Sympatric Occurrence of Two Species of the Two-Lined Salamander (Eurycea bislineata) Complex
Abstract
Genetic analyses of contact zones between closely related taxa are critical to an understanding of reproductive isolation between species. We evaluated allelic frequencies and external morphology from one such contact zone between two members of the Eurycea bislineata complex (i.e., E. cirrigera and E. wilderae). We found that, within this zone of contact, these presumed species had significantly different frequencies of alleles at three loci. In addition, these sympatric forms were significantly different in lateral mottling pattern, tail color, and length of tail stripe. These morphological patterns were identical to those used to describe the original subspecies E. b. cirrigera and E. b. wilderae. Evidence from this zone of contact supports the hypothesis that these forms are separate species. Moreover, there is evidence of ecological and/or reproductive character displacement among these species when in sympatry.
Because many groups of plethodontid salamanders have exhibited morphological conservatism in their evolution (Wake, 1991), molecular data have become increasingly important in determining genetic divergence among, and taxonomic status of, individual populations. Over the last 20 years, genetic data have facilitated the identification of many new species. The descriptions of many of these forms resulted when genetic investigations led to the partitioning of what were recognized as widely distributed species into numbers of allo-, para-, and sympatric, morphologically cryptic species. Examples include species groups discovered within Desmognathus ochrophaeus (Tilley et al., 1978; Tilley and Mahoney, 1996), Plethodon cinereus (Highton and Webster, 1976), P. dorsalis (Highton, 1979, 1997), P. glutinosus (Highton, 1989), and Eurycea bislineata (Jacobs, 1987).
Controversy has arisen, however, regarding the recognition of some of these molecularly determined species (Highton, 1998; Petranka, 1998; Wake and Schneider, 1998). Those forms that have been demonstrated to occur sympatrically with little or no genetic exchange have been more accepted as full species than those without such supporting evidence of their reproductive isolation (e.g., D. imitator and D. ocoee, Tilley et al., 1978; P. ventralis and P. websteri, Highton, 1979, 1985; P. aureolus and P. teyahalee, Highton, 1983, 1989). In particular, the use of genetic distance as the primary factor in determining taxonomic status has been questioned (Wake, 1981; Frost and Hillis, 1990). This questioning has arisen through some authors use of a “threshold” level of genetic divergence (e.g., such as a DNei ≥ 0.15) as “the” species criterion. One problem in assessing the validity of many of these newly described species is that too little is known about the nature of genetic interactions along the contact boundaries between parapatric neighbors.
One such case is among the members of the E. bislineata complex. Jacobs (1987) used allozyme data to divide this widely ranging form into three species, E. bislineata, E. cirrigera, and E. wilderae. Although the names were derived from the recognized subspecies (Mittleman, 1966), the new taxa as now constituted do not precisely correspond to the previously described forms. Eurycea cirrigera now includes populations that belonged to both E. b. bislineata and E. b. wilderae as well as to an earlier recognized subspecies, E. b. rivicola (see map in Mittleman, 1966; Jacobs, 1987:fig. 6). Morphological differentiation, as reflected by described subspecies, thus, does not match genetic divergence in this group, at least as revealed by allozyme electrophoresis. Therefore, Jacobs (1987) used genetic distance as his basis for species determination. He found no evidence of sympatry between these three forms, but he did find neighboring allopatric populations that maintained a large genetic distance. Although Jacobs (1987) showed that morphological and genetic divergence were not correlated, as in many other species groups of plethodontids, he did indicate that E. cirrigera could be distinguished from E. bislineata and E. wilderae by its longer tail stripes.
During the spring of 1988, the senior author discovered two-lined salamanders having long tail stripes as well as those having short tail stripes in a stream in Habersham County, Georgia. In addition, the long tail-stripe salamanders appeared to be browner in coloration and had sides that were mottled to produce a row of light circles. The short tail-stripe individuals were brighter in color (yellow to orange) without lateral mottling. These color and pattern characters matched those that were diagnostic for the originally recognized subspecies, E. b. cirrigera and E. b. wilderae, respectively (Mittleman, 1966). Habersham County falls along Jacobs' (1987) approximate boundary between what he recognized as E. cirrigera and E. wilderae.
Analyses of contact zones between sister taxa are critical to our understanding of species (Larson and Chippindale, 1993) and the role of natural selection in the evolution of reproduction isolation (Arnold et al., 1993). Moreover, to evaluate these evolutionary processes, one must first have confidence that the species in question occur sympatrically. The purpose of our study was to determine whether the coexistence of these forms indicates sympatry between E. cirrigera and E. wilderae.
Materials and Methods
In April 1997 and February 1998, we collected series of both forms (hereafter designated as test-cirrigera and test-wilderae) from Nancytown Creek and its tributaries in the Chattahoochee National Forest in southern Habersham and northern Banks Counties, Georgia, for morphological and allozyme analyses. This area lies in the Gainesville Ridges of the upper Piedmont (Wharton, 1975). We collected specimens (designated reference-cirrigera) from the lower Piedmont in Upson County, Georgia, well within the range delineated by Jacobs (1987) for E. cirrigera. We also took individuals (designated reference-wilderae) from the Blue Ridge in Union County, Georgia, well within the range that Jacobs (1987) gave for E. wilderae. For all “sympatric” individuals, we assigned an a priori categorization based on the presence (as in test-cirrigera) or absence (as in test-wilderae) of lateral mottling that produced a pattern of light circles. All specimens referred to as E. cirrigera had side mottling resulting in at least the suggestion of a row of light spots along each side (Fig. 1A). The specimens referred to as E. wilderae did not exhibit this row of lateral light spots (Fig. 1B). This a priori designation enabled us to test for allozyme, body size, and color pattern differences among the forms. We predicted that, if these forms simply represented alternative morphs of the same species, then allozyme frequencies would not differ.
Citation: Ichthyology & Herpetology 2000, 2; 10.1643/0045-8511(2000)000[0572:SOOTSO]2.0.CO;2
We determined body and tail color of living specimens using a paint color chart. There were nine shades of color on the chart from brown to tan to yellow to orange to red. Each shade was assigned an odd numerical value of 1–17. Body and tail colors that fell between the shades on the color chart were assigned appropriate even numerical values (2–16). Differences in color between the populations were then tested using the Mann-Whitney U-test (Zar, 1984).
We sacrificed the specimens (10% chloretone) and removed the liver, heart, a section of ventral body wall musculature, and a hind limb for electrophoretic analysis. The extracted tissues of each individual were stored separately below −70 C. Each specimen, minus these extracted tissues, was preserved in 10% formalin and stored in 35% isopropanol. Specimens were later measured for snout–vent length (SVL) using the posterior margin of the vent. For specimens with complete tails, we determined the percentage for which the unbroken tail stripe extended beyond the vent and compared those percentages between populations with a Mann-Whitney U-test (Zar, 1984). A one-way ANOVA was used to analyze the log-transformed SVL data against population (Zar, 1984). The SVL data were log-transformed to account for unequal variances among populations. Between-group comparisons were then made using the Ryan-Einot-Gabriel-Welsch multiple range test (REGWQ) at alpha = 0.05 (Day and Quinn, 1989). Because of possible sexual differences in SVL, all comparisons using this character were made using data only from mature females. Sexual maturity of females was determined by the presence of yolked oocytes seen through the venter or by a female's attendance of a clutch.
In 1997, reference-cirrigera (n = 8), test-cirrigera (n = 13), test-wilderae (n = 15), and reference-wilderae (n = 10) were analyzed with cellulose acetate electrophoresis. Frozen tissues were homogenized with 80 µl of grinding buffer (for 100 ml: 100 µl triton 100× detergent, 10 mg NADP, 100 mg DTT, and 100 ml distilled water). Homogenates were centrifuged for 5 min at 7000 rpm. The supernatant was then used for analysis. Standard methods for cellulose acetate electrophoresis were used (P. D. W. Hebert and M. J. Beaton, unpubl.) on six presumptive loci: aspartate aminotransferase (AAT-1, -2), isocitrate dehydrogenase (IDH-1, -2), lactate dehydrogenase (LDH-1), and phosphoglucomutase (PGM). Gels were initially run with a tris-citrate buffer, pH 8.0 (Richardson et al., 1986) for 35 min at 100 v. However, subsequent runs revealed that bands displayed clearer separation at a pH of 8.5. Screening revealed three loci that were polymorphic (IDH-1, IDH-2, and AAT-1) and three that were monomorphic (AAT-2, LDH-1, and PGM).
In August 1998, the hypothesized zone of sympatry was reanalyzed for allozymic differences between test-cirrigera (n = 10) and test-wilderae (n = 26). This second analysis was done to increase sample size and account for possible heterozygote deficiencies. The 1998 sample was analyzed for seven presumptive loci with starch gel electrophoresis: aspartate aminotransferase (AAT-1, -2), isocitrate dehydrogenase (IDH-1, -2), lactate dehydrogenase (LDH-1, -2), and malic enzyme (MDHP). Standard methods for horizontal starch gel electrophoresis were followed (Murphy et al., 1990; Tilley and Mahoney, 1996). Tissues were sonicated and centrifuged for 5 min at 14,000 rpm. Buffer systems were based on Tilley and Mahoney (1996). Following electrophoresis, tissue samples were stored at −70 C.
For all loci, allelic mobility was determined by measuring the distance (in millimeters) of alternative alleles to the most common allele, designated as 100. Corresponding letter designations were given to each allele, with the farthest migrating allele labeled a. Allelic frequencies were calculated for each locus across all populations. Data from the cellulose acetate and starch gel electrophoretic methods revealed identical results (i.e., same number of alleles and frequencies). Therefore, the data from the two techniques were combined for frequency and statistical analyses. We used the Fisher's exact test, which generates probabilities of occurrence, to compare allelic frequencies among populations (Zar, 1984).
Results
There were no significant differences in either body color (Mann-Whitney U = 42, P = 0.4689, Fig. 2) or tail color (U = 38.5, P = 0.3282, Fig. 2) between test-cirrigera (n = 13) and reference-cirrigera (n = 8) nor in body color (U = 64, P = 0.6506, Fig. 2) or tail color (U = 41, P = 0.072, Fig. 2) between test-wilderae (n = 16) and reference-wilderae (n = 9). There were significant differences between test-cirrigera and test-wilderae in both body color (U = 185.5, P = 0.0003, Fig. 2A) and tail color (U = 208, P = 0.0001, Fig. 2A). Test-wilderae were more yellow-orange, as opposed to yellow-brown, in body color and orange-red, as opposed to yellow, in tail color (Fig. 2A). There was no overlap in tail color between the test morphs (Fig. 2A).
Citation: Ichthyology & Herpetology 2000, 2; 10.1643/0045-8511(2000)000[0572:SOOTSO]2.0.CO;2
The one-way ANOVA indicated significant variation in mature female SVL among the four samples (P = 0.0001, Table 1). The REGWQ posthoc test revealed no significant differences between reference-wilderae (n = 3, X ± 1 SD = 37.2 ± 2.07 mm) and either test-wilderae (n = 12, X = 36.67 ± 1.51 mm) or reference-cirrigera (n = 8, X = 38.53 ± 3.06 mm). However, test-cirrigera (n = 13, X = 44.08 ± 2.56 mm) had a significantly longer SVL than reference-cirrigera, test-wilderae, or reference-wilderae. The frequency distributions of SVL for sympatric female test-cirrigera and test-wilderae are presented in Figure 3A. Frequency distributions of female SVL for allopatric populations of E. cirrigera (Poplar Cove Spring, Lafayette County, Mississippi; Marshall, 1997) and E. wilderae (Station Creek, Macon County, North Carolina; Bruce, 1988) are presented for comparative purposes (Fig. 3B).
Citation: Ichthyology & Herpetology 2000, 2; 10.1643/0045-8511(2000)000[0572:SOOTSO]2.0.CO;2
The percentage of tail covered by tail stripes varied from 9.1–52.2 in reference-wilderae (n = 6) and from 13.3–31.6 in test-wilderae (n = 7). There was no significant difference between the two (U = 12, P > 0.20). All specimens of reference-cirrigera (n = 8) and test-cirrigera (n = 6) had 100% coverage of the tail with unbroken stripes. There was a significant difference in percentage of tail stripe coverage between test-wilderae and test-cirrigera (U = 56, P < 0.001).
Allelic frequencies are shown in Table 2. Among polymorphic loci, we found that the allelic frequencies for IDH-1 were similar between the reference-wilderae and the test-wilderae (P > 0.99). The allelic frequencies for IDH-1 between test-wilderae and test-cirrigera were significantly different from one another (P < 0.0001). Test-cirrigera was almost fixed, whereas test-wilderae possessed an alternative allele in greater frequency (Table 2). However, the test-cirrigera population was significantly different (P < 0.0001) from the reference-cirrigera population. IDH-2 was not significantly different between test-wilderae and test-cirrigera (P = 0.592).
The reference-wilderae and -cirrigera samples are fixed for alternate alleles at AAT-1 (Table 2). The test populations were not fixed, but nearly so, for the alleles of their respective reference populations for AAT-1. Only three of 41 test-wilderae and two of 23 test-cirrigera possessed alleles different from their corresponding reference populations for AAT-1 (Table 2). There was a highly significant difference in allelic frequency between the test-wilderae and test-cirrigera (P < 0.0001). There was no significant difference in the frequencies of these alleles between respective reference and test samples (P = 0.580 for E. wilderae; P = 0.562 for E. cirrigera).
The populations of test-cirrigera and test-wilderae analyzed with starch gel electrophoresis revealed two additional polymorphic loci, LDH-2 and MDHP (Table 2). The test-wilderae population was fixed at the LDH-2 locus, whereas the test-cirrigera population possessed three alleles (Table 2). The test populations were significantly different from one another (P < 0.0001). There were also differences in the number of alleles between test populations at the MDHP locus. Although there were no significant differences between the populations at the MDHP locus (P = 0.376), too few test-cirrigera individuals were examined to make any firm conclusions (Table 2).
Discussion
There was no sign of morphological intermediacy in the test samples; both forms were clearly discernible by tail color, the presence or absence of light lateral spots, and the extent of striping on the tail. However, there was a narrow range of overlap in female SVL (40–43 mm SVL) between sympatric populations (Fig. 3A). The allozyme data indicate that little, if any, genetic exchange is occurring between the two forms. Data for the AAT-1 locus suggest that the reference populations may be fixed and the test populations may be exhibiting some genetic exchange similar to that reported by Guttman and Karlin (1986; see Table 2, AAT-2, hybrid populations 15 and 16). However, multiple alleles for AAT-1, IDH-1, and LDH-2 occur in both E. wilderae and E. cirrigera with a high degree of variation in allelic frequencies among populations (Guttman and Karlin, 1986; Jacobs, 1987). Furthermore, Jacobs (1987) found no fixed differences between E. wilderae and E. cirrigera. Therefore, it is possible that the alternative alleles seen in the test populations simply reflect interdemic differences with the reference populations rather than minor genetic exchange between the test populations. This could also explain the difference in allelic frequency for IDH-1 between the test and reference populations of E. cirrigera. The respective differences between the test and reference groups fall within the range of genetic variability exhibited among local populations of both E. wilderae and E. cirrigera (Jacobs, 1987).
We examined too few loci to determine genetically whether the test groups conclusively matched the reference populations. Based on both allozymic and morphological evidence, however, we suggest that the two test forms represent two distinct species. Furthermore, based on color and pattern, each of the test samples matched each of the respective reference samples. Therefore, we infer that the two forms occurring together in Habersham and Banks Counties, Georgia, are sympatric E. wilderae and E. cirrigera. These data support Jacobs' (1987) use of genetic distance data in elevating these two taxa to the level of species and contradict Petranka's (1998) conservative recommendation that they should be regarded as conspecific.
We do not know the width of the zone of sympatry. We have taken both forms together in Demorest, Habersham County, Georgia, which is 15 km north of the study site. If the zone is not much wider than this, it would be similar to the very narrow zone of sympatry exhibited by P. websteri and P. ventralis (Highton, 1985). Highton (1979, 1985) found that character displacement in color occurs between P. websteri and P. ventralis. Based on the size data from the present study, we hypothesize that ecological or reproductive character displacement may be occurring between these sympatric E. wilderae and E. cirrigera populations. The test-cirrigera were significantly larger than either the reference-cirrigera or the test-wilderae. In addition, 11 of the 13 mature females in the sample of test-cirrigera were larger than the maximum SVL reported for populations traditionally considered to comprise the subspecies E. b. cirrigera (Mittleman, 1966).
Preliminary indications are that these two species may exhibit some ecological differences as well. We found only E. wilderae in first order streams that were distinctly montane in character. We found only E. cirrigera in sections of streams in deltas or in broad valleys where the streams had a more lowland character. We found both species together where second- and third-order montane streams neared the lowland areas. However, further study is needed before any firm conclusions can be drawn concerning the ecological and reproductive relationships between these two forms in the zone of sympatry.
Eurycea cirrigera and E. wilderae from Nancytown Creek, Habersham County, Georgia in zone of sympatry. For E. cirrigera (A), note the row of distinct light-colored spots along the side and the long tail stripe. For E. wilderae (B), note the absence of light spots on the side and the short tail stripe
The relationship between body and tail color for sympatric (A) and allopatric (B) populations of Eurycea cirrigera and E. wilderae. Open squares represent E. cirrigera, whereas black squares represent E. wilderae. The color scale (i.e., 0 to 18) corresponds to a color chart ranging from brown to red (see text for explanation)
The frequency distribution of female SVL (in millimeters) for sympatric and allopatric Eurycea cirrigera and E. wilderae populations. Solid bars = E. wilderae, open bars = E. cirrigera. (A) sympatric populations from Nancytown Creek, Habersham County, Georgia (E. cirrigera, n = 27; E. wilderae, n = 26). The sympatric data are from the combined 1997 and 1998 samples. (B) allopatric populations of E. cirrigera (n = 58, from Poplar Cove Spring, Lafayette County, Mississippi; Marshall, 1997) and E. wilderae (n = 42, from Station Creek, Macon County, North Carolina; Bruce, 1988:table 6)