Assessing Risk to Birds from Industrial Wind Energy Development via Paired Resource Selection Models

Abstract

When wildlife habitat overlaps with industrial development animals may be harmed. Because wildlife and people select resources to maximize biological fitness and economic return, respectively, we estimated risk, the probability of eagles encountering and being affected by turbines, by overlaying models of resource selection for each entity. This conceptual framework can be applied across multiple spatial scales to understand and mitigate impacts of industry on wildlife. We estimated risk to Golden Eagles (Aquila chrysaetos) from wind energy development in 3 topographically distinct regions of the central Appalachian Mountains of Pennsylvania (United States) based on models of resource selection of wind facilities (n = 43) and of northbound migrating eagles (n = 30). Risk to eagles from wind energy was greatest in the Ridge and Valley region; all 24 eagles that passed through that region used the highest risk landscapes at least once during low altitude flight. In contrast, only half of the birds that entered the Allegheny Plateau region used highest risk landscapes and none did in the Allegheny Mountains. Likewise, in the Allegheny Mountains, the majority of wind turbines (56%) were situated in poor eagle habitat; thus, risk to eagles is lower there than in the Ridge and Valley, where only 1% of turbines are in poor eagle habitat. Risk within individual facilities was extremely variable; on average, facilities had 11% (SD 23; range = 0-100%) of turbines in highest risk landscapes and 26% (SD 30; range = 0-85%) of turbines in the lowest risk landscapes. Our results provide a mechanism for relocating high-risk turbines, and they show the feasibility of this novel and highly adaptable framework for managing risk of harm to wildlife from industrial development.

Use of multiple modes of flight subsidy by a soaring terrestrial bird, the golden eagleAquila chrysaetos, when on migration

“…Flight subsidized by thermal uplift is characterized by greater AGL [39] and thus lower risk. To date, modelling of risk to eagles from turbines in this region has been based simply on flight altitude, where low-altitude flight is classified as risky flight [41]. Because bird behaviour is likely to be directly linked to risk [42,43], a next generation of models could use flight classification algorithms to refine prediction risk by linking risk to specific low-altitude flight behaviours (foraging, use of orographic updraft).…”

Integrating citizen-science data with movement models to estimate the size of a migratory golden eagle population

“…Topography in Pennsylvania includes many long-linear ridges, lowland valleys, forested highlands, and mountain foothills (United States Forest Service, 2007). Local autumn weather is temperate, overcast, and characterized by westerly winds that interact with the steep topography of the ridge and valley regions to generate uplift (Miller et al, 2014). Of 32 established hawk-count sites, approximately 12 are active and regularly staffed in Pennsylvania, although others once operated intermittently.…”

“…The model was necessary because (a) hawk-count sites do not randomly sample the migrant eagle population (if they did, then we would only need detection rates to estimate population size) and (b) because telemetry data (Miller et al, 2014) are too sparse to use to estimate this parameter. We briefly describe the migration model here.…”

High-voltage power lines near wild reindeer calving areas

“…RSFs are defined as any function proportional to the probability of use of a resource unit area by an individual and have been widely used across a range of taxa including mammals (Godvik et al 2009;Peters et al 2015), birds (Miller et al 2014), and reptiles (Bauder et al 2014). The response variable was binomial (used/available) and consisted of reindeer GPS locations (taken with 3 h intervals; 10,909 locations in both areas) and an equal amount of randomly sampled positions within the merged 99 % BBMM home range for all animals all years for each area (excluding the one individual that remained at one site closer to the power line for more than a month in 2007, 2008, and 2010), representing available area.…”