Color Variation among Habitat Types in the Spiny Softshell Turtles
(Trionychidae: Apalone) of Cuatrociénegas, Coahuila, Mexico

SUZANNE E. MCGAUGH

Department of Ecology, Evolution, and Organismal Biology, Iowa State University, Ames, Iowa 50011, USA;

E-mail: smcgaugh@iastate.edu

ABSTRACT.—Ground coloration is highly variable in many reptile species. In turtles, ground color may

correspond well to the background coloration of the environment and can change over time to match new

surroundings in the laboratory. Variable carapace and plastron coloration across three habitat types were

investigated in the Black Softshell Turtle, Apalone spinifera atra, by measuring individual components of the

RGB (Red, Green, Blue) color system. In general, A. s. atra carapaces were darker in turtles from lagoons than

in turtles from playa lakes. Red and green values were significantly different among all pairs of habitat

types, but blue values differed only between the playa lakes and lagoons. Mean color components (RG only)

for each population were significantly correlated with corresponding values for the bottom substrate,

indicating a positive association of carapace and habitat substrate color components. In contrast, plastron

ground color RGB channels showed no significant differences between habitat types and no significant

correlations with substrate RGB. These results suggest that dorsal background matching in A. s. atra may be

responsible for some of the variation in this key taxonomic trait.

The color of an organism is an important
component of many aspects of an organism’s
biology and is often used as a taxonomic
character (Endler, 1990; Brodie and Janzen,
1995; Darst and Cummings, 2006). Yet, an
animal’s coloration may be plastic and can
depend on many different biotic and abiotic
factors (Endler, 1990; Bennett et al., 1994). For
example, overall habitat irradiance is a common
stimulus for physiological color change in
reptiles, presumably as a means to become
more cryptic (Norris and Lowe, 1964; Rosen-
blum, 2005; Rowe et al., 2006a). This physiolog-
ical response occurs rapidly and uses hormonal
signals to expand or contract melanophores in
the dermal layer of skin (Bartley, 1971). Inter-
estingly, laboratory tests indicate that the shells
and skin of turtles, too, can change color to
more closely match dark or light backgrounds,
although this change occurs over weeks or
months (Woolley, 1957; Bartley, 1971; Rowe et
al., 2006b).

Cuatrociénegas in Coahuila, Mexico, offers a
unique ecosystem in which to observe natural
variation in shell color of turtles across three
very different aquatic habitat types (Winokur,
1968). The region’s endemic and endangered
Softshell Turtle, Apalone spinifera atra (atra
meaning black or dark; Webb and Legler,
1960; Lovich et al., 1990, Fritz and Havaš,
2006), is taxonomically defined by its dark
pigmentation, and the validity of A. s. atra as a
full species has been challenged (Smith and
Smith, 1979). Hatchlings of A. s. atra cannot be
differentiated from a lighter conspecific, Apalone

spinifera emoryi (Winokur, 1968), but adults
show marked differences in coloration across
habitats (this study), which could be a result
of genetically based ontogenetic pigmenta-
tion variation among habitats or phenotypic
plasticity in response to substrate color varia-
tion.

Apalone spinifera atra was originally described
as an isolated species (Webb and Legler, 1960),
but canal building in the late 1800s was thought
to have opened the basin hydrologically, result-
ing in opportunities for hybridization between
A. s. atra and A. s. emoryi (Smith and Smith,
1979; Ernst and Barbour, 1992). A recent genetic
evaluation found no differentiation between
morphologically identified A. s. atra individuals
and morphologically identified A. s. emoryi
individuals within and outside the basin
(McGaugh and Janzen, in press). However,
substantial coloration differences exist among
habitats and this phenotypic variation was left
unexplained by the genetic study. Examination
of the color variation of the softshell turtles
across the basin, in relation to background
coloration of habitats (Woolley, 1957; Bartley,
1971; Rowe et al., 2006b), may provide an
important morphological perspective on A. s.
atra’s refuted species delimitation (McGaugh
and Janzen, in press). For simplicity, and
because haplotypes in the genetic study showed
no morphological (i.e., light and dark turtles
shared mitochondrial and nuclear DNA haplo-
types) or geographic grouping, the name A. s.
atra is used to refer to all Apalone in this study,
even though A. s. emoryi morphs are present in

Journal of Herpetology, Vol. 42, No. 2, pp. 347–353, 2008
Copyright 2008 Society for the Study of Amphibians and Reptiles



the basin (Smith and Smith, 1979; McGaugh and
Janzen, in press).

In this study, I evaluated the hypothesis that
background matching could be responsible for
the observed color variation across habitats in
A. s. atra. To evaluate this hypothesis, coloration
of the animals’ carapace and plastron and
locality substrate was measured. Background
matching was expected to be probable if
carapace, but not plastron coloration, was
correlated to locality substrate coloration.
Color was measured with digital photographs
using the RGB system (red, green, blue; Stevens
et al., 2007), the components of which corre-
spond to broad bands of longwave (red),
mediumwave (green), and shortwave (blue)
light (Stevens et al., 2007). Possible influences
of sex, size, and their interactions were also
evaluated.

MATERIALS AND METHODS

Study Site and Field Methodology.—Three main
habitat types, playa lakes (barrial lakes), la-
goons, and a river, were investigated within El
Área de Protección de Flora y Fauna Cuatrocié-
negas, Coahuila, Mexico. This area of high
endemism contains diverse habitats, including
hundreds of small water bodies nestled within
the Chihuahuan desert (Minckley, 1969; Meyer,
1973). Playa lakes are large, shallow lakes
(,1 m deep) with sparse vegetation, large daily
temperature fluctuations, and relatively high
mineral content (Minckley, 1969). Lagoons are
typically deeper lakes (,1–10 m) with abundant
vegetation, including waterlilies (Nymphaea),
muskgrass (Chara), pondweeds (Potamogeton),
and cattails (Typha), and relatively constant
temperatures (Minckley, 1969). Rivers are flow-
ing channels with steep banks that are up to
2.5 ms deep and have vegetation such as
Nymphaea, Chara, Potamogeton, Typha, bladder-
worts (Utricularia), and sedge (Eleocharis) where
the current is slow (Minckley, 1969). Most
aquatic habitats in Cuatrociénegas have clear
water with a visible bottom.

Six localities were examined for A. s. atra, and
localities with the same habitat types (e.g.,
lagoons) were combined in statistical analysis
except for the Pearson’s correlation analysis
(described below). All habitats used in this
study are stable bodies of water of unknown
geologic age. No aboveground aquatic connec-
tions are known between the localities used in
this study. Turtle dispersal between sites may
not be frequent because sites are separated by
distances (5.53–31.58 km) greater than other
species of softshells typically travel terrestrially
(Galois et al., 2002). Sampling for shell colora-
tion included all five drainages of the basin

(Evans, 2005) and spanned 30 days of trapping
(4 June to 5 July 2004).

Turtles were captured in lobster or hoop
traps baited with sardines. Each A. s. atra was
tattooed with a unique pattern on the plastron.
Plastron length was measured with dial calipers
to the nearest millimeter, and sex was deter-
mined by tail length (Webb and Legler, 1960),
with males identified by a much longer tail than
females relative to their body size. All A. s. atra
had plastron lengths over 73 mm. Hatchling
and juvenile turtles were not included in the
analysis. Sixty total individuals were sampled
(Lagoon1: N 5 16, Lagoon2: N 5 6, Lagoon3: N
5 13, River: N 5 10, Playa lake1: N 5 10, Playa
lake2: N 5 5).

Digital Photography and Analysis.—Digital
photographs were taken with a Canon Power
Shot G5 that was positioned directly above the
turtle. Color component measurements from
digital systems have been shown to be positive-
ly correlated with values from spectrometry,
and digital photography has many advantages
over spectrometry for data acquisition in the
field (Rowe et al., 2006a; Stevens et al., 2007). A
level on the tripod ensured that the camera lens
was parallel to the ground. The lens-to-animal
distance measure was not taken. Photos were
taken out of direct sunlight and with the flash
on and white balance off. Each photo was taken
with the highest resolution possible for the
camera (2,592 3 1,944 pixels) and converted to
JPEG at the highest quality compression level
available on the camera.

Digital quantification of color was performed
using Jasc Paint Shop Pro9 (Corel, Eden Prairie,
MN) by viewing the untransformed values of R,
G, and B in the diagnostic histogram. The HSL
(Hue, Saturation, Lightness) system was not
used in the analysis because this system may be
inaccurate (Stevens et al., 2007). In Jasc Paint
Shop Pro9, the histogram displays a distribution
graph of the separate channels of color in an
image and allows the user to analyze the
distribution. A swath of at least 100 square
pixels was selected from the right middle
portion of the carapace and the plastron of A.
s. atra. All photos were analyzed at a resolution
of 70.866 pixels per cm. Therefore, each A. s. atra
sample was $14 mm2. Some A. s. atra had a
blotched pattern on their carapaces, and these
were included in the analysis. Otherwise, their
pigmentation was generally uniform. Care was
taken during analysis only to sample areas with
a clear view of the carapace (i.e., flash glares,
algae growth, mineral deposits, or bite marks on
the animal were excluded from the swath).
Intraindividual color variability was not mea-
sured because of these various obstructions.

348 S. E. MCGAUGH



A grey color standard (paint swatch EE2054C
from Lowe’s Home Improvement Warehouse)
was placed in each photo. Digital quantification
of the paint swatch was achieved by the same
method described for the turtles. Associations
between turtle RGB-values and paint swatch
RGB-values were strong (r . 0.33 and P , 0.002
for RB-values of carapace and RGB-values of
plastron, but correlation coefficient of paint
swatch G-values and carapace G-values was
not significant [r . 0.23, P , 0.08]); therefore,
substantial light variation across photographs
occurred. Thus, standardizing the RGB values
from the turtles by the paint swatch was
necessary, and RGB-values from the paint
swatch were used as a covariate in the statistical
analyses. Some of the inconsistencies of variable
ambient field conditions and camera biases can
be accounted for by providing a common color
swatch in each picture (J. A. Endler, pers.
comm.; M. Stevens, pers. comm.; but for
detailed review, see Stevens et al., 2007).
However, this technique does not remove
inconsistencies associated with nonuniform
brightness across the photograph (J. A. Endler,
pers. comm.; Stevens et al., 2007). To reduce the
effect of these inconsistencies on the overall
analysis, the animal was consistently placed so
that a landscape photo was taken with the
posterior of the animal on the left side of the
photograph and the snout on the right side.
Finally, because only one grey color standard
was used, no linearization (also called gamma
correction) could be achieved; consequently,
dark objects may be estimated as lighter than
they are, and light objects may be estimated as
darker than they actually are (see Stevens et al.,
2007:fig. 6). Fortunately, any such biases render
the comparisons in this study conservative.

Past studies that have evaluated the relation-
ship between carapace color and habitat type
have qualitatively described the habitats as
‘‘dark’’ or ‘‘light’’ bottomed (e.g., Rowe et al.,
2006a). In this study, lagoons are dark bot-
tomed, playa lakes are light bottomed, and the
river was intermediate. To provide a more
quantitative measure of this habitat descriptor,
one wet substrate sample of approximately
200 ml in volume was taken from the bottom
of each site and was photographed and mea-
sured for RGB-values in the same manner as the
turtles. Substrate color at each site appeared
relatively uniform although this assumption
was not explicitly tested. Vegetation was not
included. However, lagoons were the only
aquatic habitats with substantial submerged
vegetation, and bare substrate makes up large
portions of the lagoon bottom (Webb and
Legler, 1960).

Statistical Analysis.—To test for habitat struc-
turing in color components, data were trans-
formed for normality (verified through Shapiro-
Wilks tests). All carapace components were log
transformed, and all paint swatch components
from the carapace photographs were raised to
the three-quarters power to achieve normality.
Plastron R and paint swatch R from the plastron
photographs were raised to the three-quarters
power to achieve normality, and other plastron
components and paint swatch components from
plastron photographs were normal (all verified
through Shapiro-Wilks tests). Each response
variable (R, G, or B from plastron or carapace)
was analyzed with analysis of covariance
(ANCOVA), using the respective paint swatch
component as the covariate and plastron length,
sex, habitat type, and interaction terms between
all combinations of these factors as fixed factors.
All factors and interaction terms that were not
significant were removed from the model, and
the ANCOVA was rerun until only significant
factors remained. The only significant interac-
tion that was detected was between habitat
type, sex, and plastron length for blue carapace
component (F2,48 5 6.17, P , 0.0038). To
incorporate this interaction term in the model
for the blue carapace color channel, all nonsig-
nificant factors and lower-order interaction
terms were left in this model. Sex was not a
significant factor for plastron or carapace RGB-
values (P . 0.09 for all ANCOVAs) and, thus,
was removed from all models, except the blue
carapace component where it was previously
explained to be important for a significant
interaction term. Plastron length was not a
significant factor for carapace color components
(P . 0.22, F1, 48 , 0.151 for all ANCOVAs) but
did show a significant, or nearly significant,
positive association with plastron color compo-
nents (R: F1, 54 5 2.82, P , 0.10; G: F1, 54 5 13.86,
P , 0.0005; B: F1, 54 5 10.36, P , 0.00022).
Therefore, for carapace color components, the
model consisted only of the color standard as a
covariate and habitat type as a factor. However,
for the blue color component as mentioned
above and for plastron color components, the
model consisted only of the color standard as a
covariate and plastron length and habitat type
as factors. This statistical analysis resulted in
some of the models being inconsistent with
others. Several additional analyses were done to
ensure that results were not an artifact of the
model. The blue color component and plastron
components were analyzed using the most basic
model for the other two carapace components
(e.g., color standard as covariate and habitat
type only), and overall results did not change.
The alternative strategy, including all nonsig-
nificant factors in the analysis of carapace red

COLOR VARIATION IN APALONE 349



and green components, resulted in important
differences across pairwise habitat types being
missed because the model was unnecessarily
overparameterized with nonsignificant terms.

Least-squares (LS) means t-tests, a method
used to assess the significance of differences
among the best linear-unbiased estimates of the
habitat means for the model design, were used
to determine which habitat types were signifi-
cantly different for certain color components.
Among-site substrate samples were compared
using LS means t-tests. Finally, Pearson’s
product-moment correlations were used to
assess the relationship between LS mean for
carapace and substrate sample RGB-values and
plastron and substrate sample RGB-values for
each locality. All statistics were performed in R
2.4.0 (R Development Core Team, 2006) and
JMP 6.0.2 (SAS Institute Inc., Cary, NC, 2006).

RESULTS

Carapace and Plastron Color Variation.—For A.
s. atra, there were significant differences among
habitat types for all carapace color components
(R: F2, 59 5 22.35, P , 0.001; G: F2, 59 5 32.60, P ,
0.001; B: F2, 58 5 7.37, P , 0.002). Red and green
color channels increased significantly from
lagoons to the river to the playa lakes (lagoons
to river: R: t57 5 2.91, P , 0.003; G: t57 5 2.37, P
, 0.011; lagoons to playa lakes R: t57 5 6.57, P ,
0.001; G: t57 5 8.07, P , 0.001; playa lakes to
river: R: t57 5 2.50, P , 0.015; G: t57 5 4.02, P ,
0.002). The B color channel was significantly
greater, or nearly so, in playa lakes than in the
lagoons and the river (lagoons to river: t48 5
0.694, P , 0.49; lagoons to playa lakes: t48 5
3.84, P , 0.001; playa lakes to river: t48 5 1.63, P
5 0.054). Consistently higher RGB values
suggest that turtles from playa lakes are overall
lighter than those from lagoons. No significant
habitat structuring of plastron color compo-
nents was observed (F2, 54 , 1.90, P . 0.16 in all
cases). Overall, results of the RGB analyses
suggest that carapace color components differ
significantly among most habitat types, whereas
plastron color components do not.

Bottom Substrate Color Variation and Relation-
ship between Substrate and Shell Colors.—Sub-
strate samples showed the same trend as the
turtle RGB-values and increased from lagoons
to the river to playa lakes. Playa lakes R and G
substrate values were significantly higher than
in lagoons (playa lakes vs. lagoons: R: t3 5 4.24,
P , 0.012; G: t3 5 3.67, P , 0.017, playa lakes vs.
river: R: t3 5 2.82, P , 0.066; G: t3 5 2.39, P ,
0.096). No significant differences existed in the
blue color channel (F2, 5 5 4.785, P , 0.12).
Correlations of turtle carapace RG-values and
substrate RG-values were strongly positive and

significant (R: r 5 0.81, P , 0.026; G: r 5 0.74, P
, 0.045). Associations for turtle carapace B and
substrate B-values were positive but not signif-
icant (B: r 5 0.39, P , 0.22). All associations of
plastron RGB components and substrate RGB
components were negative, but no significant
correlations were detected (R: r 5 20.35, P ,
0.50; G: r 5 20.51, P , 0.15; B: r 5 20.69, P ,
0.065).

DISCUSSION

Two conclusions can be drawn from the
statistical analyses performed in this study: (1)
carapace and substrate sample R and G color
components differ among habitat type compar-
isons and are significantly, positively correlated
to each other; and (2) plastron ground color
components do not differ significantly across
habitat types and do not significantly correlate
with substrate sample color components.

The color components varied across habitats
in different ways. In particular, the short
wavelength reflectance (B) of the turtles showed
a weaker trend of habitat structuring than the
long (R) and medium (G) wavelength light
(Table 1). Short wavelengths of light are often
absorbed by dissolved organic matter in the
water and are not available for illumination of
underwater objects (Markager and Vincent,
2000). Alternatively, blue is often conspicuous
in water ,2 m deep (Maan et al., 2006); hence,
may be constrained to maintain crypsis. There-
fore, ambient lighting conditions underwater
could eliminate this color channel’s relative
importance to potential cryptic coloration or
constrain the plasticity of this color channel to
match the surroundings. Although more work
remains to be done on the photic environment
experienced by these turtles, either of these
explanations may be consistent with an adap-
tive explanation for carapace color variation.

My results support a hypothesis of dorsal
background matching for A. s. atra. Although
not all comparisons were significant, the RGB
color components of the sampled localities
generally revealed an increase in carapace
ground color darkness from playa lakes to the
river to lagoons (Figs. 1, 2; Table 1), and the
correlations of R and G with substrate values
were positive and significant. In contrast,
plastron ground color did not show any
significant habitat type structuring in RGB-
values or significant correlation with substrate
RGB-values. If these observations of general
carapace darkening were the result of a passive
staining process from abiotic forces instead of
background matching, plastron and carapace
color should have changed in the same direction
(Rowe et al., 2006a). Further, background

350 S. E. MCGAUGH



matching has been noted for a wide variety of
reptiles (e.g., reviewed by Norris and Lowe,
1964; Cooper and Greenberg, 1992; Rowe et al.,
2006b), and the pattern observed here (i.e.,
dorsal, but not ventral, matching) is consistent
with expectations from a cryptic coloration
mechanism. It is unknown, and would be
interesting to investigate, whether the observed
phenotypic variability is a result of short-term
phenotypic plasticity such as that seen in
laboratory settings (Bartley, 1971; Ernst et al.,

1994) or is a result of a fixed local adaption
(Rosenblum et al., 2004).

The habitat structuring of carapace RGB-
values and plastron pigmentation in A. s. atra
is important in a taxonomic context. Apalone
spinifera atra is predominantly defined by dark
dorsal coloration (Webb and Legler, 1960; Smith
and Smith 1979; Lovich et al., 1990). Winokur
(1968) reported that darker, A. s. atra–like
animals were mainly in lagoons and that lighter,
A. s. emoryi–like specimens resided in rivers (no
playa lakes were examined in his analysis). He
then hypothesized that ecological preferences
(i.e., lagoons for the darker A. s. atra, rivers for
the lighter A. s. emoryi) potentially were in-
volved (Winokur, 1968). Here, I suggest that
background color matching, whether transient
and plastic or fixed and locally adaptive,
explains much of the phenotypic variation
among Apalone in the Cuatrociénegas basin.
This hypothesis is supported by the correlation
of carapace but not plastron, color components
to substrate samples and experimental evidence
from an earlier study that showed that dorsal
background matching occurs in the laboratory
in Apalone (Bartley, 1971; Ernst et al., 1994).

The DNA evidence of the companion study
(McGaugh and Janzen, in press), supports the
decision of Fritz and Havaš (2006) to demote A.
s. atra from species to subspecies rank but leaves
unexplained the morphological diversity seen
in this species across the basin. The preliminary
morphological data presented here suggests
that background matching may explain the
phenotypic diversity seen across habitats in
the basin. However, the mechanism by which
this substantial morphological variation is
achieved remains enigmatic. Additional work,

TABLE 1. Statistics on RGB-values of carapaces for Apalone spinifera atra in Cuatro Ciénegas, Mexico. Carapace
coloration across three habitat types was analyzed by ANCOVA and LS means t-tests. Significant comparisons
are denoted by an asterisk. Raw data were log transformed (carapace) and taken to the three-fourths power
(color standard for carapace photographs) for all A. s. atra. Data from soil samples and their color standards
were not transformed. Abbreviations are R 5 red (long wavelength), G 5 green (medium wavelength), and B 5
blue (short wavelength).

Habitat Soil Apalone

Lagoon LS mean LS mean Playa River
R 69.24 1.936 , 0.001* , 0.003*
G 68.34 1.884 , 0.001* , 0.011*
B 64.68 1.781 , 0.001* 0.490
Playa Lake
R 187.22 2.149 – , 0.015*
G 172.56 2.109 – , 0.002*
B 156.72 1.939 – 0.054
River
R 78.00 2.043 , 0.015* –
G 80.00 1.960 , 0.002* –
B 81.00 1.773 0.054 –

FIG. 1. Boxplots of carapace RGB-values for Apa-
lone spinifera atra in different habitats in Cuatrociéne-
gas, Coahuila, Mexico. The box indicates quartiles,
and the median is indicated by the heavy line within
the box. The lines illustrate points falling within 1.5
times the box size. Outliers, which were more extreme
than the lines, are not shown.

COLOR VARIATION IN APALONE 351



such as a reciprocal transplant experiment, is
needed to determine whether the pigmentation
variation is plastic in the field, as is seen in the
lab (Bartley, 1971; Ernst et al., 1994), or could
potentially be a result of fixed genetic differ-
ences between populations.

Acknowledgments.—K. Lundquist helped dig-
itize the images. R. Jeppesen, N. Hernandez, E.
Bonnell, C. McKinney, J. Howeth, J. Siegrist, and
D. Hendrickson provided valuable field assis-
tance. Special thanks to J. Endler, M. Stevens,
and D. Adams for help with the evaluation of
color parameters; F. Janzen, J. Tucker, and the
Janzen Lab for useful comments on the manu-
script; and R. Rapp for valuable discussion.
Support was provided by Chelonian Research
Foundation’s Linnaeus Fund Research Grants,
Iowa State University’s Professional Advance-
ment Grant in nondissertation research, and the
Gaige Fund of the American Society of Ichthy-
ologists and Herpetologists. Work was done
under permit DAN00739. Field methods were
approved by the Committee on Animal Care
from Iowa State University (Protocol 5-03-5442-
J). SEM was supported by a Graduate Research
Fellowship from the National Science Founda-
tion and by National Science Foundation IBN-
0212935 to F. Janzen.

LITERATURE CITED

BARTLEY, J. A. 1971. A histological and hormonal
analysis of physiological and morphological chro-
matophore responses in the soft-shelled turtle
Trionyx sp. Journal of Zoology 163:125–144.

BENNETT, A. T. G., I. C. CUTHILL, AND K. J. NORRIS. 1994.
Sexual selection and mismeasure of color. Amer-
ican Naturalist 144:848–860.

BRODIE, E. D., III, AND F. J. JANZEN. 1995. Experimental
studies of coral snake mimicry: generalized avoid-
ance of banded snake patterns by free-ranging
avian predators. Functional Ecology 9:186–190.

COOPER, W. E., AND N. GREENBERG. 1992. Reptilian
coloration and behavior. In C. Gans and D. Crews
(eds.), Biology of the Reptilia. Volume 18,
pp. 298–422. Smithsonian Institution Press, Wash-
ington, DC.

DARST, C. R., AND M. E. CUMMINGS. 2006. Predator
learning favours mimicry of a less-toxic model in
poison frogs. Nature 440:208–211.

ENDLER, J. A. 1990. On the measurement and classifi-
cation of color in studies of animal colour patterns.
Biological Journal of the Linnean Society 41:
315–352.

ERNST, C. H., AND R. W. BARBOUR. 1992. Turtles of the
World. Smithsonian Institution Press, Washington,
DC.

ERNST, C. H., J. E. LOVICH, AND R. W. BARBOUR. 1994.
Turtles of the United States and Canada. Smithso-
nian Institution Press, Washington, DC.

EVANS, S. B. 2005. Using Chemical Data to Define Flow
Systems in Cuatro Ciénegas, Coahuila, Mexico. Un-
publ. master’s thesis, University of Texas, Austin.

FRITZ, U., AND P. HAVAŠ. 2006. Checklist of chelonians
of the world. Vertebrate Zoology 57:149–368.

GALOIS, P., M. LÉVEILLÉ, L. BOUTHILLIER, C. DAIGLE, AND S.
PARREN. 2002. Movement patterns, activity, and
home range of the Eastern Spiny Softshell Turtle
(Apalone spinifera) in Northern Lake Champlain,

FIG. 2. Pictorial illustration of Apalone spinifera atra
(above) from the three habitat types described in this
study. Samples are representative of the habitat type
and are not outliers.

352 S. E. MCGAUGH



Québec, Vermont. Journal of Herpetology 36:
402–411.

LOVICH, J. E., W. R. GARSTKA, AND C. J. MCCOY. 1990.
The development and significance of melanism in
the Slider Turtle. In J. W. Gibbons (ed.), Life
History and Ecology of the Slider Turtle,
pp. 233–254. Smithsonian Institution Press, Wash-
ington, DC.

MAAN, M. E., K. D. HOFKER, J. J. M. VAN ALPHEN, AND O.
SEEHAUSEN. 2006. Sensory drive in cichlid specia-
tion. American Naturalist 167:947–954.

MARKAGER, S., AND W. F. VINCENT. 2000. Spectral light
attenuation and the absorption of UV and blue
light in natural waters. Limnology and Oceanog-
raphy 45:642–650.

MCGAUGH, S. E., AND F. J. JANZEN. In press. The status
of Apalone atra populations in Cuatrociénegas,
Coahuila, Mexico: preliminary data. Chelonian
Conservation and Biology.

MEYER, E. R. 1973. Late-Quaternary paleoecology of
the Cuatro Ciénegas basin, Coahuila, México.
Ecology 54:983–995.

MINCKLEY, W. L. 1969. Environments of the Bolsón of
Cuatro Ciénegas, Coahuila, Mexico, with special
reference to the aquatic biota. Texas Western Press,
El Paso.

NORRIS, K. S., AND C. H. LOWE. 1964. An analysis of
background color matching in amphibians and
reptiles. Ecology 45:565–580.

R DEVELOPMENT CORE TEAM. 2006. R: A Language and
Environment for Statistical Computing [Internet].
R Foundation for Statistical Computing, Vienna,
Austria, Available from: http://www.R-project.
org. Accessed 10 March 2006.

ROSENBLUM, E. B. 2005. The role of phenotypic
plasticity in color variation of Tularosa Basin
lizards. Copeia 2005:586–596.

ROSENBLUM, E. B., H. E. HOEKSTRA, AND M. W. NACHMAN.
2004. Adaptive reptile color variation and the
e v o l u t i o n o f t h e M C 1 R g e n e . E v o l u t i o n
58:1794–1808.

ROWE, J. W., D. L. CLARK, AND M. PORTER. 2006a. Shell
color variation of Midland Painted Turtles (Chrys-
emys picta marginata) living in habitats with
variable substrate colors. Herpetological Review
37:293–298.

ROWE, J. W., D. L. CLARK, C. RYAN, AND J. K. TUCKER.
2006b. Effect of substrate color on pigmentation
in Midland Painted Turtles (Chrysemys picta
m a r g i n a t a ) a n d R e d - E a r e d S l i d e r T u r t l e s
(Trachemys scripta elegans). Journal of Herpetology
40:358–364.

SMITH, H. M., AND R. B. SMITH. 1979. Guide to Mexican
Turtles Bibliographic Addendum III, Synopsis of
the Herpetofauna of Mexico, Volume 6. John
Johnson, North Benington, VT.

STEVENS, M., C. A. PÁRRAGA, I. C. CUTHILL, J. C.
PARTRIDGE, AND T. S. TROSCIANKO. 2007. Using digital
photography to study animal coloration. Biological
Journal of the Linnean Society 90:211–237.

WEBB, R. G., AND J. M. LEGLER. 1960. A new soft-
shell turtle (genus Trionyx) from Coahuila,
Mexico. University of Kansas Science Bulletin
40:21–30.

WINOKUR, R. M. 1968. The Morphology and Relation-
ships of the Soft-Shelled Turtles of the Cuatrocié-
negas Basin, Coahuila, Mexico. Unpubl. master’s
thesis. Arizona State University, Tempe.

WOOLLEY, P. 1957. Colour change in a Chelonian.
Nature 179:1255–1256.

Accepted: 9 December 2007.

COLOR VARIATION IN APALONE 353