DISCUSSION

The salamander assemblage recovered from the GFS represents the earliest Cenozoic record of salamander fossils from the Appalachians (a region well-known for its current salamander diversity). The fauna contains at least five distinct taxa, and represents the earliest record of Ambystomatidae and Plethodontidae east of the Mississippi River, and the earliest record of the salamandrid genus Notophthalmus east of the Mississippi River and north of Florida.

The difficulty with identifying fossil plethodontids below the family level is not resolved by this study; however, the use of observed subfamily characters allows for reference to 'type'. Attribution to species for Ambystoma and Notophthalmus is avoided in this report to reflect a departure from the tendency to identify fossil taxa based on locally endemic modern species.

Paleoecology

Our paleoecological interpretation based on the salamanders assumes that the described taxa co-occurred. At this point we feel this assumption is justified because all excavations have been in the same type of lake deposit, and represented vertebrate and plant species are consistent from one pit to another.

In holding with Schubert and Wallace (2006), paleoecological inferences are based primarily on phylogenetic bracketing (i.e., inferring habitat preference based on family membership) when taxa cannot be identified to genus (as is the case with the GFS plethodontids), and by more direct analogy in cases where genus can be determined (i.e., Ambystoma sp. and Notophthalmus sp.). Extant Notophthalmus viridescens inhabit coniferous and deciduous forests, with immature larvae and adult newts living in small bodies of freshwater and the juvenile "eft" stage inhabiting the shorelines and woodland habitats around these bodies of water (Petranka 1998). GFS Notophthalmus sp., assuming analogous habitat preference, supports the wooded-pond environment interpretation of DeSantis and Wallace (2006, 2008). Extant Ambystoma species can be found in upland mixed and coniferous forests that are sufficiently damp and have bodies of water suitable for breeding, and places where they are able to burrow in the soil or find burrows made by other animals (Petranka 1998). Thus, the presence of Ambystoma sp. garners more support for the wooded-pond interpretation of the GFS. Plethodontid salamanders (plethodontine and spelerpine alike) in general prefer wooded moist habitats, particularly along streamsides, whether or not they have aquatic larvae (Petranka 1998; Wells 2007). This paleoenvironmental interpretation based on the salamanders is no more resolved than that of Schubert and Wallace's (2006) analysis utilizing the entire herpetofauna. This lack of resolution is due to our inability to identify the recovered salamander vertebrae below genus level.

The faunal composition as represented by the preliminary sample is predominated by terrestrial plethodontids and underscored by ambystomatids (with a mixed population of both neotenic and metamorphosed adults) and desmognathine plethodontids (which may or may not be semi-aquatic to aquatic in habit). The Ambystoma sp., Desmognathus sp., and some Plethodon sp. fall into a similarly large size range (estimated around 20 cm in total length by comparison with modern specimens), presenting the potential for adult niche partitioning and a base for ecological interpretation revolving around partitioning of similar sized taxa. This may support the idea that terrestrial competition was the impetus for the origin of the genus Desmognathus and reacquisition of aquatic larvae in more derived desmognathines (see Lombard and Wake 1986; Chippindale and Wiens 2005; Vieites et al. 2007). Neoteny in part of the Ambystoma sp. population is likely attributable to lack of large-bodied predatory fish (Woodward, personal commun., 2010) and pond permanence. Pond permanence is evidenced by the non-existence of mud cracks in the strata and supported by the abundance of large bodied Rana that are indistinguishable from R. catesbeiana (bullfrog). Today bullfrogs require permanent bodies of water to survive their larval stage, which may last as long as 3 years, and always requires overwintering (Bruening 2002; Wells 2007).

Paleobiogeography and Evolutionary History

The Pleistocene and Holocene fossil record for ambystomatids, plethodontids, and salamandrids is well represented in the southern Appalachians and much has been reported on biogeography during this time frame (see references in Holman 2006). Therefore, the presence of these families and genera in the Mio-Pliocene of the southern Appalachians is not surprising. While these new records represent the earliest occurrences of these taxa in the region, they provide limited information from a paleobiogeographical or evolutionary history standpoint. In essence, the only taxon present at the GFS that warrants special biogeographic/evolutionary attention is the Desmognathus sp., being the oldest specimens referable to the genus Desmognathus. The genus Desmognathus was proposed by Tihen and Wake (1981), and supported by Chippindale et al. (2004), to have split from the rest of the Plethodontinae 7 million years ago. The Desmognathus sp. at the GFS may support this date by its presence. In terms of evolution, this taxon stands out as potential support for an Appalachian origin for the group; though the possibility that this group could have its origin outside of the region cannot be discounted given the scant nature of plethodontid remains from this period (or any period prior to the Pleistocene for that matter).