Population genetics and landscape modeling in water snakes

John Clark Marshall, Purdue University

Abstract

Habitat loss and isolation is pervasive in the Midwest U.S. Wetlands are experiencing particularly dramatic declines, yet there is a paucity of information on the responses to these losses for most obligate wetland taxa. It is particularly important to understand both the genetic and landscape distributional consequences, and importantly, the mechanisms responsible for interspecific differences. Both dispersal ability and niche breadth have been shown to be important mechanisms in shaping patterns of gene flow and in sensitivity to habitat fragmentation in terrestrial taxa. Water snakes provide an opportunity to investigate these mechanisms in wetland obligate animals. Two sympatric water snakes; one a highly mobile anuran specialist (Nerodia erythrogaster neglecta), one a sessile generalist (Nerodia sipedon), were genotyped at six polymorphic loci (n = 198 & 101, respectively). Bayesian clustering analyses support the niche breadth hypothesis. Posterior log probabilities were greatest when K (i.e. number of populations) was set at four and two for N. e. neglecta and N. sipedon, respectively. FST estimates of genetic differentiation between populations also support the niche breadth hypothesis. The FST estimate for N. sipedon was one-third that of N. e. neglecta, 0.05 (0.07–0.04, 95% CI) and 0.14 (0.20 – 0.08 95% CI), respectively. Furthermore, the standardized FST estimate was nearly twice that of N. e. neglecta than for N. sipedon, 0.45 and 0.24, respectively. Estimated levels of gene flow were higher in N. sipedon, indicating that niche breadth has a stronger effect than dispersal ability. Microsatellite variation was further quantified for the federally protected copperbelly water snake (Nerodia erythrogaster neglecta). A total of 228 copperbelly water snakes were genotyped representing populations from three states. Moderate genetic differentiation exists among all three regions (FST = 0.12, P < 0.001), with evidence for low levels of differentiation within the federally protected Ohio region (FST = 0.025, P = 0.007), and moderate to strong differentiation within the Indiana region (FST = 0.23, P < 0.001). Differentiation among sampling sites did not appear to be related to geographic distance, but rather depended on the quality of terrestrial corridors used for dispersal. Mode shifts in allele frequencies and excess heterozygosity tests were negative, while M-ratio tests were nearly all positive, indicating the likelihood of historical rather than contemporary population bottlenecks. However, potential subspecific intergradation in the Kentucky region may have artificially lowered the M-ratio. Landscape modeling of an assemblage of Midwestern water snakes revealed species-specific scale-dependent responses to wetland habitat. The copperbelly water snake abundance responded strongest to the amount of habitat at the smallest spatial scale (i.e. 500m), the midland water snake (Nerodia sipedon pleuralis) did not exhibit an abundance-habitat area relationship, and the diamondback water snake (Nerodia rhombifer rhombifer) abundance responded strongest to the amount of habitat at the largest spatial scale (i.e. 6000m). The proportion of wetland types also had a significant impact on water snake abundance. Copperbelly water snake abundance was negatively correlated with deep water habitats, while diamondback water snake abundance was positively correlated with the proportion of deep water habitat, which increased as a function of scale. These results, and those from genetic analyses suggest that diet breadth is a significant mechanism in shaping both patterns of gene flow and distribution in a fragmented landscape.

Degree

Ph.D.

Advisors

Kingsbury, Purdue University.

Subject Area

Ecology

COinS