The Tsinling and Daba Mountains do not present significant genetic barriers to the Chinese wood frog
A high level of genetic diversity was detected at microsatellite DNA loci in Chinese wood frogs. Two commonly used indices of genetic diversity, allelic diversity and heterozygosity, are comparable to or higher than those found in other similar studies of anurans. For example, Funk et al. (2005) found 5 – 16 alleles per locus and expected heterozygosity of 0.23 – 0.66 in various populations of the Columbian spotted frog (Rana luteiventris) , while Johansson et al. (2006) found 10 – 34 alleles per locus (mean = 19) in Rana temporaria across Sweden . We found 14 – 47 alleles per locus (mean = 25.5) and H
of 0.504 to 0.855 in different populations.
Low to moderate genetic differentiation between the ten central sites was observed, with pairwise F
ST ranging from 0.0175 to 0.1625, and with an average of 0.0878 (see Additional file 2). Considering the large geographical distance (up to 337.1 km) and two large mountain chains within the region, it is somewhat surprising that greater population subdivision was not found, as these values are much lower than other studies at similar or even smaller geographical scales. Using microsatellite DNA markers, Funk et al. (2005) reported F
values of 0.153–0.242 between Columbian spotted frog populations (Rana luteiventris) from different sides of mountain ridges within distance of only 10 kilometers . Furthermore, Spears et al. (2005) reported pairwise F
ST values up to 0.453 between populations within a major mountain valley and a distance of 60 kilometers in tiger salamanders (Ambystoma tigrinum) . While higher F
values were observed when the two peripheral sites were also included (up to 0.3130), these values may have been inflated by the low genetic diversity of the peripheral sites.
A moderate to high level of migration is the most likely explanation for the low to moderate differentiation between sites. Despite geographical distances of 50 – 100 kilometers and a major mountain range between them, the migration rates between site-pairs from different sides of the Tsinling Mountains ranged from 0.17 to 1.26 individuals per generation with an average of 0.79 (immigration rate; Additional file 2). Studies of migration rate in other anurans have usually found much lower values. For example, Stevens et al. (2006) reported N
of 0.03 to 0.34 between sites within distance of 15 kilometers for the natterjack toad (Bufo calamita) . Additionally, the large allele numbers of these microsatellite DNA loci may partially contribute to the relatively low F
values. Hedrick (1999) analytically demonstrated that, when the amount of within-population variation becomes high, the F
values necessarily become low . Other evolutionary processes, such as recent population colonization, may also contribute to the low to moderate differentiation between sites. However, the Chinese wood frog populations of the Tsinling and Daba region are unlikely to have been recently established, as recently established populations typically have low levels of genetic diversity  whereas these wood frog populations have very high genetic diversity (see Additional file 1). Furthermore, the Tsinling region was not covered by ice during Pleistocene glaciations , and therefore, there was no post-glaciation colonization.
The Tsinling and Daba Mountain ranges do not present significant barriers to gene flow in Chinese wood frogs. Both AMOVA and Monmonier's maximum difference algorithm have proven to be powerful tools for detecting the impact of landscape features on population genetic structure [3, 23]. Nevertheless, our AMOVA attributed most variation to the within-site component (Table 3). Furthermore, much of the variation among sites can be explained by isolation-by-distance, as the correlation between geographical and genetic distance was very strong (Figure 2). None of the predicted genetic barriers concurred with the Tsinling Mountains (Figure 3). In only one case did a predicted genetic barrier correspond to the location of the Daba Mountains, when only the ten central populations were included (Figure 3B). When all populations were included, however, the association disappeared (Figure 3A). The Daba Mountains may impose a weak genetic barrier to the populations of the Chinese wood frog. Whether the Tsinling and Daba Mountains present significant barriers (or not) to other species is an interesting topic for future study.
These results prompt the question: Given that mountains have repeatedly been shown to have strong effects on gene flow in amphibians, how did a major geologic divider such as the Tsinling Mountains not impose any significant genetic barrier to the Chinese wood frog? Large gene flow among neighboring sites (high connectivity) is probably the primary cause of the diminished mountain effect, and extensive juvenile dispersal may be responsible for such gene flow. Most mature female Chinese wood frogs produce 400 to 900 eggs annually, with some producing up to 2000 eggs, which is higher than other closely related wood frog species [16, 24]. Thus, Chinese wood frogs have the capacity to produce a very large number of juveniles. Although a high ratio of mortality was observed at larval or juvenile stages, the number of emerging juveniles is still extremely large (Fu, personal observation). With a large number of juveniles, even a small percentage of successful migrants will create sufficient gene flow to counter population differentiation. Furthermore, there are many available breeding sites between our sampling sites, which may facilitate the dispersal. The relatively high N
values in this study support this explanation (see Additional file 2). In a closely related and biologically similar species, Brede & Beebee (2004) also found high connectivity among subpopulations of R. temporaria , although other studies revealed substantial population structure . Funk et al(2005) also found a large number of juvenile Columbia spotted frogs dispersing a large distance (> 5 km) .
Chinese wood frogs are capable of using slow moving mountain streams as breeding sites at high elevations and are not restricted to breeding in ponds. For example, at one of our collecting sites, site 12, the frogs bred in a stream at an elevation of 2480 meters a.s.l. This life history trait is perhaps crucial for the Chinese wood frog to maintain a low to moderate population differentiation with the presence of major mountain ridges. Mountain ridges have significant impact on pond breeder population structure, which is well established by numerous studies [6–11]. Nevertheless, whether the same impact occurs on stream breeders is still an open question. Additionally, the Chinese wood frog may use mountain passes and valleys as corridors to cross the seemingly impossible high mountain ridges. Lu et al. (2006) reported that the Chinese wood frog hibernates at an elevation of 2000 meters a.s.l. . There are many mountain passes lower than 2000 meters a.s.l. along the chain of Tsinling Mountains.