PATTERNS OF GEOGRAPHIC VARIATION IN FLORIDA SNAKES
STEVEN P. CHRISTMAN
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REREQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
During the course of my graduate training at the University of
Florida I have received financial support from the Society of Sigma
Xi, and I held a National Defense Education Act Fellowship for one
year. Funds for computer analysis were made available by the North-
east Regional Data Center. I thank all those people who loaned me
specimens or data for analysis in this study. I especially thank my
wife, Sheila, for financial, moral and physical help in all phases of
my academic work.
TABLE OF CONTENTS
ACKNOWLEDGMENTS. . . . . . . . . ... . . .... ii
ABSTRACT . . . . . . . ... . . . . . . iv
INTRODUCTION . . . . . . . ... ..... .... 1
MATERIALS AND METHODS. . . . . . . . . ... . .. 6
The Problem of Samples and Populations . . . . . . 8
The Mapping Procedure. . . . . . . . ... ... 9
Multivariate Analysis. . . . . . . . ... ... 10
Comparison of Mapped Data. . . . . . . . . .. 10
Geography. . . . . . . . . ... . . . 11
RESULTS. . . . . . . . . ... . . . . .... 14
The Species. . . . . . . . . ... ....... 15
The Patterns . . . . . . . . ... .. .... . 54
The Correlations . . . . . . . . ... ..... 59
DISCUSSION . . . . . . . . ... . . . . . . 331
Pattern of Variation . . . . . . . . ... . 331
Phylogenetic Considerations. . . . . . . . ... 343
SUMMARY. . . . . . . . . ... . . . . .... 371
A CHARACTERS EXAMINED. . . . . . . . . ... . 373
B ENVIRONMENTAL VARIABLES EXAMINED . . . . . . ... 388
LITERATURE CITED . . . . . . . . ... . . .... 389
BIOGRAPHICAL SKETCH. . . . . . . . . ... ..... 394
Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
PATTERNS OF GEOGRAPHIC VARIATION IN FLORIDA SNAKES
Steven P. Christman
Chairman: Archie Carr
Major Department: Zoology
I analyzed geographic variation in fifteen species of Florida
snakes. Machine-produced contour maps were created for each of over
200 morphologic variables and 17 climatic variables. One hundred of
the maps were factor analyzed, and seven major patterns of geographic
variation extracted. These seven patterns were found to account for
over 60% of the information contained in the original contour maps.
Each of the patterns of geographic variation can be explained in terms
of natural selection by past or present environments. Disjunct popu-
lations showing phenetic similarities are the result of an earlier
widespread phenotype followed by differentiation in geographically
intermediate regions. Recourse to land bridge hypotheses and retro-
gressive evolution are not necessary to explain polytopic phenotypes.
Correlations between the patterns of variation and environment are
discussed, but experimental verification of cause and effect relation-
ships are not provided. The geographic localities of primitive character
states and/or primitive species are not the centers of origin for the
groups, but are considered to be areas in which evolution has proceeded
relatively more slowly.
Darwin's theory of natural selection has as its keystone one
important requirement: variation. Implicit in any interpretation of
natural selection as the guiding force behin-d organic evolution is
the assumption that organisms are not all alike. The members of a
population of sexually-reproducing plants or animals are in fact
(except for identical twins) all different, both genetically and
phenotypically. Mutations and recombinations of existing genes insure
the continuance of this variability.' Selection acts on the differences
between organisms by eliminating unfit phenotypes (and hence their
associated genotypes) from the reproductive effort of a population.
However, phenotypes which are unfit in one part of a species' geographic
distribution may be quite fit in another region. Thus geographic
variation in selective pressures, brought about by geographic variation
in environment, predisposes organisms to vary geographically in morpho-
logical, physiological and behavioral traits, even within a single
The process of speciation begins when differential selective
pressures act on populations of a species in remote parts of its geo-
graphic range. Natural selection creates different phenotypes in
response to these different selective regimes. When the degree of
IBut see Murray (1972) for a discussion of genetic diversity
maintained by natural selection.
phenotypic divergence includes also reproductive incompatibility, the
populations involved are said to have reached the species level.
Populations of organisms change phenotypically and genetically
through time as they become better adapted to their environment and as
their environment changes. This process is called evolution. Popu-
lations of organisms also change phenotypically and genetically through
space as they adapt to different environments. This phenomenon has been
called geographic variation, but it is really just another form of
evolution. Albert Einstein has shown the equivalence of space and
time in the physical world. Preston (1960, etc.) has pointed out
analogies between space and time in ecology and species diversity. In
this study, it is assumed that character variation through space is a
form of organic evolution just as character variation through time is
It is currently impossible to study the environmental factors
responsible for character variation through time. We do not have
accurate data on temperature or rainfall variations throughout the
evolutionary history of any species. We are not able to assess the
relative importance of various selective pressures that have brought
about the species of today. However, we can study evolution through
space. We do have accurate environmental measurements taken at many
geographic points, and should be able to correlate these with character
variation as we see it throughout a species' distribution. Assuming
an analogy between space and time, I believe that patterns of character
variation in space are brought about and maintained in the same ways as
patterns of character variation through time (evolution, in the Darwinian
Thus the key to an understanding of organic evolution lies in
an understanding of geographic variation and the environmental factors
responsible for its maintainence.
Although the study of geographic variation in living systems is
not a new one, the use of modern multivariate methods to describe and
compare patterns of variation has hardly begun. Most previous studies
have simply described or illustrated the geographic variation of each
of a set of characters pertaining to a single plant or animal species.
A few very recent studies have utilized multivariate techniques to
analyze the degree and type of covariation between characters of a
single species. Still fewer investigators have attempted to demon-
strate correlations between environmental parameters and intraspecific
geographic variation. No previous study, to my knowledge, has attempted
to quantitatively investigate patterns of geographic variation common
to several species, and to compare these patterns with environmental
variation. The present study does just that.
Florida is for the most part a peninsula with a warm temperate
climate grading into that of a subtropical region. It has such diverse
habitats as large swamps and marshes (prairies), forested pine lands,
and scrubby chaparral-like deserts. It thus seems surprising that most
studies concerned with the biogeography of Florida have concentrated on
historical causation and left little credit to the power of natural
selection as a factor in establishing patterns of geographic variation.
Changing sea levels, "Ocala Islands," Suwannee Straits and the like have
all been cited as influencing the patterns of distribution and variation
in Florida plants and animals. No doubt many of the distributional
patterns and much of the variation seen in Florida organisms have been
influenced by these historical phenomena. However, I believe that most
of the variation within species represents adaptation to the present
environment, brought about by natural selection that is still at work.
Recent work by several authors has shown that demes or micro-geographic
populations can and do differentiate as theybecome adapted to their
own specialized environment. Species, in the conventional sense, are
not panmictic. Rather, they are groups of populations, each adapted to
its own particular portion of the total species distribution. There-
fore, any analysis of geographic variation should include tests for
correlation with components of the environment. Only after all possible
correlations with environmental parameters have been eliminated, should
historical phenomena be suggested as causation in biogeography.
Leon Croizat, in a series of works spanning the last thirty years
(especially 1958, 1962) has advocated a method of biogeographical analysis
which begins without a priori assumptions, and lets the data speak for
itself. His method is to plot the distribution of the species of a
given group on a map, and to connect the disjunct ranges with straight
lines. When many groups are treated in this way, it is found that the
lines of connection (tracks) do not form a random network over the map,
but rather, they tend to follow the same routes. These routes, an
aspect of the data, and in no way influenced by preconceived ideas of
past geography or climate, can and must be interpreted as representing
remnants of former wide-spread distributions.
By plotting the geographic patterns of morphologic variation of
a particular species on a map, and then considering many such maps
together, one is, in effect, using a modification of Croizat's pan-
biogeographic analysis on a micro scale. We begin with no assumptions
of past dispersal routes, climate or geography. Instead, the data are
mapped and the patterns emerging interpreted in the most parsimonious
The method is simple: Computerized contour maps are produced
depicting the geographic variation of each character to be investigated,
one character per map. Next the maps are compared, and a smaller number
of underlying patterns extracted and mapped. These patterns must be
explained. By means of correlation analysis it is possible to compare
the pattern maps with maps of environmental variables such as rainfall
or temperature. In this manner, patterns of organic geographic variation
can be identified and compared with geographic variation in environmental
factors. Patterns of morphologic variation which do not correlate with
the environmental variables may be correlated with untested environmental
variables, or they may have received their present shape by past paleo-
climatic or geographic factors.
MATERIALS AND METHODS
I analyzed character variation data on 3578 specimens of 15 species
of Florida snakes. The species were chosen on the basis of availability
of specimens or data, and, more importantly, because they are for the
most part, wide-ranging, rather ubiquitous species with representatives
taxonomicallyy distinct or not) in the southwestern United States and
Mexico. It was assumed that species with wide distributions across the
southern United States would be phenetically (and genetically) plastic
throughout their ranges as they adapt to different environments. Geographic
variation in a wide-ranging species is inevitable if we believe in evolu-
tion, and a species' ability to adapt to its environmental surroundings.
Furthermore, it was hoped that some light could be shed on the postulated
"Gulf Coast Corridor" phenomenon of Auffenberg and Milstead (1965) by
studying character variation in species associated with the Gulf Coastal
The characters examined were primarily standard taxonomic variables
with a history of application for the species studied. Although color
is probably the single most important aspect of a snake's phenotype at
the infraspecific level, this character was largely ignored, because
of unpredictable color changes that specimens undergo in preservative.
Aspects of pattern, carination and scutellation constitute the bulk of
the characters employed in the present analysis. See Appendix A for a
description of the characters examined.
Once the data were accumulated, they were coded, qualitative
characters were ranked, and everything was punched on computer Holerith
cards. Various combinations of characters were utilized to produce new
characters, such as ventrals plus caudals or percent tail, etc.
In addition to the data on snake morphology, I also recorded and
analyzed summary data on Florida climate from the Climatic Summary of
the United States, U.S. Department of Commerce. These data were also
punched on computer cards. The environmental variables investigated are
listed in Appendix B. Mean data from 196 weather stations in Florida,
southern Georgia and southern Alabama were used. The average period of
record on which means were based was 29 years.
In order to associate data from an individual snake or weather
station with the appropriate geographic locality, it was necessary to
develop a row and column coordinate system. This was done using as a
base map, the 1972 edition of the American Automobile Association road
map for Florida. Each specimen and weather station was assigned a
latitude and longitude value representing its locality on the base
map. These coordinates were punched on the cards along with the data.
Although the use of the AAA road map for a base may seem limiting, it
is not, because the coordinates can easily be changed to any other
system with the appropriate conversion factor. Actual latitude and
longitude values could not be used because these vary in inter-line
distance with the curvature of the earth, and the computational
algorythms employed require that the grid system be uniform.
The Problem of Samples and Populations
In every previous study of geographic variation that I know of,
data on organisms from more or less nearby localities have been pooled
and averaged, with the resultant average values applied to some point
or area representing all the individuals included. This seems to me to
be an arbitrary and unrealistic approach to the problem. Whether the
method employed lumps specimens by state, county, circular or square
grids, or the strange "splotch" system of Rossman (1963) and his students
(Blaney, 197Ta; Williams, 1970; etc.), the implication is that the values
associated with the geographic units represent values for populations.
There is no reason to believe that populations assume regular shapes
any more than shapes determined by political boundaries. Nor is there
any valid reason for lumping specimens with "similar" character states
from "adjacent" localities. What are "similar" and "adjacent?"
The method used in the present study is, I believe, less arbitrary,
and does not assume a knowledge of either population structure or geog-
raphy. The character state for each specimen is plotted on a map at the
point where the specimen was collected. If several specimens are avail-
able from the same point, they are averaged; otherwise, each specimen is
plotted independently. The next step involves calculating hypothetical
values for intermediate localities based only on the data available.
Character values are weighted inversely according to the square of their
distance from the point being considered. In this way values for all
points on the map (or any fraction thereof) are calculated based on the
points for which there are actual values. Contour lines can be con-
structued, and the areas of character change readily identified. Instead
of a priori assumptions of geographic structure which are implicit in any
scheme involving the means of putative populations, this method allows
the patterns of geographic variation to emerge from considerations of
The Mapping Procedure
This method of character mapping still suffers from the one draw-
back common to all methods: the map is only as good as the data.
Obviously, the more points on the map with actual data values, the less
interpolation will be necessary to identify the areas of character change.
Ideally, we would like to have our specimens from regularly-spaced
localities gridded over the entire map surface. This is simply not
possible in the majority of cases, and probably never possible when
Throughout the following discussion, it is necessary to keep in
mind that the mapping procedure employed does nothing more than map the
data as they appear, and interpolate between data points, just as a
cartographer would do, mentally, in constructing a contour map by hand.
There is nothing mysterious or even very sophisticated in the method. It
is faster than mapping manually, and it removes the element of bias from
contour line placement.
The actual computer program used in the production of character
variation maps was the SYMAP program developed by the Harvard University
Laboratory for Computer Graphics. This program has been used previously
in geographic variation studies by Jackson (1970) and Johnston and
Selander (1971), and is described in some detail in Peucker (1972).
The SPSS factor analysis procedure (Nie et at., 1970) was employed
to extract factor scores from the data set for each species studied. If
morphological characteristics of an organism co-vary, then there is some
underlying component or "factor" that will explain the variation in the
suite of characters that are varying together. This is the assumption
behind factor analysis. First a matrix of product-moment correlation
coefficients between the variables was calculated. R-type factor
analysis was employed to extract a smaller number of summarizing axes
that explain the covariation in the characters. These axes represent
a "summary" of the variation in a larger number of variables. Based
on these factor axes, factor scores were assigned to each specimen, and
these values mapped like univariate characters. The finished product
is a map of the geographic variation of a factor, which is, by definition,
a map of the underlying trends of geographic variation in a suite of
characters, where such underlying trends exist. A more detailed discus-
sion of Factor Analysis may be found in Harmon (1962).
Comparison of Mapped Data
The problem of comparing mapped data is a complex one. Prior to
this study there were no available methods for analyzing maps or perform-
ing correlation analysis between maps. If the variables to be analyzed
were located geographically at the same point, simple product-moment
correlation analysis could be performed and any correlative tendencies
readily discovered. However, when the maps to be compared have different
data points, a preliminary step to standardize the reference points is
necessary. That is, given two maps of Florida, one with X data points
(set A) and the other with Y data points (located at different points,
set B), it is necessary to interpolate each data set to a standard grid
system, or to interpolate one data set so that its new grid system corre-
sponds with the other. If the data are treated in this manner, it is
possible to compare climatic variables from U.S. weather stations, with
morphological variables of snakes from wherever they were collected
(usually not weather stations).
The SYMAP mapping routine, discussed above, calculates estimated
data values for a finite number of points on a map based on the values
associated with actual data points. It is possible to output the values
so calculated for one data set, and compare these with values output from
another data set run in the same way at another time. In this way,
multiple sets of geographic data can be standardized to a grid system
by non-arbitrary, reproducible methods (SYMAP) and the resulting grids can
be treated as matrices in correlation, cluster, or factor analysis.
A computer program to handle the SYMAP-produced data point values
and store these on magnetic tape for later analysis was written for this
study by William Ingram.
Regional and place names used in this study are located on Fig. 1.
The major rivers of Florida are shown on Fig. 2. Figs. 1 and 2 should
enable the reader not familiar with Florida to follow easily the results
and discussions presented below.
PANHANDLE OKEFENOKEE SWAMP
Figure 1. Regional and place names referred to in the text.
V," O A
Pearson Product-Moment correlation coefficients were calculated
between every character and sex and snout-vent length for each species.
Unless otherwise noted, all characters mapped showed no correlation with
sex or body size. Note that the mapping procedure employed does nothing
more than assign a given value to the appropriate locality, then inter-
polate between the localities to predict the location of a change in
character state. Each specimen is weighted equally, so geographically
isolated specimens may contribute more heavily in the analysis than
individual specimens from better-collected regions. That is, if only
a single specimen is available from region A, then that region will be
shaded as though the population had the characteristics of that specimen.
If 20 specimens were available from the same region, the shading would
reflect the average of those 20 specimens. When interpreting the maps
in this section, it is important to bear this in mind, and to refer to
the specimen locality maps presented for each species.
Results are presented first by species, and compared with previous
studies. Patterns of geographic variation are discussed in the second
section, and finally, a third section deals with correlations between
observed patterns and environmental factors. A description of each of
the characters examined appears in Appendix A. Actual raw values may
be obtained from the author.
Storeria dekayi (Holbrook)
I examined 151 specimens of Storeria dekayi from Florida and
southern Georgia (Map 1) for possible geographic variation in each of
21 characters (Appendix A). The numbers of supralabials and infra-
labials are essentially unvarying over the study area. More than 93%
of the specimens examined had 14 supralabial combining both sides, and
87.4% had 14 infralabials. The number of postocular scales was usually
four (88.3%). None of the characters examined was found to be size-
correlated. The following characters showed apparent geographic
Number of ventrals. The number of ventral scales in Florida
Storeria dekayi was only slightly correlated with sex, with the males
usually having lower counts (r = 0.3559). Nevertheless, the sexes were
mapped separately as shown in Maps 2 and 3. In general, the number of
ventrals increases southward on the peninsula and drops again on the
Lower Keys. The highest ventral counts are observed on snakes from the
Everglades region, while the lowest values are found on the Lower Keys
and in the Panhandle. A reasonable degree of concurrent variation between
the sexes suggests that they are responding similarly in ventral count
Number of subcaudals. Males tend to have more subcaudals than
females (r = 0.5167). Geographic variation in the number of subcaudals
in Florida S. dekayi is shown in Maps 4 and 5. Variation is similar to
that described for ventrals, above. Snakes from the Panhandle and from
the Lower Keys have the lowest subcaudal counts. Otherwise, the
variation is clinal, increasing southward on the Florida peninsula.
Brown snakes from the Apalachicola Valley may be more similar to snakes
from Central Florida as regards this character. Again, concordance in
the patterns of variation observed in the sexes is very good.
Number of ventrals plus caudals. Although both the number of
ventrals and the number of subcaudals are correlated with sex (see above)
their sum is not (r = 0.1522). Therefore, the sexes can be lumped to
increase sample size, producing Map 6. Geographic variation in this
character consists of increasing counts southward on the peninsula, and
a major drop on the Lower Keys. Snakes from west of the Suwannee River
in the Panhandle also have low values for this character. (Note that the
area west of the Yellow River is represented by only a single specimen,
and is therefore an inadequate sampling.)
Percent tail. Males tend to have proportionately longer tails
(r = 0.7214), and so the sexes have been mapped separately in Maps 7 and
8. Brown snakes from the southern mainland have the longest tails
proportionate to body length. Snakes from the Lower Keys have somewhat
shorter tails, more like specimens from northern Florida.
Dorsal scale rows. Dorsal scales were counted at the standard
three places along the body, but all three varied the same way, so only
midbody scale rows are mapped (Map 9). Brown snakes from the Panhandle
east to the vicinity of the Suwannee River have 17 scale rows around the
body, while those from the remainder of the state have 15.
Preocular scales. Generally, Storeria dekayi has a single preocular
scale on each side of the head. However, individuals from the Lower Keys
typically have two preoculars on each side. In addition, a high pro-
portion of specimens from near Jacksonville and Gainesville in northern
Florida have two preoculars on each side. See Map 10.
Ventral dark pigmentation. Geographic variation in the qualitative
assessment of the amount of dark pigment ventrally in Florida S. dekayi is
presented in Map 11. Generally speaking, brown snakes from the south-west
coast of the peninsula have the darkest bellies, while those from the Lower
Keys have the lightest.
Temporal pigmentation. Dark pigment on the temporal scale in S.
dekayi may be in a tear-shaped blotch with one end wider and continuing
onto adjacent scales (see Appendix A). The wider end may be directed
posteriorly as in most snakes from the Panhandle, or it may be directed
to the front as in specimens from the Florida peninsula. Brown snakes
from the Lower Keys often have the temporal pigment so reduced as to have
no obvious orientation, but when present, the wider end is directed to
the front. See Map 12.
Subocular pigmentation. The number of supralabial scales contained
within the subocular dark blotch in Florida S. dekayi varies geographically
as shown in Map 13. Generally, snakes from the peninsula tend to have
larger subocular spots than specimens from the Panhandle. Many individuals
from the Everglades region have very small or absent subocular spots.
Subocular spots on specimens from the Lower Keys are diffuse and faint,
but cover three to five supralabials.
Factor 1. The first factor accounts for 24% of the variation in 18
characters. It accounts for most of the variation in the three dorsal
scale row counts, the sum of ventrals plus caudals, temporal pigmentation,
ventral pigmentation, and supralabial pigmentation. The first factor
discriminates between two phenotypes of Storeria in Florida: snakes
with 17 dorsal scale rows, fewer than 187 ventrals plus caudals, a
temporal blotch with its wider end directed posteriorly, reduced ventral
pigment, and fewer supralabials with black pigment occur in the Pan-
handle, while snakes of a contrasting phenotype occur on the peninsula.
See Map 14.
Thamnophis sirtalis (Linnaeus)
Data on 192 specimens of Thamnophis sirtalis from Florida (Map 15)
were analyzed for possible geographic variation in each of 13 characters
(Appendix A). Dorsal scale rows varied little throughout the study area,
with most specimens having 19-19-17 scale rows (94.7%, 95.7%, and 96.8%
respectively). Ninety-one percent of the T. sirtalis examined had 14
supralabials and 92.4% had 20 infralabials. None of the characters
examined was size- or sex-correlated. The following characters appear
to show trends of geographic variation within Florida.
Numbers of ventrals. Geographic variation in the number of ventral
scales in T. sirtalis is shown in Maps 16 and 17. Although differentiation
within Florida for this character is minimal, some trends are apparent. In
general, garter snakes from Lake Okeechobee southward have the highest
ventral counts. Snakes from the Panhandle tend to have low counts. It
is not possible, however, to discern a well-developed dine within the
state of Florida.
Number of subcaudals. Trends in the geography of subcaudal count
variation in Florida garter snakes are not clear. Maps 18 and 19 show
the general lack of concordance between the sexes for this character.
Patterns like these may imply that the sexes are responding differently
in subcaudal count expression, or they may be the result of sampling
bias due to inadequate sample sizes (51 males, 62 females). Alternatively,
the number of subcaudal scales in Florida T. sirtaZis may not correspond
with anything that varies geographically.
Number of ventrals plus caudals. The summation of the proceeding
two characters varies geographically as shown in Map 20. Garter snakes
tend to have the highest counts in the northern peninsula and extreme
southern peninsula, with lower counts found in the intervening region,
and in the Panhandle.
Percent tail. Tail length divided by total length varies as shown
in Maps 21 and 22. In general, specimens from Apalachicola Valley, the
northern peninsula, and the southern peninsula have proportionately
longer tails than snakes from elsewhere. Although the pattern appears
complex, the concordance between the sexes is quite good, supporting the
reality of the trends, based, as they are, on different sets of specimens.
Dorsal spotting. That some individuals of Thamnophis sirtalis are
marked with black spots dorsally is well known. Linnaeus (1766) described
the phase as Coluber ordinatus; Cope (1900) considered it a subspecies of
T. sirtalis, and more recent authors have considered it little more than
an occasional pattern variant without geographic correlation (Rossman,
1965). Maps 23, 24, and 25 show the geographic variation in this character
for Florida specimens. Garter snakes from the western Panhandle usually
have a well-developed pattern of dorsal checks, sometimes to the complete
exclusion of longitudinal stripes. Additionally, the west coast of the
peninsula and parts of the Central Highlands support populations of T.
sirtalis with dorsal checking. Specimens from the southern Everglades
are almost invariably heavily spotted.
Parietal spots. Rossman (1963) considered the nature of the paired
parietal light spots important to the taxonomy of the genus Tharmophis.
The pattern of geographic variation in thischaracter is remarkably
similar to that seen in the previous character, dorsal spotting. Maps
26, 27, and 28 show the geographic variation as interpreted in the present
study. Again, the Panhandle, the southwest coast of the peninsula and the
southern Everglades stand out as areas with higher states for this
character. Concordance between the sexes is good.
Factor 1. The first factor extracted from the correlation matrix
of eleven characters accounts for 18.5% of the total variation, and much
of the variation in dorsal spotting and parietal spots. Map 29 shows
how factor 1 varies in Florida. The western half of the Panhandle, the
southwestern coast of the peninsula, and much of the Central Highlands
and southern Everglades support populations of garter snakes characterized
by well-developed dorsal and parietal spots.
Thamnophis sauritus (Linnaeus)
I analyzed data on 279 specimens of Thamnophis sauritus from Florida
and southern Georgia (Map 30) for possible geographic variation in each of
13 characters (Appendix A). Another 12 specimens from the Lower Florida
Keys were examined and included in the discussion, but were not available
at the time of mapping. The number of infralabials was essentially un-
varying throughout the study area (92.6% had ten on each side). None
of the characters investigated was correlated with snout-vent length.
Those characters which showed trends of geographic variation follow.
Number of ventrals. Males and females do not differ significantly
in ventral counts (r = 0.2910). Maps 31 and 32 show the apparent clinal
increase in ventral counts for both sexes southward on the Florida
peninsula. Higher ventral counts tend to extend farther northward along
the coasts, and specimens from the Gulf Hammock region on the east coast
have ventral counts comparable to snakes from the most southerly localities.
Data from Paulson (1968) and from the present study indicate that T.
sauritus from the Lower Keys have ventral counts like those seen in
specimens from the southern mainland. Seven males from the Lower Keys
averaged 163.4 ventrals and five females averaged 160.2.
Number of subcaudals. Males usually have more subcaudal scales
than females (r = 0.5801), and the sexes have been mapped separately in
Maps 33 and 34. Ribbon snakes from the Panhandle west of the Chocta-
whatchee River tend to have more subcaudals than specimens from the
remainder of the Panhandle and northern peninsula. Snakes from the
peninsula usually have higher counts than specimens from the Panhandle,
and there seems to be a weakly differentiated dine of increasing counts
southward on the peninsula. Specimens examined from the Lower Florida
Keys have higher subcaudal counts than ribbon snakes from anywhere else
in Florida. Seven males averaged 137.9 subcaudals and three females
Number of ventrals plus caudals. The summation of the two previous
counts was found to correlate with sex (r = 0.5201), with males having
more total ventral and subcaudal scutes. Maps 35 and 36 show how this
character varies in Florida. The variation tends to be clinal, with values
increasing southward. Higher values extend farther north along the coasts
of the peninsula. The highest values observed in Florida ribbon snakes
are associated with specimens from the Lower Keys. Seven males from the
Lower Keys had an average ventral plus caudal value of 301.3, and three
females averaged 286.2 ventrals plus caudals.
Percent tail. Ribbon snakes with the longest tails relative to
snout-vent length tend to occur in the central part of the peninsula,
with specimens having lower values occurring both north and south. In
addition, the few specimens available from the extreme western Panhandle
and the Lower Keys suggest that snakes from these areas also have longer
relative tail lengths.
Supralabials. Ribbon snakes from the Panhandle generally have
seven upper labials on each side, while those from the peninsula have
eight. Although not mapped, the ribbon snakes from the Lower Keys
occasionally have seven supralabials as well. Map 37 shows the geo-
graphic variation in supralabial number for both sexes of Thcanophis
sauritus from Florida.
Dorsal stripe edge. Geographic variation in the width and
development of the black border of the dorsal stripe is figured in
Maps 38, 39, and 40. Although the pattern is complex, congruence between
the sexes is good. Thamnophis sauritus from the Panhandle west of the
Apalachicola River and from parts of the central peninsula tend to have
well-developed dorsal stripe edges. Snakes examined from the Lower Keys
have extremely well-developed black dorsal stripe borders. Many specimens
from the northern half of the Florida peninsula lack a stripe border
altogether, and some lack even the mid-dorsal yellow stripe (Rossman,
Parietal spot. Although Rossman (1963) stated that the nature of
the paired parietal light spots in T. sauritus does not vary geograph-
ically, my analysis of his data indicates that it does. Maps 41, 42, and
43 show this variation. Most Florida T. sauritus lack a distinct parietal
spot. However, specimens from the extreme northern Florida peninsula, at
the edge of the Okefenokee Swamp, the area east of Tampa Bay, and the
southern tip of the peninsula have distinct parietal spots. The concord-
ance between the sexes in the geographic pattern of variation observed
in this character support the reality of the pattern. The few specimens
available from the Lower Keys have weakly developed parietal spots.
Ground color. Most ribbon snakes from Florida have a tan or light
brown ground color. However specimens from the Gulf Hammock region usually
have a very dark brown or black ground color. Some specimens from the
Everglades and southern mainland also have a darker ground color, but
specimens seen from the Lower Keys are light brown or tan. See Map 44.
Factor 1. The first factor extraced from the correlation matrix
of ten characters accounts for 22.3% of the total variation, and most of
the variation in parietal spot distinctiveness and parietal spot size.
Map 45 shows how the first factor varies geographically. Higher states
are associated with snakes from the northern peninsula, the area east of
Tampa Bay, and the southern tip of the peninsula.
Coluber constrictor Linnaeus
I examined 440 specimens of Coluber constrictor from Florida
(Map 46) for possible geographic variation in each of 18 characters
(Appendix A). The numbers of supralabials and infralabials were
practically nonvariant with seven and eight respectively on each side.
Aspects of the color and pattern were scored only on adult snakes (over
70 cm). With ontogenetic change in color and pattern thus taken out of
the picture, none of the characters investigated was found to correlate
with snout-vent length. Those variables that showed trends of geographic
variation within the study area are discussed below.
Number of ventrals. There is no significant difference between
males and females in ventral counts (r = 0.2253), and the sexes have been
lumped to produce Map 49. Maps 47 and 48 show the good degree of con-
cordance between the patterns of variation in the two sexes. There is
a well-developed clinal increase in number of ventrals as one proceeds
southward down the peninsula. Specimens from the Lower Keys do not
follow this trend, however, having much lower ventral counts than
specimens from the southern tip of the mainland. It is also noteworthy
that occasional specimens from the Apalachicola River Valley have higher
than expected ventral counts, being more like specimens from farther
south on the peninsula in this regard.
Number of subcaudals. Males tend to have more subcaudal scales
than females (r = 0.3898), although the correlation is not a strong one.
Maps 50 and 51 show the geographic variation in subcaudal counts for
Florida Coluber constrictor. Specimens from the Panhandle usually have
lower counts than those from the peninsula. On the peninsula, there
may be a very weakly defined clinal increase in subcaudal numbers for
each sex, but any tendency in that direction is well masked.
Number of ventrals plus caudals. Map 52 shows the dine on the
Florida peninsula of increasing ventral scute counts southward, except
for a dip on the Lower Keys. The sexes could be lumped in this map
because ventrals plus caudals is independent of sex (r = 0.1618).
Supralabial loreal contact. Auffenberg (1955) first noted
the variability of this character. In some specimens of Coluber,
especially in the south-eastern United States, the first supralabial
is in contact with the loreal. Although Auffenberg did not believe
that this character varied with any degree of geographic regularity,
my analysis has shown that it does. Map 53 shows that specimens from
extreme northern Florida, the area east of Tampa Bay and extreme southern
Florida, including the Middle Keys tend to have the first supralabial
in contact with the loreal more frequently than specimens from the
remainder of the state. This pattern occurs repeatedly in Florida snakes
and must be considered indicative of some geographic factor based on
present or past environments.
Ventral white. Geographic variation in the amount of white on
the ventral surface in Florida C. constrictor is shown in Maps 54, 55,
and 56. Snakes under 70 cm have been excluded from this analysis.
Black snakes from south Florida and the Everglades region have almost
totally white undersides. Snakes with the darkest bellies (i.e. the
least white) are found on the Lower Keys and extreme northern Florida.
Lighter colored ventrums seem to be associated with the coastal and
treeless areas of the state. The black snakes on the Upper Keys are
intermediate between Lower Keys and south Florida mainland specimens.
The correspondence between the sexes is excellent.
Gular brown pigmentation. Map 57 displays the geographic variation
in the presence or absence of brown pigment on the gular scales in Florida
Coluber. Most black snakes from the Lower Apalachicola River Valley have
brown pigment on the gulars. Snakes from the remainder of the state
seldom have such pigment. When the sexes were mapped separately, the
patterns were virtually identical.
Supralabial brown pigmentation. See Maps 58, 59, and 60 for the
geographic variation in this character. Black snakes from the lower
Apalachicola Valley have the most brown pigment on the supralabial scales.
Snakes from the southern part of Florida peninsula and the Everglades tend
to have some supralabial brown, but not as much. Black snakes from the
extreme southern tip of the peninsula and from the Lower Keys generally
do not have any brown pigment on the upper labials.
Gular black pigmentation. The presence or absence of black on the
gular scales in Florida Coluber varies geographically as shown in Map 61.
The extreme northern base of the peninsula just south of the Okefenokee
Swamp, the area east of Tampa Bay, and the Lower Keys support populations
of C. constrictor with black on the gular scales.
Supralabial black pigmentation. The amount of black pigmentation
on the supralabial scales varies geographically much like the previous
character (see Maps 62, 63, and 64). The congruence between the sexes
is remarkable, and can be considered as very strong evidence for the
reality of the pattern. Black snakes from extreme northern Florida, the
area east of Tampa Bay and the Lower Florida Keys have more black pigment
on the supralabial scales than specimens from anywhere else in Florida.
Factor 1. The first factor accounts for 17.4% of the total vari-
ation in 16 characters. This factor accounts for most of the variation
in gular black pigmentation and supralabial black pigmentation as well
as much of the variation in ventral white pigmentation. Its geography
(Map 65) shows regions with similar phenetic responses in northern
Florida, the area east of Tampa Bay and the Lower Florida Keys.
Factor 3. This factor accounts for 13.2% of the total variation
in 16 characters, and most of the variation in gular and supralabial
brown pigmentation as well as part of the variation in the number of
ventrals plus caudals. Snakes with high factor 3 scores (higher values
for brown pigmentation characters and lower ventral plus caudal counts)
tend to occur in the Apalachicola River Valley, especially the lower
valley, and the Everglades region of south Florida. See Map 66.
Masticophis flagellwum (Shaw)
I analyzed data on 85 specimens of Masticophis flagewlum from
Florida, southern Alabama and southern Georgia (Map 67) for possible
geographic variation in each of nine characters (Appendix A). Over 96%
of the specimens examined had 16 supralabials. None of the characters
was size-correlated. Those which appear to vary geographically follow.
Number of ventrals. Male coachwhips usually have more ventrals
than females (r = 0.4010). Maps 68 and 69 show that in general, the
highest ventral counts are found in snakes from the peninsula, and
especially the western half of the peninsula. Correspondence between
the sexes is weak.
Number of subcaudals. Males and females are not significantly
different in subcaudal counts (r = 0.2425), and have been lumped to
produce Map 70. It appears that specimens from the Panhandle and western
parts of the peninsula usually have more subcaudal scales than snakes from
the eastern peninsula.
Ventrals plus caudals. The summation of the proceeding two charac-
ters is not correlated with sex (r = 0.3219) and varies geographically as
in Map 71. Again, the Panhandle and western part of the peninsula are
characterized by coachwhips with higher ventral counts.
Percent tail. Tail length divided by total length is not corre-
lated with sex (r = 0.2422), and the sexes have been lumped to produce
Map 72. Longer tails seem to be associated with specimens from the
Panhandle and western parts of the peninsula.
Color phase. Most Masticophis from the Panhandle of Florida are
of the light color phase, and most from the peninsula are dark. However,
as shown in Maps 73, 74, and 75 there is some individual variation in this
character. Furthermore, there appears to be two areas in peninsular
Florida where snakes of the light phase are more common. This same
pattern appears in both sexes, lending credence to its reality. Un-
fortunately, the sample sizes are very small.
Factor 1. The first factor accounts for 43.0% of the total varia-
tion in five characters, and most of the variation in ventrals plus caudals,
and part of the variation in number of infralabials. Its geography,
depicted in Map 76, shows a pattern of high factor 1 scores in the Pan-
handle and northern peninsula separated from another area of high values
in southern Florida by an intervening region where the coachwhips tend
to have fewer ventral scutes and infralabial scales.
Opheodrys aestivus (Linnaeus)
I examined 176 specimens of Opheodrys aestivus from Florida (Map
77) for geographic variation in each of 16 characters (Appendix A).
Supralabial and infralabial counts remained essentially constant through-
out the study area with 87.4% having 14 supralabials and 75.6% having 16
infralabials. Dorsal scale rows were found to be 17-17-15 for over 95%
of the specimens examined. Three measurements of the frontal scale were
found to correlate with snout-vent length even after standardization by
dividing by snout-vent length. The characters which appear to show
geographic variation are discussed below.
Number of ventrals. Female green snakes typically have more ventral
scales than males (r = 0.3991). There is no obvious well-developed dine
in ventral counts for Florida Opheodrys. However, specimens from the
southern parts of the state, and especially the south-west usually do
have the highest counts.
Number of subcaudals. Green snake males tend to have higher sub-
caudal counts than females (r = 0.5513). Specimens from the western
Panhandle east to the Apalachicola River tend to have the most subcaudal
scales, while specimens from the Everglades and southern peninsula have
Ventrals plus caudals. Although both ventrals and caudals were
found to be correlated with sex, their summation was not (r = 0.2807).
Map 78 shows the geographic variation in this character for the combined
sexes. Highest values are associated with snakes from the western Pan-
handle and parts of South Florida. Snakes from the Everglades region
have the lowest ventral scute values. The variation is complex, and it
is not possible to discern a simple north-south dine in this character.
Percent tail. Relative tail length in Florida green snakes varies
with sex. Males usually have proportionately longer tails (r = 0.5211).
Concordance between the sexes is poor. In general, however, both sexes
tend to have slightly longer tails in the Panhandle, and shorter tails
south of Lake Okeechobee. In addition, the males show a pattern of
longer tails on the Lower Keys, while the females do not.
Keeling of the second dorsal scale row. Maps 79, 80, and 81 show
the geographic variation in the development of the keel on the scales of
the second dorsal row at midbody. The maps for the two sexes indicate that
the variation is very similar. North of central Florida and in the Pan-
handle green snakes tend to have reduced or no keeling on these scales.
In central Florida, south throughout the peninsula, states for this charac-
ter tend to be higher, with the highest states frequently associated with
snakes from coastal regions. On the Lower Keys, the males have keeled
scales on the second row, but the females apparently do not.
Supralabial pigmentation. Maps 82, 83, and 84 show the geographic
variation in the amount of dark pigmentation on the upper labial scales in
Florida Opheodrys aestivus. The strong degree of concordance between the
sexes may be taken as evidence for the reality of the pattern, since the
two maps are based on separate sets of specimens. In general, snakes with
more dark pigment on the supralabials occur in the Panhandle, the northern
parts of the peninsula, and southward along the west coast. Green snakes
from the Keys have very light colored upper labials.
Factor 3. This multivariate character varies geographically as shown
in Map 85. It accounts for 12.6% of the total variation in ten characters,
and most of the variation in keels of the second dorsal scale row and supra-
labial pigmentation. Green snakes from the southern half of the peninsula
have less supralabial pigment, and better developed keels on their second
scale fow. Snakes from extreme northern Florida may have higher Factor 3
scores, like specimens from the south.
Elaphe guttata (Linnaeus)
I examined 455 specimens of Elaphe guttata from Florida, southern
Georgia, and southern Alabama (Map 86) for possible geographic variation
in each of 16 characters (Appendix A). Some of the characters varied
little throughout the study area. Dorsal scale rows counted at three
points along the body were almost always 25-27-19. Measurements of the
frontal scale were divided by snout-vent length, and found to correlate
with snout-vent length. These characters were excluded from the analysis
Number of ventrals. Females usually have more ventral scales than
males (r = 0.5933). In both sexes, ventral counts increase southward on
the Florida peninsula, with the highest values associated with snakes from
the Lower Keys. Coastal areas also seem to support E. guttata with higher
ventral counts. Maps 87 and 88 depict the geographic variation in ventral
numbers for E. guttata from Florida.
Number of subcaudals. Males tend to have more subcaudal scales than
females (r = 0.5015). The number of subcaudals in Florida corn snakes
appears to increase southward on the peninsula, but the trend is not as
clear-cut as in the proceeding character. See Maps 89 and 90.
Ventrals plus caudals. The summation of ventrals and caudals is
not correlated with sex (r = 0.1118), and so the sexes could be lumped
to increase sample size (Map 91).
The highest counts occur on snakes from the Florida Keys and
adjacent mainland. Higher counts are frequently associated with coastal
areas as well. The lowest ventral plus caudal counts occur on snakes
from the northern peninsula and Panhandle.
Number of body blotches. EZaphe guttata from the Lower Keys have
the highest dorsal blotch counts. Specimens from the Panhandle and
northern peninsula have the lowest. Higher counts reach farther north-
ward on the peninsula along both coasts. Maps 92 and 93 show the geo-
graphic variation in number of body blotches in Florida E. guttata.
The rather high degree of individual variation in this character
partially obscures the clinal nature of its geography.
Number of tail blotches. Corn snakes from the Lower Keys and
south Florida mainland have the highest tail blotch counts. Maps 94
and 95 show the geography of this character. The lowest tail blotch
counts are seen in snakes from the Panhandle, and the increase south-
ward is probably clinal, although complicated by individual variation
from snake to snake.
Blotch border. The red blotches on the dorsum of Florida corn
snakes are usually surrounded by a narrow black border. The width of
this border varies geographically as shown in Map 96. Snakes from the
extreme south Florida mainland and parts of the western Panhandle have
wider blotch borders.
Lateral blotch shape. Corn snakes from the Panhandle of Florida
east to the Aucilla River frequently have the border of the lateral
blotch open ventrally, suggesting an inverted U rather than a complete
circle. The character occurs sporadically throughout Florida, but is
almost universal among Panhandle specimens. See Map 97.
Ventral pigmentation. Corn snakes from the Lower Florida Keys
have the least dark pigment ventrally. Snakes in coastal areas and
other off shore islands have reduced pigment, while the majority of
specimens from the remainder of the peninsula and Panhandle tend to
have much black pitment on their ventral surfaces. Maps 98, 99, and
100 show the geographic variation of this character in Florida E.
Ventral check shape. The ventral dark pigmentation is generally
confined to discrete rectangles. Snakes from the Lower Florida Keys and
coastal regions (especially the south-west coast) have small, often
square pigment spots ventrally. Specimens examined from more interior
regions, especially in the northern parts of the state, have their
ventral pigment in elongated rectangles. Many specimens from the
Panhandle and extreme northern Florida have wide rectangular pigment
spots covering entire ventral scutes.
Factor 1. The first factor accounts for 19% of the total varia-
tion in 12 characters. Factor 1 accounts for most of the variation in
ventrals plus caudals, number of body blotches and number of tail
blotches. Map 101 shows how this multivariate character varies geo-
graphically in Florida. Snakes from the Florida Keys and southern parts
of the peninsula tend to have high factor 1 scores, while the lowest
values are associated with Panhandle specimens.
Factor 2. The second factor accounts for 15.0% of the total
variation, and most of the variation in the amount of ventral dark
pigmentation, and the shape of the ventral pigment blotches. Low
values (that is, lighter bellies with smaller blotches) are associated
with corn snakes from coastal areas, and the lowest values of all are
seen in specimens from the Lower Keys. See Map 102.
Elaphe obsoleta (Say)
I analyzed data on 370 specimens of Elaphe obsoleta from Florida
(Map 103) for potential geographic variation in each of six characters
(Appendix A). None of the characters investigated was correlated with
size. The sex of the specimens was not determined.
Number of dorsal blotches. The number of body blotches in E.
obsoleta increases clinally on the Florida peninsula to the south.
Map 104 shows the variation in this count.
Ground color. The dorsal ground color of Florida E. obsoleta
varies geographically as shown in Map 105. This character attempts
to describe the amount of melanin or dark pigmentation on the dorsum
of these snakes. In general, specimens from coastal areas have the
lightest ground colors. Most chicken snakes from the interior of the
peninsula, the Panhandle, and the Upper Keys have darker dorsums. (The
species does not occur on the Lower Keys.)
Stripe development. Elaphe obsoleta from the Panhandle east to
the vicinity of the Suwannee River generally lack dorsal stripes. With-
in the peninsula, specimens from more northerly regions have the most
well-defined stripes, while most examples from southern Florida have
moderately-developed dorsal stripes. See Map 106.
Blotch development. Chicken snakes from the Florida Panhandle
east to the vicinity of the Suwannee River, and southward along the
west coast to the vicinity of the Withlacoochee River have dorsal
blotches. Those specimens south of the Suwannee have dorsal stripes
as well, and are recognized as the subspecies E. obsoleta williaasi.
In addition, E. obsoleta from the Upper Keys and extreme southern
Florida mainland have fairly well-developed dorsal blotches, and have
been called E. o. deckerti. See Map 107.
Ventral pigmentation. Map 108 shows the geographic variation in
the amount of dark pigment in the ventral pattern of Florida chicken
snakes. Specimens from the Panhandle east to the Suwannee River have
the darkest bellies.
Supralabial pigmentation. The geographic variation in the amount
of dark pigment on the upper labial scales in this species varies geo-
graphically as in Map 109. Its variation is almost identical to that
seen in ventral pigmentation except that snakes with dark supralabials
occur also on the Upper Keys.
Factor 1. The first factor accounts for 51.6% of the total varia-
tion in four characters, and most of the variation in ventral and supra-
labial pigmentation, as well as part of the variation in number of
dorsal blotches. Map 110 shows how this character varies within Florida.
Snakes from the Panhandle receive the highest factor scores, implying
darker labial and ventral pigment and fewer dorsal blotches.
Lampropeltis getulus (Linnaeus)
I analyzed data on 207 specimens of Lampropeltis getuZus from
Florida, southern Georgia and southern Alabama (Map 111). A total of
eleven characters were investigated for geographic variation (Appendix
A). Head length was divided by snout-vent length and the resulting
ratio found to correlate with snout-vent length. Further consideration
of this character has been omitted. The characters which seem to vary
geographically are discussed below.
Number of ventrals. The number of ventral scales in Florida king-
snakes does not correlate with sex (r = 0.1392). Therefore, the sexes
could be lumped to increase the sample size. Kingsnakes from the Panhandle
tend to have fewer ventrals than those from the peninsula. The tendency
is not pronounced, however, and if any geographic pattern of variation
exists within the peninsula, it does not emerge from examination of the
Number of subcaudals. Male L. getulus usually have more subcaudals
than females (r = 0.6306). There is little geographic concordance between
the sexes. There is a very generalized tendency for subcaudal counts to
be higher in the southern parts of the peninsula.
Percent tail. Tail length divided by total length varies as shown
in Maps 112 and 113. Males have proportionately longer tails than females
(r = 0.4837), thus the sexes have been treated separately in the geograph-
ical analysis. Snakes with proportionately longer tails occur in extreme
southern Florida and apparently the Apalachicola River Valley.
Dorsal scale rows. The number of scale rows at midbody varies geo-
graphically as shown in Map 114. Kingsnakes from the Panhandle east to
the vicinity of the Aucilla River have 21 midbody scale rows, while
specimens from the peninsula usually have 23. It is apparent that a
large percent of kingsnakes from extreme southern Florida have 21 scale
rows as well.
Number of infralabials. Most Florida kingsnakes have 18 lower
labial scales, counting both sides. However, many from extreme northern
parts of the state have 19 or 20 infralabial scales. In addition, king-
snakes from the east coast of the peninsula frequently have 20 lower
labials. See Map 115.
Number of cross bands. Maps 116, 117, and 118 show the geographic
variation in the number of dorsal precaudal cross bands in Florida L.
getulus. The concordance between the sexes is remarkable, and suggests
that the sexes are responding similarly to whatever environmental factor
selects for cross band counts. There is a very weak (r = 0.4038)
correlation between snout-vent length and number of cross bands, implying
that larger snakes tend to have fewer cross bands. In the present case,
however, there is no correlation between snout-vent length and latitude
in Florida (r = 0.1678), so the variation described is not due to
variation in snout-vent length, but to inherent variation in cross
band counts. In other words, snout-vent length does not vary geograph-
ically in Florida, based on the present sample. The number of cross
bands increasesin regular dine southward on the Florida peninsula.
Dorsal pattern. The amount of light pigment in the dorsal pattern
of Florida kingsnakes varies geographically as shown in Maps 119, 120,
and 121. Congruence between the sexes is quite good. Areas in which
the kingsnakes tend to have more light pigment dorsally include the
Everglades and southern peninsula, the Lower Apalachicola Valley, and
extreme northern Florida. Frequently, the kingsnakes from the region
to the east of Tampa Bay are light colored, as well.
Factor 2. The second factor accounts for 27% of the total varia-
tion in five characters, and much of the variation in number of ventrals
and midbody scale rows. Map 122 shows the geographic variation for this
multivariate character. Specimens with higher factor 2 scores tend
usually to occur in the peninsula as opposed to the Panhandle.
Factor 3. Factor 3 accounts for 23% of the total variation in
five characters, and much of the variation in dorsal scale rows and
light pigment in the pattern. Snakes from extreme northern Florida,
parts of southern Florida, and certain other regions in the peninsula
have high factor 3 scores. See Map 123.
Lampropeltis triangulum (Lacepede)
I analyzed data on 120 specimens of La'propeltis triangulum from
Florida, southern Georgia and southern Alabama (Map 124) looking for
trends of geographic variation in each of 15 characters (Appendix A).
None of the characters investigated showed correlations with snout-vent
length. Those characters which seem to vary geographically are discussed
Number of ventrals. Males and females do not differ appreciably in
number of ventral scales (r = 0.2330). However they do seem to differ in
their patterns of geographic variation. Sample sizes are admittedly small,
and trends uncovered by the mapping techniques may be a result of sampling
bias. However, based on the data available (Maps 125 and 126) male milk
snakes seem to have higher ventral counts in northern Florida and the
peninsula, while females have lower counts there.
Number of subcaudals. Males generally have more subcaudal scales
than females (r = 0.6352). Maps 127 and 128 show the geographic variation
of this character. Concordance between the sexes is fair. If there is
any geographic tendency, it is for snakes from the Panhandle and parts of
southern Florida around Lake Okeechobee to have higher subcaudal counts.
Ventrals plus caudals. The summation of the proceeding two charac-
ters does not correlate with sex (r = 0.1106), and samples could be lumped
to produce Map 131. In addition, Maps 129 and 130 are provided to show
the degree of correspondence between the sexes. Generally speaking,
milk snakes from the Panhandle have higher ventral and caudal counts
than specimens from most of the peninsula. However, snakes from the
east coast and areas around Lake Okeechobee seem to have high ventral
plus caudal counts as well.
Percent tail. Relative tail length varies geographically as
shown in Maps 132 and 133. Males have proportionately longer tails
(r = 0.6537). The variation does not have a clear-cut pattern, but there
seems to be a slight tendency for relative tail length to increase clinally
to the south on the Florida peninsula.
Dorsal scale rows. Female L. trianguZwn from Florida usually have
19 midbody scale rows, while males may have either 19 or 17. Maps 134
and 135 show the geographic variation in midbody scale rows. Male
specimens from the Panhandle usually have 19 scale rows at midbody, and
males from the peninsula may have either 19 or 17.
Body bands. The number of red precaudal cross bands varies geo-
graphically as shown in Maps 136, 137, and 138. High band counts are
associated with milk snakes from northern Florida, including the Pan-
handle, the area east of Tampa Bay on the central Highlands, and apparently,
the region around Miami in southern Florida.
Tail bands. Maps 139, 140, and 141 show the geographic variation
in number of red tail cross bands in Florida milk snakes. In general,
specimens from the Panhandle and northern peninsula have higher tail
band counts than those from the remainder of the state.
Total bands. The summation of precaudal and caudal red bands
varies geographically as shown in Map 142. Highest counts occur on
snakes from northern Florida, the area east of Tampa Bay and the south-
eastern peninsula region. Milk snakes from the southern half of the
peninsula (not counting the Miami rim area) have low band counts com-
pared with specimens from the Panhandle and northern half of the penin-
Factor 3. This multivariate character accounts for 16.6% of the
total variation in 10 characters and most of the variation in the number
of red cross bands on the body and the tail. Its variation is shown in
Map 143 and is very similar to the variation seen in total band counts.
Diadophis punctatus (Linnaeus)
I examined 295 specimens of Diadophis punctatus from Florida,
southern Georgia and southern Alabama (Map 144) for potential geographic
variation in each of 23 characters (see Appendix A). None of the
characters examined was found to be correlated with body size. Those
characteristics which show apparent patterns of geographic variation
are discussed below.
Number of ventrals. Maps 145 and 146 show the geographic variation
in ventral numbers for male and female ringneck snakes from Florida. The
correlation coefficient between ventral counts and sex was found to be
0.6314. Snakes from the Panhandle and northern peninsula have the highest
ventral counts. Concordance between the sexes is very good.
Number of subcaudals. Males have more subcaudal scales than females
(r = 0.7425). The geographic variation in subcaudal counts is presented
in Maps 147 and 148. Its variation is more complex than that observed in
ventral count variation.
Ventrals plus caudals. The summation of ventrals and subcaudals
is not correlated with sex (r = 0.0624), and the samples could be lumped
to produce Map 149. The highest ventral scute counts occur in snakes
from the Panhandle and northern peninsula. Snakes with intermediate
values occur in the southern tip of the peninsula, and specimens with
low ventral plus caudal counts are found on the Lower Keys and most of
the south-central Florida peninsula.
Percent tail. Relative tail length was found to be correlated
with sex (r = 0.8058), and the males and females have been mapped
separately (Maps 150 and 151). Generally speaking, Diadophis from the
southern half of the peninsula and the Keys have proportionately longer
tails than specimens from the remainder of the state.
Numer of supralabials. Most southeastern ringneck snakes have 16
total supralabial scales. Occasional specimens have only 14 upper labials,
and these tend to occur more frequently in the Panhandle and northern parts
of the Florida peninsula. See Map 152.
Subcaudal spots. The number of small black spots on the underside
of the tail varies geographically as shown in Map 153. Diadophis
punctatus from the Florida Panhandle, the Gulf Hammock region and the
Lower Keys often have such spots. Specimens from the remainder of the
state usually lack these black spots.
Labial pigmentation. Pigment on the labials of Florida D. punctatus
may be in discrete spots, diffuse smudges or lacking altogether. Map 154
shows the geographic variation of this character for both sexes of Florida
ringnecks. Specimens from the Lower Keys have no such pigment or it is
very diffuse. Specimens from the southern Everglades and parts of the
Gulf Hammock region usually have diffuse labial pigment. The majority
of Florida ringneck snakes have labial pigment confined to discrete
Pigmented supralabials. The number of upper labial scales with
black pigment varies geographically as in Maps 155, 156, and 157.
Specimens from South Florida and the Keys as well as many from the Gulf
Hammock region score high for this character. Snakes from the Panhandle
have the lowest states for this character, while specimens from the
remainder of the state receive intermediate scores for number of pigmented
Pigmented labials. This character is the summation of labials,
supra- and infra, with black pigment. Its geography is depicted in
Map 158. Ringneck snakes from the Gulf Hammock region and the southern
tip of the peninsula including the Keys have the highest states for
this character. Specimens from the Panhandle have the least pigment
on the labials and most peninsular specimens are intermediate for pig-
Ring interruption. Most ringneck snakes from the Florida peninsula
have a middorsal break in the neck ring. Map 159 shows, however, that
the geographic variation in this character is more complex. Specimens
from the Lower Keys and the Gulf Hammock may lack a nuchal ring alto-
gether. Many ringnecks from coastal areas do not have a middorsal ring
interruption. Most Diadophis from the Panhandle have complete neck
Ring width. The width of the nuchal ring varies geographically as
shown in Maps 160, 161, and 162. Diadophis from the Panhandle and northern
parts of the peninsula have the widest nuchal rings.
Ring displacement. The position of the nuchal ring relative to
the parietal scales varies geographically as shown in Map 163. Snakes
from the Panhandle and northern peninsula tend to have their neck rings
originating nearer to the parietals than do more southerly specimens.
Factor 2. The second factor extracted from a matrix of 19
characters accounted for 13.2% of the total variation, and much of the
variation in number of pigmented supralabials and infralabials as well
as labial pigmentation type. Map 164 shows that Diadophis from the
Gulf Hammock region and the southern peninsula including the Keys have
the highest scores for factor 2. Snakes from the Panhandle have the
lowest scores, and specimens from the remainder of the peninsula have
Factor 4. This multivariate character describes 7.8% of the
variation in 19 characters as well as part of the variation in numbers
of labials with pigment. Again, high factor 4 scores are associated with
snakes from the Gulf Hammock region and the Keys. Snakes from most of
the Panhandle and northern peninsula get low scores for this factor,
implying wider, interrupted, more anterior rings. Map 165 shows the
geographic variation of factor 4.
Cemophora coccinea (Blumenbach)
I analyzed data on 90 specimens of Cemophora coccinea from Florida
and southern Georgia (Map 166) for geographic variation in each of 17
characters (Appendix A). The sample size is unfortunately small. The
characters examined which seem to have patterns of geographic variation
are discussed below.
Number of ventrals. The number of ventral scales in Florida
scarlet snakes correlates with sex (r = 0.4144), and the sexes have been
mapped separately (Maps 167 and 168). The correspondence between the
sexes is very good. Sample sizes are small, but scarlet snakes from
the Panhandle and the southern end of the peninsula tend to have fewer
ventral scales than specimens from the middle of the peninsula.
Number of subcaudals. Maps 169 and 170 show the geographic variation
for male and female Cemophora subcaudal counts. The number of subcaudals
correlates only weakly with sex (r = 0.3385), with males having the higher
counts. In both sexes, snakes from the Panhandle tend to have fewer
subcaudals than specimens from the peninsula. The sample sizes are
small, but it seems that specimens from the southern tip of the state
also have low subcaudal counts.
Ventrals plus caudals. This character summarizes nicely the
variation apparent in the two proceeding counts, and is not correlated
with sex (r = 0.2240). Map 171 shows the pattern of geographic varia-
tion in ventral plus caudal counts for both sexes of Florida C. coccinea.
Lowest counts are associated with snakes from the Panhandle east to the
Suwannee River and from the southern end of the peninsula, south of
Lake Okeechobee. Scarlet snakes from the remainder of the peninsula
between the Suwannee River and Lake Okeechobee have higher ventral scute
Percent tail. Tail length divided by total length varies geograph-
ically as shown in Maps 172 and 173. There is no significant difference
between relative tail lengths in the two sexes for this species. In
general, scarlet snakes from more southern localities tend to have pro-
portionately longer tails, although the tendency is not sharply defined.
In the Panhandle, females have relatively longer tails, but males have
shorter tails. The result of this is that sexual dimorphism is pronounced
in the few Panhandle specimens examined but lacking in the peninsular
specimens. The two Panhandle females have longer tail length/snout vent
ratios (mean = 0.158) than the three males available (mean = 0.140).
Infralabials. Florida C. coccinea may have 14, 16 or 18 lower
labial scales. Specimens from the Everglades Region frequently have
14 infralabials, while many specimens from the central and north-central
peninsula have 18.
Supralabials. Map 174 shows the geographic variation in the number
of supralabial scales in Florida Cemophora. Snakes from the Panhandle
west of the Aucilla River, northeast Florida, and the Everglades usually
have 11 to 13 upper labials. Snakes from the central peninsula tend to
have 14 or 15.
Number of red body bands. Map 181 depicts geographic variation in
the number of red cross bands in Florida scarlet snakes. Snakes from
northern Florida and the Osceola National Forest region, just south of
Tampa Bay, and the Everglades have fewer cross bands than snakes from
the rest of the state.
Number of red tail bands. The number of cross bands on the tail in
Florida Cemophora correlates with sex (r = 0.5179). Maps 182 and 183
show the nature of variation in this character. Although concordance
between the sexes is poorly developed, a general tendency for scarlet
snakes from more northern localities to have more tail bands is apparent.
Dorsal scale rows. Ninety and four-tenths percent of Florida
Cemophora have 19 dorsal scale rows at mid body. Specimens studied from
the Everglades have 17 anterior scale rows, and specimens from extreme
northern Florida tend to have 21, while those from the remainder of
Florida usually have 19 anterior scale rows. All Cemophora examined
from the Panhandle have 19 posterior scale rows. In the peninsula,
many specimens have fewer than 19. See Maps 175 and 176.
Length of the white bands. Maps 177, 178, and 179 show the vari-
ation in the length (in scale-lengths) of the first and fifth white
bands in Florida Cemophora. The geographic patterns will be discussed
multivariately under factor 1.
Length of red bands. There seems to be three areas where longer
red bands are the rule: northern Florida, especially the northern parts
of the Panhandle, around Lake Okeechobee, and the extreme southern
Everglades. See Map 180.
Number of red cross bands. The only consistent tendency between
the sexes in the geography of the variation in number of red body bands
is that specimens from the Everglades region usually have fewer such bands
(Map 181). The number of tail bands varies geographically as shown in
Maps 182 and 183. In this character, it is difficult to discern a pattern
that is common to both sexes.
Factor 1. The first factor accounts for 24.7% of the total variation
in 14 characters, and 80% of the variation in length of the first white
band and 64% of the variation in length of the fifth white band. Thus
factor 1 describes variation in the length of the white bands in C.
coccinea. Its geographic variation is figured in Map 184. Snakes from
the western Panhandle, north central peninsula, and west coast tend to
have longer white bands.
Factor 2. The second factor represents a multivariate assessment
of red bands in Cemophora. It accounts for much of the variation in
number of red body bands, and length of the first and fifth red bands.
It represents 16% of the total variation, and is mapped in Map 185.
The southern half of the peninsula tends to support populations of
scarlet snakes with fewer but longer red bands, while the pattern in the
rest of Florida is unclear.
Factor 3. Accounting for much of the variation in number of supra-
and infralabials and ventrals plus caudals, the third factor represents
13.5% of the total variation, and is figured in Map 186. The Panhandle
and north Florida as well as the Everglades are separated out with low
values for factor 3. These areas have cemophora with fewer labials and
fewer ventrals plus caudals.
Factor 4. This factor accounts for 10.3% of the total variation,
and gets high loadings from the three dorsal scale row variables.
Cemophora with higher scale row counts occur in much of the Panhandle
and northern half of the peninsula. See Map 187.
Tantilla coronata (Baird and Girard), Tantilla reZicta Telford, and
Tantilla oolitica Telford
I analyzed data on 198 specimens of Tantilla from Florida (Map 188)
for possible geographic variation in each of 11 characters (Appendix A).
Those characters which seem to vary geographically are discussed below.
Number of ventrals. Females generally have more ventrals than males
(r = 0.4150). The sexes have been mapped separately in Maps 189 and 190.
In both sexes, the number of ventral scales decreases southward on the
peninsula until the southernmost localities in Florida. Specimens examined
from Miami and Key Largo have high ventral counts, reminiscent of northern
Number of subcaudals. The highest subcaudal counts occur on snakes
from the Big Bend region of northern Florida and decrease quickly to the
west and more gradually to the south. Maps 191 and 192 show the variation
for males and females which differ in mean subcaudal counts (r = 0.4194).
Ventrals plus caudals. The summation of the proceeding two counts
produces a variable which does not correlate with sex (r = 0.0783). Maps
193, 194, and 195 show how this character varies geographically. Tantilla
from the Panhandle and northern peninsula have the highest counts, decreas-
ing clinally southward on the peninsula. Tantilla from the Miami area are
more like northern populations in this character, with higher ventral
Percent tail. Relative tail length varies geographically as shown
in Maps 196 and 197. Males usually have proportionately longer tails than
females (r = 0.6339). Both sexes vary geographically with relatively
longer tails occurring in populations inhabiting the Big Bend region and
the western coast of the peninsula.
Supralabials. Tantilla from Highlands County often have six
upper labials on each side, while specimens from the remainder of the
state usually have seven. Map 198 shows this variation.
Parietal pigmentation. TantiZta examined from the Miami area and
parts of the Big Bend region of northern Florida lack light spots on the
parietal scales. Maps 199, 200, and 201 show the geographic variation
of this character. Specimens from the Panhandle east to the Ochlockonee
River, Highlands County, and the lower east coast of the peninsula have
extensive parietal light markings, often forming a partial nuchal ring.
Snakes with intermediate amounts of light pigment in the parietal region
occur throughout most of the Florida peninsula.
Snout pigmentation. Map 202 shows the geographic variation in the
amount of light pigment on the snout in Florida Tantilla. Most specimens
examined from the lower east coast of the peninsula have a large white
spot on the rostral and internasal scales. Tantilla from Highlands
County have some light pigment on the snout, while the majority of
Tantilla populations studied lack this characteristic.
Factor 2. This multivariate character accounts for 18.1% of the
total variation in eight characters and most of the variation in number
of ventrals plus caudals and parietal and snout pigmentation. Map 203
summarizes the geographic variation in factor 2. The specimens examined
from the Miami area and the Big Bend region receive the highest factor 2
scores, implying reduced light pigmentation and more ventrals plus caudals
in these areas. Panhandle Tantilla and those from Highlands County have
the lowest factor 2 scores. Intermediate values are seen in specimens from
the remainder of the state.
Sistrnuus miliarius (Linnaeus)
I examined 320 specimens of Sistrurus miliarius from Florida,
southern Georgia and southern Alabama (Map 204) for possible variation
in each of 22 characters (see Appendix A). Much of the variation studied
in this species was very noisy owing to a large degree of individual
variation among snakes from nearby localities. Smoothing algorhythms
might help clarify the picture by plotting averages of adjacent specimens.
However, the mapping procedure employed here does not smooth, but rather
plots the data exactly as they appear. Three ratios concerned with the
frontal scale and snout-vent length were correlated with snout-vent
length and excluded from further consideration. Those characters for
which geographic trends were noted are discussed below.
Number of ventrals. Females tend to have more ventral scales than
males (r = 0.4672), and the sexes have been treated separately in Maps
205 and 206. The lowest ventral counts are apparently associated with
snakes from the Panhandle, while the highest occur in snakes from the
Number of subcaudals. Males usually have more subcaudal scales
than females (r = 0.6267). Maps 207 and 208 show the geographic variation
in subcaudal counts for Florida pigmy rattlesnakes. The pattern is like
that seen in the variation of ventral counts. Specimens from the Panhandle
have the lowest counts, and specimens from the central part of the penin-
sula tend to have the most subcaudal scales. As with the proceeding
character, pigmys from the lowland region south of Lake Okeechobee tend
to have lower counts than specimens occurring just north and south of
Ventrals plus caudals. Although ventrals and caudals are both
correlated with sex, their summation is not (r = 0.0570). The variation
in this character is depicted in Map 209 for the combined sexes. Snakes
from the Panhandle east to the vicinity of the Aucilla River and in the
Everglades region south of Lake Okeechobee tend to have the lowest ventral
scute values. Sistrzrus from the peninsula excluding the Everglades seem
to show a clinal increase in ventral and subcaudal counts to the south.
Dorsal scale rows. Maps 210 and 211 show the geographic variation
in the number of dorsal scale rows at two points along the body in Florida
Sistrurus. The lowest scale counts tend to be associated with specimens
from the Florida Panhandle and the highest with specimens from coastal
areas along the peninsula. Snakes with intermediate counts occur in
most of the interior peninsula. A lot of individual variation among
snakes from closely separated localities creates the noisy surfaces
depicted in the maps.
Number of blotches. Pigmy rattlesnakes from the Panhandle west of
the Ochlockonee River and from many coastal areas on the peninsula
frequently have more dorsal body blotches than specimens from the
remainder of the study area. Maps 212 and 213 show the patterns of
geographic variation for male and female S. miliarius.
Spot/space ratio. This character varies geographically as shown
in Map 214. Pigmys occurring west of the Ochlockonee River in the Pan-
handle and in the Everglades usually have crossband-like body blotches
that are more narrow than their inter-blotch spaces. Snakes from coastal
areas in the peninsula tend to have larger blotches with very narrow
spaces between. Many specimens from interior localities on the penin-
sula have spot/space ratios intermediate between these.
Spot shape. When the length (in scale lengths) of a typical mid-
body dorsal blotch is divided by its width, the resulting ratio varies
geographically as depicted in Map 215. Sistrurus from the Panhandle and
Everglades tend to have dorsal blotches that are more like crossbands.
Specimens from most of the peninsula have blotches that are more roundish.
Dorsal contrast. Sist-urcs from the Panhandle and parts of south-
ern Florida including the Everglades have more contrast between their
dorsal blotches and their ground color, resulting in a more distinctive
dorsal pattern. There may also be a tendency for snakes from coastal
areas to have higher values for this character as well. See Map 216.
Ventral pigmentation. Pigmy rattlers from much of the Panhandle
west of the Ochlockonee and from South Florida tend to have more white
ventrally than specimens from the remainder of the state. The geographic
variation of this character is displayed in Maps 217 and 218.
Crotalus adamanteus Beauvois
I examined 194 specimens of Crotalus admnanteus from Florida (Map
219) for possible geographic variation in each of 18 characters (Appendix
A). There is a large amount of individual variation in most of the
characters examined. Patterns of geographic variation are obscured by
this individual variation, with the result that many of the maps appear
noisy. Several of the characters examined appeared not to vary geograph-
ically within the study area. Those for which geographic patterns could
be recognized are discussed below.
Number of ventrals. Female rattlesnakes usually have more ventral
scales than males (r = 0.6808). The only geographic trend apparent in the
variation of ventral counts is that snakes from the Florida Keys and adja-
cent mainland consistently have more ventrals than specimens from elsewhere
in the study area. There is apparently no dine within Florida.
Number of subcaudals. Males have more subcaudals than females (r =
0.7971). The variation in subcaudal counts within Florida is complex and
seems to have no geographic component based on the sample available.
Dorsal scale rows. I can discern no geographic trend in the vari-
ation of dorsal scale row counts in Florida C. adamanteus. Map 220 shows
the spotty occurrence of specimens with higher scale row counts.
Number of infralabials. The variation in this character is also
noisy, but some geographic trends appear to exist. Maps of the variation
in each sex are provided (Maps 221 and 222), and some correspondence is
apparent. Rattlesnakes from northern Florida and from the Lower Keys
typically have more lower labials than specimens from the central penin-
sula. Map 223 shows the variation in infralabial counts for the com-
Dorsal blotches. The number of diamonds on Florida Crotatus varies
geographically as shown in Map 224. There seems to be a tendency for
specimens from the Panhandle, the Keys, and the west coast of the penin-
sula to have fewer dorsal blotches than specimens from interior localities
on the peninsula.
Labial pigmentation. The number of immaculate supralabials in
Florida C. adamanteus varies geographically as shown in Maps 225, 226,
and 227. There is a tendency for specimens from the western Panhandle,
the Upper Keys and the Everglades region to have dark pigmentation on all
upper labial scales. Rattlesnakes from the Central Ridge east of Tampa
Bay and from the Lower Keys frequently have the most immaculate labials.
Ventral pigmentation. Maps 228, 229, and 230 depict the geographic
variation in the degree of ventral dark smudging in Florida rattlesnakes.
There is a tendency for Crotalus from the Panhandle, the Everglades region
and the western coast of the peninsula to have lighter ventral patterns
than specimens from the interior of the peninsula. Snakes from the
Lower Keys usually have darker ventral patterns, more like specimens from
Factor 1. The first factor accounts for 16.7% of the total variation
in 14 characters, and most of the variation in labial pigmentation and the
number of immaculate supralabials. Map 231 shows its geographic variation
in Florida. Snakes from the western Panhandle, the Everglades region and
the Keys typically have darker supralabials. Specimens examined from the
Central Ridge consistently have the least labial pigmentation.
Factor 2. The second factor accounts for 12.8% of the total varia-
tion in 14 characters, and most of the variation in infralabial number.
Factor 2 also receives a contribution from the variation in the number of
supralabial scales. Map 232 shows how factor 2 varies geographically.
Specimens from the Florida Keys and parts of northern Florida tend to
receive higher factor 2 scores, and specimens from the southern half of
the peninsula usually score the lowest.
I mapped environmental data from 196 weather stations in Florida,
Georgia and Alabama (Map 233). Maps 234 through 245 show the results of
the variables examined.
I used factor analysis to compare one hundred maps of morphological
variation in fifteen species of Florida snakes. The SPSS factor analysis
procedure extracted 24 factors (patterns of geographic variation) which
accounted for 76.9% of the total information in the maps. The maps
(characters) analyzed and their communalities (percentage of variation
accounted for) are presented in Table 1. Table 1 also gives the factor
(pattern of geographic variation) with which each character is most
Most of the geographic variation observed in Florida snakes is
distributed along a north-south axis. Factor 1 accounted for 26.7% of
the variation explained by the factor analysis procedure, and most of
the variation in those characters which showed distinct north-south
changes in character states. Of the 100 maps analyzed, factor 1 accounted
for more than half of the variation in ten maps, and was the principal
pattern of variation for 38 characters (Table 1). In all the maps,
factor 1 accounts for that part of the geographic variation that is
north-south oriented. Factor 1 actually identifies two important
patterns of geographic variation. The Suwannee River Pattern is shared
by those species which demonstrate distinct character state changes
occurring in the region of the present Suwannee River. The North-South
Pattern includes those species which show clinal changes in character
states southward on the Florida peninsula.
Table 2 presents the characters (maps) and their factor loadings
on factor 1. These factor loadings may be interpreted as the relative
importance of the various characters (i.e. maps) to the definition of the
factor. Their square is the percentage of variation accounted for by
the factor. The important observation to be made here is that nearly
all of the characters analyzed have a significant portion of their vari-
ation which may be described as a north-south gradient or a character
shift along north-south lines.
The North-South Pattern is the most important pattern of geographic
variation observed in Florida snakes. Table 2 presents the factor load-
ings for all maps analyzed. Even characters which are primarily varying
in some other pattern usually have some component of their variation which
can be described as north-south oriented. Many characters (e.g. ventrals
plus caudals in Storeria dekayi, Thamnophis sirtalis, Coluber constrictor,
Elaphe guttata, Lampropeltis getulus, Lampropeltis triangulrn, Cemophora
coccinea, Diadophis punctatus, and Sistrurus miliarius, and blotches or
crossbands in Elaphe guttata, Elaphe obsoleta, and Lwnpropettis getulus)
vary primarily in the North-South Pattern. Figure 3 diagrammatically
represents this important pattern of geographic variation.
The Suwannee River Pattern is presented in Figure 4, and is best
exemplified by the geographic variation observed in such characters as
midbody scale rows in Storeria dekayi, number of supralabials in
Thamnophis sacritus, midbody scale rows in Lampropeltis getulus, and
dorsal blotch development in Elaphe obsoleta. Like the North-South
Pattern, this pattern is occasionally superimposed on other patterns
of geographic variation (e.g. number of labial spots in Diadophis
The second factor extracted from the matrix of 100 maps accounted
for another large proportion of the information. Table 3 presents
factor loadings on factor 2 for the variables which were important in
its construction. This pattern of geographic variation may be called the
Everglades Pattern and is the principal pattern of variation for such
characters as the number of immaculate labials in Crotalus adamanteus,
development of the dorsal blotch border in Elaphe guttata, and number
of crossbands, number of infralabials, and length of the red bands in
Cemophora coccinea. A diagrammatic representation of the Everglades
Pattern is provided in Figure 5.
Factor 3 describes a North Florida-Lower Keys Pattern, and is
best illustrated by the variation seen in supralabial and gular black
pigmentation in Coluber constrictor, and number of preocular scales in
Storeria dekayi. This pattern is concerned with a phenetic resemblance
between populations in northern Florida, the region east of Tampa Bay
and the Lower Florida Keys. Ventral white in Coluber constrictor,
ventrals plus caudals in Storeria dekayi, and supralabial-loreal contact
in Coluber constrictor also have elements of this pattern in their
geographic variation. Factor loadings for significant characters are
presented in Table 4, and Figure 6 illustrates the Lower Keys Pattern.
The fourth factor is the principal pattern of variation for number
of infralabials in Masticophis flagellum, number of infralabials in
Lampropeltis triangulzu, and the amount of white in the ventral pattern
of Sistrurus miliarius. The Panhandle-Everglades Pattern describes the
situation in which infraspecific populations in the Florida Panhandle
and the Everglades region are more similar to each other than either is
to geographically intermediate populations. Spot shape in Sistrurus
miliarius, supralabial brown pigment in Coluber constrictor, and ring
separation in Diadophis punctatus also have elements of this pattern
in their geographic variation. Table 5 gives the factor loadings for
characters with significant contributions to factor 4, and Figure 7
illustrates the Panhandle-Everglades Pattern.
The Coastal Pattern is defined by factor 5, and illustrated in
Figure 9. Table 6 presents the factor loadings for important characters
varying in a coastal manner. Such characters as ventral pigmentation
in Elaphe guttata, dorsal blotch border in the same species, labial
pigmentation in Diadophis punctatus, dorsal scale rows in Sistrurus
miliarius, and body blotches in Sistrurus load highly on factor 5.
In this pattern, populations from coastal regions and the Florida Keys
(when they occur there) tend to form a phenetic entity distinct from
populations farther inland.
Factor 6 represents the Okeechobee Pattern and is illustrated in
Figure 8. This pattern of geographic variation is seen in the number
of midbody scale rows in Lcnpropeltis trianguZun, number of supralabial
scales in Florida tantilla, and nuchal pigmentation in Tantilla. The
Okeechobee Pattern is characterized by geographic variation in which the
region around Lake Okeechobee, and especially the high ridge immediately
to the west, is inhabited by snakes phenetically different from con-
specifics to the north and to the south. Table 7 gives the factor
loadings for important characters in factor 6.
The remaining factors produced by the statistical procedure account
for smaller amounts of the information in the maps, and their interpre-
tation is omitted.
The multivariate analysis of contour maps of geographic variation
in Florida snakes has shown that the variation can be reduced to seven
major patterns. There are other patterns, but these are of less importance,
and are shared by fewer species. For example, Thawnophis sau~itus ground
color is distinctive in the Gulf Hammock region of central Florida. This
character fell out of the analysis in factor 22, which also received
significant loadings from Crotalus adamanteus ventral pigmentation and
T. sauritus parietal spot size. However, factor 22 accounted for only
1.2% of the variation explained in the analysis. Tables 8 through 11
present the factor loadings on factors 7 through 10 for the characters
which showed correlations with these factors.
The major patterns of geographic variation in Florida snakes
(North-South Pattern, Suwannee River Pattern, Everglades Pattern, North
Florida-Lower Keys Pattern, Panhandle-Everglades Pattern, Coastal Pattern,
and Okeechobee Pattern) account for 60.4% of the information contained in
the original contour maps. The remaining information is partitioned into
lesser patterns, and in some cases represents variation unique to a
particular species or character.
In order to search for correlations between the patterns of geographic
variation and environmental factors, I analyzed 17 parameters of environ-
mental variation. When these variables were factor analyzed, three
factors were extracted that accounted for 74.4% of the variation. These
factors are clearly defined as 1) average temperature, 2) maximum summer
temperatures, and 3) average rainfall. When the factor analysis of the
snake morphological data was carried out, representative variables from
the climatic data matrix were included.
Average annual temperature loads vary highly on factor 1 (Table 1).
Inspection of Map 235 reveals that average annual temperature belongs in
the North-South Pattern. Thus the North-South Pattern of geographic
variation is highly correlated with mean annual temperature. Similarly,
The Panhandle-Everglades Pattern is correlated with mean annual rainfall,
and the Coastal Pattern is correlated with maximum summer temperatures
Although correlation does not prove cause and effect, its existence
does suggest the possibility of just such a relationship. Until experi-
mental falsification is at hand, I would hypothesize that the influence of
mean annual temperature on morphological variation in Florida snakes is
a great one. Using the same line of reasoning, very high summer maximum
temperatures may be responsible for the Coastal Pattern of geographic
variation, and mean annual rainfall may influence snake morphology to
vary in the Panhandle-Everglades Pattern.
The remaining patterns of variation are not highly correlated with
any environmental variables tested. These patterns may be maintained
by some other environmental variables (biotic or physical) that were
not examined, or they may be remnants of previous environmental conditions.
If the latter were the case, then we would have to believe that present
selective regimes have been insufficient to direct changes in phenotypes
towards adaptation to present conditions. For example, the Suwannee
River Pattern does not seem to correlate with any of the environmental
variables tested. Snakes showing this pattern of geographic variation
may be a result of past adaptations to an insular environment during
Pleistocene high sea levels. The Florida peninsula is no longer an
island, but selection has not been strong enough to eliminate (or
"spread out") the abrupt character changes in the region of the former
"Suwannee Straits." Furthermore, adaptations acquired during periods
of former isolation may have been accompanied by reduced reproductive
compatibility with mainland populations. This process speciationn) would
tend to preserve character states in both populations even after they
became geographically rejoined. Such a pattern would not necessarily be
expected to correlate with any present environmental parameter, and
would have to be interpreted as a remnant of some past selective
Table 1. Variables used in the factor analysis, their
communalities, and the principal pattern to which
each belongs. Note that most of the variables
belong to several patterns, but only the principal
pattern is included here.
CoZuber. Gulbl k
Table 1 Continued.
Communality Principal Pattern
.51104 Factor 7
.76185 North-South, Factor 7
.54415 Factor 24
.40211 Factor 16
.70248 Suwannee River
.81187 Suwannee River
.87962 Suwannee River
.84859 Suwannee River
.86965 Suwannee River
.70095 North-South, Everglades
.68396 Factor 10
.74186 Factor 7
.74450 Factor 7
.65063 Factor 11
.65221 Factor 20
.86221 Suwannee River, Everglades
.77072 Factor 11
Table 1 Continued.
Suwannee River, Panhandle-Everglades
North-South, Suwannee River
Table 1 Continued.
Table 2. North-South pattern and Suwannee River
pattern (Factor 1). Variables followed by
an asterisk (*) have factor 1 as their
Table 2 Continued
Table 2 Continued
Sistrru. Sclsmi d
Table 2 Continued
Variable Factor Loading
Florida.Over 90 0.23942
Table 3. Everglades pattern (Factor 2). Variables
followed by an asterisk have factor 2 as
their principal pattern.
Table 4. Keys pattern (Factor 3). Variables followed
by an asterisk have factor 3 as their principal
Table 5. Panhandle-Everglades pattern (Factor 4).
Variables followed by an asterisk (*) have
factor 4 as their principal pattern.
Table 6. Coastal pattern (Factor 5). Variables followed
by an asterisk (*) have factor 5 as their
Table 7. Okeechobee pattern (Factor 6). Variables
followed by an asterisk (*) have factor 6
as their principal pattern.
Table 8. Factor 7. Variables followed by an asterisk
(*) have factor 7 as their principal pattern.
Sistruru. Sc smid
Table 9. Factor 8. Variables followed by an asterisk
(*) have factor 8 as their principal pattern.
Table 10. Factor 9. Variables followed by an asterisk
(*) have factor 9 as their principal pattern.
Table 11. Factor 10. Variable followed by an
asterisk (*) has factor 10 as its principal
Figure 3. North-South Pattern (Factor 1, in part).
Figure 4. Suwannee River Pattern (Factor 1, in part).
Figure 5. Everglades Pattern (Factor 2).
Figure 6. Keys Pattern (Factor 3).
Figure 7. Panhandle-Evergladers PattPrn (Factcr 4).
Figure 8. Okeechobee Pttern (Factor 5).
Figure 9. Coastal Pattern (Factor 6).
--------------- ------------------ ---
------- ------------- :::::::
--- ------ --- -- ;-1 1r,-- :----:-:_:: :_:::_ ::::_
..... .......-----------.... .......... .. . . ...-.-.................. ..
i] ] .............. .............. ]]]] ]:2 i_]]]]]] ]]]]]]-]]i] ]]]]K ]]]K ] -:!]K ]]] ] ]
.. . .. . .... .. . .. .... .. ... -. .
...................... ..~.~. ............. ...........
~~~~~ ~~.. ............ ...
- -- - - -
~.------- --- -----------
~~.--- ------------- -------------------------------- ---------- ----- ----------
------- ------ ----- ---- - ------------------------- ---------------- --
S.~. ...... ~ -.--.... ~...........~............................ ........................~ ~
RAP 1. Localities of 151 Storeria dekayi specimens
examined. An asterisk represents one specimen; an S
represents more than one.
I ------------ "--- *i i *-- --- --- -------.-..- - -
as- - -
MAP 2. Geographic variation in number of ventral scales in
Storeria dekayi, males only. Levels by increased shading are:
124-131, 132-138, 139-145, 146-153. Based on 69 specimens.
________ ___F____~__ r___l_____l___ __F______ lCI__~_1___
- - -.
MAP 3. Geographic variation in number of ventral scales in
Storeria dekayi, females only. Levels by increasing shading
are: 126-i31, 132-139, 139-145, 116-152. Based on 80 speci-
..-~c...-.. r-.-...~- -----.--- c-~--.---~------- ----------------- ----- -----:
I ----- ------
^ i l l l l l l l l i l::: . . .. . . .: :... .
................. ................ .........
..... :* :... "
MAP 4. Geographic variation in number of subcaudal scales in
Storeria dekayi, males only. Levels by increasing shading are:
50-56, 57-63, 64-70, 71-77. Eased on 63 specimens.
________--__ A -
-- --------------- --------.--------
moo= 7:. .. .. ii
--. --. -- --.-..
MAP 5. Geographic variation in number of subcaudal scales
in Storeria dekayi, females only. Levels by increasing
shading are: 43-9, 50-55, 56-61, 62-68. Based on 77
O ml .......... ... .. ................
MAP 6. Geographic variation in the sum of ventrals plus
caudals in Storeria dekayi, both sexes. Levels by increasing
shading are: 169-186, 187-200, 201-215, 216-227. Based on
----------------------------- ----- ----- -
MAP 7. Geographic variation in tail length divided by total
length in Storeria dekayi, males only. Darker shading repre-
sents proportionately longer talIs. Based on 67 specimens.
MAP 7. Geographic variation in tail length divided by total
length in Storeria dekayi, males only. Darker shading repre-
sents proportionately longer tails. Based on 67 specimens.
MAP 8. Geographic variation in tail length divided by
total length in Storeria dekayi, females only. Darker shading
represents proportionately longer tails. Based on 75 specimens.
-- ------------- ------ --------- --- ---------- ---- ------------ -------
_~_~... ........ .... 1/1.i
1^11... ,,~_L----~1 H,. ,,,, H..
.... in|,., .. ., ,u..
~I-m--___ I_-___i-- ll-..., ...... IIIIII 1-_iiiii
~^-IIlfL_ L~--1-11 1--.___1 I--lil- Ls--l. .I. iiiiiiii iii
C-~~~~~~ ~ ~ ~ -I'"_-_-II i.iiiiii It-._i_'--__~ -C
---------- I-----------.---- ----- ----
represents 15 scale rows; darker shading represents 17.
Based on 151 specimens.
---------- ---- ----- ------------* ---+ --- ---- I----- ---*---4------- --- ---r--- ----- --- ---- ----a
MAP 10. Geographic variation in number of preoculars (both
............iiii: ......... :::: ..... i
sides) in Storeria dekavi, both sexes. Lighter shading repre-
-------- ----- ------------- --------- ----------------------------- --------~---- -----------
MAP 10. Geographic variation in number of preoculars (both
sides) in Storeria dekavi, both sexes. Lighter shading reore-
sents 2 preocular scales; darker shading represents 3 or 4.
Based on 149 specimens.
.aaa..-i.e,aa .,.. ......... I.......*
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ... ...s ........... --- ---s--*--s --' -- *--*- i- ---*- - -
--------- ---------- _
MAP 11. Geographic variation in extent and development of
ventral dark pigment in Storeria dekayi, both sexes. Increased
shading represents increased dark pigment on the ventral sur-
face. Based on 149 specimens.