Complementary Network-Based Approaches for Exploring Genetic Structure and Functional Connectivity in Two Vulnerable, Endemic Ground Squirrels

Abstract

The persistence of small populations is influenced by genetic structure and functional connectivity. We used two network-based approaches to understand the persistence of the northern Idaho ground squirrel (Urocitellus brunneus) and the southern Idaho ground squirrel (U. endemicus), two congeners of conservation concern. These graph theoretic approaches are conventionally applied to social or transportation networks, but here are used to study population persistence and connectivity. Population graph analyses revealed that local extinction rapidly reduced connectivity for the southern species, while connectivity for the northern species could be maintained following local extinction. Results from gravity models complemented those of population graph analyses, and indicated that potential vegetation productivity and topography drove connectivity in the northern species. For the southern species, development (roads) and small-scale topography reduced connectivity, while greater potential vegetation productivity increased connectivity. Taken together, the results of the two network-based methods (population graph analyses and gravity models) suggest the need for increased conservation action for the southern species, and that management efforts have been effective at maintaining habitat quality throughout the current range of the northern species. To prevent further declines, we encourage the continuation of management efforts for the northern species, whereas conservation of the southern species requires active management and additional measures to curtail habitat fragmentation. Our combination of population graph analyses and gravity models can inform conservation strategies of other species exhibiting patchy distributions.

Keywords: Sciuridae; Urocitellus [Spermophilus]; functional connectivity; gene flow; graph theory; gravity model; landscape genetics.

Figure 1

Sampling locations of genetic data for northern Idaho ground squirrel (Urocetillus brunneus; NIDGS) and southern Idaho ground squirrel (U.endemicus; SIDGS). Also shown are the NIDGS probable historic distribution (U.S. Fish and Wildlife Service, 2003) and the current known range for SIDGS (Idaho Game and Fish Department). Individuals were sampled from 2002 to 2006 (Hoisington-Lopez et al., 2012). Background hillshade map was produced from the National Elevation Dataset (http://ned.usgs.gov). Full site names and sample sizes can be found in Table S1.

Figure 2

Illustration of the node removal procedure used to simulate population extinction events in NIDGS (top) and SIDGS (bottom). In each step, a randomly selected node (in red), representing a sampling location, is removed from the network along with the edges connecting it to additional nodes. Network mean betweenness values are given on top. Following the removal of 2 nodes, the SIDGS network fails to create a single component, becoming fragmented.

Figure 3

Network diagrams representing the genetic relationships between northern (a) and southern (b) Idaho ground squirrel sampling locations. Networks are pruned using conditional genetic distance (cGD; Dyer and Nason, 2004). Individuals were sampled during 2002–2006. Node colors differ by cluster assignment with the Girvan-Newman algorithm. Nodes are placed according to geographic location and scaled to reflect coreness, a network metric that quantifies proximity to the core in a core/periphery model (Table S3). Edge width is proportional to the genetic flow between sampling locations.

Figure 4

The effects of random sequential removal of nodes on proportion of connected graphs (A) and the largest remaining network component (B) for northern Idaho ground squirrels (solid line) and southern Idaho ground squirrels (dashed line). Proportions, means and 95% confidence intervals were calculated for up to 100 simulated networks in each scenario. Idaho ground squirrels were sampled from 23 locations during 2002–2006 and genotyped using 8 microsatellite loci (Hoisington-Lopez et al., 2012).

Figure 5

Cumulative AICc weights of several factors examined with gravity models as candidates for driving NIDGS (red circles) and SIDGS (blue circles) gene flow. Panels include cumulative AICc weights for each factor in all examined models (A) and only the top-ranked model for each factor (B). Positive and negative values represent the direction and magnitude of relative support for each factor's influence.