Scale‐dependent influence of the sagebrush community on genetic connectivity of the sagebrush obligate Gunnison sage‐grouse

Abstract

Habitat fragmentation and degradation impacts an organism's ability to navigate the landscape, ultimately resulting in decreased gene flow and increased extinction risk.Understanding how landscape composition impacts gene flow (i.e., connectivity) and interacts with scale is essential to conservation decision-making. We used a landscape genetics approach implementing a recently developed statistical model based on the generalized Wishart probability distribution to identify the primary landscape features affecting gene flow and estimate the degree to which each component influences connectivity for Gunnison sage-grouse (Centrocercus minimus). We were interested in two spatial scales: among distinct populations rangewide and among leks (i.e., breeding grounds) within the largest population, Gunnison Basin. Populations and leks are nested within a landscape fragmented by rough terrain and anthropogenic features, although requisite sagebrush habitat is more contiguous within populations.Our best fit models for each scale confirm the importance of sagebrush habitat in connectivity, although the important sagebrush characteristics differ. For Gunnison Basin, taller shrubs and higher quality nesting habitat were the primary drivers of connectivity, while more sagebrush cover and less conifer cover facilitated connectivity rangewide. Our findings support previous assumptions that Gunnison sage-grouse range contraction is largely the result of habitat loss and degradation. Importantly, we report direct estimates of resistance for landscape components that can be used to create resistance surfaces for prioritization of specific locations for conservation or management (i.e., habitat preservation, restoration, or development) or as we demonstrated, can be combined with simulation techniques to predict impacts to connectivity from potential management actions.

A habitat‐centered framework for wildlife climate change vulnerability assessments: Application to Gunnison sage‐grouse

“…Gunnison has the largest population—containing almost 85% of individuals—and the other “satellite” populations are perilously small, generally declining, and in urgent need of habitat improvements (USFWS, 2019). The populations have substantial differences in environmental characteristics, land ownership, and genetic adaptation, highlighting the need for population‐specific management (USFWS, 2019; Zimmerman et al, 2022). USFWS has designated 7775 km 2 of critical habitat for Gunnison sage‐grouse (Figure 2, black outlines; USFWS, 2014).…”

Conservation genomics of an endangered montane amphibian reveals low population structure, low genomic diversity and selection pressure from disease

“…We investigated landscape connectivity using spatial autoregressive models (SAM; Peterson et al., 2019) and generalized Wishart modelling (GWM; Hanks & Hooten, 2013) using the rwc package in R (Hanks, 2018) as specified in Zimmerman et al. (2022). Landscape surfaces were developed in ArcGIS V.10.7 and consisted of 11 environmental factors hypothesized to be important for boreal toad movement and gene flow across the SRM (Table 1 and Figure 1).…”

“…We used a minimally correlated (Pearson's r < 0.5) set of variables for GWM multivariate modelling by retaining variables with the lowest single‐covariate deviance information criterion (DIC; Spiegelhalter et al., 2002) among correlated sets of variables (Dormann et al., 2013; Row et al., 2017; Zimmerman et al., 2022). Calculation of the covariance parameter of GWM probability distribution is complicated by non‐obvious correlation among sampled locations of raster covariates and can prevent models from converging (Zimmerman et al., 2022). All model fitting was performed in R with two independent chains of 150,000 Markov chain Monte Carlo (MCMC) iterations, with the first 75,000 iterations discarded as a burn‐in period, using a random‐walk Metropolis‐Hastings sampler.…”

“…Calculation of the covariance parameter of GWM probability distribution is complicated by non-obvious correlation among sampled locations of raster covariates and can prevent models from converging (Zimmerman et al, 2022). All model fitting was performed in R with two independent chains of 150,000 Markov chain Monte Carlo (MCMC) iterations, with the first 75,000 iterations discarded as a burn-in period, using a random-walk Metropolis-Hastings sampler.…”

“…We used a minimally correlated (Pearson's r < 0.5) set of variables for GWM multivariate modelling by retaining variables with TA B L E 1 Landscape variables used for connectivity analyses of boreal toads in the southern Rocky Mountains (SRM), including data source, spatial resolution and ecological justification. Spiegelhalter et al, 2002) among correlated sets of variables (Dormann et al, 2013;Row et al, 2017;Zimmerman et al, 2022).…”