Habitat models to predict wetland bird occupancy influenced by scale, anthropogenic disturbance, and imperfect detection

Abstract

Abstract. Understanding species-habitat relationships for endangered species is critical for their conservation. However, many studies have limited value for conservation because they fail to account for habitat associations at multiple spatial scales, anthropogenic variables, and imperfect detection. We addressed these three limitations by developing models for an endangered wetland bird, Yuma Ridgway's rail (Rallus obsoletus yumanensis), that examined how the spatial scale of environmental variables, inclusion of anthropogenic disturbance variables, and accounting for imperfect detection in validation data influenced model performance. These models identified associations between environmental variables and occupancy. We used bird survey and spatial environmental data at 2473 locations throughout the species' U.S. range to create and validate occupancy models and produce predictive maps of occupancy. We compared habitatbased models at three spatial scales (100, 224, and 500 m radii buffers) with and without anthropogenic disturbance variables using validation data adjusted for imperfect detection and an unadjusted validation dataset that ignored imperfect detection. The inclusion of anthropogenic disturbance variables improved the performance of habitat models at all three spatial scales, and the 224-m-scale model performed best. All models exhibited greater predictive ability when imperfect detection was incorporated into validation data. Yuma Ridgway's rail occupancy was negatively associated with ephemeral and slow-moving riverine features and high-intensity anthropogenic development, and positively associated with emergent vegetation, agriculture, and low-intensity development. Our modeling approach accounts for common limitations in modeling species-habitat relationships and creating predictive maps of occupancy probability and, therefore, provides a useful framework for other species.

Species‐Habitat Relationships and Priority Areas for Marsh‐Breeding Birds in Ontario

“…The time period covered by these models is, therefore, typically a single field season, or a single year over which the closure assumption can be assumed to hold. Despite the descriptor of single season, these models are not necessarily limited to time periods within a single season or year (Glisson et al 2017). In the context of our 3‐year study, we extended the closure assumption across all years, such that we considered the 6 surveys at each sample site (i.e., 2 surveys at each sample site each year) to be repeated surveys in the single‐season models.…”

Predicting species distributions: unifying model selection and scale optimization for multi‐scale occupancy models

“…Following Glisson et al. (), we used the following categories of the NWI classification system to describe characteristics of wetland habitats: wetland system–subsystem, wetland class, and water regime and special modifiers (Appendix ). Wetland attributes for each of these categories were characterized within each polygon, resulting in each wetland being assigned a categorical value for each of the four categories listed above (e.g., a spatial polygon could be assigned to a category of emergent wetland or forested wetland for NWI class; Appendix ).…”

“…Detailed descriptions of NWI classifications for each habitat variable are provided by Glisson et al. (). We used a total of 21 NWI wetland cover variables as habitat predictor variables in our analyses (Appendix ).…”

“…These data were described by rasters with a spatial grain of 30‐m pixels, and we also rasterized all NWI polygon data in order to measure all variables at the same spatial resolution, as described by Glisson et al. ().…”

“…These scales correspond to the distance at which birds were commonly detected during field surveys (typically <225 m) and cover a range of extents at which marsh birds commonly respond to habitat features (Glisson et al. , ). We characterized the amount of habitat surrounding each 30‐m pixel on the landscape at each spatial extent by using moving window analyses with radii of 100 m, 224 m, and 500 m, respectively.…”