to microhabitat conditions, surveys provided an unbiased
sample of conditions at the forest floor and could be compared to capture fre-
quencies of each species at those locations. Only traps associated with linear
transects were considered in this analysis.

At each trap station, we established a circular plot with a 2.5-m radius.
Within each plot we measured the diameter and length of all coarse woody debris
greater than 4 cm in diameter. We also measured the greatest length and width
of all rocky debris, and the diameter at the forest floor of all stumps within each
plot. These measurements yielded an index to the abundance of fallen logs,
rocks, and stumps surrounding each pitfall trap. We also measured the distance
from the pitfall trap to the nearest fallen log, rock, or stump.

Microhabitat measurements were compared to shrew capture frequen-
cies using Pearson product-moment correlations. We regressed capture frequen-
cy of each species against distance to nearest structure (Neter et al. 1990).

RESULTS
METHOD COMPARISON

In 3,360 trapnights (TN) at selective transects in 1994 we collected 358
individuals representing four species (Table 1). In 3,360 TN at linear transects
we collected 126 individuals from the same four species. Sorex cinereus was the
most commonly captured shrew, followed by S. fumeus, Blarina brevicauda, and
S. hoyi. Sorex hoyi was uncommon at our sites and, therefore, was omitted from
all statistical analyses. We captured 2.8 times as many individuals in selective as
in linear traps, and this ratio was relatively constant among species (Table 1).
Consequently, the distribution of shrew captures (relative abundance of each
species) did not differ between these two methods (G2 = 0.722; P = 0.697; df =
2).

114

Timothy S. McCay, Joshua Laerm, M. Alex Menzel
and William M. Ford

Table 1 . Number of captures of four species of shrews using two pitfall transect
designs, selective and linear, at the Coweeta Hydrologic Laboratory, July 1994.
Traps in selective transects were positioned 5-m apart next to structures such as
down logs and rocks. Traps in linear transects were set 5-m apart in a straight
line.

Transect Type

Selective

Linear

Selective:Linear

Species

(TN = 3,360)

(TN = 3,360)

Total

Ratio

Blarina brevicauda

25

8

3

3.1:1

Sorex cinereus

208

78

286

2.7:1

Sorex fumeus

116

36

152

3.2:1

Sorex hoyi

9

4

13

2.3:1

Total

358

126

484

2.8:1

In 2,240 TN at selective transects in 1995 we captured 124 individuals
of the same four species captured in 1994 (Table 2), whereas linear transects
yielded 52 individuals. Similar to the 1994 data, we captured 2.4 times as many
individuals in selective as in linear transects; however, there was greater varia-
tion in this ratio among species than in 1994. In 4,040 TN at drift-fence arrays
we captured 81 individuals of the same 4 species. Capture frequencies observed
at drift-fence arrays differed from both types of transects, and ratios involving
drift-fence arrays varied markedly (Table 2). Consequently, the distribution of
shrew captures among sampling methods varied in 1995 (G2 = 17.849; P = 0.001;
df = 4).

We were able to construct four independent, four-fold tables using the
data collected in 1995. Two of the tables compared captures of S. cinereus and
S. fumeus between the two types of transects (G2 = 0.021 ; P = 0.884; df = 1) and
between transects (both combined) and arrays (G2 - 3.049; P = 0.081; df = 1). In
neither case did the data support dependence; therefore, capture frequency of
these species was not markedly affected by trapping method.

The remaining four-fold tables compared Sorex spp. to B. brevicauda
with respect to trap arrangement. The first of these, a table comparing Sorex spp.
and B. brevicauda captures by transect type, indicated that capture frequencies
for these two taxa differed between the two methods (G2 = 6.061; P - 0.014; df
= 1). The odds ratio for this table was greater than unity (9 = 3.17; Z = 2.409;
df = 163), indicating that Sorex spp. were more likely than B. brevicauda to be
captured using selective transects. The second of these tables compared Sorex

Survey Shrews

115

o 3

"o o

<o >

ex, o

1/3 Z

s -a

**-< <->

O c/5

S3 3>

fa 3

3 <

c t «
3 < 05

<L>

.2: 2? o

& t3 «3

— < qc

00

> 13 o
— 3 otf

Z .y

13 <H

4> CN

o

> <N
O ™

£ II

do ^ do P "^i
d ^ c4 £ o»

-* OO _ ^4- t-

© O en *-H j£

a
| &

H H H — i

si si £ a

- o o o o
OQ Co co co H

-ft .C

116 Timothy S. McCay, Joshua Laerm, M. Alex Menzel

and William M. Ford

spp. and B. brevicauda captures between transects and arrays. Likewise, this
table supported dependence (G2 = 8.718; P = 0.003; df = 1). The odds ratio for
this table was less than unity (0 = 0.168; Z = -2.375; df = 242), indicating that
Sorex spp. were more likely than B. brevicauda to be captured using arrays.

MICROHABITAT USE

None of the shrews examined showed strong relationships with the
abundance of rocks, logs, and stumps within 2.5 m of the trap stations. There is
some evidence that Sorex fumeus selectively used habitat structure, as the cap-
ture success of this species was significantly correlated with the abundance of
rocks (r2 = 0.017; P = 0.050) and logs (r2 = 0.018; P = 0.047). Sorex cinereus
was correlated only with the abundance of rocks (r2 = 0.026; P = 0.016). Final-
ly, B. brevicauda was not correlated with any of the habitat features examined.
It should be noted that the correlations presented above are, while statistically
significant, exceedingly weak. For example, the abundance of rocks accounts for
only 1.7% of the variability in S. fumeus capture.

In agreement with our microhabitat correlations, capture success of S.
fumeus showed a highly significant, although weak, relationship with proximity
to structure (r2 = 0.034; P = 0.006; df = 218). Sorex cinereus capture success was
not significantly related to proximity to structure (r = 0.009; P = 0.158; df =
218), nor was the capture success of B. brevicauda (r2 = 0.002; P = 0.481; df =
218).

Thus, S. fumeus showed the strongest relationship with habitat structure
and the greatest positive differences between selective transects and linear tran-
sects in 1994 (220%) and 1995 (180%; Tables 1 and 2). Captures of S. cinereus
were less strongly related to habitat structure and showed smaller, positive dif-
ferences between selective and linear transects in 1994 (170%) and 1995 (160%).
Blarina brevicauda was not correlated with the abundance of any structural habi-
tat features or proximity to structure and was the only species to exhibit a nega-
tive difference between selective and linear transects (-20% in 1995), reflecting
a higher capture success at traps placed without regard to microhabitat features
than those traps placed selectively.

DISCUSSION

The relative capture frequencies of Sorex fumeus and S. cinereus, when
considered with respect to each other, were not significantly affected by trap
arrangement. This suggests that any of the three methods considered would pro-
vide a similar depiction of the relative abundance of these species in similar habi-
tats. Furthermore, the capture rates of these species using transects were similar
over a two-year period. Thus, our data for S. fumeus and S. cinereus suggest that
in comparisons over time, selective and linear transects provide estimates of rel-

Survey Shrews 117

ative abundance that are similar to each other and reasonably stable over two
years.

Comparisons involving Blarina brevicauda must be considered with
some caution due to low sample sizes. However, our study provides some evi-
dence that B. brevicauda was less likely than Sorex spp. to be captured with trap
arrangements utilizing natural or artificial structures to direct movement. On
encountering a linear structure, B. brevicauda may not have followed the struc-
ture lengthwise, which was necessary for capture. It is also noteworthy that B.
brevicauda is largely fossorial (George et al. 1986) and may not spend as much
time moving across the forest floor and encountering drift-fences or natural habi-
tat structures.

Largely because of B. brevicauda we found that drift-fence arrays pro-
vided a different depiction of the southern Appalachian shrew community than
did either of two types of transects. Mitchell et al. (1993) and Dowler et al.
(1985) also found differences in species richness and numbers of shrews collect-
ed using pitfalls set singly and in conjunction with drift-fence arrays. In 1,750
TN at each trap arrangement, Dowler et al. (1985) captured 47 S. cinereus at
arrays compared to 29 in isolated pitfalls, but they only captured 2 B. brevicau-
da at arrays compared to 3 at isolated pitfalls. Again, inferences are tenuous due
to small capture frequencies, and further studies into the movement patterns and
behavior of Blarina are recommended.

Overall capture success with transects was 7.2% in 1994 and 3.9% in
1995. This disparity provided evidence that numbers of shrews may have
decreased between the two trapping periods, perhaps due to the removal of ani-
mals during 1994. Thus, for the purposes of these analyses we have had to make
the assumption that this change in overall shrew abundance did not affect pat-
terns of shrew microhabitat use.

Among the 3 shrews we studied (Sorex hoyi excluded), S. fumeus and S.
cinereus exhibited weak, but significant, relationships with structures on the for-
est floor, whereas B. brevicauda did not. We are aware of no previous studies of
microhabitat use by S. fumeus. The observations of MacCracken et al. (1985) in
southeastern Montana support the importance of litter cover (dead plant parts) to
S. cinereus; however, they did not separate downed logs from other types of
debris. In contrast, Yahner (1986) found that the mean length and density of logs
were lower at trap stations where S. cinereus was captured than where this
species was not captured, and Getz (1961) concluded that microhabitat features
have little or no effect on distributions of S. cinereus, emphasizing the impor-
tance of moisture. Our results suggest that selective use of microhabitat features
by S. cinereus may be so weak as to require a very large sampling effort to detect.

Our results agree with Getz (1961) and Yahner (1982) who found no
evidence for microhabitat selection in B. brevicauda. Conversely, Seagle (1985)

118 Timothy S. McCay, Joshua Laerm, M. Alex Menzel

and William M. Ford

found that B. brevicauda seemed to avoid areas with few fallen logs in decidu-
ous forests in Tennessee.

Our perception of the relative abundance of three shrew species was
partially a function of the trap arrangement we used to capture them. Each sam-
pling method takes advantage of certain patterns of microhabitat use, which vary
among species. We suggest that caution be used when comparing the results of
surveys using different trap arrangements, as well as different traps.

ACKNOWLEDGMENTS - We thank D. Krishon and J. Boone for help in the field
and two anonymous reviewers for comments and suggestions that improved the
manuscript. This project was funded by the Museum of Natural History at the
University of Georgia and NSF Grant BSR 901 1661.

LITERATURE CITED

Agresti, A. 1990. Categorical data analysis. John Wiley & Sons, New York,
New York.

Brillhart, D. B., and D. W. Kaufman. 1991. Influence of illumination and sur-
face structure on space use by prairie deer mice (Peromyscus maniculatus
bairdii). Journal of Mammalogy 72:764-768.

Bury, R. B., and P. S. Corn. 1987. Evaluation of pitfall trapping in northwest-
ern forests: trap arrays with drift-fences. The Journal of Wildlife Manage-
ment 51:112-119.

Dowler, R. C, H. M. Katz, and A. H. Katz. 1985. Comparison of live trapping
methods for surveying small mammal populations. Northeastern Environ-
mental Science 4:165-171.

George, S. B., J. R. Choate, and H. H. Genoways. 1986. Blarina brevicauda.
Mammalian Species 261:1-9.

Getz, L. L. 1961. Factors affecting the local distribution of shrews. American
Midland Naturalist 65:67-88.

Handley, C. O., Jr., and M. Varn. 1994. The trapline concept applied to pitfall
arrays. Pages 285-287 in Advances in the biology of shrews (J. F. Merritt,
G. L. Kirkland, Jr., and R. K. Rose, editors). Carnegie Museum of Natural
History, Special Publication No. 18, Pittsburgh, Pennsylvania.

Kalko, E. K. V, and C. O. Handley, Jr. 1993. Comparative studies of small
mammal populations with transects of snap traps and pitfall arrays in south-
west Virginia. Virginia Journal of Science 44:3-18.

Kirkland, G. L., Jr., and P. K. Sheppard. 1994. Proposed standard protocol for
sampling small mammal communities. Pages 277-283 in Advances in the
biology of shrews (J. F. Merritt, G. L. Kirkland, Jr., and R. K. Rose, editors).
Carnegie Museum of Natural History, Special Publication No. 18, Pitts-
burgh, Pennsylvania.

Survey Shrews 119

Lancaster, H. O. 1949. The derivation and partition of x2 in certain discrete dis-
tributions. Biometrika 36: 117-1 29.

MacCracken, J. G., D. W. Uresk, and R. M. Hansen. 1985. Habitat used by
shrews in southeastern Montana. Northwest Science 59:24-27.

Mitchell, J. C, S. Y. Erdle, and J. F. Pagels. 1993. Evaluation of capture tech-
niques for amphibian, reptile, and small mammal communities in saturated
forested wetlands. Wetlands 13:130-136.

Neter, J., W. Wasserman, and M. H. Kutner. 1990. Applied linear statistical
models. Third edition. Richard D. Irwin, Inc., Homewood, Illinois.

Otis, D. L., K. P. Burnham, G. C. White, and D. R. Anderson. 1978. Statistical
inference for capture data from closed populations. Wildlife Monograph 62.

Seagle, S. W. 1985. Patterns of small mammal microhabitat utilization in cedar
glade and deciduous forest habitats. Journal of Mammalogy 66:22-35.

Swank, W. T., and D. A. Crossley, Jr. 1988. Introduction and site description.
Pages 3-16 in Forest hydrology and ecology at Coweeta (W. T Swank, and
D. A. Crossley, Jr., editors). Ecological Studies, volume 66, Springer- Ver-
lag, New York, New York.

Wharton, C. H. 1977. The natural environments of Georgia. Georgia Depart-
ment of Natural Resources, Atlanta, Georgia.

Williams, D. F., and S. E. Braun. 1983. Comparison of pitfall and convention-
al traps for sampling small mammal populations. The Journal of Wildlife
Management 47:841-845.

Yahner, R. H. 1982. Microhabitat use by small mammals in farmstead shelter-
belts. Journal of Mammalogy 63:440-445.

Yahner, R. H. 1986. Microhabitat use by small mammals in even-aged forest
stands. Journal of Mammalogy 115:174-180.

Received 29 August 1996
Accepted 4 November 1996

The Water Shrew, Sorex palustris Richardson
(Insectivora:Soricidae), and Its Habitat in Virginia

John F. Pagels, Leonard A. Smock, and Stephen H. Sklarew

Department of Biology, Virginia Commonwealth University

Richmond, Virginia 23284-2012

ABSTRACT - The water shrew, Sorex palustris, known from a single
Virginia locality in Bath County as recently as 1991, has been found at
four additional sites, all in Highland County. The five sites were above
900 m (mean = 1,000 m; range = 902-1,128 m) elevation. All sites were
fully-canopied first or second order streams with habitat characteristics
and a macroinvertebrate community typical of relatively pristine, high-
altitude, headwater streams. All streams were high gradient (7-14%
slope) and had a variety of flow and depth regimes and a predominate-
ly well-sorted cobble substrate with abundant woody debris. Channel
banks were fully vegetated and had extensive undercut areas and many
crevices. Riparian canopy trees at all sites were primarily northern
hardwoods, including yellow birch (Betula lutea) and sugar maple
{Acer saccharum). The macroinvertebrate community of the streams
was dominated by stoneflies (Plecoptera), mayflies (Ephemeroptera),
and midges (Diptera).

Although the water shrew, Sorex palustris Richardson, enjoys a broad
distribution, including much of Canada, southwestern Alaska, and northern and
high elevation regions of the United States (Hall 1981, Beneski and Stinson
1987), the Appalachian water shrew, S. p. punctulatus Hooper, is rare and found
only in highly-localized boreal environments in the southern Appalachian Moun-
tains. In an overview of distribution and diversity of Virginia mammals, Hand-
ley (1992:165) described S. palustris as a "high/medium boreal species" that
occurs as a relict and is in danger of extirpation in Virginia. Laerm et al. (1995)
summarized records of S. p. punctulatus, showing its known distribution as only
14 counties in a seven state area extending from southwestern Pennsylvania to
northern Georgia.

The water shrew was first collected in Virginia in northwestern Bath
County in 1974 (Pagels and Tate 1976) at the then proposed site of the upper
reservoir for a pumped-storage electrical generating facility. Although feared
lost from this site when the valley was flooded, the shrew was found in an undis-

120

Water Shrew 1 2 1

turbed area just above the reservoir (Pagels 1987). Pagels and Handley
(1991:564) subsequently observed, "The water shrew, extremely rare in Virginia,
is known to occur in only one small watershed in the state." The Appalachian
water shrew was declared a state endangered species in 1990, and a recovery
plan was prepared (Pagels et al. 1991). The primary objective of that plan is to
prevent extirpation of the water shrew in Virginia, with critical aspects being
determination of its distribution, description of its habitat, and investigation of
factors that might adversely impact the species.

Herein we report on records of the water shrew and examine biotic and
abiotic features at all sites where it has been collected in Virginia. Our objective
is to provide basic information on the water shrew that will be of interest to biol-
ogists and resource managers who might be concerned with the distribution,
ecology, and protection of this species.

MATERIALS AND METHODS
SMALL MAMMAL SAMPLING

Efforts to locate the water shrew in Virginia were intensified in the late
1980s in conjunction with development of the recovery plan. Based on pub-
lished information and knowledge obtained on S. palustris habitat during visits
to sites where it had been collected in Maryland and West Virginia, a profile of
apparently suitable habitat was developed for use in this study. Subsequently,
numerous sites were sampled in northern hardwood or northern hardwood-
conifer forests along mountain streams at or above about 900-m elevation in Vir-
ginia. Emphasis was placed on sampling in Bath, Highland, and Rockingham
counties in western Virginia, and Grayson, Smyth, and Washington counties in
southwestern Virginia. In addition to efforts by several individuals over many
years, portions of approximately 45 first and second order streams were sampled
for S. palustris by Pagels in the period 1989-1996.

Traps used included Museum Special snap-back traps and Sherman live
traps, but most sampling in 1991 to present was completed with use of 2-L plas-
tic pitfall traps. Traps were set as near the water as possible in "most likely"
spots, for example, under overhanging banks or root masses of trees, but often a
trap was placed in a given spot only because the substrate allowed placement of
a pitfall. Number of traps, space between traps, and length of sampling period
varied greatly among sites, but whenever trapping was completed in anticipation
of perturbative activities, i.e. timbering, pitfalls were usually kept in place for a
minimum of 30 days. Most sampling was between April to November as dictat-
ed by winter weather and road conditions. All specimens are deposited in the
Virginia Commonwealth University Mammal Collection.

122

John F. Pagels, Leonard A. Smock and Stephen H. Sklarew

Table 1. Habitat parameters measured at streams where S. palustris was found
in Virginia, 1974-1993. General methodology follows that of Platts et al. (1983).

Parameter

Method

Stream order
Elevation
Channel gradient
Channel width
Wetted width
Water depth

Water velocity

Substrate type

Riffle-pool prevalence

Bank undercutting

Bank water depth
Wood abundance

Debris dams

PH

Conductivity
Riparian vegetation

method of Strahler (1964)

U.S.G.S. 1:24,000 topographic maps

altimeter

bank-to-bank distance

width of the flowing stream

determined from measurements at points

along each transect across a stream

determined at points along each transect

across a stream with a Marsh-McBirney

impulse flow meter at six-tenths water

depth

visual estimate of the proportion of each of

six sediment particle sizes in the channel

(bedrock; boulder = >256 mm; cobble = 64-

256 mm; pebble = 16-64 mm; gravel = 2-16

mm; sand = 0.06-2 mm.

proportion of the stream channel in riffles,

pools and glides

distance that the bank overhangs the

channel

depth of water at the bank

volume of wood in the channel, calculated

using the method of Wallace and

Benke(1984)

number of large accumulations of woody

debris per 100 m of stream

field pH meter

conductivity meter

percentage composition of canopy trees

within 8 m of both banks

HABITAT CHARACTERIZATION

Two approaches were used to describe the in-stream and riparian habi-
tat at each stream where S. palustris was found. The first approach described 1 6
general geomorphic and physico-chemical characteristics deemed as potentially
significant habitat variables for the shrews (Table 1). Habitat characteristics

Water Shrew

123

were quantified over a 100-m stretch following the general methodology of
Platts et al. (1983). Values for in-channel parameters are reported as the mean of
measurements made at 5- or 10-m intervals along the entire 100-m stretch. Com-
position of the riparian vegetation was made from surveys of canopy trees along
both banks. All habitat analyses were conducted in a three-day period in June
1995 when streams were at base flow.

The second approach to characterizing stream and riparian habitat was
a formal assessment using the U.S. Environmental Protection Agency's (EPA)
Rapid Habitat Assessment Protocol (Plafkin et al. 1989). Twelve metrics were
used to score habitat condition at each stream (Table 2). All metrics were scored
on a 20-point scale, 20 points indicting the best or preferred condition. A portion
of the riparian zone and near-channel watershed of one of the sites at which S.
palustris was collected was timbered between the time that the shrews were col-
lected and the habitat analysis was conducted. Habitat data from this site were
not included in our summary.

Table 2. Parameters measured for the habitat assessment of streams where S.
palustris was found in Virgina. Methodology follows that of Plafkin et al. (1989).

Parameter

General Description

In-stream cover
Epifaunal substrate

Embeddedness

Velocity /depth ranges

Channel alteration
Sediment deposition
Frequency of riffles
Channel flow status

Condition of banks
Bank vegetative cover

Disturbance pressure

Riparian vegetation

abundance of submerged logs, undercut

banks, and other forms of stable habitat

abundance of the "most productive

benthic habitat," typically riffle areas

and/or submerged snags

degree to which the primary substrate

was surrounded by fine sediment

variety of water velocity and depth

regimes in the stream

evidence of stream channelization

evidence of recent sediment deposition

prevalence and size of riffles

percentage of the channel bed that was

wetted

evidence of bank stability versus erosion

percentage of bank that was covered with

vegetation

degree of disruption of riparian vegetation

by grazing or other processes

width of riparian zone that was vegetated

and with minimal human disturbance

124

John F. Pagels, Leonard A. Smock and Stephen H. Sklarew

Fig. 1. General location of sites where S. palustris was found in Virginia, 1974-
1993. B and H indicate Bath and Highland counties. Insert map (after Webster
1987:38) shows the range of the species in North America; its range in the south-
ern Appalachian Mountains is solid dark.

Table 3. Virginia Commonwealth University catalog number, sex, date of capture, and
general location of captures of 5. palustris in Virginia, 1974-1993. Bath County sites
(B) include those of Pagels and Tate (1976) and Pagels (1987). Highland County sites
are comprised of the private property location (HP) and three sites, HI, H2, and H3,
in the Laurel Fork area of George Washington and Jefferson National Forests.

Catalog Number

sex

Date

Location

557

F

3 August 1974

B

558

F

5 October, 1974

B

559

F

4 October 1974

B

560

F

4 October 1974

B

4715

F

29 August 1986

B

4742

F

5 August 1986

B

10025

F

2 July 1992

HP

10491

F

11 September 1992

HI

10492

F

11 September 1992

HI

10940

F

15 June 1993

H2

11261

M

16 July 1993

H3

11262

F

16 July 1993

H3

11263

F

16 July 1993

H3

Water Shrew 125

AQUATIC MACROINVERTEBRATES

Aquatic macroinvertebrate communities were sampled in June 1995 at
each stream to provide information on the species composition and relative abun-
dance of the shrew's primary food source. A D-frame dip net was used to collect
organisms along a 100-m stretch. All primary habitats were sampled, including
riffles, pools, bank areas, and woody debris; the material collected was compos-
ited into one sample per stream.

Invertebrates were removed from the samples in the laboratory under a
stereo-microscope after addition of rose bengal to facilitate the sorting process.
A minimum of 200 organisms was randomly picked from the samples; picking
ending when no new taxa were observed. Organisms were identified with the
taxonomic keys in Merritt and Cummins (1996) and Pennak (1989).

RESULTS
DISTRIBUTION

Sorex palustris was collected along four streams of the Potomac River
drainage in Highland County. Along with the original Little Back Creek site
(Pagels and Tate 1976, Pagels 1987) in the James River drainage in Bath Coun-
ty, it is now known from five Virginia localities, all on Allegheny Mountain.
Three of the new sites are in the George Washington and Jefferson National
Forests in the Laurel Fork area of northwestern Highland County. The other site
is on private property just west of Hightown, Virginia, and is the site since tim-
bered by the land owner. The continued existence of the water shrew at this site
after timbering has not been confirmed. General locations of the sites where S.
palustris has been captured in Virginia are given in Figure 1 .

Four other species of shrews, the short-tailed shrew {Blarina brevicau-
da), masked shrew (Sorex cinereus), smoky shrew (S. fumeus), and rock shrew
(S. dispar), were captured on traplines with S. palustris at all Highland County
sites except the private property site where no S. dispar was taken. Sorex fumeus
represented 66.5%, S. cinereus 17.9%, S. dispar 6.9%, B. brevicauda 4.6%, and
S. palustris 4.0% of the 173 shrews taken at the four Highland County sites
where S. palustris was captured. A summary of all S. palustris captured in Vir-
ginia, including records in Pagels and Tate (1976), and Pagels (1987), is given in
Table 3.

HABITAT

Habitat characteristics of all streams were typical of relatively pristine
headwater streams of the Virginia mountains (Table 4). Streamwater was cool,
had a circum-neutral pH, and low to moderate conductivity. All streams were
either first or second order and hence had a narrow channel and wetted width.
Channel gradients were a relatively steep 7-14%. A variety of flow regimes
Table 4. Means, standard errors, and ranges of geomorphic, hydrologic, and

126 John F. Pagels, Leonard A. Smock and Stephen H. Sklarew

physico-chemical characteristics of the streams where S. palustris was found in
Virginia, 1974-1993.

Parameter

Mean

SE

Range

Stream order

1

<0.1

1-2

Elevation (m)

1,000

37

902-1,128

Channel gradient (%)

10

1.4

7-14

Channel width (m)

4

0.7

2-10

Wetted width (m)

3

0.4

1-7

Water depth (cm)

8

1.3

1-1,510

Water velocity (m3/s)

0.15

0.01

0-0.80

Bank undercut (cm)

18

0.5

0-99

Wood volume (m3/m2)

0.53

0.1

0.37-0.66

Debris dams (#/100m)

3

0.9

2-6

pH

7.0

0.1

6.7-7.2

Conductivity (uS/cm)

24

9

12-52

existed in the streams, as reflected by the ranges in water velocity and depths
(Fig. 2). Also, although riffles were the predominant habitat (Fig. 3), a distinct
riffle-pool geomorphology occurred in all of the streams.

In-stream cover was abundant in all streams. Cobble was the predomi-
nant substrate, with boulder, pebble and gravel also being relatively common
(Fig. 3). The sediment was well sorted, with little embeddedness by fine-grained
particles (Table 5). Woody debris and resulting debris dams were common in the
channels of all streams, and growth of moss in the channel was evident at most
sites.

The riparian areas of the streams were undisturbed, vegetated, and sta-
ble (Table 5). All channel banks had extensive undercut areas (Fig. 4) and areas
that were rocky with many crevices and downed wood. Most of the undercuts
were not in contact with the flowing stream but rather with dry channel sediment.
One site had a broad floodplain and many side channels where water flowed
freely through interstitial areas in the rocky floodplain floor.

Yellow birch {Betula luted) was the most abundant canopy tree along
the riparian areas of all streams (Table 6). Other common canopy species along
the streams were black birch (B. lenta), sugar maple (Acer saccharum), bass-
wood (Tilea americana), black cherry (Prunus serotina), red maple (A. rubrum),
American beech (Fagus grandifolia), and Eastern hemlock (Tsuga canadensis).
The understory was generally undeveloped, typical of mature forests.

Water Shrew

127

Fig. 2. Frequency distribution (mean ± SE) of water velocity and depth in four
streams at which S. palustris was found in Virginia.

60

^ 40

>-
O

z

LU

O
g 20

Li.

a. Water Velocity

zEz=l

K

Z.

0.1 0.2 0.3 0.4 0.5
WATER VELOCITY (cm/s)

>0.5

60

- 40

>-
O
2
LU

O
£ 20

b. Water depth

YM

vVa r^^ 'v%n

0-5 6-10 11-15 16-20 21-25 >25
WATER DEPTH (cm)

128

John F. Pagels, Leonard A. Smock and Stephen H. Sklarew

Fig. 3. Frequency distribution (mean ± SE) of riffle, pool, and glide areas and
sediment particle sizes in four streams at which S. palustris was found in Vir-
ginia. BED = bedrock, BLD = boulder, COB = cobble, PEB = pebble, GRV =
gravel, SND = sand.

100

C^ 75

50

O

z

LU

Z)

o

LU

8: 25

a. Geomorphology

VA V^F/A

RIFFLE POOL GLIDE

50

40

>
o

30

2

LU

3

O

20

LU

tt

10

b. Sediment Particle Size

v-fcn.

BED BLD COB PEB GRV SND
SEDIMENT TYPE

Water Shrew

129

Table 5. Means, habitat assessment metric scores (after Plafkin et al. 1989), stan-
dard errors and ranges for five streams in Virginia where S. palustris was found,
1974-1993. Values range from 0-20, with 20 indicating the highest quality.

Parameter

Mean

SE

Range

In-stream cover 20

Epifaunal substrate 20

Embeddedness 1 8

Velocity /depth ranges 15

Channel alteration 20

Sediment deposition 18

Frequency of riffles 20

Channel flow status 16

Bank condition 19
Bank vegetative protection 19

Disturbance pressure 20

Riparian vegetation 19

0.0
0.0
0.2
0.2
0.0
1.2
0.0
1.0
0.2
0.2
0.0
1.0

20-20
20-20
18-19
15-16
20-20
15-20
20-20
13-18
19-20
18-19
20-20
16-20

Fig. 4. Frequency distribution (mean ± SE) of the extent of bank undercutting in
four streams at which S. palustris was found in Virginia.

0-10 11-20 21-30 31-40 41-50 >50
BANK UNDERCUT DEPTH (cm)

130

John F. Pagels, Leonard A. Smock and Stephen H. Sklarew

Table 6. Mean percent abundance, standard error, and range of riparian canopy
trees at the streams where S. palustris was found in Virginia, 1974-1993.

Species

Mean

SE

Range

Betula lutea

29

3

22-35

Betula lenta

13

4

0-24

Acer saccharum

10

2

4-17

Tilia americana

9

4

0-22

Prunus serotina

7

3

0-14

Acer rubrum

6

4

0-18

Fagus grandifolia

6

3

0-14

Tsuga canadensis

6

2

0-9

Fraxinus sp.

3

2

0-8

Quercus rubra

3

1

0-7

Robinia pseudo-acacia

2

1

0-6

Carya sp.

<1

<1

0-2

Picea rubens

<1

<1

0-2

Standing dead trees

6

2

0-9

MACROINVERTEBRATE COMMUNITY

Taxonomic composition and richness were very similar among all
streams (Table 7). Taxonomic richness ranged from 26 to 28 taxa, but would
have been considerably higher had individuals in the dipteran family Chirono-
midae (midges) been identified to genus. The macroinvertebrate communities of
all of the streams were dominated by species of midges (Diptera), stoneflies (Ple-
coptera) and mayflies (Ephemeroptera). Non-midge taxa common at most
streams included stoneflies in the families Leuctridae and Perlodidae; the mayfly
families Heptageniidae, Ephemerellidae, and Baetidae; and the caddisfly fami-
lies Philopotamidae and Hydropsychidae. Although biomass of macroinverte-
brates was not measured, these taxa no doubt dominated the macroinvertebrate
biomass because of their generally large size.

DISCUSSION
In Virginia, Pleistocene remains of S. palustris are known from Natural
Chimneys in Augusta County, elevation 414 m (Guilday 1962), and Clarks Cave
in Bath County, elevation 456 m (Guilday et al. 1977). Indicative of the boreal
nature of the sites where S. palustris now occurs, certain other boreal species
with highly disjunct populations in the southern Appalachians remain associates
of S. palustris in Virginia, but also no longer occur in the environs of the Natural

Water Shrew 131

Table 7. Taxonomic richness and mean percent contribution (1 SE and range) of
macroinvertebrate taxa occuring in four streams at which S. palustris was found
in Virgina, 1974-1993.

Taxon

Number of
Genera

Mean Percent
Abundance

SE

Range (%)

Hydracarina

1

<1

<1

0-1

Decapoda

1

1

<1

0-2

Ephemeroptera

6

26

10

6-54

Odonata

1

1

<1

1-2

Plecoptera

6

26

2

20-31

Trichoptera

8

22

4

14-31

Coleoptera

3

3

2

1-8

Diptera

8

22

4

11-32

Pelecypoda

1

<1

<1

0-1

Chimneys or Clarks Cave sites. These include the rock vole {Microtus chrotor-
rhinus), in the Little Back Creek area of Bath County (Pagels 1990), and the
northern flying squirrel (Glaucomys sabrinus), at the Highland County sites
(Pagels et al. 1990). Handley (1992:159) suggested that these species, along
with a few others, probably "...represent the last stages of the recoil of high bore-
al species from southern latitudes into higher latitudes in the United States and
Canada— north and west of Virginia."

Sorex palustris was found only at sites where cool, mesic conditions
occurred along with considerable cover on the banks of swift-flowing streams.
Beneski and Stinson (1987) observed that although it is found in a variety of
habitats, rocky crevices, logs, and abundant overhanging areas along stream
banks are typical of S. palustris habitat throughout much of its range, and likely
are critical in warmer, southern areas to provide cool, mesic microhabitats for the
shrews. Also, the full canopy of mature forests, as occurred at the sites where
the shrews were found, probably is essential for maintaining the cool conditions.

Habitat conditions also are important in affecting the food resources
available to the shrews. Both terrestrial and aquatic invertebrates are consumed
by water shrews (Beneski and Stinson 1987). Wrigley et al. (1969) suggested
that a mesic microhabitat along stream banks was important for supporting ter-
restrial invertebrates that at times can compose a significant portion of the diets
of the shrews (Hamilton 1930, Whitaker and Schmeltz 1973). The primary
aquatic organisms consumed by shrews, including stoneflies, mayflies, and cad-
disflies (Conoway 1952, Conoway and Pfitzer 1952, Sorenson 1962, Linzey and

132 John F. Pagels, Leonard A. Smock and Stephen H. Sklarew

Linzey 1973) are most abundant in streams with fast current and cobble sub-
strate. Those conditions were prevalent at the five sites at which S. palustris was
found and these aquatic insect orders were common.

All streams where shrews were found had habitat characteristics indica-
tive of relatively pristine conditions. The Envrionmental Protection Agency
habitat assessment metric scores (Table 4) show the high quality of in-stream,
bank, and riparian habitat at all sites. All metrics were scored at 15 or higher,
and many scores were at or near the maximum score of 20. Total metric scores
of 220-229 points where shrews were found reflect high habitat quality at the
streams. The taxonomic composition of macroinvertebrates also is indicative of
relatively undisturbed streams. Most taxa we collected are generally intolerant
of low habitat or water quality.

We cannot state assuredly that S. palustris does not occur along streams
that we sampled unsuccessfully, or on streams that possess the habitat conditions
reported herein. Laerm et al. (1995:49) observed that despite the great increase
in knowledge of shrew species that were formerly thought to be very rare, includ-
ing both Sorex hoyi and S. dispar that are now known to be more common than
was earlier thought, "...the water shrew appears to be the rarest and most local-
ized shrew in the southeastern United States." New records such as those report-
ed herein, and a low altitude site (808 m) in northern Georgia (Laerm et al. 1995),
provide hope that S. palustris will be found at additional locations. Baseline data
on suitable habitat will make searching for new sites more efficient, and will aid
in the development of management programs for protection of known and poten-
tial sites.

ACKNOWLEDGEMENTS. - Individuals who assisted in field work included J.
Baker, J. Blevins, T. Blevins, B. Glasgow, D. Kirk, S. Klinger, J. McGovern, J.
E. Pagels, J. Pound, R. Reynolds, S. Rinehart, C. Thomas, and S. Whitcomb. K.
Terwilliger (formerly of the Virginia Department of Game and Inland Fisheries),
M. Fies and R. Reynolds of the Virginia Department of Game and Inland Fish-
eries, R. Glasgow (formerly of the U.S. Forest Service, George Washington and
Jefferson National Forests) and J. Bellemore, S. Croy, and S. Klinger of the U.
S. Forest Service were supportive in numerous aspects of the study. E. Thomson
and D. Feller, Maryland Natural Heritage Program, and C. Stihler, West Virginia
Department of Natural Resources, graciously led visits to sites in their states, and
D. Webster provided unpublished observations on habitat in North Carolina. C.
Handley, Jr. and the late J. Guilday were inspirational leaders. Throughout most
of the study funding to JFP was received from the Nongame and Endangered
Species Program of the Virginia Department of Game and Inland Fisheries, and
more recently also from matching funds from the U. S. Forest Service, George
Washington and Jefferson National Forests.

Water Shrew 1 3 3

LITERATURE CITED

Beneski, J. T., Jr., and D. W. Stinson. 1987. Sorexpalustris. Mammalian Species
296:1-6.

Conoway, C. H. 1952. Life history of the water shrew (Sorex palustris naviga-
tor). American Midland Naturalist 48:219-248.

Conoway, C. H., and D. W. Pfitzer. 1952. Sorex palustris and Sorex dispar from
The Great Smoky Mountains National Park. Journal of Mammalogy
33:106-108.

Guilday, J. E. 1962. The Pleistocene local fauna of the Natural Chimneys,
Augusta County, Virginia. Annals of Carnegie Museum 36:87-122.

Guilday, J. E, P. W. Parmalee, and H. W. Hamilton. 1977. The Clark's Cave
bone deposit and the Late Pleistocen paleoecology of the central Appalachi-
an Mountains of Virginia. Bulletin of Carnegie Museum of Natural History
2:1-86.

Hall, E. R. 1981. The mammals of North America. Second edition. John Wiley
and Sons, Inc., New York, New York.

Hamilton, W. J., Jr. 1930. The food of the Soricidae. Journal of Mammalogy
11:26-39.

Handley, C. O., Jr. 1992. Terrestrial mammals of Virginia: Trends in distribu-
tion and diversity. Virginia Journal of Science 43:157-169.

Laerm, J. L., C. H. Wharton, and W. M. Ford. 1995. First record of the water
shrew, Sorex palustris Richardson (Insectivora: Soricidae), in Georgia with
comments on its distribution and status in the southern Appalachians. Brim-
leyana 22:47-51.

Linzey, D. W., and A. V. Linzey. 1973. Notes on food of small mammals from
Great Smoky Mountains National Park, Tennessee-North Carolina. Journal
of the Elisha Mitchell Scientific Society 89:6-14.

Merritt, R. W., and K. W. Cummins. 1996. An introduction to the aquatic
insects of North America. Kendall/Hunt, Dubuque, Iowa.

Pagels, J. F. 1987. The pygmy shrew, rock shrew and water shrew: Virginia's
rarest shrews (Mammalia: Soricidae). Virginia Journal of Science 38:364-
368.

Pagels, J. F. 1990. First record of the rock vole, Microtus chrotorrhinus (Miller)
(Rodentia: Cricetidae), in Virginia. Brimleyana 16:1-3.

Pagels, J. E, R. P. Eckerlin, J. R. Baker, and M. L. Fies. 1990. New records of
the distribution and internal parasites of the endangered northern flying
squirrel, Glaucomys sabrinus, in Virginia. Brimleyana 16:73-78.

Pagels, J. F., M. L. Fies, and R. Glasgow. 1991. Recovery plan: Appalachian
water shrew, Sorex palustris punctulatus Hooper. Virginia Department of
Game and Inland Fisheries. Unpublished manuscript.

Pagels, J. E, and C. O. Handley, Jr. 1991. Water shrew: Sorex palustris punc-
tulatus Hooper. Pages 564-565 in Virginia's endangered species (K. Ter-

134 John F. Pagels, Leonard A. Smock and Stephen H. Sklarew

williger, coordinator). McDonald and Woodward Publishing Company,
Blacksburg, Virginia.

Pagels, J. F., and C. M. Tate. 1976. The shrews (Insectivora: Soricidae) of the
Paddy Knob-Little Back Creek area of western Virginia. Virginia Journal
of Science 27:202-203.

Pennak, R. W. 1989. Fresh-water invertebrates of the United States. Third Edi-
tion. John Wiley, New York, New York.

Plafkin, J. L., M. T. Barbour, K. D. Porter, S. K. Gross, and R. M. Hughes. 1989.
Rapid bioassessment protocols for use in streams and rivers: benthic
macroinvertebrates and fish. U.S. Environmental Protection Agency,
PA/444/4-89-011. Washington, D.C.

Platts, W. S., W, F. Megahan, and G. W. Minshall. 1983. Methods for evaluat-
ing stream, riparian and biotic conditions. U.S. Forest Service, General
Technical Report INT-138, Ogden, Utah.

Strahler, A. N. 1964. Quantitative geomorphology of drainage basins and chan-
nel networks. Section 4-2 in (V T. Chow, editor). Handbook of Applied
Hydrology, McGraw-Hill, New York, New York.

Sorenson, M. W. 1962. Some aspects of water shrew behavior. American Mid-
land Naturalist 68:445-462.

Wallace, J. B., and A. C. Benke. 1984. Quantification of wood habitat in sub-
tropical Coastal Plain streams. Canadian Journal of Fisheries and Aquatic
Science 41:1643-1652.

Webster, W. D. 1987. Sorex palustris punctulatus Hooper. Pages 36-38 in
Endangered, threatened, and rare fauna of North Carolina (M. K. Clark,
Editor) Part I. A re-evaluation of the mammals. Occasional papers of the
North Carolina Biological Survey, 1987-3, North Carolina State Museum of
Natural Sciences, Raleigh.

Whitaker, J. O., Jr., and L. L. Schmeltz. 1973. Food and external parasites of
Sorex palustris and food of Sorex cinereus from St. Louis Co., Minnesota.
Journal of Mammalogy 54:283-285.

Wrigley, R. E. 1969. Ecological notes on the mammals of southern Quebec.
Canadian Field Naturalist 83:201-211.

Received 13 September 1996
Accepted 3 December 1996

Occurrence of the Brook Silverside, Labidesthes sicculus
(Atheriniformes: Atherinidae), in North Carolina

Mary L. Moser

Center for Marine Science Research

7205 Wrightsville Avenue

Wilmington, NC 28403

Fred C. Rohde

North Carolina Division of Marine Fisheries

127 Cardinal Drive Extension

Wilmington, NC 28405-3845

Rudolf G. Arndt

Faculty of Natural Sciences and Mathematics

The Richard Stockton College of New Jersey

Pomona, NJ 08240

and

Keith W. Ashley

North Carolina Wildlife Resources Commission

102 Hillcrest Drive

Elizabethtown, NC 28337

ABSTRACT- The brook silverside has never been reported from
Atlantic slope drainages north of the Santee River in South Carolina.
We collected 42 specimens at 6 sites in the Waccamaw River drainage
in North and South Carolina and 26 specimens at 5 sites in the Lynches
River, South Carolina. We also report the collection of five specimens
from the Lumber River and four from the Cape Fear River in North
Carolina. Extensive sampling from 1961-1992 in the Waccamaw River
drainage failed to collect this species. Therefore, we propose that the
brook silverside has expanded its range northward on the Atlantic slope.

The brook silverside, Labidesthes sicculus (Cope), occurs in the St.
Lawrence River drainage and southern Great Lakes tributaries, the Mississippi
River basin, and Atlantic and Gulf drainages from the Santee River in South Car-

135

1 3 6 Mary L. Moser, Fred C. Rohde, Rudolf G. Arndt,

and Keith W. Ashley

olina to the Sabine River in Texas (Etnier and Starnes 1994). Reported occur-
rences closest to the state of North Carolina are from eastern Tennessee, in the
Little Tennessee River drainage (Etnier and Starnes 1994), and from eastern
South Carolina, in the Santee River drainage (Rohde et al. 1994). Our sampling
indicates that the brook silverside in the southeastern United States has expand-
ed its range northward.

Fig. 1. Distribution of the brook silverside, Labidesthes sicculus, in southeast-
ern North Carolina and adjacent South Carolina.

N

0 1 35

i i

KILOMETERS \

^^

\ <s>

\ ^

\ ?

LAKE
WACCAMAW %

HI [^

V ^ / KINGSTON
\ *> / LAKE SWAMP

X V^v WET ASH

V * **

• . JL ^ SWAMP

S z. /

\^ V

^Tiy^

Jol

^yT

\ o

/ t/

?*/

ATLANTIC
OCEAN

WIN YAH (IS

bay i^y/)

Brook Silverside 137

SURVEY AREA
The Waccamaw River originates at Lake Waccamaw in Columbus
County, North Carolina, and flows approximately 63 km south-southwest to the
North Carolina/South Carolina border (Fig. 1). The river has a total length of
approximately 225 km and joins the Pee Dee River at Winyah Bay, South Car-
olina (Shute et al. 1981). In North Carolina the Waccamaw River drains 3,255
km2 of primarily forested land and has a relatively low ratio of base flow to
runoff (Bales and Pope 1996). Tributary streams in this system are generally
highly colored, low pH, low flow to stagnant, blackwater swamps. Many of
these streams exhibit large seasonal changes in depth and flow, with concomi-
tance changes in discharge. Recent studies of streamflow characteristics in this
drainage indicate that flow variability has increased the past decade (Bales and
Pope 1996).

METHODS AND MATERIALS

We sampled at 39 sites in the Waccamaw River drainage from May
1994-mid-November 1996. Standardized sampling was conducted at each site
using a backpack electroshocker, seine, and/or rotenone. For sites in North Car-
olina (n = 18), we used coated nylon nets (50 m, 1.3 cm-mesh) to block off a 33
m or 66 m reach. Three passes were made through the blocked off area using
either a backpack electroshocker or a 3 m x 1.2 m, 3.2-mm mesh seine. For
rotenone samples, fish were dipnetted for up to two hours following introduction
of rotenone at 1 ppm. Rotenone was then neutralized using a 1 ppm potassium
permanganate solution. South Carolina collections (n = 21) were made with a 3
m x 1.2 m, 3.2-mm mesh seine, except for one backpack electroshocker collec-
tion. Fishes taken, and data on habitat (stream depth, width, and substrate type;
current speed; air and water temperatures; pH; dissolved oxygen concentration)
were recorded at each site.

All fishes were preserved in 10% formalin upon capture for subsequent
examination. Brook silversides were measured to the nearest mm standard
length (SL) and deposited at the North Carolina State Museum of Natural Sci-
ences in Raleigh.

RESULTS AND DISCUSSION

In North Carolina we collected 36 brook silversides. We took 18 spec-
imens in the Waccamaw River 50 m below the Lake Waccamaw dam, Columbus
County, on the 14th (5 specimens by electroshocker) and 18th (13 by rotenone)
of September 1995 (Fig. 1). Size range of these fish was 29-48 mm SL, and 2-
3 size classes were represented. Current velocity at this site was 0. 17 m/sec with
a pH of 6.4 and dissolved oxygen of 7.2 mg/L. We also took two specimens (34,
40 mm SL) by electroshocker in Wet Ash Swamp, a Waccamaw River tributary,
at a point 50 m below the State Route 1300 bridge in Brunswick County, on 14
September 1995 (Fig. 1). Current there was 0 m/sec, pH 5.8, and dissolved oxy-

1 3 8 Mary L. Moser, Fred C. Rohde, Rudolf G. Arndt,

and Keith W. Ashley

gen concentration was 8.8 mg/L. A second Waccamaw River tributary, Shingle-
tree Swamp, yielded three juveniles on 18 November 1996, and three juveniles
and one adult on 1 1 April 1997. In addition, survey sampling with a boat elec-
troshocker produced five specimens in the Lumber River at the NC Route 711
bridge in Robeson County on 8 May 1995 (Fig. 1). Size range was 57-65 mm
SL. Four additional fish (48, 53, 60, and 78 mm TL) were collected with the boat
electroshocker from the Cape Fear River at rkm 76 on 24 January 1997.

In South Carolina we collected 15 specimens of the brook silverside
from the Waccamaw River drainage and 26 specimens from the Lynches River
with a seine. Four specimens were taken on 27 May 1994 in the Waccamaw
River at Conway in Horry County. We took eight more fish from Kingston Lake
Swamp, a tributary of the Waccamaw River, near Conway, approximately 54 rkm
downstream of the North Carolina border, on 10-11 May 1994. Three specimens
were also collected in Stanley Creek, a tributary stream approximately 11.8 km
northeast of Conway on 18 March 1996. From 13-17 May 1996, we sampled 10
sites in the Lynches River, a tributary of the Pee Dee River. Brook silversides (n
= 26) were collected at five of these sites from the Route 403 bridge, Flo-
rence/Lee counties north to the Route 15 bridge, Kershaw/Darlington counties, a
distance of approximately 50 rkm. These localities fill in a gap between the pre-
viously known South Carolina records in the Santee River drainage and the new
North Carolina records (Rohde et al. 1994).

Is the distribution of the brook silverside expanding, or has the species
been overlooked in North Carolina? The sites in the North Carolina portion of
the Waccamaw River, as well as other sites in this drainage, have been sampled
frequently by E. Menhinick, J. and P. Shute, Wildlife Resources Commission
biologists, and by F. Rohde. Twenty-nine collections were made by these biolo-
gists at the Waccamaw River site from 1979-1992, and 5 collections in Wet Ash
Swamp during 1961-1989. Since 1960, numerous other sites have been sampled
on multiple occasions with rotenone (39 sites), seines (38), and electroshockers
(10). This intensive sampling over time argues for an increase in brook silver-
side distribution. In addition, the fact that this species is fragile suggests that the
observed range expansion did not result from "bait-bucket" introductions. We
propose that the brook silverside has expanded its range northward and may be
longer lived than previously thought.

ACKNOWLEDGMENTS- For assistance in field work we thank those students
too numerous to mention here individually from The Richard Stockton College
of New Jersey and the University of North Carolina at Wilmington, and Tom
Rachels of the N.C. Wildlife Resources Commission. A. Braswell, North Car-
olina State Museum of Natural Sciences, answered questions on specimens in his
care. The Graphics Production Department of Stockton College, and L.M.
Neveu, helped to prepare the figures. This research was supported in part by a

Brook Silverside 1 39

grant from the North Carolina Division of Water Resources by the Cape Fear
River Program, and by travel funds and vehicles provided by Stockton College.
These surveys were conducted under North Carolina Wildlife Resources Scien-
tific Collecting Permit Number 39.

LITERATURE CITED
Bales, J. D., and B. F. Pope. 1996. Streamflow characteristics of the Waccamaw

River at Freeland, North Carolina, 1940-94. Water Resources Investigations

Report 96-4093, U.S. Geological Survey, Raleigh, North Carolina.
Etnier, D.A., and W.C. Starnes. 1994. The fishes of Tennessee. University of

Tennessee Press, Knoxville.
Rohde, F.C., R.G. Arndt, D.G. Lindquist, and J.F. Parnell. 1994. Freshwater

fishes of the Carolinas, Virginia, Maryland, and Delaware. The University

of North Carolina Press, Chapel Hill.
Shute, J.R., P.W. Shute, D.G. Lindquist. 1981. Fishes of the Waccamaw River

drainage. Brimleyana 6:1-24.

Received 10 October 1996
Accepted 6 November 1996

Winter Mortality in the Green Anole, Anolis carolinensis
(Lacertilia: Polychridae)

Jane K. Distler

Michael E. Dorcas

J. Whitfield Gibbons

Karen L. Kandl

Savannah River Ecology Laboratory

Drawer E

Aiken, South Carolina 29802

and

Kevin R. Russell

Department of Forest Resources

Clemson University

Clemson, South Carolina 29634-1003

ABSTRACT- Winter behaviors of lizards are poorly documented.
Most available information pertains to the formation of aggregations to
escape freezing temperatures. During cold weather, lizards may seek
shelter under bark, felled trees, or rotting stumps. However, such refu-
gia may not provide adequate protection during abnormally cold condi-
tions. We discovered the remains of 12 adult green anoles (Anolis car-
olinensis) within a Carolina bay located on International Paper Timber
Company land in Marion County, South Carolina. The anoles presum-
ably were killed by severe winter temperatures during the winter of
1996.

The winter behaviors of lizards have been poorly documented (Neill
1948, Weintraub 1968, Vitt 1974). Consequently, available information pertains
only to the formation of winter aggregations, apparently to escape freezing tem-
peratures. Several species of lizards are known to form these aggregations. Both
Hamilton (1948) and Neill (1948) described the occurrence of five-lined skinks
(Eumeces fasciatus) within felled trees and rotted logs and stumps. Worthington
and Sabath (1966) documented winter aggregations of tree lizards {Urosaurus
ornatus) within limestone outcroppings in Texas, and Weintraub (1968)
described aggregations of over 37 individual granite spiny lizards (Sceloporus

140

Winter Mortality

141

orcutti) in rock crevices in California. Green anoles {Anolis carolinensis) have
been reported to seek shelter, both individually and in aggregations, underneath
felled trees and rotted stumps (Hamilton 1948, Neill 1948).

During severe cold winter spells, otherwise adequate habitat may not
provide suitable refugia for winter protection. Lacking protection, lizards may
die from freezing temperatures inside their chosen refugia. Such deaths have
been documented. Worthington and Sabath (1966) found skeletal remains of
over sixteen tree lizards in limestone fragments in Texas. Vitt (1974) found thir-
teen dead tree lizards and one banded sand snake {Chilomeniscus cinctus) with-
in a rotted stump in Arizona. Weintraub (1968) found the remains of granite
spiny lizards of all age classes within granite crevices.

Fig. 1 . Remains of green anoles {Anolis carolinensis) on a Cypress stump within a
Carolina bay in the PeeDee River region, South Carolina.

142 Jane K. Distler, Michael E. Dorcas, J. Whitfield Gibbons,

Karen L. Kandl, and Kevin R. Russell

During a herpetofaunal survey on 15 March 1996, we discovered the
remains of twelve adult green anoles within a Carolina bay located between the
Great Pee Dee and Little Pee Dee Rivers in northeastern South Carolina (Fig. 1).
The 1.2-ha Carolina bay was dominated by bald cypress (Taxodium distichum)
and was semipermanently flooded. The bay was surrounded by a one-year-old
clear-cut. The anoles were found beneath the bark of a rotted bald cypress that
was still upright within the bay. Eleven of the specimens appeared to have been
mummified. One anole skeleton was also found along side the other anoles,
indicating that at least one individual had died previously in the same location.
All twelve specimens were on the southeast side of the tree.

Although there are several possible explanations for the anole deaths
(e.g., disease), the anoles were most likely killed by freezing temperatures dur-
ing winter, which leads to the question of "why did the anoles choose this par-
ticular site to overwinter?" One possibility is that there was no other suitable
habitat available (Worthington and Sabath 1966, Weintraub 1968, Vitt 1974).
However, the presence of living anoles in the immediate vicinity (within the
same Carolina bay) indicates at least some lizards were able to find suitable over-
wintering habitat. Long-distance migration is unlikely to account for the pres-
ence of living anoles, because they were found within the bay shortly after the
uncharacteristically cold weather.

Another, more likely explanation is that, under normal circumstances,
this particular site would have provided suitable overwintering habitat. Howev-
er, below-normal temperatures during winter 1995-1996 might have been too
extreme for lizard survival in this particular location. According to the South
Carolina State Climatology Office, average temperatures during winter 1995-
1 996 were lower than the average of all temperatures for the Pee Dee region from
1948 to 1996. More importantly, December 1995 and January 1996 each had
more than 20 consecutive days with low temperatures below 0 C; February 1996
experienced 15 days below 0 C. Prolonged low temperatures are infrequent in
most years in this region. January and February 1996 rank highest in the num-
ber of days since 1988 when temperatures fell below the deep freeze point (-2 C).
We hypothesize that some wintering habitat, which was suitable in years with
normal winter temperatures, might have proven to be unsuitable during the par-
ticularly cold winter of 1995-1996. In addition, Anolis carolinensis commonly
seeks relatively superficial cover (Palmer and Braswell 1995), thus exposing
itself to harsher temperatures than other species that select areas with greater pro-
tection. As a result, winter temperatures are most likely a controlling factor in
the northern distribution of Anolis carolinensis.

ACKNOWLEDGMENTS- We would like to thank A. Braswell and B. Wigley for
providing comments that substantially improved the manuscript. We thank Inter-
national Paper for providing access to the study area and the funding for this pro-

Winter Mortality 143

ject. Other funding agencies include the National Fish and Wildlife Foundation
and the National Council of the Paper Industry for Air and Stream Improvement
(NCASI). Manuscript preparation was supported by financial assistance award
# DE-FC09-96SR 18546 from the U.S. Department of Energy to the University
of Georgia Research Foundation.

LITERATURE CITED

Hamilton, W. J. 1948. Hibernation site of the lizards Eumeces and Anolis in

Louisiana. Copeia 3:211.
Neill, W. T 1948. Hibernation of amphibians and reptiles in Richmond County,

Georgia. Herpetologica 4:107-1 14.
Palmer, W. M., and A. L. Braswell. 1995. Reptiles of North Carolina. The Uni-
versity of North Carolina Press. Chapel Hill, North Carolina.
Vitt, L. J. 1974. Winter aggregations, size classes, and relative tail breaks in the

tree lizard, Urosaurus ornatus (Sauria: Iguanidae). Herpetologica 30:182-

183.
Weintraub, J. D. 1968. Winter behavior of the granite spiny lizard, Sceloporus

orcutti Stejneger. Copeia 4:708-712.
Worthington, R. D., and M. D. Sabath. 1966. Winter aggregations of the lizard

Urosaurus ornatus ornatus (Baird and Girard) in Texas. Herpetologica

22:94-96.

Received 15 November 1996
Accepted 18 March 1997

Abundance And Size Of Dominant Winter-Immigrating Fish
Larvae At Two Inlets Into Pamlico Sound, North Carolina

William F. Hettler, Jr.

National Marine Fisheries Service

National Oceanic and Atmospheric Administration

Southeast Fisheries Science Center

101 Pivers Island Road

Beaufort, North Carolina, USA 28516-9722

ABSTRACT— Weekly sampling for the larvae of six species of ocean-
spawning, estuarine-dependent fishes was conducted from October
1994 to April 1995 inside Oregon Inlet and Ocracoke Inlet, two major
inlets into Pamlico Sound, North Carolina. Atlantic menhaden,
Brevoortia tyrannus, were similar in average density at both inlets;
Atlantic croaker, Micropogonias undulatus, and summer flounder, Par-
alichthys dentatus, were more abundant at Oregon Inlet; spot, Leiosto-
mus xanthurus, pinfish, Lagodon rhomboides, and southern flounder, P.
lethostigma, were more abundant at Ocracoke Inlet. Atlantic croaker
were significantly larger at Oregon Inlet at the beginning and end of the
ingress season, whereas Atlantic menhaden were significantly smaller
at Ocracoke Inlet at the end of the season (ca. 12 mm vs. 27 mm). Abun-
dance data from Oregon and Ocracoke inlets were compared with abun-
dance data collected during the same period at Beaufort Inlet and with
data from a previous monthly survey conducted six years earlier at the
same stations at Oregon and Ocracoke inlets. Winter temperatures were
similar at both inlets, but Ocracoke Inlet was warmer during spring.
Oregon Inlet was less saline than Ocracoke Inlet at every sampling
event.

Pamlico Sound, the largest barrier island estuary in the United States
(5,200 km2), supports numerous fisheries either indirectly as juvenile habitat or
directly as fishing grounds. Major fisheries include species of Clupeidae, Paral-
icthyidae, and Sciaenidae. Most species of these families spawn in the ocean,
after which their larvae pass through inlets before reaching estuarine nurseries.
Data on the ingress through inlets of larvae of these species are essential in
understanding variability in annual recruitment. The only publication describing
the seasonal abundance of fish larvae in inlets to Pamlico Sound was based on

144

Abundance And Size

145

once-monthly sampling (Hettler and Barker 1993). Since that study, analysis of
a daily sampling experiment at Beaufort Inlet concluded that sampling weekly or
more often significantly increases confidence in larval abundance estimates
(Hettler et al. 1997) . The objective of my study was to sample weekly at two of
the three inlets connecting Pamlico Sound directly to the Atlantic Ocean (Ore-
gon Inlet and Ocracoke Inlet) to compare their relative contribution as larval fish
pathways to the marine species nursery grounds in the sound and adjacent tribu-
taries as identified by Epperly and Ross (1986).

METHODS
Oregon Inlet is the only inlet into Pamlico Sound north of Cape Hatteras
and lies in the temperate Virginian Province near the southern end of the
Labrador Current (Fig. 1). Ocracoke Inlet, the largest inlet in North Carolina and
one of two inlets connecting Raleigh Bay (located between Cape Hatteras and
Cape Lookout) with Pamlico Sound, lies in the subtropical Carolinian Province.
These inlets were sampled for 27 consecutive weeks between October 1994 and
April 1995 during the ingress of larvae of six targeted species of fall-winter
spawning fishes, five of which contribute 85% of the total commercial fish catch
in North Carolina (Miller et al. 1984).

Fig. 1. Study location.

Cape
Hatteras

0 10 20 30

146

William F. Hettler, Jr.

Fig. 2. Mean water column temperature and salinity at Oregon Inlet (solid line,
squares) and Ocracoke Inlet (dashed line, triangles), North Carolina, for each
weekly sampling trip during the 1994-95 larval fish immigration period. Solid
symbols indicate ebb tide samples; open symbols indicate flood tide samples.

S 10

2>

I 15

X As,

'A- A * -A

01OCT 01NOV 01DEC 01JAN 01FEB 01MAR 01APR 01MAY

Inside each inlet a single sampling station was established in the center
of the main flood-tide channel (Oregon Inlet station: 35E 46.3'N, 75E 33.5'W ;
Ocracoke Inlet station: 35E 06.4'N, 75E 59.5'W). The deepest water at each sta-
tion was 7 m and the channel width was about 300 m. Inlets were sampled one
night each week on adjacent nights (quasi-synoptic). Each night's sampling con-
sisted of 12 repetitive tows, about 10 minutes apart, with a 0.8-m2, 800 micron-
mesh-net on a 1-m-diameter, sled-mounted, aluminum frame towed at a net
speed of 1 m/sec . A tow consisted of actively towing the net in the deepest water
along the axis of the channel down to the bottom and back to the surface. Tows
were always made into the current. A flow-meter measured the volume of water
passing through the net. Each tow took 4 minutes, filtering approximately 200 m3

Abundance And Size

147

of water. Preceding each tow, temperature and salinity casts were taken with a
SeaBird 19 CTD and direction of tidal flow was recorded. CTD data were aver-
aged for the entire water column and all tows on a given date (Fig. 2), because
the oblique net tows integrated the larval catch from throughout the water col-
umn and the vertical distribution of the larvae was unknown. However, the sur-
face and bottom values were compared to show the amount of temperature and
salinity stratification in the channel at each station (Fig. 3). As observed from the
vessel, the channel currents were flooding on 15 of the 27 dates at Oregon Inlet
and on 20 of the 27 dates at Ocracoke Inlet.

Fig. 3. Difference (anomaly) between the surface and the bottom temperature and
salinity at Oregon Inlet (solid line, squares) and Ocracoke Inlet (dashed line, tri-
angles), North Carolina, during the 1994-95 larval fish immigration period. Pos-
itive values indicate warmer or more saline water at the surface; negative values
indicate warmer or more saline at the bottom. Solid symbols indicate ebb tide
sampling; open symbols indicate flood tide sampling.

01OCT 01NOV 01DEC 01JAN 01FEB 01MAR 01APR

148 William F. Hettler, Jr.

On board the vessel, larvae were preserved in 70% ethyl alcohol. In the
laboratory, larvae were sorted by species and counted. Up to 10 larvae of each
species from each tow were measured to the nearest 0. 1 mm standard length. Lar-
val abundance was calculated as the number per 100 m3 and plotted as the week-
ly mean density (± 1 standard error) of the individual tow densities by inlet and
date. Lengths were plotted as the mean standard length of up to 1 20 larvae of
each species at each inlet each week (± 1 standard error).

Wilcoxon rank sum tests were used to compare densities of species
between inlets. To examine the relative contribution by inlet for each species, the
seasonal weekly density by species for Oregon Inlet and Ocracoke Inlet was
compared with data collected during the same period in a separate study at Beau-
fort Inlet (Warlen 1994; S. Warlen, NMFS Beaufort Laboratory, personal com-
munication). For this comparison, it should be recognized that the Beaufort Inlet
study results are used as proxy data in the absence of data collected with the same
methods as at Oregon and Ocracoke inlets. In the Beaufort study, a 2-m2, 1000-
micron-mesh neuston net was fished passively in the tidal current at the surface.
In both studies, however, the data were standardized to densities per unit volume
by the use of flow meters.

RESULTS AND DISCUSSION
TEMPERATURE AND SALINITY

The inlets were similar in temperature, except that Ocracoke Inlet
warmed at a faster rate after late February than did Oregon Inlet (Fig. 2). Dur-
ing February, when abundance of most larval species was low, temperature at
both inlets dropped to less than 5C.

Salinity as high as 33 ppt was observed twice at Ocracoke Inlet, once
in late autumn and once in early spring, a time when salinity at Oregon Inlet was
about 20 ppt. Salinity at Oregon Inlet was always 5-20 ppt lower than Ocracoke
Inlet and in February was as low as 4 ppt. Salinities lower than 10 ppt in Oregon
Inlet in combination with low temperatures occurred eight times. The physio-
logical consequences of low salinities and temperatures on ocean-spawned lar-
vae is only partially known. For example, Brevoortia tyrannus (Atlantic men-
haden) larvae died in laboratory experiments at salinities <5 ppt and tempera-
tures <5 C. In these experiments, however, 50% mortality in <48 hours also
occurred at high salinity (30 ppt) and low temperatures (<5 C) (Lewis, 1966). In
other laboratory experiments, Leiostomus xanthurus (spot) were determined to
be more cold sensitive at 10 C than Micropogonias undulatus (Atlantic croaker),
but test salinities were not given (Hoss et al. 1988). Their study concluded that
during severe winters many early arriving larvae in estuaries are killed and that
only late arriving larvae survive for recruitment into the fishery.

Twice at each inlet, the temperature difference between the surface and
bottom water equaled or exceeded 1 C in the 7-m-deep channel, but generally

Abundance And Size 149

there was little thermal stratification (Fig. 3). On several occasions the water col-
umn was colder at the surface when strong, cold winds were present. On the
other hand, salinity was often positively stratified, as much as 6 ppt less saline at
the surface. At Oregon Inlet during early February, when there was a 5 ppt dif-
ference between the surface and bottom, the surface was 1C colder than the bot-
tom. At this time, the current direction at the surface was ebbing.

Table 1. Average weekly densities (number per 100 m3 ± 1 standard error) at
Oregon Inlet and Ocracoke Inlet (0.8-m2 net, this study) compared with Beaufort
Inlet (2-m2 net, S. Warlen, NMFS, Beaufort Laboratory, personal communica-
tion) during the October 1994 - April 1995 immigration season (n=27 weeks).
Values connected with a dashed line are not significantly different (Wilcoxon
rank sum test, a =0.05).

Species Oregon Inlet Ocracoke Inlet Beaufort Inlet

Brevoortia tyrannus 43.2 (± 4.1) 43.5 (± 4.9) 22.9 (± 8.4)

Lagodon rhomboides 0.6 (±0.1) 1.7 (±0.3) 12.4 (± 3.9)

Leiostomus xanthurus 4.4 (±1.0) 21.1 (" 4.8) 4.8("18.7)

Micropogonias undulatus 155.5(± 27.1) 26.9 (± 3.9) 25.7 (± 6.1)

Paralichthys dentatus 1.0 (± 0.2) 0.3 (± 0.1) 0.3 (± 0.2)

Paralichthys lethostigma 0.1 (± 0.1) 0.5 (± 0.1) 0.8 (± 0.3)

ABUNDANCE

Unlike the other five selected species, Atlantic menhaden were not sig-
nificantly different in average weekly density at any inlet, although fewer
appeared to be caught at Beaufort Inlet during the year (Table 1). Spot were less
abundant at Oregon Inlet than the other inlets, but Atlantic croaker were most
abundant at Oregon Inlet. Pinfish {Lagodon rhomboides) and southern flounder
(P. lethostigma) were different in density among all inlets. Spot, pinfish, and
southern flounder increased in density towards the south, whereas Atlantic
croaker and summer flounder decreased, which is the expected pattern based on
the known distribution of these species (Fahay 1983). North Carolina is the cen-
ter of the known spawning range of Atlantic menhaden (Freidland et al. 1996),

150

William F. Hettler, Jr.

and similar densities at these inlets is not surprising even though the spawning
locations contributing Atlantic menhaden larvae to each inlet is unknown.

Fig. 4. Mean densities of six selected species of fish larvae at Oregon Inlet (solid
line) and Ocracoke Inlet (dashed line), North Carolina, during the 1994-95 larval
fish immigration period. Error bars equal ± 1 standard error.

P. lethostigma

:

j
\

16 1 P. dentatxLS

2500
2000
1500
1000
500
0

M. urtdxUattis

360

B. tyrannus

I

240

I

120

I

0

/ i \ il

l.i

yx\

(r\

k ,

L. xartthrurus

w

01OCT 01NOV 01DEC OUAN 01FEB 01MAR 01APR 01MAY

01OCT 01NOV 01DEC OIJAN OTFEB 01MAR 01APH 01MAY

One or more prominent peaks in densities of each species occurred at
one or both inlets during the season (Fig. 4). Atlantic croaker were dominant
during the early season at Oregon Inlet with a weekly mean density of >2000
per 100 m3 in late October. In one tow on 29 October 1994, the catch density was
3000 larvae per 100 m3. Another pulse of Atlantic croaker entered Oregon Inlet
in early December, a week after summer flounder peaked in density at that inlet.
Peak summer flounder densities at Oregon Inlet preceded the period of peak
recruitment into Ocracoke by more than 3 months. Summer flounder were found
to peak in Beaufort Inlet in February (Burke et al. 1991). The peak abundance
of Atlantic croaker and summer flounder larvae observed early in the season at
Oregon Inlet compared to the two inlets south of Cape Hatteras, suggests that

Abundance And Size 151

these species are coming from spawning areas that have cross-shelf transport
routes north of Cape Hatteras. Southern flounder, which were not abundant at
Oregon Inlet, peaked at Ocracoke Inlet in mid-February, the same period as
reported earlier for Beaufort Inlet by Burke et al. (1991). Gulf flounder (P.
albigutta), an abundant paralichyid south of Cape Hatteras, were not caught at
Oregon Inlet and therefore are not considered further. The largest numbers of
pinfish were caught at both inlets in mid-January. Spot also were most abundant
in mid-January, but only at Ocracoke. Early in the season Atlantic menhaden
were more abundant at Oregon Inlet than at Ocracoke, but both inlets had high
numbers in mid-December and mid-January. The high densities of Atlantic men-
haden at Oregon Inlet in November, a month before significant ingress into Ocra-
coke, suggests that spawning or favorable cross-shelf transport currents supply-
ing these larvae took place north of Cape Hatteras. In early October, concentra-
tions of Atlantic menhaden larvae have been reported as far south as Currituck
Beach, North Carolina, about 60 km north of Oregon Inlet (Kendall and Reintjes
1974). If this distribution also occurred in October 1994, larvae would have been
in position for transport to the inlet by November. The largest densities of
Atlantic menhaden observed during the season came into Ocracoke Inlet in mid-
March. Except for southern flounder, the abundance of all other species was low
in February at both inlets.

Seasonal density patterns in 1994-1995 were different than those report-
ed for 1988-1989 (Hettler and Barker 1993). In 1988-1989, sampling was con-
ducted monthly with the same 0.8-m2, 800 micron-mesh-net on a 1-m-diameter
frame at the same stations as in 1994-1995. Because large variablity in density
estimates can occur as a result of infrequent sampling, monthly densities proba-
bly do not represent average monthly values (Hettler et al. 1997). However, in
that earlier study, Atlantic menhaden were most abundant at Ocracoke Inlet in
February (92 per 100 m3) and at Oregon Inlet in March (222 per 100 m3), where-
as in the present study density was highest in mid-March at Ocracoke Inlet and
mid-December at Oregon Inlet. Warlen (1994) also recorded peak menhaden
density (130 per 100 m3) in February 1989 at Beaufort Inlet, earlier that year than
any other year between 1986 and 1992. In 1989, spot densities were less than
10% of their 1995 values at Ocracoke Inlet (27 per 100 m3). Flounder densities
at any month were low during 1988-1989 (< 1 per 100m3) for either species.
Southern flounder and pinfish were taken in 1988-1989 only at Ocracoke Inlet.

SIZE

For all species, significant differences in body size occurred between
inlets on many sampling dates (Fig. 5). Average lengths of Atlantic menhaden at
Oregon Inlet decreased in length during November and then rapidly increased by
about 10 mm in mid-December. Increasing density and decreasing size of
Atlantic menhaden larvae in early November at Oregon Inlet indicated that

152

William F. Hettler, Jr.

spawning schools moving south for the winter were approaching the vicinity of
the inlet. At Ocracoke larvae increased from about 17 mm in early December to
27 mm by early January. During the remainder of winter, 25-28 mm Atlantic
menhaden were caught at both inlets until the end of the season at Ocracoke
when the size of larvae decreased to as small as 10 mm. These small menhaden
in April probably resulted from spawning south of Ocracoke Inlet by northerly-
moving adults. Small Atlantic menhaden were not collected at Oregon Inlet or at
Beaufort Inlet in April.

Fig. 5. Mean standard length of six selected species offish larvae at Oregon Inlet
(solid line) and Ocracoke Inlet (dashed line), North Carolina, during the 1994-95
larval fish immigration period. Error bars equal ± 1 standard error.

P. lethostxgma

P. dentai

01OCT OINOV OIDEC 01JAN 01FEB 01MAR OIAPR OlMAY

01OCT 01NCV OIDEC 01JAN 01FEB OlMAR OIAPR OlMAY

M. xindulafrus

L. rhomboides

i-1 v^^^r-il

,j"'V

F

' Y Y

OlOCT OINOV OIDEC 01 JAN 01FEB 01 MAR OIAPR OlMAY

01OCT OINOV OIDEC OIJAN 01FEB 01MAR OIAPR OlMAY

B. tyrartrvus

L. xanthurus

OlOCT OINOV OIDEC OIJAN 01FEB 01MAR OIAPR OlMAY

OlOCT OINOV OIDEC OIJAN 01FEB 01MAR OIAPR OlMAY

Atlantic croaker increased >50% in length at both inlets between late
October and late December. Spawning of Atlantic croaker near Cape Hatteras
begins at least by early September, peaks in October, and is reduced by late
December with perhaps another peak in the spring (Morse 1980). Near Beaufort
Inlet, in Onslow Bay, Atlantic croaker were reported to spawn between mid Sep-

Abundance And Size 153

tember and late February, with the majority of spawning between late September
through November (Warlen 1982). Evidence of summer spawning was present-
ed by Hettler and Barker (1993) who caught 7 mm Atlantic croaker at both inlets
in late August 1989. Atlantic croaker this size are probably about 30-days old
(Warlen 1982). In April at Ocracoke Inlet, the size of croaker dropped to less
than 10 mm due possibly to inshore spawning. The corresponding density data,
however, did not indicate the arrival of significant numbers of newly spawned
larvae.

Spot increased in length about 2 mm per month after early January and
were nearly identical in size at both inlets. At Oregon Inlet no further increase in
length was noted until mid-March, when a few early juveniles (>17 mm)' were
caught. It is difficult to determine if these juveniles had just entered following
ocean transport, or were established residents in the inlet or nearby estuary.
Juvenile spot (20-26 mm) have been collected in that inlet in May and June 1989
(Hettler and Barker 1993). Spot size data before January and after late March are
probably not useful, as few larvae were caught.

The mean lengths of pinfish and both species of flounders increased
during the sampling period. Pinfish were typically about 1 mm smaller at Ocra-
coke than at Oregon Inlet and showed a slight increase in average size at both
inlets from December to February. After January, few pinfish were caught at Ore-
gon Inlet. The average lengths of both species of flounder at both inlets increased
about 2 mm from December to February. In mid-March at Ocracoke when den-
sities of southern flounder were highest, this species was about 13 mm. When
summer flounder peaked in density at Oregon Inlet in mid-December, they also
averaged 13 mm.

CONCLUSIONS

From these quasi-synoptic weekly abundance and size estimates of the
winter-immigrating marine fish larvae at two major inlets to Pamlico Sound, it
appears that Oregon Inlet imported larger-sized individuals and higher densities
of Atlantic menhaden, Atlantic croaker, and summer flounder (important com-
mercial species) significantly earlier than at Ocracoke. In winters with mild tem-
peratures, cohorts of older, larger larvae that establish in the nursery areas with-
in Pamlico Sound early in the season may have a survival advantage over cohorts
of larvae entering later through either inlet; in severe winters the converse
would apply as inferred by Hoss et al. (1988).

The relative value of each inlet as a larval pathway for future juvenile
production and recruitment into the fisheries cannot be extrapolated from these
data without comparing analyses of the daily age structure of immigrating larvae
with juveniles emigrating Pamlico Sound nurseries. Towards this goal, larval
specimens furnished from this study are now undergoing age and growth analy-
ses: Atlantic menhaden (J. Rice, North Carolina State University); Atlantic

154

croaker and spot (C. Jones, Old Dominion University). In the mean time, the den-
sity data provided above should be useful in evaluating the effects of any future
anthropogenic modifications (e.g., jetties) to Oregon or Ocracoke inlets on immi-
grating fish larvae.

ACKNOWLEDGEMENTS - 1 thank P. Crumley, D. Fuss, M. Greene, D. Hoss, M.
Johnson, C. Lund, R. Robbins, L. Settle, D. Peters, H. Walsh, and S. Warlen for
assistance with sampling. Larval fish processing was done by M. Greene and H.
Walsh. Early drafts of the manuscript were reviewed by D. Ahrenholz, D.
Peters, and A. Powell of the Beaufort Laboratory and J. Rice, North Carolina
State University. Support was provided by the National Oceanic and Atmos-
pheric Administration's Coastal Ocean Program / Coastal Fisheries Ecosystems
Studies.

LITERATURE CITED

Burke, J. S., J. M. Miller, and D. E. Hoss. 1991. Immigration and settlement pat-
tern of Paralichthys dentatus and P. lethostigma in an estuarine nursery
ground, North Carolina. Netherlands Journal of Sea Research 27:393-405.

Fahay, M. P. 1983. Guide to the early stages of marine fishes occurring in the
western North Atlantic Ocean, Cape Hatteras to the southern Scotian Shelf.
Journal of Northwest Atlantic Fishery Science 4:1-423.

Epperly, S. P. and S.W. Ross. 1986. Characterization of the North Carolina Pam-
lico-Albemarle estuarine complex. NOAA Technical Memorandum NMFS-
SEFC-175.

Friedland, K. D., D. W. Ahrenholz, and J. F. Guthrie. 1996. Formation and sea-
sonal evolution of Atlantic menhaden juvenile nurseries in coastal estuaries.
Estuaries 19:105-114.

Hettler, W. F., Jr., and D. L. Barker. 1993. Distribution and abundance of larval
fishes at two North Carolina inlets. Estuarine and Coastal Shelf Science
37:161-179.

Hettler, W. F, Jr., D. S. Peters, D. R. Colby, and E. H. Laban. 1997. Daily vari-
ability in abundance of larval fishes at Beaufort Inlet. Fishery Bulletin
95:477-493.

Hoss, D.E., L. Coston-Clements, D.S. Peters, and P. A. Tester. 1988. Metabolic
responses of spot, Leiostomus xanthurus, and Atlantic croaker, Micropogo-
nias undulatus, larvae to cold temperatures encountered following recruit-
ment to estuaries. Fishery Bulletin 86:483-488.

Kendall, A. W., Jr., and J. W. Reintjes. 1974. Geographic and hydrographic dis-
tribution of Atlantic menhaden eggs and larvae along the middle Atlantic
coast from RV Dolphin cruises, 1965-66. Fishery Bulletin 73:317-335.

Lewis, R.M. 1966. Effects of salinity and temperature on survival and develop-
ment of larval Atlantic menhaden, Brevoortia tyrannus. Transactions of the

Abundance And Size 15 5

American Fisheries Society 95:423-426.

Miller, J. M, J. P. Reed, and L. J. Pietrafesa. 1984. Patterns, mechanisms and
approaches to the study of migrations of estuarine-dependent fish larvae and
juveniles. Mechanisms of migrations in fishes, pages 209-225 in J. D.
McCleve, G. P. Arnold, J. J. Dodson, and W.H. Neill, (editors). Plenum,
New York, New York.

Morse, W. A. 1980. Maturity, spawning, and fecundity of Atlantic croaker,
Micropogonias undulatus, occurring north of Cape Hatteras, North Caroli-
na. Fishery Bulletin 78:190-195.

Warlen, S. M. 1982. Age and growth of larvae and spawning time of Atlantic
croaker in North Carolina. Proceedings of the Annual Conference of the
Southeastern Association of Fish and Wildlife Agencies 34:204-214.

Warlen, S. M. 1994. Spawning time and recruitment dynamics of larval Atlantic
menhaden, Brevoortia tyrannus, into a North Carolina estuary. Fishery Bul-
letin 92:420-433.

Received 8 December 1996
Accepted 25 February 1997

First Record of Nutria, My ocas tor coy pus (Mamalia:Rodentia),

in Tenneessee

Michael L. Kennedy and Phyllis K. Kennedy

Department of Biology and Edward J. Meeman Biological Station,

The University of Memphis, Memphis, Tennessee 38152

ABSTRACT — We document the first record of nutria (Myocastor coy-
pus) in Tennessee. A specimen was collected in Shelby County in 1996.

In 1899, nutria (Myocastor coypus) were first introduced into the Unit-
ed States in California as a furbearer (see Willner 1982). The occurrence of the
species in the Southeast probably stemmed from animals that escaped or were
released in Louisiana during the late 1930s (Ashbrook 1948, Willner 1982).
Choate et al. (1994) pointed out that through new introductions and natural dis-
persal, nutria have spread rapidly, and the species is common to abundant in Gulf
coastal marshes and along major waterways of the Coastal Plain in the south-
central United States. Introductions or wild specimens are known from several
states in proximity to Tennessee (e.g., Mississippi, Kennedy et al. 1974; Ken-
tucky, Barbour and Davis 1974; Missouri, Schwartz and Schwartz 1981; North
Carolina, Lee et al. 1982; Illinois, Hoffmeister 1989; Arkansas, Sealander and
Heidt 1990). Sealander and Heidt (1990) showed records of nutria from coun-
ties in the Gulf Coastal Plain of Arkansas that bordered or were near the Missis-
sippi River adjoining Tennessee. However, at this time, no previous record exists
for this species in Tennessee.

On 22 January 1996, a female nutria was collected (by shooting) from
a water-control ditch (ca. 7 m in width) on Eagle Lake Refuge, Shelby County,
Tennessee. The adult animal weighed ca. 6 kg. External measurements (mm)
were as follows: total length, 903; length of tail, 403; length of hind foot, 140;
length of ear, 24. Vegetation along the water-control ditch was early succes-
sional grasses and weeds. The ditch was used to control the water level in adja-
cent agricultural fields and was only a short distance from the Mississippi River.

Since the collection of a single specimen in January 1996, Refuge per-
sonnel and sportsmen have reported additional sightings of nutria in adjacent Shel-
by Forest Wildlife Management Area. Some nutria activity has been observed on
beaver lodges in cypress swamps. However, because this species does not endure
extremely cold temperatures (Schwartz and Schwartz 1981), we doubt Nutria will
become very numerous in Tennessee. The specimen reported on herein is deposit-
ed in the biological collections at The University of Memphis (No. 16628).

156

Nutria 1 5 7

ACKNOWLEDGMENTS - The specimen was collected by J. W. Freeman. D.
Fuqua and R. Byron (Tennessee Wildlife Resources Agency) made the specimen
available to the investigators and provided information relating to their sightings
of nutria activity in the area as well as sightings by sportsmen.

LITERATURE CITED

Ashbrook, F. G. 1948. Nutrias grow in the United States. The Journal of
Wildlife Management 12:87-95.

Barbour, R. W, and W. H. Davis. 1974. Mammals of Kentucky. The Universi-
ty of Kentucky Press, Lexington.

Choate, J. R., J. K. Jones, Jr., and C. Jones. 1994. Handbook of mammals of the
south-central states. Louisiana State University Press, Baton Rouge.

Hoffmeister, D. F. 1989. Mammals of Illinois. University of Illinois Press,
Urbana.

Kennedy, M. L., K. N. Randolph, and T. L. Best. 1974. A review of Mississip-
pi mammals. The National Science Research Institute, Eastern New Mexico
University, Portales.

Lee, D. S., J. B. Funderburg, Jr., and M. K. Clark. 1982. A distributional sur-
vey of North Carolina mammals. Occasional Papers of the North Carolina
Biological Survey 1982-10. North Carolina State Museum, Raleigh.

Schwartz, C. W., and E. R. Schwartz. 1981. The wild mammals of Missouri.
University of Missouri Press, Columbia.

Sealander, J. A., and G. A. Heidt. 1990. Arkansas mammals, their natural his-
tory, classification, and distribution. The University of Arkansas Press,
Fayetteville.

Willner, G. R. 1982. Nutria. Pages 1059-1076 in Wild Mammals of North
America: Biology, management, and economics (J. A. Chapman and G. A.
Feldhamer, editors). The Johns Hopkins University Press, Baltimore,
Maryland.

Prey Selection by Five Species of Vespertilionid Bats on
Sapelo Island, Georgia

Timothy C. Carter, Michael A. Menzel, Diane M. Krishon

Daniel B. Warnell

School of Forest Resources

and

Joshua Laerm

Museum of Natural History

University of Georgia, Athens Georgia 30602

ABSTRACT. - Prey items obtained from fecal samples of 132 individ-
uals representing five species of vespertilionid bats were compared to
available prey as determined by insect light trapping in foraging habi-
tats on Sapelo Island, Georgia. Four orders of insects dominated the
diet of these bats: Coleoptera, Hymenoptera, Lepidoptera, and
Hemiptera. Homoptera and Diptera were present in smaller propor-
tions. All five bat species exhibited significant selection for or against
certain insect orders. The evening bat (Nycticeius humeralis) consumed
Coleoptera and Hymenoptera in proportion to their availability, but sig-
nificantly fewer Homoptera than were available in the foraging habitats.
Differences in feeding selectivity were observed between sexes and age
groups. Adult male and juvenile evening bats consumed significantly
fewer Coleoptera and more Hymenoptera than were available in the for-
aging habitats; adult females showed little feeding selectivity. The
Seminole bat (Lasiurus seminolus) consumed Homoptera and Diptera
in significantly lower proportion to their availability. The eastern pip-
istrelle (Pipistrellus subflavus) consumed significantly more Lepi-
doptera and fewer Coleoptera and Homoptera in proportion to their
availability. The big brown bat (Eptesicus fuscus) fed mostly on
Coleoptera and Lepidoptera, whereas the northern yellow bat (L.inter-
medius) consumed only Coleoptera and Hymenoptera.

Twelve species of bats occur in Lower Coastal Plain ecosystems of
South Carolina, Georgia, and northern Florida (Barbour and Davis 1969, Hall
1981). With the exception of Zinn and Humphrey's (1981) study of prey avail-
ability and prey selection of the southeastern bat, Myotis austroriparius (Rhoads
1897), only anecdotal comments regarding foraging behavior of bats inhabiting

158

Prey Selection 159

these regional ecosystems are available (Harper 1927, Sherman 1935, 1939,
Moore 1949, Ivey 1959, Golley 1962, 1966, Neuhauser and Baker 1974, Sanders
1978, Schacher and Pelton 1979).

In conjunction with mist netting studies relating to roost site selection
and habitat use of bats on Sapelo Island, Georgia (Menzel et al. 1995), we under-
took a comparative study of prey selection based on analysis of fecal pellet con-
tents collected from five species of bats captured on the island: evening bat, Nyc-
ticeius humeralis (Rafinesque 1818), Seminole bat, Lasiurus seminolus (Rhoads
1895), eastern pipistrelle, Pipistrellus subflavus (Cuvier 1832), northern yellow
bat, L. intermedius Allen 1862, and big brown bat, Eptesicus fuscus (Beauvois
1796). To determine the degree of prey selectivity by the bats from among
potential prey, we compared fecal pellet contents to available insects collected at
vegetational community types on the island where bats foraged.

STUDY AREA

The study was conducted on Sapelo Island, Mcintosh County, Georgia
from 19 June through 24 July 1995. Sapelo Island is located approximately 63
km south of Savannah and 5.5 km off shore (31°27'N, 81°16'W). The island is
approximately 16 by 3.2 km in size and is typical of barrier islands of the south-
eastern Atlantic Bight (Johnson et al. 1974). Seven well-defined vegetational
community types characteristic of regional lower Coastal Plain ecosystems are
present on the island (Shaw and Fredine 1956). Bats are known to forage in all
seven of these communities. Longleaf pine stands (Pinus palustris) are restrict-
ed to the northern third of the island. The remaining vegetational community
types are located throughout and include stands of pond pine (P. serotina), loblol-
ly-slash pine (P. taeda and P. elliottii), mixed pine-oak (upland oaks comprise
less than 25 % of the overstory), mixed oak-pine (pines comprise less than 25 %
of the overstory), oak stands dominated by live oak {Quercus virginiana), and
high marsh. Further descriptions of the floral associates of these vegetational
communities are provided by Johnson et al. (1974).

The climate of Sapelo Island is characterized by long, warm summers
and short, mild winters. Average temperatures for June and July are 26.3 and
27.7 C, respectively. Average monthly rainfall for June and July is 14.58 and
15.65 cm, respectively (National Climatic Center 1983, Johnson et al. 1974).

METHODS

Capture Techniques - Bats were captured throughout the study using 3
x 12 m mist nets set over or near ponds in all seven major vegetational commu-
nity types on the island. Nets were opened from dusk until 0200 hours. Bats
are known to forage over dunes, marshes, and open salt water. However, no
effort was made to mist net in these areas. Bats netted throughout the night were
held in a 32-ounce cup, and fecal pellets were collected. All bats were released

1 6 0 Timothy C. Carter et al.

within an hour, whether fecal pellets were collected or not. Data recorded from
bats included species, sex, and age class (juvenile or adult). Age classes were
determined by back-lighting finger joints to examine the level of epiphyseal dia-
physeal fusion (Anthony 1988).

Insect Sampling - A variety of methods are available to sample insects.
All of these have inherent biases (Kunz 1988). While light traps are biased
toward phototrophic insects (Bowden 1982), they have been shown to be satis-
factory in foraging studies of bats (Taylor and Carter 1961, Brack and LaVal
1985, Jones 1990, Lacki et al. 1995).

Seven, 10- watt, black light insect traps were powered by automotive
batteries. One was placed in each vegetational community type. Traps were sus-
pended from 1 to 3 m above the ground and positioned to be visible from most
points within a 60-m radius. Traps were operated each night between 2100 and
0300 hours at the same time bats were tracked using telemetry. Insects were
removed each night and frozen for subsequent identification. The size of the
insects considered to be consumable ranged from 2 to 25 mm for all bats (Gould
1955, Ross 1961, Black 1974, Feldhamer et al. 1995). A total of 8,753 insects
in this size range was identified to order, and proportions of orders present were
calculated. Regression analysis indicated no changes in relative insect abun-
dance in the respective habitat types over our sampling period. Therefore, we
combined data for insects in each habitat type over our sampling period.

Fecal Analysis - Fecal samples were placed in a petri dish with 70%
ethanol solution and teased apart using probes and forceps (Whitaker 1988). All
fecal pellets collected from a single individual were examined together using a
dissecting microscope. To eliminate researcher bias, fecal samples were exam-
ined using identification numbers that were referenced to the species, age, and
sex of the bats. A reference collection of insects collected during the study was
used to help identify fecal matter (Whitaker 1988). Most insects were identified
to order, some to family or species. Percent volume of prey taxa was visually
estimated for each sample, and percent occurrence was calculated. Lepidopter-
ans were often only represented in fecal samples by scales. Therefore, percent
volume of this order was estimated using a modified version of Black's method
(1972), and were not considered if present in small numbers.

Selectivity - Whitaker (1994) noted that to distinguish between oppor-
tunistic versus selective feeding by insectivorous bats, it is necessary to assess
the insect taxa available to the bats and compare these to prey items actually
eaten. We followed Whitaker (1994) in assessing prey taxa availability by sam-
pling insects in the habitats in which the bats were foraging (see below). We then
compared prey taxa availability in different habitats to the insect taxa found in

Prey Selection 161

the fecal pellets. If prey availability at sampling sites differed significantly from
prey taxa obtained in fecal samples, we assumed the bats were feeding selec-
tively.

One-way analysis-of-variance (ANOVA) and the Bonferonni multiple
range test revealed that the diet of the bats remained constant over our sampling
period. Therefore, we compared the fecal samples of each species of bat to the
samples of insects collected throughout the summer.

Since it is likely that bats feed in more than one vegetational communi-
ty type and the proportion of available prey may differ between vegetational
community types, we again followed Whitaker (1994) by prorating the time
spent foraging in different vegetational community types. We used telemetry
data to determine the time each bat species spent in each vegetational commu-
nity type and multiplied this by the proportion of insect taxa collected in that
vegetational community type. The prorated time spent in each vegetational com-
munity type was then summed to obtain the total proportion of insect taxa in the
bat's hypothetical foraging area. Only fecal samples from bats captured while
foraging in areas where insects were collected were used in this analysis. Dif-
ferences between expected and actual diet were determined using an ANOVA
(Sokal and Rohlf 1987). Significance was accepted at the p < 0.05 level.

RESULTS AND DISCUSSION

Fecal samples from 132 individual bats were examined: 99 N. hwner-
alis (Table 1), 24 L. seminolus, 4 P. subflavus, 3 E. fuscus, 2 L. intermedius
(Tables 2). Due to the large sample size of N. hwneralis, we were also able to
analyze this species in three groups: adult males, adult females, and juveniles.
Table 1 and 2 summarize fecal analysis data and prey availability comparisons.
Samples were collected from bats netted in all vegitational community types
except pine-oak, a community in which no bats were captured. Due to the size
of insect fragments found in fecal pellets, identification of only six major orders
was possible: Coleoptera, Hymenoptera, Lepidoptera, Hemiptera, Homoptera,
and Diptera. Other orders may have been present in lower quantities. Percent
volume and percent occurrence of insect orders consumed varied among species
(Tables 1 and 2). While previous studies suggest that small, insectivorous bats
are opportunistic feeders (Kunz 1974, Fenton and Morris 1976, Swift et al.
1985), each of the five species we studied demonstrated statistically significant
feeding selectivity for certain insect orders.

Nycticeius humeralis

Fecal samples from 99 evening bats were examined (Table 1). Six
orders of prey items were found. Coleoptera were present in 9 1 % of the fecal
samples followed by Hymenoptera (69%), Lepidoptera (48.5%), Hemiptera
(40.5%), Homoptera (7%), and Diptera (8%). No significant depar-

162

Timothy C. Carter et al.

*£*

E -

£6

«J >

X <

is

£A II

8 8

o-s

«5

*l!

uu O

I <

w-i «o

O t>

p

:3

©

00

5 s
<£ 8

- c

S 8

II

« T3

o 53

& 8 i .£H

Ja a k q

Prey Selection

163

8

at

2
So

c

o

Io

O

<•-*

-a

o

S3

>

d

o

fe

8,

a>

a

E

on

3

c

O

O

>

1/5

3

3

o

JO

«

T3

JC

i

00

c

'5b

d

«

O

£

_c

3

u.

— ;

K <

81

Oh Q>

: «>

•xs ^

•IS

*£:?

8 8

so

4S

s

§ II

3 C

5 ii

% S

X <

u o

u. >

*&8

•is

45

(S 8 8

u- O

•II

8 8

o o o o

>n <n On cs

O "2

o m

O — O O en

moo
<s <o —

u

o

>

ON

o

vn

<<*■

Csl

<N

>n

d

d

<u

u

tu O

<N

o

_

wo

OS

l/~.

r-

CM

o
£

o

X

£ &

Q *

1 6 4 Timothy C. Carter et al.

tures between fecal volume and prey availability were noted except that
Homoptera occurred in significantly lower volume (p < 0.001) in fecal samples
than were available in the environment. The relatively large number of evening
bats netted allows for a comparison between adult males (n = 16), adult females
(n = 41), and juveniles (n=42) of the percent volume of insect taxa in fecal sam-
ples to the percent availability in environment (Table 1).

Regardless of sex and age, the fecal volume of evening bats was com-
posed primarily of Coleoptera, Hymenoptera, and Hemiptera (combined these
groups comprised 88% or more of the diet). Adult males consumed significant-
ly fewer Coleoptera (p = 0.001) and Homoptera (p < 0.001) than were available
in the environment (Table 1). The large proportion of Hymenoptera in fecal sam-
ples of males compared to availability was not significant (p = 0.071), but sug-
gests a feeding preference. Adult males consumed smaller proportions of Lepi-
doptera, Hemiptera, and Diptera than did males or juveniles. Combined, these
taxa constituted only 23.5 % of the diet and were consumed in roughly equiva-
lent proportion to their availability in the environment (21%). Adult females
showed little feeding selectivity. Homoptera differed significantly (p = 0.004)
between the percent volume in the fecal samples and their availability in the
environment. However, these comprised only a very small portion (0.5 %) of the
diet. The taxa comprising over 99.5% of the fecal contents were consumed in
equal proportion to their availability in the environment. Juvenile evening bats
consumed similar prey to that of adult males and females. However, their prey
consisted of significantly fewer Coleoptera (p = 0.056), Homoptera (p < 0.001),
and Diptera (p < 0.001) than were available in their environment.

Significant differences in fecal volume of prey species were observed
between adult male and female evening bats. Male evening bats consumed sig-
nificantly fewer Coleoptera (p = 0.025) than females and also significantly fewer
than in proportion to their availability (Table 1, Figure 1). Males also consumed
significantly more Hymenoptera (p = 0.039) than females, and in significantly
higher proportion to their availability. This might be related to differences in the
physiological state and metabolic requirements of males and females during the
time of the year of our study (e.g., parturition and lactation). Adult females are
expected to be under high levels of nutritional stress and coupled with time con-
straints imposed by offspring, might not be able to be as selective in their diets
as adult males. Juveniles are not as constrained by time or energy as they are by
their lack of foraging experience. Juveniles may be less selective, eating what-
ever they can catch. Adult males are not restricted by time constraints, experi-
ence or energy demands, allowing them more dietary selectivity.

A few reports on the foraging habits of the evening bat are available
(Ross 1967 in Freeman 1981, Whitaker 1972, Zinn 1977, Whitaker and Clem
1992, Feldhamer et al. 1995). Most of these studies had small sample sizes, and
none compared diet to relative prey abundance. Coleoptera were generally

Prey Selection

165

Fig. 1. Comparison of available insect sample percentages (white) with fecal
analysis from juvenile (vertical bars), adult male (horizontal bars), and adult
female (diagonal bars) Nycticeius humeralis captured between 19 June and 24
July 1995, on Sapelo Island, Georgia.

1

Q)

I

CO
.=3

.8

I

1

2

SS8S

-ffl

S

I

I

E

O

a.
2 §

I s

e «

0 O

1 ©

Q.
O

■g

o

c

0)

E
>«

I

CO
X

CO

«■*

>.
o

auiniOA %

1 6 6 Timothy C. Carter et al.

reported to be the most important food source. Although we found Coleoptera
to be present in 91% of the fecal samples, we found that significantly fewer
Coleoptera were fed upon by adult males and juveniles than were available in
environment. Zinn (1977) and Ross (1967 in Freeman 1981) also reported
Hymenoptera from fecal pellets of the evening bat. We also found Hymenoptera
(mostly flying ants - Formicidae) to be a major food source.

Lasiurus seminolus

Six orders of prey items were also found in the fecal samples of 24
Seminole bats (Table 2). In fecal samples of Seminole bats 94.5% of the diet
was from three orders: Coleoptera, Hymenoptera, and Lepidoptera. The percent
volume of these taxa in fecal samples was not significantly different from their
availability in the environment. Smaller proportions of Hemiptera, Homoptera,
and Diptera (combined, constituting only 5.5% of the diet) were also found. Per-
cent volumes of Homoptera (p < 0.001) and Diptera (p < 0.001) in the fecal sam-
ples were significantly lower than the percent available in the environment.
There have been two reports of this species gleaning (Sherman 1935, Barbour
and Davis 1969). Sherman (1939) found Coleoptera, Homoptera, and Diptera in
the contents of a single stomach. Zinn (1977) found Coleoptera, Odonata, and
Hymenoptera to be food items. These observations combined with our results
confirm the importance of Coleoptera and Hymenoptera in the diet of Seminole
bats.

Pipistrellus subflavus

The fecal samples obtained from four eastern pipistrelles suggest the
most dramatic foraging selectivity of the five species of bats studied. Five taxo-
nomic orders were present in fecal samples (Table 2). Lepidoptera were present
in 100% of the fecal samples. They constituted only 5% of prey taxa available
in the environment, but made up 74% of the volume of prey items in fecal sam-
ples. Coleoptera, on the other hand, were present in only 25% of the fecal sam-
ples. They constituted 66% of taxa available in environment, but made up only
6% of the volume in the fecal samples. Differences for both Lepidoptera (p =
0.007) and Coleoptera (p < 0.001) were highly significant. No Homoptera were
found in the fecal samples, although they made up 12% of the prey available in
the environment (p < 0.001). Hymenoptera were present in 50% of the fecal
samples, and Hemiptera and Diptera in 25% of the samples. Differences
between the respective percent volume of these taxa in fecal samples and their
percent availability in the environment were not significant.

Whitaker (1972) found that the 23 eastern pipistrelles he examined con-
sumed nearly 30% Coleoptera and only 7.3% Lepidoptera. Other researchers
have found Coleopterans present in lower proportions or entirely absent (Sher-
man 1939, Ross 1967 in Freeman 1981, Zinn 1977, Swift et al. 1985). Sherman

Prey Selection 167

(1939) and Swift et al. (1985) both reported Diptera to be the most important
food source for the eastern pipistrelle.

Eptesicus fuscus

Three taxonomic orders of prey were observed in the fecal samples of
three big brown bats. Coleoptera and Lepidoptera were found in 100% of the
fecal samples, whereas Diptera was only observed in one (Table 2). The diet of
these bats was dominated by Coleoptera (78%). Beetles were fed upon in pro-
portions equal to their availability. Lepidoptera appeared to be selectively fed
upon. They comprised 21% of the fecal volume, compared to 5% of available
insects sampled in the environment (p = 0.056). The small proportion of Diptera
observed in fecal samples was not significantly different from their availability.
Hymenoptera, Hemiptera, or Homoptera were not observed in the fecal samples.
The diet of the bats we examined was similar to that reported in previous stud-
ies, in that Coleoptera predominated in the diet (Hamilton 1933, Phillips 1966,
Ross 1967 in Freeman 1981, Whitaker 1972, Whitaker 1995). Whitaker (1972)
found that 4.3% of the diet was composed of non-flying insects, suggesting that
big brown bats may occasionally glean from the ground or foliage.

Lasiurus intermedius

The fecal samples of the two northern yellow bats captured were com-
posed entirely of Coleoptera and Hymenoptera (Table 2). Coleoptera and
Hymenoptera made up 3 1 % and 69% of the fecal samples by volume, respec-
tively. No significant difference between percent fecal volume and percent avail-
ability was observed for Coleoptera. However, differences between fecal vol-
ume and availability of Hymenoptera approached significance (p = 0.067), sug-
gesting a feeding preference for this taxa.

Previous studies reported Coleoptera as the most frequently consumed
prey taxa (Sherman 1939, Zinn 1977). Hymenoptera were also found in lower
volumes. Ivey (1959) reported observing northern yellow bats foraging in back
dune depressions where mosquitoes and flies were abundant. However, in con-
trast to Webster et al. (1980), he did not actually witness bats consuming these
insects.

CONCLUSION
Despite small sample sizes, we found significant differences among
available and consumed prey in all five species of bats studied. Although there
are some biases associated with any type of sampling (Taylor and Carter 1961,
Rabinowitz and Tuttle 1982); the comparison of the available prey and prey that
represented in the fecal samples gives us a greater insight into the complex for-
aging habits of some of the bat species found in the Southeast.

1 6 8 Timothy C. Carter et al.

ACKNOWLEDGMENTS - We thank E. Higgins, R. P. Carter, B. T. Bond, R. R.
Smith, G. Westmoreland, J. W. Pesaturo, and T. Oput for field assistance, and
particularly the Sellers family for their hospitality. We also thank C. L. Smith,
B. R. Chapman, K. V. Miller, R. J. Warren, and G. Balkcomb for their profes-
sional support and advice. We especially acknowledge A. G. Chalmers and J. D.
Gould for their advice and assistance with the global positioning equipment and
Arc Info/View data base we used. This project was funded by the Georgia
Department of Natural Resources Nongame-Endangered Wildlife Program and
the University of Georgia Museum of Natural History. Light traps were supplied
by the D. B. Warnell School of Forest Resources.

LITERATURE CITED

Anthony, E. L. P. 1988. Age determination. Pages. 47-57 in Ecological and
behavioral methods for the study of bats (T. H. Kunz, editor). Smithsonian
Institution Press, Washington, D.C.

Barbour, R. W., and W. H. Davis. 1969. Bats of America. The University Press
of Kentucky, Lexington.

Black, H. L. 1972. Differential exploitation of moths by the bats Eptesicus fus-
cus and Lasiurus cinereus. Journal of Mammalogy 53:598-601.

Black, H. L. 1974. A north temperate bat community: structure and prey popu-
lations. Journal of Mammalogy 55:138-157.

Brack, V. Jr., and R. K. LaVal. 1985. Food habits of the Indiana bat in Missouri.
Journal of Mammalogy 66:308-315.

Feldhamer, G. A., J. O. Whitaker, Jr., J. K. Krejca, and S. J. Taylor. 1995. Food
of the evening bat (Nycticeius humeralis) and red bat {Lasiurus borealis)
from southern Illinois. Transactions of the Illinois State Academy of Sci-
ence 88:139-143.

Fenton, M. B., and G. K. Morris. 1976. Opportunistic feeding by desert bats
(Myotis spp.). Canadian Journal of Zoology 54:526-530.

Freeman, P. W. 1981. Correspondence of food habits and morphology in insec-
tivorous bats. Journal of Mammalogy 62:166-173.

Golley, F. B. 1962. Mammals of Georgia. University of Georgia Press, Athens.

Golley, F. B. 1966. The Mammals of South Carolina. Contributions from the
Charleston Museum XV, South Carolina.

Gould, E. 1955. The feeding efficiency of insectivorous bats. Journal of Mam-
malogy 36:399-407.

Hall, E. R. 1981. Mammals of North America. Second edition. John Wiley and
Sons, New York, New York.

Hamilton, W. J. 1933. The insect food of the big brown bat. Journal of Mam-
malogy 14:155-156.

Harper, F. 1927. The mammals of the Okefinokee Swamp region of Georgia.
Proceedings of the Boston Society of Natural History 38:191-396.

Prey Selection 169

Ivey, R. D. 1959. The mammals of Palm Valley, Florida. Journal of Mammal-
ogy 40:585-591.

Johnson, A. S., H. O. Hillestad, S. F. Shanholtzer, and G. F. Shanholtzer. 1974.
An ecological survey of the coastal region of Georgia. National Park Ser-
vice Scientific Monograph Series 3:1-233.

Jones, G. 1990. Prey selection by the greater horseshoe bat {Rhinolophus fer-
rumequinum): optimal foraging by echolocation? Journal of Animal Ecolo-
gy 59:587-602.

Kunz, T. H. 1974. Feeding ecology of a temperate insectivorous bat {Myotis
velifer). Ecology 55:693-711.

Kunz, T. H. 1988. Methods of assessing the availability of prey to insectivorous
bats. Pages 191-209 in Ecological and behavioral methods for the study of
bats (T. H. Kunz, editor). Smithsonian Institution Press, Washington, D.C.

Lacki, M. J., L. S. Burford, and J. O. Whitaker , Jr. 1995. Food habits of gray
bats in Kentucky. Journal of Mammalogy 76:1256-1259.

Menzel, M. A., T. C. Carter, and D. M. Krishon. 1995. Roosting, foraging, and
habitat use by bats of Sapelo Island, Georgia. Final Report to the Georgia
Nongame Wildlife Program, Georgia Department of Natural Resources.

Moore, J. C. 1949. Putnam County and other Florida mammal notes. Journal
of Mammalogy 30:57-65.

National Climatic Center. 1983. Climate normals for the United States (Base
1951-1980), First edition. Gale Research Company Book Tower, Detroit,
Michigan.

Neuhauser, H. N., and W. W. Baker. 1974. Appendix 5. Annotated list of the
mammals of the coastal islands of Georgia. Pages. 197-207 in An ecologi-
cal survey of the coastal region of Georgia (A. S. Johnson, H. O. Hillestad,
S. F. Shanholtzer, and G. F. Shanholtzer, editors). National Park Service Sci-
entific Monograph Series 3:233.

Phillips, G. L. 1966. Ecology of the big brown bat (Chiroptera: Vespertilion-
idae) in northeastern Kansas. The American Midland Naturalist 75:168-
198.

Rabinowitz, A. R., and M. D. Tuttle. 1982. A test of the validity of two cur-
rently used methods of determining bat prey preferences. Acta Theriologi-
ca 27:283-293.

Ross, A. 1961. Notes on food habits of bats. Journal of Mammalogy 42:66-71.

Sanders, A. E. 1978. Mammals (of the coastal zone of South Carolina). Pages
296-308 in An annotated checklist of the biota of the coastal zone of South
Carolina (R. G. Zingmark, editor). University of South Carolina Press,
Columbia.

Schacher, W. H., and M. R. Pelton. 1979. The mammals of Kiawah Island,
South Carolina. Proceedings South Carolina Endangered Species Sympo-
sium 1(1976): 184-195.

1 7 0 Timothy C. Carter et al.

Shaw, S. P., and C. G. Fredine. 1956. Wetlands of the United States. U.S. Fish
and Wildlife Service 39, Washington, D. C.

Sherman, H. B. 1935. Food habits of the Seminole bat. Journal of Mammalo-
gy 16:224.

Sherman, H. B. 1939. Notes on the food of some Florida bats. Journal of Mam-
malogy 20:103-104.

Sokal, R. R., and R. J. Rohlf. 1987. An introduction to the study of biostatis-
tics. W. H. Freeman and Company, New York, New York.

Swift, S. M., P. A. Racey, and M. I. Avery. 1985. Feeding ecology of Pipistrel-
lus pipistrellus (Chiroptera: Vespertilionidae) during pregnancy and lacta-
tion. II. Diet. Journal of Animal Ecology 54:217-225.

Taylor, L. R., and C. I. Carter. 1961. The analysis of number and distribution in
an areal population of Macrolepidoptera. Transactions of the Entomologi-
cal Society of London 113(12):369-386.

Webster, W. D., J. K. Jones, and R. J. Baker. 1980. Lasiurus intermedius. Mam-
malian species 132:1-3.

Whitaker, J. O., Jr. 1972. Food habits of bats from Indiana. Canadian Journal
of Zoology 50:877-883.

Whitaker, J. O., Jr. 1988. Food habits analysis of insectivorous bats. Pages 171-
189 in Ecological and behavioral methods for the study of bats (T. H. Kunz,
editor). Smithsonian Institution Press, Washington, D.C.

Whitaker, J. O., Jr. 1994. Food availability and opportunistic versus selective
feeding in insectivorous bats. Bat Research News 35:75-77.

Whitaker, J. O., Jr. 1995. Food of the big brown bat Eptesicus fuscus from
maternity colonies in Indiana and Illinois. The American Midland Natural-
ist 134:346-360.

Whitaker, J. O., and P. Clem. 1992. Food of the evening bat Nycticeius humer-
alis from Indiana. The American Midland Naturalist 127:211-214.

Zinn, T. L. 1977. Community ecology of Florida bats with emphasis on Myotis
austroriparius. M.S. Thesis, University of Florida, Gainesville.

Zinn, T. L., and S. R. Humphrey. 1981. Seasonal food resources and prey selec-
tion of the southeastern brown bat {Myotis austroriparius) in Florida. Flori-
da Scientist 44:81-90.

Received 21 January 1997
Accepted 18 March 1997

171

DATE OF MAILING

Brimleyana 24 was mailed on 20 June 1997.
Brimleyana 25 was mailed on 1 September 1998.

172

INFORMATION FOR CONTRIBUTORS

Submit an original and three copies of manuscripts to Editor, Brim-
ley ana, North Carolina State Museum of Natural Sciences, P.O. Box 29555,
Raleigh, NC 27626. In the case of multiple authors, indicate correspondent.
Include a cover letter indicating that the paper has been exclusively submitted to
Brimleyana.

Preparation of Manuscript — Generally adhere to the Council of Biolo-
gy Editors Style Manual, Sixth Edition. Use medium-weight bond paper, 8.5 x
1 1 inches. Leave at least an inch margin on all sides. All typewritten material
including tables and literature cited should be double-spaced. To expedite pub-
lication, typesetting is done electronically from a computer disk copy of the man-
uscript. Therefore, the final version of papers accepted for publication must be
submitted on a floppy disk. Specify word processing software and file name on
the disk.

The first page should contain only the title of the article, the name(s)
and address(es) of the author(s), and any footnotes such as present address(es) as
necessary. When appropriate, the title should indicate at least two higher taxa.
For example: Nest and larvae of the Neuse River waterdog, Necturus lewesi
(Brimley) (Amphibia: Proteidae).

A brief informative abstract on a separate sheet follows the title page.
Footnotes, other than those in tables, should be used only when absolutely nec-
essary, and should be numbered consecutively throughout the paper.

Individuality of writing style and text organization are encouraged, but
most papers are best presented with first-level headings in this form: introduc-
tion, materials and methods, results, discussion, and literature cited. Make first-
level headings centered and in all capital letters. Second-level headings should
be flush left, in all capital letters, and on a separate line; third headings are
indented, upper and lower case, underlined, and followed by a dash. An
acknowledgments section should precede the literature cited. Use no more than
three levels of headings.

Scientific names in taxonomic papers should include the author when
first used in the text. Descriptions of new taxa must be in accordance with the
requirements of established international codes. Presentations of etymologies
are desirable.

Format for Literature Cited — Authors, not the editor, are responsible
for verifying references. Do not use an excessive number of citations; do not
abbreviate journal titles, names of publishers, cities, or states; and do not capi-
talize works in titles (except proper names and places). List citations in alpha-
betical order by author's last names, then by date if more than one citation is by
the same author(s). For example:

173

Adams, J. J. 1977. Food habits of the masked shrew, Sorex cinereus (Mam-
malia: Insectivora). Brimleyana 7:32-39.

Adams, J. J. 1988. Animals in North Carolina folklore. Second edition. Uni-
versity of North Carolina Press, Chapel Hill.

Barnes, R. G. 1986. Range, food habits, and reproduction in Glaucomys sabri-
nus in the southern Appalachian Mountains of North Carolina and Tennessee.
Ph.D. Thesis. North Carolina State University, Raleigh.

Barnes, R. G. 1989. Northern flying squirrel. Pages 203-230 in Mammals of
the southeastern United .States (J. J. Adams and J. M. Smith, Jr., editors). Harp-
er and Row, New York, New York.

Citing References in Text — Use parenthetical format, for example:
Adams (1988) said . . ., or flying squirrels (Barnes 1989). For a publication with
more than two authors, use the first author and et al. (not italicized). When cit-
ing more than one source, list them chronologically, then alphabetically.

Tables — Prepare each table on a separate sheet and arrange in numeri-
cal order at the end of the manuscript. Do not use vertical lines in tables. For
footnotes in tables, use superscript numbers beginning left to right, then down.
Indicate in the margin of the text in pencil where tables and figures should be
placed.

Figures — Number illustrations, including maps, graphs, charts, draw-
ings, and photographs, consecutively as figures. Size should not be larger than
8.5 x 11 inches. Proportions should make optimal use of the page format of
Brimleyana. Do not send material as collections of individual items to be
arranged by the editorial staff. Do not mount photographs. The author's name,
figure number, and notation "top" should be placed on the back of each figure.
Lettering should be large enough to permit proper reduction and still be readable.
Do not type on illustrations. Figure legends should be typed, double-spaced, on
a separate sheet and accompany the figures.

Page Changes, Reprints, and Proofs — A charge of $30 per page is
expected from authors who have available funds. Ability to pay page charges is
not considered as a part of the publication acceptance procedure. Contributors
who pay full page charges will be furnished 100 free reprints. Reprint order
forms will be sent with page proofs to authors. Page proofs are to be corrected,
signed, and returned to the Editor within 7 days.

STATE LIBRARY OF NORTH CAROLINA

3 3091 00748 4751

BRIMLEYANA NO. 25, JULY 1998
CONTENTS

Turtles (Reptilia: Testudines) of the Ardis Local Fauna Late Pleistocene (Rancholabrean)
of South Carolina. Curtis C. Bentley and James L. Knight 3

Observations of Freshwater Jellyfish, Craspedacusta sowerbyi Lankester (Trachylina:
Petasidae), in a West Virginia Reservoir. Ted R. Angradi 34

Distribution and Status of Selected Fishes in North Carolina, with a New State Record.
Fred C. Rohde, Mary L. Moser, and Rudolf G. Arndt 43

Status of the River Frog, Rana heckscheri (Anura: Ranidae), in North Carolina.
Jeffrey C. Beane 69

Wood Ducks, Aix sponsa (Anseriformes: Anatidae), and Blackwater Impoundments in
Southeastern North Carolina. Craig A. Harper, James F. Parnell, and Eric G. Bolen . . .80

Relationship of Mass to Girth in Raccoons, Procyon lotor (Mammalia: Procyonidae),
from West Virginia. Troy A. Ladine 91

A Multiscale Approach to Capture Patterns and Habitat Correlations of Peromyscus leu-
copus (Rodentia: Muridae). Troy A. Ladine and Angela Ladine 99

Methods Used to Survey Shrews (Insectivora: Soricidae) and the Importance of
Forest-Floor Structure. Timothy S. McCay, Joshua Laerm, M. Alex Menzel, and
William M. Ford 110

The Water Shrew, Sorex palustris Richardson (Insectivora: Soricidae), and Its Habitat in
Virginia. John F. Pagels, Leonard A. Smock, and Stephen H. Sklarew 120

Occurrence of the Brook Silverside, Labidesthes sicculus (Atheriniformes:
Atherinidae) in North Carolina. Mary L. Moser, Fred C. Rohde, Rudolf G.
Arndt, and Keith W. Ashley 135

Winter Mortality in the Green Anole, Anolis carolinensis (Lacertilia: Polychridae). Jane
K. Distler, Michael E. Dorcas, J. Whitfield Gibbons, Karen L. Kandl, and Kevin R. Rus-
sell 140

Abundance and Size of Dominant Winter-Immigrating Fish Larvae at Two Inlets Into
Pamlico Sound, North Carolina. William F. Hettler, Jr. 144

First Record of Nutria, Myocastor copyus (Mammalia: Rodentia), in Tennessee. Michael
L. Kennedy and Phyllis K. Kennedy 156

Prey Selection by Five Species of Vespertilionid Bats on Sapelo Island, Georgia. Timothy
C. Carter, Michael A. Menzel, Diane M. Krishon, and Joshua Laerm 158

Miscellany 171