Climatic variation and the distribution of an amphibian polyploid complex


Clint Otto, Towson University, Department of Biological Sciences, 8000 York Road, Towson, MD 21215, USA. Fax: +1 410 704 2405. E-mail:


  • 1The establishment of polyploid populations involves the persistence and growth of the polyploid in the presence of the progenitor species. Although there have been a number of animal polyploid species documented, relatively few inquiries have been made into the large-scale mechanisms of polyploid establishment in animal groups. Herein we investigate the influence of regional climatic conditions on the distributional patterns of a diploid–tetraploid species pair of gray treefrogs, Hyla chrysoscelis and H. versicolor (Anura: Hylidae) in the mid-Atlantic region of eastern North America.
  • 2Calling surveys at breeding sites were used to document the distribution of each species. Twelve climatic models and one elevation model were generated to predict climatic and elevation values for gray treefrog breeding sites. A canonical analysis of discriminants was used to describe relationships between climatic variables, elevation and the distribution of H. chrysoscelis and H. versicolor.
  • 3There was a strong correlation between several climatic variables, elevation and the distribution of the gray treefrog complex. Specifically, the tetraploid species almost exclusively occupied areas of higher elevation, where climatic conditions were relatively severe (colder, drier, greater annual variation). In contrast, the diploid species was restricted to lower elevations, where climatic conditions were warmer, wetter and exhibited less annual variation.
  • 4Clusters of syntopic sites were associated with areas of high variation in annual temperature and precipitation during the breeding season.
  • 5Our data suggest that large-scale climatic conditions have played a role in the establishment of the polyploid H. versicolor in at least some portions of its range. The occurrence of the polyploid and absence of the progenitor in colder, drier and more varied environments suggests the polyploid may posses a tolerance of severe environmental conditions that is not possessed by the diploid progenitor.
  • 6Our findings support the hypothesis that increased tolerance to severe environmental conditions is a plausible mechanism of polyploid establishment.


Polyploidy is a relatively common mode of sympatric speciation in plants, but ranges from common to very rare among animal groups (deWet 1980; Futuyma 1998; Otto & Whitton 2000; Mable 2004). Speciation by polyploidy involves two steps: (1) formation of the polyploid condition by hybridization between two species (allopolyploidy) or the union of unreduced male and female gametes of a single species (autopolyploidy), or some combination of these events; and (2) establishment and persistence of viable polyploid populations in the face of potential competition with the progenitor species. Persistence of the polyploid requires some degree of competitive superiority over the progenitor or the ability of the polyploid to colonize or adapt to environments outside the progenitor's niche (niche separation: deWet 1980; Thompson & Lumaret 1992).

A predominant role of niche separation in promoting polyploid establishment is supported by patterns of differential distribution or habitat use among closely related diploid and polyploid cytotypes (Schlosser et al. 1998; Pagano et al. 2001). Often these habitats are considered to be environmentally severe and not tolerated by the progenitor species. The occurrence of many polyploid species in harsh habitats such as arid, arctic and alpine communities provides evidence that polyploids may posses an increased tolerance for severe environmental conditions (Wright & Lowe 1968; Levin 1983; Beaton & Hebert 1988). Although there have been several studies that have compared the environments of polyploids with those of their progenitors, most of these studies were based on qualitative descriptions of habitat rather than quantitative comparisons between species distributions and environmental conditions (Rothera & Davy 1986; Lumaret et al. 1987; Felber-Girard, Felber, & Buttler 1996). In addition, most of these qualitative comparisons have been made using plants as the model system, leaving most animal polyploids unstudied. The only animal polyploid complexes for which habitat use and environmental conditions have been compared are the hybridogenic European water frogs (Rana sp.; Semlitsch, Hotz & Guex 1997; Pagano et al. 2001; Plenet et al. 2005) and gynogenic minnows (Phoxinus sp.; Schultz 1977; Doeringsfeld et al. 2004). Although the above-referenced studies compared polyploid distributions to environmental conditions among habitat patches over small spatial scales (a single watershed or individual wetlands within a 100-km2 landscape), to our knowledge no studies have quantitatively compared the distribution of any animal polyploid complex to environmental variation across large spatial scales (>10 000-km2 landscape).

The gray treefrog complex is composed of two independently reproducing bisexual species, allowing the tetraploid Hyla versicolor LeConte to establish viable populations outside the range of the diploid progenitor, Hyla chrysoscelis Cope. In general, H. versicolor is distributed in large areas of allopatry across the north-eastern USA, while H. chrysoscelis exhibits a more southerly distribution (Bogart 1980; Romano et al. 1987). Zones of sympatry are common over a range of latitudes, and the geographical distribution of this species pair suggests that large-scale climatic factors may be influencing their distribution. Herein we compare the distribution of the gray treefrog complex, a diploid/tetraploid species pair, to climatic variability in the mid-Atlantic region of North America. We predict the distribution of this complex will be partitioned along a gradient of climatic conditions, with the polyploid species occurring in areas of greater climatic severity. We utilize treefrog distribution data from Maryland and Virginia, USA because of (1) the extensive treefrog distribution records collected by D.C.F., J.C.M. and R.W.M., and (2) the extensive climatic and elevation variation that exists in these states.


study species

Both members of the gray treefrog complex inhabit the eastern USA from the coastal states to as far west as Texas and Minnesota (Bogart 1980; Romano et al. 1987). Hyla versicolor, the tetraploid, is also found as far north as southern Canada, while H. chrysoscelis, the diploid, occurs in allopatry as far south as Florida. The most recent evidence suggests the tetraploid has originated multiple times from the extant diploid and two other extinct lineages of treefrogs (Holloway et al. 2006). Hyla versicolor and H. chrysoscelis are virtually identical based on morphology, hence positive identification requires erythrocyte measurements, chromosome counts or advertisement call analysis (Wasserman 1970; Jaslow & Vogt 1977; Matson 1990a). During April–June, males of both species congregate in temporary or permanent wetlands and perform their breeding calls to advertise for mates. Interbreeding within syntopic wetlands is probably prevented by different microhabitat preferences and the ability of the females to distinguish the species-specific advertisement call (Ptacek 1992; Gerhardt 1994; Keller & Gerhardt 2001). These unique and discernible advertisement calls make field identification of H. chrysoscelis and H. versicolor possible.

calling surveys

During March–August 1979–83 (Virginia) and 1986–93 (Maryland) D.C.F., J.C.M. and R.W.M. visited a total of 432 wetland sites to document the presence of H. versicolor and H. chrysoscelis. Seventeen sites were eliminated from the extreme south-western portion of Virginia to confine the analysis to areas where geographical barriers (Appalachian Mountains) would not limit the distribution of either species. Species identifications were made by listening to male advertisement calls in the field. The calls of 774 males from 181 sites were verified electronically in the laboratory. Sonograms of male advertisement calls were produced with a Kay Electronic Sound Spectrograph Model 6061B and Multigon Uniscan II spectrum analyser (Multigon Industries Inc., Yonkers, New York, USA). Temperature-adjusted pulse rates (Gayou 1984) determined directly from the sonograms were used to confirm species identifications. Field identification agreed with laboratory identification in all cases.

climate model generation

To examine the influence of climate severity on the distribution of these two hylid species, 12 climatic models and one elevation model were constructed for the states of Maryland and Virginia (Table 1). We define areas of climatic severity from the standpoint of all amphibians as areas with relatively cooler and drier conditions and greater yearly variation in temperature and precipitation. Weather and elevation data were obtained from the South-eastern Regional Climate Center (SRCC) website ( for weather stations in Maryland and Virginia, and all other stations within 25 km of the Maryland or Virginia state boundary. To be included in the models, a station had to have ≥19 years of climate data within the years 1948–93. A total of 233 weather stations recording precipitation and 173 weather stations recording temperature met these criteria and were used in the modelling process. Elevation data from the 233 weather stations that had records for precipitation were used to construct the elevation model.

Table 1.  Descriptions and abbreviations of climatic variables used to describe climatic variation across Maryland and Virginia
AbbreviationDescription of variablesMonths includedUnitsSite type mean (SD)
Hyla chrysoscelisSyntopicHyla versicolor
  1. Mean (± SD) of climatic variable, site scores resulting from principal components analysis (PCA), and canonical analysis of discriminants (CAD) for each site type are given on the right. For CAD scores, means with different letters differed significantly based on individual Wilcoxon pairwise comparison after a sequential Bonferroni correction for multiple comparisons.

AATEMPAverage annual temperatureall°C 14·12 (0·06) 13·36 (0·11) 12·30 (0·08)
CVATEMPCoefficient of variation for AATEMP between yearsAll°C 17·33 (0·05) 17·74 (0·08) 16·92 (0·09)
ABTEMPAverage temperature during breeding seasonApr–Jun°C 18·36 (0·05) 17·84 (0·10) 16·73 (0·08)
CVBTEMPCoefficient of variation for ABTEMP between yearsApr–Jun°C  4·60 (0·01)  4·69 (0·02)  5·06 (0·03)
AWTEMPAverage temperature during overwintering periodNov–Feb°C  5·15 (0·07)  4·13 (0·15)  3·20 (0·09)
CVWTEMPCoefficient of variation for AWTEMP between yearsNov–Feb°C 25·04 (0·38) 30·69 (1·25) 42·83 (1·19)
AAPRECIPAverage annual precipitationAllcm111·33 (0·28)108·33 (0·25)104·49 (0·47)
CVAPRECIPCoefficient of variation for AAPRECIP between yearsAllcm  4·52 (0·01) 4·70 (0·02)  5·23 (0·04)
ABPRECIPAverage precipitation during breeding seasonApr–Juncm 27·18 (0·05) 27·32 (0·12) 27·38 (0·07)
CVBPRECIPCoefficient of variation for ABPRECIP between yearsApr–Juncm 31·09 (0·08) 32·62 (0·15) 30·99 (0·16)
AWPRECIPAverage precipitation during overwintering periodNov–Febcm 33·20 (0·09) 32·70 (0·13) 30·17 (0·23)
CVWPRECIPCoefficient of variation for AWPRECIP between yearsNov–Febcm 25·23 (0·05) 25·58 (0·14) 25·99 (0·12)
ElevationElevationnam 54·09 (5·51) 97·94 (6·74)334·30 (15·92)
PCA I scoresScores on axis I resulting from PCA of climatic variables and elevationnana  0·54 (0·75) –0·17 (0·56) –0·93 (0·80)
PCA II scoresScores on axis II resulting from PCA of climatic variables and elevationnana  0·09 (0·72)  0·80 (0·58) –0·45 (1·3)
PCA III scoresScores on axis III resulting from PCA of climatic variables and elevationnana –0·06 (0·94)  0·07 (1·01)  0·09 (1·10)
CAD I scoresScores on axis I resulting from CAD among site typesnana  1·00 (0·84)A –0·07 (0·75)B –1·78 (1·30)C
CAD II scoresScores on axis II resulting from CAD among site typesnana –0·18 (0·89)A  1·53 (0·87)B –0·20 (1·21)A

The climate-related values we obtained from the SRCC website were monthly precipitation and temperature averages for all years between 1948 and 1993. These data were used to obtain a mean value of precipitation and temperature for all 12 months (annual, abbreviated AATEMP and AAPREC), breeding-season months (April–June, abbreviated ABTEMP and ABPREC), and overwintering months (November–February, abbreviated AWTEMP and AWPREC). Winter and breeding-season conditions were treated separately because climate during these periods may affect different stages of the gray treefrog lifecycle. Lower temperatures and precipitation levels during the breeding season (April–June) may lead to lower recruitment of gray treefrog tadpoles, while lower temperatures and precipitation levels during the overwintering period (November–February) may increase mortality rates among adults and juveniles. We considered lower temperatures and precipitation levels during these two periods to represent severe environmental conditions for larval and adult gray treefrogs. We also calculated the coefficient of variation (CV) for all climatic variables to model variation in climate between years. We predicted that areas occupied by the polyploid species would have more unstable weather patterns and therefore exhibit greater between-year variation in climate than areas occupied by the progenitor. The 12 variables we elected to use are summarized in Table 1.

Weather station data (average and CV) and elevation data were incorporated into a geographic information system running ARCMAP software ver. 8.1 (Environmental Systems Research Institute 2001) and the geostatistical extension package to generate climate and elevation models. We used ordinary, predictive kriging to predict climate and elevation values for each of the gray treefrog sites surveyed.

statistical analyses

Principal components analysis (PCA) was used to summarize relationships among climatic variables and elevation, independent of the gray treefrog distribution. The correlation matrix was used as input for the PCA and an orthogonal varimax rotation was performed to facilitate interpretation of axes. Canonical analysis of discriminants (CAD) was used to describe relationships between climatic variables, elevation and distribution of the two cytotypes. For analysis purposes, sites were classified by three types of occurrence: (1) allotopic H. chrysoscelis; (2) allotopic H. versicolor; (3) syntopic site. Because efforts to transform the raw data were not successful in meeting the multivariate normality assumption of the CAD, the CAD analysis was considered to be descriptive and a set of non-parametric Wilcoxon pairwise comparisons was used to test for differences in canonical variable scores among occurrence site types. Because the Wilcoxon pairwise comparisons were not independent, P values for these tests were corrected using the stepdown Bonferroni method available in the MULTITEST procedure of sas (SAS Institute, Cary, NC, USA).


distribution of H. chrysoscelis and h. versicolor

Within the study area, H. chrysoscelis was restricted to lower elevations of the Atlantic Coastal Plain and lower Piedmont in eastern Maryland and Virginia, with the exception of three occurrences in the upper Piedmont and five occurrences in the Blue Ridge Mountains of extreme south-western Virginia. In contrast, H. versicolor was generally restricted to the upper Piedmont, Blue Ridge and Allegany Mountains, and Ridge and Valley of western Maryland and Virginia. However, the occurrence of H. versicolor extended on to the Atlantic Coastal Plain south and north of Baltimore, Maryland and along the eastern shore of the Chesapeake Bay at its northern extreme.

Distributions of H. chrysoscelis and H. versicolor consisted of large areas of allopatry with a band of sympatry in the central portion of the study region (Fig. 1). Two main clusters of syntopic sites occurred, one at the head of the Chesapeake Bay, and one including a string of sites extending from north-east to south-west in south-central Virginia. These syntopic sites occurred along the periphery of the species’ ranges within the study area.

Figure 1.

Map of the study area (Maryland and central and eastern Virginia) showing the distribution of the gray treefrog complex and general climatic conditions in the region. Symbols indicate sites where Hyla chrysoscelis (circles), H. versicolor (triangles), or both (crosses) were present. Climate patterns are represented by site scores on axis I (a) and II (b) resulting from principal components analysis (PCA) of climatic variables. (a) Darker-shaded regions indicate warmer, wetter, less varied climatic conditions; (b) darker-shaded regions indicate areas with greater yearly variation in precipitation.

general climatic conditions

The first three axes resulting from PCA of climatic variables and elevation all had eigenvalues >1 and cumulatively accounted for 85% of the variation in climatic conditions across the study region (Table 2). The first axis was related to a north-west–south-east trend in climate variation across the study sites (Fig. 1a) and accounted for 55% of the variation in climatic conditions. Variable loadings on PCA axis I suggested that this axis was related to a gradient of higher and more stable temperatures in the lower-elevation, south-eastern portion of the study area and along the Chesapeake Bay (Table 2). AATEMP, AWTEMP and ABTEMP had positive loadings, and CVATEMP, CVWTEMP, CVBTEMP and elevation had negative loadings on PCA axis I. Additionally, AAPREC and AWPREC showed a similar pattern of increase from north-west to south-east as indicated by positive loadings on PCA axis I. An increase in PCA scores in the Ohio River Valley of extreme western Maryland suggests there is a trend of increased precipitation and lower variation in precipitation in those regions.

Table 2.  Loadings of 12 climate variables and elevation on the first three axes resulting from principal components analysis (PCA) of gray treefrog breeding-site conditions and the percentage variation accounted for by each axis
VariablePCA axis IPCA axis IIPCA axis III
  1. Abbreviations for variables are given in Table 1

AAPREC0·8525  0·4770
ABPREC   0·9424
AWPREC0·8312  0·4182
CVAPREC 0·7261 
CVBPREC 0·8987 
Percentage variation55·119·011·2

The second two axes resulting from PCA of climatic variables and elevation were related to ABPREC, CVBPREC and CVAPREC, which varied more on a much smaller scale (Table 2). Specifically, high PCA scores on axis II indicated a relatively high degree of variation in breeding season and annual precipitation (both CVBPREC and CVAPREC had loadings >0·7 on PCA axis II) that occurred in central Virginia and central Maryland near Baltimore and the head of the Chesapeake Bay (Fig. 1b). The only variable to load heavily on PCA axis III was ABPREC. In general, higher relative annual precipitation occurred in the Virginia portion of the Shenandoah Valley, on the Atlantic Coastal Plain of southern Virginia, and in central and eastern Maryland north of Washington DC (not shown in figures).

relationships between climate and gray treefrog distributions

Results from the CAD identified a gradient of colder, drier and more varied conditions in the north-western portion of the study area, to warmer, wetter and more stable conditions in the south-eastern portion of the study area. Along this gradient there was a high degree of separation between allotopic H. versicolor, allotopic H. chrysoscelis, and syntopic sites (Fig. 2). Nine of the climatic variables and elevation loaded heavily on CAD axis I (Table 3). Variables related to mean climatic conditions were positively correlated with CAD axis 1, while variables related to variation in climatic conditions (CV) and elevation were negatively related to CAD axis I. Sites where only the polyploid was found had relatively low scores on CAD axis I, indicating an association with colder, drier and more variable climatic conditions (Figs 2a and 3). In contrast, sites where only the progenitor was found had relatively high scores on CAD axis I, indicating an association with warmer, wetter and relatively stable climatic conditions. In relation to sites where only one species occurred, syntopic sites tended to have intermediate climatic conditions and CAD axis I scores. Mean CAD axis I scores were significantly different between H. chrysoscelis and H. versicolor sites (P < 0·0001), H. chrysoscelis and syntopic sites (P < 0·0001), and H. versicolor and syntopic sites (P = 0·003; Table 1).

Figure 2.

Map of the study region showing the relationship between gray treefrog site types and canonical analysis of discriminants (CAD) axis I (a) and II (b) site scores. (a) Darker-shaded regions indicate warmer, wetter, less varied climatic conditions; (b) darker-shaded regions indicate areas with greater variation in annual temperature and precipitation during the breeding season. Symbols indicate sites where Hyla chrysoscelis (circles), H. versicolor (triangles), or both (crosses) were present.

Table 3.  Loadings of 12 climate variables and elevation on the first two axes resulting from canonical analysis of discriminants (CAD) of gray treefrog breeding-site conditions in relation to site type as defined by the presence or absence of Hyla chrysoscelis and H. versicolor
VariableCAD axis ICAD axis II
  1. Abbreviations for variables are given in Table 1

CVATEMP 0·4337
CVBPREC 0·6745
Figure 3.

Histogram of canonical analysis of discriminants (CAD) site scores for each gray treefrog site type.

The second axis of the CAD was related to variation in annual temperature (CVATEMP) and breeding season precipitation (CVBPRECIP; Table 3), which discriminated syntopic from non-syntopic sites (Fig. 3). Variation in annual temperature (CVATEMP) and breeding season precipitation (CVBPRECIP) were both positively related to CAD axis II. Syntopic sites tended to have higher scores on CAD axis II when compared with non-syntopic sites, indicating greater variation in annual temperature and breeding season precipitation at sites where both species occurred (Figs 2b and 3). Although there was no significant difference in mean CAD axis II scores between H. chrysoscelis and H. versicolor allotopic sites (P = 0·049), mean scores for syntopic sites were significantly higher when compared with both types of allotopic sites (P < 0·0001 for both cases; Table 1).


Our data suggest a role of niche separation in promoting the persistence of the polyploid gray treefrog, H. versicolor, within our study region. The polyploid and the progenitor, H. chrysoscelis, occupy broad areas of allopatry separated by a narrow band of sympatry running from north-eastern Maryland to south-central Virginia. Predictions concerning polyploidy suggest that areas occupied by polyploids are considered to be environmentally severe (Bogart & Wasserman 1972; Otto & Whitton 2000; Mable 2004). In accordance, we found the polyploid species occupied areas of greater climatic severity, specifically sites that exhibited cooler, drier and more varied climatic conditions.

In our analysis, we assumed that a relatively cold, dry and variable climate represented severe environmental conditions for gray treefrogs. It has been inferred from the distributional patterns of temperate amphibians that colder temperatures associated with northern latitudes are, in part, responsible for setting the range limit for a number of amphibian species (Schall & Pianka 1978; Duellman & Sweet 1999). Furthermore, areas of low annual precipitation often have lower amphibian diversity (Schall & Pianka 1978; Duellman & Sweet 1999). The expected tolerance of H. versicolor to cold temperatures is supported by their larger body (Jaslow & Vogt 1977; Kamel, Marsden & Pough 1985; Matson 1990b; Ptacek 1996) and cell size (Cash & Bogart 1978; Chaffin & Trauth 1987; Matson 1990a) relative to H. chrysoscelis. In general, larger body size in many vertebrates, including amphibians, is positively correlated with habitats of increased latitude and elevation, and thus lower temperatures (Ashton, Tracy & de Queiroz 2000; Ashton & Feldman 2003; Morrison & Hero 2003). It remains unclear if the larger body size of H. versicolor provides it with an increased tolerance of cooler or drier climates. The limited number of laboratory studies conducted on both cytotypes show H. versicolor is not more tolerant of environmental proxies such as freezing conditions (Lukose & Reinert 1998; Irwin & Lee 2003) and desiccation (Ralin 1981). Hence evidence suggesting that the polyploid has the physiological capability to inhabit severe environments remains largely speculative. Discrepancies between previous laboratory studies and our study would best be resolved through a series of controlled field experiments that monitor the survival of both species along a gradient of environmental conditions.

Biogeographical evidence of polyploid tolerance of severe environments is supported by high incidences of many plants (Stebbins 1950), zooplankton (Beaton & Hebert 1988), fish (Leggatt & Iwama 2003; Le Comber & Smith 2004) and amphibian (Bogart 1980) polyploids in subarctic and alpine environments. Furthermore, patterns of habitat segregation between diploid and polyploid cytotypes consistently show that polyploids occur in relatively severe or poor-quality habitats (Wright & Lowe 1968; Schlosser et al. 1998; Johnson, Husband & Burton 2003). Establishment of polyploid cytotypes into areas of higher-quality habitat is probably prevented by competition with the progenitor, while range expansion of the progenitor into areas occupied by the polyploid is thought to be limited by the physical environment (Vrijenhoek 1994; Otto & Whitton 2000; Adamowicz et al. 2002).

The restriction of H. versicolor to relatively severe environments and H. chrysoscelis to relatively benign environments provides evidence that the distribution of H. versicolor is limited by competition with H. chrysoscelis. Our investigation highlighted two areas of sympatry that have the potential for competitive interactions between the two cytotypes. These areas are characterized by having high between-year variation in breeding season precipitation and annual temperature. This high degree of climatic variability provides evidence that the outcome of competitive interactions may be influenced by local climatic conditions; however, determining what factors shape the distribution of both cytotypes will ultimately require transplant (Baack & Stanton 2005) and competition (Semlitsch et al. 1997) experiments under varying environmental conditions. The high degree of climatic variation occurring in the zones of sympatry also suggest the outer range limits of both species may shift in response to changes in regional weather patterns. We recommend the distribution of this polyploid complex be monitored over a long temporal scale to determine if climate change in the mid-Atlantic region (Parmesan & Galbraith 2004) has caused shifts in its distribution pattern.


We thank Christopher A. Pague for his help in generating much of the data from Virginia and Ronald Heyer for providing facilities and use of the sonogram. Funding in Virginia was provided by grants from the Virginia Academy of Science and the Non-game and Endangered Species Program of the Virginia Department of Game and Inland Fisheries. Funding in Maryland was provided by Towson University, Department of Biological Sciences and the Towson University Graduate School. We also extend our gratitude to the numerous field assistants who aided in data collection and Evan Grant and two anonymous reviewers for comments on an earlier version of this manuscript. This study is in compliance with all current laws of the USA.