Mapping breeding bird species richness at management‐relevant resolutions across the United States

Abstract

Human activities alter ecosystems everywhere, causing rapid biodiversity loss and biotic homogenization. These losses necessitate coordinated conservation actions guided by biodiversity and species distribution spatial data that cover large areas yet have fine-enough resolution to be management-relevant (i.e., ≤5 km). However, most biodiversity products are too coarse for management or are only available for small areas. Furthermore, many maps generated for biodiversity assessment and conservation do not explicitly quantify the inherent tradeoff between resolution and accuracy when predicting biodiversity patterns. Our goals were to generate predictive models of overall breeding bird species richness and species richness of different guilds based on nine functional or life-history-based traits across the conterminous United States at three resolutions (0.5, 2.5, and 5 km) and quantify the tradeoff between resolution and accuracy and, hence, relevance for management of the resulting biodiversity maps. We summarized 18 years of North American Breeding Bird Survey data (1992-2019) and modeled species richness using random forests, including 66 predictor variables (describing climate, vegetation, geomorphology, and anthropogenic conditions), 20 of which we newly derived. Among the three spatial resolutions, the percentage variance explained ranged from 27% to 60% (median = 54%; mean = 57%) for overall species richness and 12% to 87% (median = 61%; mean = 58%) for our different guilds. Overall species richness and guild-specific species richness were best explained at 5-km resolution using $24 predictor variables based on percentage variance explained, symmetric mean absolute percentage error, and root mean square error values. However, our 2.5-km-resolution maps were almost as accurate and provided more spatially detailed information, which is why we recommend them for most management applications. Our results represent the first consistent, occurrence-based, and nationwide maps of breeding bird richness with a thorough accuracy assessment that are also spatially detailed enough to inform local management decisions. More broadly, our findings highlight the importance of explicitly considering tradeoffs between resolution and accuracy to create management-relevant biodiversity products for large areas.

Using satellite data to assess spatial drivers of bird diversity

“…In this paper, we model bird diversity across Great Britain (GB) using Landsat data and determine the drivers of observed spatial variation. This builds on previous work (Bonthoux et al, 2018;Carrasco et al, 2018;Fuller et al, 2005 and complements recent studies (Carroll et al, 2022) by incorporating Landsat-derived measures of habitat extent, and monthly heterogeneity and productivity into the modelling. To do this, RFR trained using bird survey data was used to assess the extent to which a combination of satellite-derived measures of habitat heterogeneity and habitat productivity could explain the variation of bird diversity across GB.…”

Anthropogenic habitat modification causes nonlinear multiscale bird diversity declines

“…Exploratory analyses, however, confirmed that at a 5 km resolution, eBird gamma diversity (i.e. total species richness of all checklists within a pixel) was positively correlated with species richness from an external source, the Breeding Bird Survey (Carroll et al 2022, Supporting information). Finally, we computed beta diversity (β) for each diversity metric using Whittaker's (1960) multiplicative formulation: , where is the average of the diversity metric across all the checklists in a given grid cell.…”

Mapping multiscale breeding bird species distributions across the United States and evaluating their conservation applications

“…Of the total species represented in the BBS in our study area, we successfully modeled 56% (192 of 341 species) at the 5-km resolution, 50% (160 of 317 species) at the 2.5-km resolution, and 28% (80 of 282 species) at the 0.5-km resolution due to data constraints (Carroll et al, 2023). Of the species with at least 100 observations (our modeling cutoff based on our number of predictors), we successfully modeled 96% (192 of 199 species) at the 5-km resolution, 98% (160 of 163 species) at the 2.5-km resolution, and 99% (80 of 81 species) at the 0.5-km resolution-see Appendix S3.…”

“…Species distribution models (SDMs) are typically used to generate such products (Elith & Leathwick, 2009; Franklin, 2010; Peterson et al, 2011). However, many existing distribution maps are based on guild or other aggregated occurrence data due to species‐level occurrence data scarcities, efforts to better capture community patterns and ecosystem complexity, or both (Carroll et al, 2022; D'Amen et al, 2017; Ferrier & Guisan, 2006).…”

“…We aimed to model and map single‐species distributions and determine which species had sufficient data for individual species mapping at different management‐relevant resolutions (0.5, 2.5, 5 km). We also sought to compare aggregates of our single‐species maps (stacked maps, Figure 1b) with those generated using functional guild richness data (guild maps, Carroll et al, 2022) to determine where, spatially, stacked maps differed from guild maps across resolutions and by how much. We expected that we would be unable to model a small proportion of species surveyed by the North American Breeding Bird Survey (BBS) at the coarsest resolution (5 km) due to insufficient sample sizes, and many more species would lack sufficient observations for distribution models at the finest resolution (0.5 km).…”