A species' genetic structure often varies in response to ecological and landscape processes that differ throughout the species' geographic range, yet landscape genetics studies are rarely spatially replicated. The Cope's giant salamander (Dicamptodon copei) is a neotenic, dispersal-limited amphibian with a restricted geographic range in the Pacific northwestern USA. We investigated which landscape factors affect D. copei gene flow in three regions spanning the species' range, which vary in climate, landcover and degree of anthropogenic disturbance. Least cost paths and Circuitscape resistance analyses revealed that gene flow patterns vary across the species' range, with unique combinations of landscape variables affecting gene flow in different regions. Populations in the northern coastal portions of the range had relatively high gene flow, largely facilitated by stream and river networks. Near the southeastern edge of the species' range, gene flow was more restricted overall, with relatively less facilitation by streams and more limitation by heat load index and fragmented forest cover. These results suggested that the landscape is more difficult for individuals to disperse through at the southeastern edge of the species' range, with terrestrial habitat desiccation factors becoming more limiting to gene flow. We suggest that caution be used when attempting to extrapolate landscape genetic models and conservation measures from one portion of a species' range to another.
Dicamptodon copei microsatellite genotype data
Dicamptodon copei tissue samples were collected from the field in 2006-2008. DNA was extracted using DNeasy tissue kits (Qiagen) according to the manufacturer’s protocols. Eleven microsatellite loci were amplified using PCR with fluorescently labeled primers, using PCR conditions from Steele et al. (2008) (Table S1, Supporting Information). Microsatellite products were run on an ABI 3730XL automated sequencer (Applied Biosystems) at the Washington State University LBB1 core facility and genotyped using ABI GENEMAPPER 3.7 software. The data include microsatellite fragment lengths for each study site (rows) and (diploid) locus (columns).
Dcopei_microsatdata_10242012.csv
Dcopei_landscape_distances
These data are landscape distances between study sites calculated using least cost path (LCP) analysis (columns B-JJ) and Circuitscape resistance analysis (columns JL-KB). LCP data represent the weighted averages of all landscape layers along a particular LCP (indicated in the row below the region name), and all LCPs were corrected for topography. Landscape cost surfaces were developed using the following GIS datasets downloaded from online databases (including abbreviations and source of data citations from Trumbo et al. 2012): stream networks (USGS 2001) (str10 = 10:1 cost of movement outside streams, str100 = 100:1 cost of movement outside streams), canopy cover (cc) (Homer et al. 2004), frost-free periods (ffp) (Rehfeldt 2006), growing season precipitation (gsp) (Rehfeldt 2006), landcover (lc) (Homer et al. 2004), and heat load index (hli) (McCune and Keon 2002). For streams, a "no" following abbreviation means that large rivers were excluded from the stream network. For all continuous variables, a "1" following the abbreviation means an exponentially increasing non-linear response, such that the transformed surface T1 = 100^ (original cost surface). A "2" following the abbreviation means an exponentially decreasing non-linear response such that the transformed surface T2 = 100 - 100^(1- original cost surface). The "strghtline" data are null models of isolation by Euclidean distance alone.