Todd B Cross

Characterizing genetic structure across a species’ range is relevant for management and conservation as it can be used to define population boundaries and quantify connectivity. Wide-ranging species residing in continuously distributed habitat pose substantial challenges for the characterization of genetic structure as many analytical methods used are less effective when isolation by distance is an underlying biological pattern. Here, we illustrate strategies for overcoming these challenges using a species of significant conservation concern, the Greater Sage-grouse (Centrocercus urophasianus), providing a new method to identify centers of genetic differentiation and combining multiple methods to help inform management and conservation strategies for this and other such species. Our objectives were to (1) describe large-scale patterns of population genetic structure and gene flow and (2) to characterize genetic subpopulation centers across the range of Greater Sage-grouse. Samples from 2,134 individuals were genotyped at 15 microsatellite loci. Using standard STRUCTURE and spatial principal components analyses, we found evidence for four or six areas of large-scale genetic differentiation and, following our novel method, 12 subpopulation centers of differentiation. Gene flow was greater, and differentiation reduced in areas of contiguous habitat (eastern Montana, most of Wyoming, much of Oregon, Nevada, and parts of Idaho). As expected, areas of fragmented habitat such as in Utah (with 6 subpopulation centers) exhibited the greatest genetic differentiation and lowest effective migration. The subpopulation centers defined here could be monitored to maintain genetic diversity and connectivity with other subpopulation centers. Many areas outside subpopulation centers are contact zones where different genetic groups converge and could be priorities for maintaining overall connectivity. Our novel method and process of leveraging multiple differ- ent analyses to find common genetic patterns provides a path forward to characterizing genetic structure in wide-ranging, continuously distributed species.

Wildfire-generated snags provide key habitat for wildlife associated with recently disturbed forests, offering nesting and foraging resources for several woodpecker species. Snag harvest through post-fire salvage logging provides economic value but reduces habitat in recently burned forests. Managers of recently burned forests often must identify suitable woodpecker habitat within timeframes precluding field surveys. We developed nesting habitat suitability models for the Lewis's woodpecker (Melanerpes lewis)—a species of conservation concern potentially impacted by salvage logging—to inform post-fire management planning that includes habitat conservation objectives. Using weighted logistic regression, we selected models that maximized predictive performance measured by spatial cross-validation to verify model applicability to novel wildfire locations. From 1994 to 2018, we monitored 348 nest sites, ≤5 years post-fire, at four wildfire locations across the Inland Pacific Northwest, USA. We used both exclusively remotely sensed covariates (to support habitat mapping) and a combination of remotely sensed and field collected covariates (to inform logging prescriptions). Top models proved predictive of nest distribution across novel wildfire locations, describing positive relationships with local and landscape-scale burn severity, landscape-scale open (0–10%) and low (10–40%) canopy cover, large nest-tree diameters, locally high snag densities, and southeast-facing slopes. The top combination model was more discriminating of nest from non-nest sites compared to the top remotely sensed model (RPI = 0.976 vs. 0.733, AUC= 0.794 vs. 0.716), indicating the value of vegetation field measurements for quantifying habitat. Herein is exemplified an effective process for developing and evaluating predictive habitat models, broadly applicable, and useful for prioritization of post-fire forest management objectives.

Functional connectivity, quantified using landscape genetics, can inform conservation through the identification of factors linking genetic structure to landscape mechanisms. We used breeding habitat metrics, landscape attributes, and indices of grouse abundance, to compare fit between structural connectivity and genetic differentiation within five long-established Sage-Grouse Management Zones (MZ) I-V using microsatellite genotypes from 6,844 greater sage-grouse (Centrocercus uropha-sianus) collected across their 10.7 million-km 2 range. We estimated structural con-nectivity using a circuit theory-based approach where we built resistance surfaces using thresholds dividing the landscape into "habitat" and "nonhabitat" and nodes were clusters of sage-grouse leks (where feather samples were collected using non-invasive techniques). As hypothesized, MZ-specific habitat metrics were the best predictors of differentiation. To our surprise, inclusion of grouse abundance-corrected indices did not greatly improve model fit in most MZs. Functional connec-tivity of breeding habitat was reduced when probability of lek occurrence dropped below 0.25 (MZs I, IV) and 0.5 (II), thresholds lower than those previously identified as required for the formation of breeding leks, which suggests that individuals are willing to travel through undesirable habitat. The individual MZ landscape results suggested terrain roughness and steepness shaped functional connectivity across all MZs. Across respective MZs, sagebrush availability (<10%-30%; II, IV, V), tree canopy cover (>10%; I, II, IV), and cultivation (>25%; I, II, IV, V) each reduced movement beyond their respective thresholds. Model validations confirmed variation in predictive ability across MZs with top resistance surfaces better predicting gene flow than geographic distance alone, especially in cases of low and high differentiation among lek groups. The resultant resistance maps we produced spatially depict the strength and redundancy of range-wide gene flow and can help direct conservation actions to maintain and restore functional connectivity for sage-grouse.

Connectivity of populations influences the degree to which species maintain genetic diversity and persist despite local extinctions. Natural landscape features are known to influence connectivity, but global anthropo-genic landscape change underscores the importance of quantifying how human-modified landscapes disrupt con-nectivity of natural populations. Grasslands of western North America have experienced extensive habitat alteration , fragmenting populations of species such as black-tailed prairie dogs (Cynomys ludovicianus). Population sizes and the geographic range of prairie dogs have been declining for over a century due to habitat loss, disease, and eradication efforts. In many places, prairie dogs have persisted in the face of emerging urban landscapes that carve habitat into smaller and smaller fragments separated by uninhabitable areas. In extreme cases, prairie dog colonies are completely bounded by urbanization. Connectivity is particularly important for prairie dogs because colonies suffer high probabilities of extirpation by plague, and dispersal permits recolonization. Here we explore con-nectivity of prairie dog populations using analyses of 11 microsatellite loci for 9 prairie dog colonies spanning the fragmented landscape of Boulder County, Colorado. Isolation by resistance modeling suggests that wetlands and high intensity urbanization limit movement of prairie dogs. However, prairie dogs appear to move moderately well through low intensity development (including roads) and freely through cropland and grassland. Additionally, there is a marked decline in gene flow between colonies with increasing geographic distance, indicating isolation by distance even in an altered landscape. Our results suggest that prairie dog colonies retain some connectivity despite fragmentation by urbanization and agricultural development.

Dispersal can strongly influence the demographic and evolutionary trajectory of populations. For many species, little is known about dispersal, despite its importance to conservation. The Greater Sage-Grouse (Centrocercus urophasianus) is a species of conservation concern that ranges across 11 western U.S. states and 2 Canadian provinces. To investigate dispersal patterns among spring breeding congregations, we examined a 21-locus microsatellite DNA dataset of 3,244 Greater Sage-Grouse sampled from 763 leks throughout Idaho, Montana, North Dakota, and South Dakota, USA, across 7 yr. We recaptured ~2% of individuals, documenting 41 instances of breeding dispersal, with 7 dispersal events of 50 km, including 1 of 194 km. We identified 39 recaptures on the same lek up to 5 yr apart, which supports the long-held paradigm of philopatry in lekking species. We found no difference between the sexes in breeding dispersal distances or in the tendency to disperse vs. remain philopatric. We also documented movements within and among state-delineated priority areas of conservation importance, further supporting the need to identify movement corridors among these reserves. Our results can be used to better inform the assumptions of count-based population models and the dispersal thresholds used to model population connectivity.

Understanding population structure is important for guiding ongoing conservation and restoration efforts. The greater sage-grouse (Centrocercus urophasianus) is a species of concern distributed across 1.2 million km2 of western North America. We genotyped 1499 greater sage-grouse from 297 leks across Montana, North Dakota and South Dakota using a 15 locus microsatellite panel, then examined spatial autocorrelation, spatial principal components analysis, and hierarchical Bayesian clustering to identify population structure. Our results show that at distances of up to ~240 km individuals exhibit greater genetic similarity than expected by chance, suggesting that the cumulative effect of short-range dispersal translates to long-range connectivity. We found two levels of hierarchical genetic subpopulation structure. These subpopulations occupy significantly different elevations and are surrounded by divergent vegetative communities with different dominant subspecies of sagebrush, each with its own chemical defense against herbivory. We propose five management groups reflective of genetic subpopulation structure. These genetic groups are largely synonymous with existing priority areas for conservation. On average, 85.8 % of individuals within each conservation priority area assign to a distinct subpopulation. Our results largely support existing management decisions regarding subpopulation boundaries.

We propose a spatially-explicit approach for modeling genetic variation across space and illustrate how this approach can be used to optimize spatial prediction and sampling design for landscape genetic data. We propose a multinomial data model for categorical microsatellite allele data commonly used in landscape genetic studies, and introduce a latent spatial random effect to allow for spatial correlation between genetic observations. We illustrate how modern dimension reduction approaches to spatial statistics can allow for efficient computation in landscape genetic statistical models covering large spatial domains. We apply our approach to propose a retrospective spatial sampling design for greater sage-grouse (Centrocercus urophasianus) population genetics in the western United States.