Geographic Variation In The Frog Genus Vanzolinius (Anura: Leptodactylidae)

28 Pages·1997·13.9 MB·English

by  W R Heyer

PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
110(3):338-365. 1997.
Geographic variation in the frog genus Vanzolinius
(Anura: Leptodactylidae)
W. Ronald Heyer
Department of Vertebrate Zoology, National Museum of Natural History, Smithsonian Institution,
Washington, D.C. 20560, U.S.A.
—
Abstract. Vanzolinius discodactylus, a forest-dwelling frog species of west-
ern Amazonia in South America, varies in characters of color pattern, mor-
phology, and advertisement call. Analysis of this variation indicates that very
local (site) differentiation results in mosaic patterns of differentiation, largely
obfuscating larger geographic patterns. Comparison of available genetic esti-
mates of differentiation for V. discodactylus are consistent with the morpho-
A
logically and advertisement call-based conclusions. previously studied forest-
dwelling lizard and another forest-dwelling frog also demonstrate local differ-
entiation patterns suggesting that the variation in V. discodactylus may repre-
sent a general pattern for forest-dwelling amphibians and reptiles in Amazonia.
—
During examination of specimens for a straight; juvenile condition of gonads in-
study of Leptodactylus species (Heyer determinate (in some cases, gonads had
1994), several Vanzolinius specimens were been removed by previous workers).
encountered. Dr. Claude Gascon found Analyses differ depending on the type of
Vanzolinius to be relatively common along data gathered for the characters examined.
the Rio Jurua in Brazil and used the species The following descriptions of characters are
to test the riverine barrier hypothesis (Gas- arranged by analytical groups.
A
con et al. 1996). cursory examination of Color patterns and external morpholo—g-
these additional materials suggested that ical features of adult form individuals.
there was considerable variation, which These qualitative traits are categories re-
might profitably be studied. The purpose of corded as either binary or multistate char-
this paper is to analyze geographic variation acters. For the latter, states were added to
in Vanzolinius. the series as they were encountered during
the data-taking phase. The states within
Materials and Methods each series have no intended or implied re-
lationships or transformation series. Inter-
As many adults, larvae, and recordings mediate conditions between states were re-
of advertisement calls as possible were as- corded with the first letter of the state that
sembled from major museum collections most nearly approached the condition ob-
(Appendix 1). served in the specimen examined.
The sex of individuals was determined Dorsal snout pattern: Three basic states
either by examination of vocal slits, or dis- were encountered: a relatively uniform dark
section to examine gonads. The following pattern (Fig. 1A); a variegated pattern (Fig.
—
categories are used: adu—lt male vocal slits IB); and a uniform light pattern (Fig. 1C).
A
present; juvenile male testes present, but Light postorbital eye stripe: series of
vocal—slits not broken through; adult fe- symbols define the distinctiveness of the
male oviduc—t folded at least in part; ju- postorbital eye stripes: — [absent]; (+) [in-
venile female ovaries present, but oviduct distinct]; + [distinct]; +! [sharply defined].
!
VOLUME NUMBER
110, 3 339
Fig. 1. Dorsal snout pattern standards.
In cases where the two sides of the head eyes on f—ull extent of dorsum (Fig. 2D);
differed, both conditions were recorded. State D-l as previous state, except sides
Light subocular bar: The distinctiveness irregular.
of the bar was noted by the same symbols Dark mid-dorsal pin stripe: An interrupt-
as for the previous character, except the + ed dark mid-dorsal pin stripe was recorded
category was not encountered. as either present or absent.
—
Dorsal pattern: Dorsal pattern variation Throat and chest pattern. Variation in
forms a continuum among the more dis- this character is continuous—among the
tinctive states reco—rded. The states recog- states encountered: State A— variegated
A
nized are: State either an uniform dor- pattern (Fig. 3A); State A-l as previous
—
sum (brown or tan) or in—determinately state, but light; State B —uniform light pat-
blotched (Fig. 2A); State B the dorsum tern (Fig. 3B); State B-l as previou—s state,
with very distinct dark markings in the in- but lateral portions darker; State C dark
—
terorbital and interscapular areas (Fig. 2B); speckled pattern (Fig. 3C); State C-l as
—
State C distinct interorbital blotch, well previous state, bu—t dark spotting more ex-
D
defined chevron markings anteriorly and tensive;—State dark pattern (Fig. 3D);
blotches posteriorly on the body (Fig. 2C); State E dark pattern with light spots (Fig.
—
State C-l as previous state except chev- 3E).
—
D
rons continuous; State a distinct darker Belly pattern: Variation in this character
straight edged band extending from behind is continuous among the states encountered:
Fig. 2. Dorsal pattern standards.
—
340 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
Fig. 3. Throat and chest pattern standards.
—
A
State speckled pattern (Fig. 4A); State Posterior thigh pattern: Variation in this
—
A-—l almost uniform cream pattern; State character is continuou—s among the states en-
B indistinctly mottled, more intense an- countered: State A indistinctly mottled
—
teriorly (Fig. 4B); State B-l, as previous (Fig. 5A); State B indistinctly mottled
—
state, but lighter; State C distinctly mot- with indistinc—t dark longitudinal band (Fig.
tled, rather uniform over belly (Fig. 4C); 5B); State C distinctly mottled (Fig. 5C);
— —
D
State C-l—as previous state, but lighter; State speckled with indistinc—t dark lon-
State C-2 as State C but dark anteriorly gitudinal band (Fig. 5D); State E speckled
D
and no melanophores posteriorly; State with distinct dark longitudinal band (Fig.
—
distinctly variegated dark and light pattern 5E); State F speckled with dark longitu-
(Fig. 4D). dinal band bordered above by light longi-
VOLUME NUMBER
110, 3 341
Fig. 4. Belly pattern standards.
—
tudinal stripe (Fig. 5F); State F! as for Male secondary sexual characters: All
State F except light stripe very distinct. males lack secondary sexual characters of
Outer tarsal pattern: Data were taken on thumb or chest pads or spines or male arm
the distinctiveness of the outer tarsal pattern hypertrophy as found in Leptodactylus.
relative to the dorsal tarsal pattern. How- Male vocal sac: Variation in this char-
ever, variation turned out to be minimal and acter is minimal and difficult to evaluate in
scoring could not be done consistently. terms of preservation artifact. In most
These data are not analyzed further. males, the vocal sac is single and internal;
Dorsolateral fold condition: There is rel- in a few males, the single vocal sac has ex-
atively little variation in this character and ternal indications of weak lateral folds.
the variation that exists is difficult to eval- Variation of this character is not analyzed
uate in terms of the impact preservation has further.
on recognition of fold condition. Most in- Textures: Data were taken on textures of
dividuals have no dorsolateral folds. In a the dorsum, the upper shank, the outer tar-
few individuals, a short ridge or elongated sus, and sole of foot. In all cases the degree
warts lie in the dorsolateral fold region pos- of development of shagreen and tubercles
terior to the eye. The variation in this char- was difficult to categorize consistently and
acter is not analyzed further. differentiate from preservation artifact. The
Fig. 5. Posterior thigh pattern standards.
342 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
75 individuals, the third toe dorsal outline
is perceptibly more expanded than the
fourth. As the third toe tip is a bit more
expanded in the total sample, data are an-
alyzed only for the third toe tip.
Dorsal toe grooves: The dorsal surfaces
of the toes typically have from 2-5 grooves
involving the epidermis and dermis. The
grooves are almost parallel to each other
Fig. 6. Digit tip dorsal outline standards.
along the long axis of the toe, but radiate
slightly outward from the proximal toe tip
dorsum is tuberculate, with either small to the distal tip. Usually the grooves extend
black-tipped or white-tipped tubercles, usu- almost the entire length of the expanded
ally more densely packed posteriorly. The portion of the toe tip, but the grooves do
dorsum may also be somewhat granular, not reach the tip of the toe. Counting the
have a shagreen, or be smooth. The upper exact number of grooves is not always pre-
shank and outer tarsus consistently have tu- cise as the grooves are sometimes incom-
bercles, black and/or white tipped, and the plete and preservation artifact can obscure
surfaces may also be shagreened. The foot the definition of the grooves. Data were
is either smooth or the outer margin has a taken for both the third and fourth toes. As
few black/white tipped tubercles and/or a for the dorsal outline, the conditions within
shagreen. Variation for these characters is individuals are similar, but typically the
not analyzed further. third toe has one more groove than the
Finger tip dorsal outline: For both finger fourth. For 96 individuals, the third and
and toe tip shapes, the outline shapes stan- fourth toes have the same number of
dardized by Savage (1987) for Eleuthero- grooves, for 18 individuals, the fourth toe
dactylus were used. For both finger and toe has more grooves than the third, and for
tip dorsal outlines, only three of Savage's 132 individuals the third toe has more
(1987) shapes were encountered corre- grooves than the fourth. Because the raw
sponding to his unexpanded even (Fig. 6A), data indicate that the variation in the fourth
expanded even (Fig. 6B), and expanded toe mirrors that seen in the third, only the
pointed or lanceolate (Fig. 6C) states. The data for the third toe are analyzed further.
third finger tip is the most expanded and is Analysis of preceding characters: The
the digit from which data were taken. Very preceding characters are recorded as dis-
little variation was encountered, suggesting crete entities even though variation is main-
that observer variation in interpreting shape ly continuous. Because the data are discrete,
was probably as large as actual variation. chi-square analyses are used to determine
The conditions recorded ranged from not whether occurrence frequencies of states
expanded, just A, A, A-B, or A-C. Variation differ significantly. The 0.05 convention is
is not analyzed further for this character. used to determine significance. Data were
Toe tip dorsal outline: The same stan- adjusted, when necessary, to reach a mini-
mum
dards were used for toe tips as for finger cell size of an average expected fre-
tips (Fig. 6). Data were recorded for both quency of 5 (Hayek 1994:239). Data were
the third and fourth toe tips. The conditions first examined to determine whether states
for the third and fourth toe dorsal outlines for adult males and females differed signif-
are very similar. For 165 individuals, the icantly. If they did not, then data recorded
conditions are identical, for 9 individuals for juveniles were added to both the male
the fourth toe dorsal outline is perceptibly and female data to provide more robust data
more expanded than for the third, and for sets for statistical analysis among geograph-
VOLUME NUMBER
110, 3 343
ic regions (see definitions of regions be- Brazil: Acre; Nova Vida, Rio Jurua,
USNM
low). — Tape 256 Cut 12. Recorded at 1900
Measurement data and analyses. Mea- h on 17 March 1992 by Claude Gascon at
surements were taken on the following vari- a temperature of 25°C, no voucher speci-
ables, as defined by Gascon et al. (1996): men.
snout-vent length (SVL), nostril separation, Brazil: Amazonas; Altamira, Rio Jurua,
USNM
eye width anterior, eye width posterior, Tape 255 Cut 2. Recorded at 1915
head width, head length, eye to nostril dis- h on 17 November 1991 by Claude Gascon
tance, thigh length (=femur length of Gas- at a temperature of 25°C, voucher INPA
con et al. 1996), shank length (=tibia length 5021.
of Gascon et al. 1996), foot length, tym- Brazil: Amazonas; Barro Vermelho, Rio
panum diameter (=tympanum height of Jurua, USNM Tape 254 Cut 5. Recorded at
Gascon et al. 1996), eye length, maximum 1900 h on 27 October 1991 by Claude Gas-
width of third finger, and maximum width con at a temperature of 24.4°C, voucher
of fourth toe. Measurements were taken INPA 3352.
with a Helios dial calipers and recorded to Brazil: Amazonas; Jainii, Rio Jurua,
the nearest 0.1 mm. USNM Tape 254 Cut 13. Recorded at 1740
Only adult specimens are used for the h on 2 November 1991 by Claude Gascon
measurement data analyses. As males and at a temperature of 26.1°C, no voucher
females are sexually dimorphic in size, they specimen.
USNM
are analyzed separately (L. C. Hayek, C. Ecuador: Napo; Limoncocha,
Gascon, and W. R. Heyer (unpublished Tape 18 Cut 1. Recorded at 2000-2034 h
data) discuss multivariate analyses on mor- on 9 July 1971 by Ronald Heyer at an air
phometric data on Vanzolinius.) The data temperature of 23.4°C, water temperature
LACM
are analyzed using the software program 23.6°C, voucher 92001.
SYSTAT 5 (Wil—kinson et al. 1992). Calls were analyzed using Canary 1.2
Larval data. To my knowledge, the software (Charif et al. 1995) on a Power
only larvae available are those reported on Macintosh 8500 computer. Calls were dig-
by Duellman (1978) from a single locality itized for analysis at a sample rate of 22050
in Ecuador (Mera, Pastaza). Specimens KU Hz and a sample size of 16 bits. Call rate
121362-121363 are larvae ranging from was determined directly from recordings for
KU
Gosner stages 30-38. Specimens periods ranging from 45 to 180 s per re-
121360-121361 are just metamorphosed cording. Other call parameters were taken
individuals. There is no internal evidence from a combination of waveform, audios-
from study of these specimens to either es- pectrogram (=spectrogram as used in Ca-
tablish that they are Vanzolinius or they are nary manual), and spectrum analyses based
not. Dr. John Lynch collected the specimens on ten calls for each individual. Most of the
and informs me (pers. comm.) after he con- recordings had considerable noise. Many
sulted his field notes, "... it appears that parameters were taken from filtered calls.
I guessed on the identification ... on the The filter around option was used for de-
USNM
basis of habitat selection. Hence, don't trust termining some parameters for Tape
the identification." As nothing can be de- 256 Cut 12 (filtered around 520-5000 Hz),
USNM
termined about geographic variation based Tape 255 Cut 2 (filtered around
USNM
on these specimens (even assuming they are 500-5000 Hz), Tape 254 Cut 5 (fil-
USNM
Vanzolinius discodactylus), larvae are not tered around 330-4000 Hz), and
treated further in this paper. — Tape 18 Cut 1 (filtered around 500-4500
Advertisement call data and analyses. Hz).
Recordings of single individuals from five Definition ofgeographic areas for anal-
—
localities are available for analysis. ysis. The primary purpose of this study is
344 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
to determine the nature of geographic vari- As there are but five individuals available
ation found within Vanzolinius discodacty- from Region D, (1 female, 1 male, 3 juve-
lus. Sample sizes are insufficient to analyze niles), Region D data are omitted from the
the data from each locality independently. analyses by areas, unless otherwise noted.
Localities were plotted on a map and lo-
calities were grouped on the basis of geo-
Results
graphic proximity (Fig. 7). The rationale —
used for grouping localities involved trying Dorsal snoutpattern. This was the only
to maximize three criteria simultaneously: character for which some individuals clear-
to have as many groups as possible in order ly demonstrated two states (this exception
to characterize geographic variation; to involved only conditions B and C both oc-
have as many individuals as possible in curring in the same individual), suggesting
each group to permit robust statistical anal- partial independent genetic control of this
yses; and to maintain geographic integrity. character. The states (Appendix 2) were
With respect to geographic integrity, a rule collapsed for analysis to three: (1) pure A,
of thumb of keeping localities within the (2) any B, and (3) any C. Because several
same major river drainage basins was gen- individuals had both states (2) and (3), the
erally applied (Areas A, B, C, D, F, G, H), total number of state conditions analyzed
but not exclusively so (Area E). Initially 10 exceeds the number of individuals exam-
geographic area groupings were made. ined for this character only.
When the data were examined for these The chi-square analysis by sex was not
groupings to see if they were sufficient, significant (x2 = 1.35; df = 2; P = 0.50 >
three ofthe groups lacked sufficient data for 0.30). Thus, male, female, andjuvenile data
analysis. Two Colombian localities were were combined to analyze by geographic
sufficiently isolated from each other as well area. The chi-square analysis by geographic
as other samples to be placed in their own area is significant (x2 — 69.92; df = 12; P
groups; unfortunately, the specimens from < 0.001). In partitioning the results, regions
both localities are faded such that data are A+B are distinct from regions E+F+G+H.
incomplete for them. Thus, data for the Co- Region C is not distinct from either of the
lombian localities of Caldas; Villa Maria other two area groupings. —
and Caqueta; Florencia, are not included in Light postorbital eye stripe. For indi-
the geographic analyses (Fig. 7, upper two viduals having different states on either side
A
squares). single Peruvian locality (Uca- of the head, the more distinctive state was
yali, Yarinacocha) also formed a distinct scored for statistical analysis (e.g., an in-
geographic group by itself and contains one dividual recorded as having state (+) on
faded specimen, unsuitable for further geo- one side of the head and + on the other
graphic analysis (Fig. 7, lower square). was treated as having the + state for statis-
Eight geographic groupings remain and are tical analysis). The chi-square analysis by
identified by letter in further discussion: (A) sex was significant (x2 = 6.43; df = 2; P =
northern Amazonian Ecuador; (B) southern 0.05 > 0.02); therefore the variation among
Amazonian Ecuador; (C) Amazonian Peru; geographic areas has to be analyzed sepa-
(D) the Brazil-Colombia border region; (E) rately by sex. For females, the chi-square
easternmost known localities for Vanzoli- analysis by geographic area is significant
nius in Amazonian Brazil; (F) the mid-re- (X2 = 42.43; df = 12; P < 0.001). In par-
B+E
gion of the Rio Jurua of Brazil; (G) the up- titioning the significance, regions are
A+C+F+G+H.
per region of the Rio Jurua in the Brazilian distinct from regions In
State of Amazonas; and (H) the upper re- order to meet the minimum expected cell
gion of the Rio Jurua in the Brazilian State size criterion for statistical robustness for
of Acre. males, the data had to be collapsed by rec-
VOLUME NUMBER
110, 3 345
Fig. 7. Map of northwestern South America showing known distribution of Vanzolinius discodactylus. Guy-
ana and Brazil are truncated by the 60°W meridian; Bolivia and Brazil by the 15°S parallel. Squares are single
localities excluded from analysis of geographic variation. Circles and ellipses labelled A-H indicate groupings
of localities (dots) used for analysis of geographic variation. A dot may represent more than one locality.
346 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
ognizing but two character states: absent (— C+E+F+G. For males, the chi-square
state) or present (combined (+), +, + analysis of character states by area is not
!
states). The chi-square results are signifi- significant (x2 = 9.43; df = 6; P = 0.20 >
cant (x2 = 27.80; df = 6; P < 0.001). In 0.10). —
partitioning the results, the significance is Throat and chest pattern. As for dorsal
due entirely to the distinctiveness of the pattern, the first state recorded for inter-
Area B specimens. — mediate conditions is scored for the statis-
Light subocular bar. For individuals tical analyses. State E was scored for only
having different states on either side of the one individual and is not considered in the
head, the more distinctive state was scored analyses. As only two known-sex individ-
for statistical analyses. The chi-square anal- uals have State B-l, and only one known-
ysis by sex was significant (x2 — 6.85; df sex individual has State C-l, State B-l is
= 2; P = 0.05 > 0.02). In order to meet combined with State B and State C-l is
the assumption of minimum expected cell combined with State C in the analyses. The
H
size for females, the data for Area were chi-square analysis by sex is significant (x2
deleted. The resultant chi-square analysis is = 23.20; df = 4; P < 0.001). To satisfy the
not significant (x2 = 15.49; df = 10; P = minimum expected cell frequency assump-
0.20 > 0.10). To meet the minimum ex- tion for females, drastic manipulations are
pected cell size for the male data, the states required. Areas B and H are excluded.
A
were collapsed to two: absent (— state) and States A-l and are combined. The resul-
present ((+) and + states). The resultant tant chi-square analysis is significant (x2 =
chi-square analysis is not significant (x2 = 23.39; df=S;P = 0.01 > 0.001). In par-
= P = > E+F
9.59; df 6; —0.20 0.10). titioning the results, only areas are dis-
Dorsal pattern. For statistical analyses, tinct. Drastic manipulations are also re-
the first state recorded for intermediate con- quired to analyze the male variation. States
ditions is used (e.g., a specimen recorded as C and C-l are deleted. Areas C, E, G, and
having condition A—B is scored as having H are deleted. States A-l and A are com-
A
condition for the statistical tests). The bined. The resultant chi-square test is not
chi-square analysis by sex is not significant significant (x2 = 1.90; df = 4; P = 0.90 >
(X2 = 2.93; df = 3; P = 0.50 > 0.30); fe- 0.50). —
male, male, and juvenile data were com- Belly pattern. As for dorsal pattern, the
bined to analyze by geographic area. To first state recorded for intermediate condi-
meet the assumption for minimum expected tions is scored for statistical analyses. To
cell size, States D and D-l are combined; meet the minimum expected cell frequency
D
however, Area is included. The chi- assumption for the analysis by sex, the fol-
square result among areas is significant (x2 lowing states are combined: A and A-l; B
= 203.09; df = 28; P < 0.001). In parti- and B-l; C, C-l, and C-2; State D is delet-
tioning the results the following area group- ed. The resultant chi-square analysis is sig-
ings are distinct from each other: A; B; nificant (x2 = 8.60; df = 2; P = 0.02 >
C+D+E; F+G+H. — 0.01). To analyze females, the same state
Dark mid-dorsal pin stripe. The chi- combinations described above are used and
square analysis by sex is significant (x2 = Areas B and H are deleted to meet the min-
5.86; df = 1; P = 0.02 > 0.01). To meet imum expected cell size assumption. The
the minimum expected cell size criterion for chi-square result is statistically significant
females by area, Area H was deleted. The (X2 = 56.20; df=S;P< 0.001). Partition-
chi-square result is significant (x2 = 11.47; ing the results indicates that the following
df = 5; P = 0.05 > 0.02). Partitioning the area groupings are distinct from each other:
results indicates the following area group- A; C+E; F+G. To meet the minimum ex-
ings are distinct from each other: A+B; pected cell size assumption for males, only
VOLUME NUMBER
110, 3 347
the combined states (as above) from Areas groupings are distinct from each other: C;
B, E, and F could be analyzed. The chi- E; A+B+F+G+H.
—
square results are significant (x2 — 24.43; Measurements. Because only two adult
df = 4; P = 0.001); however, there are too males represent Area G, they are deleted
few areas involved for meaningful parti- from the analyses.
MANOVA
tioning. — analyses were run on the two
Posterior thigh pattern. As for dorsal data sets (male and female). All univariate
pattern, the first state recorded for interme- F tests were statistically significant (P <
diate conditions is scored for statistical anal- 0.001). The multivariate tests are also sig-
yses. As only three individuals have State nificant for both data sets (e.g., Wilks'
F!, it is combined with State F The chi- Lambda for male data = 0.022, F(70) 231 =
square analysis by sex is not significant (x2 4.136, P < 0.001; for female data Wilks'
= 9.95; df = 5; P = 0.10 > 0.05); females, Lambda = 0.025, F = 3.501, P <
(84) 312
males, and juveniles are combined to ana- 0.001). Thus, there is significant variation
lyze character states by geographic areas. of the measurement data among the geo-
The chi-square analysis by area is significant graphic areas.
(X2 = 101.30; df = 30; P < 0.001). Parti- In order to understand the nature of the
tioning the results indicates the following geographic variation, discriminant function
geographic area groupings are distinct from analyses were performed on untransformed
each other: A+B; C; E; F+G; H—. data and are discussed separately for the
Third toe tip dorsal outline. Only one male and female data.
individual was scored with any indication The male data post-classification results
of State C, and it was an intermediate State (Table 1) indicate that sizes and shapes are
B—
C; for statistical analyses, that individ- distinctive for each area, with lesser dis-
ual was scored as having State B. As only tinctiveness in morphologies between Areas
A A
two individuals had the unexpanded state, and B. plot of the first two canonical
that state is combined with the "Just A" scores (Fig. 8) indicates that the Area A and
State for analyses. The chi-square analysis B samples overlap each other to a greater
by sex is not significant (x2 — 6.90; df — extent than any of the other samples. The
4; P = 0.20 > 0.10); females, males, and male measurement data support the follow-
juveniles are combined for analysis of state ing area groupings as distinct from each
frequencies among geographic areas. The other: A+B; C; E; F; H.
chi-square analysis by area is significant (x2 The female data post-classification re-
= 158.78; df = 24; P < 0.001). In parti- sults (Table 2) also indicate that the mor-
tioning the results, the following area phologies are distinctive within each re-
groupings are distinct from each other: A; gion, with less differentiation between Ar-
B+C; E; F+G+H. — eas C and E and among Areas F, G, and H.
Third toe tip dorsal grooves. Because The plot of the first two canonical scores
so few individuals had only one groove (Fig. 9) demonstrates extensive overlap of
(Appendix 2), the individuals are combined specimen data for Areas C and E and Areas
with those having two grooves. The result- F, G, and H. The female measurement data
ing chi-square analysis by sex is not signif- indicate the following area groupings to be
icant (x2 = 0.58; df = 3; P = 0.80 > 0.70); distinct from each other: A; B; C+E;
females, males, and juveniles are combined F+G+H.
for analysis of state frequencies among geo- Comparison of the male and female data
graphic areas. The resultant chi-square anal- underscore that the males and females from
ysis is significant (x2 = 34.90; df = 18; P Area C differ markedly in their association
= 0.01 > 0.001). Partitioning the results in- with the other samples. The Area C males
H
dicates that the following geographic are quite similar to those of Area but dif-