Telemetry-Derived Artificial Intelligence SDMs: A Generalizable Conservation Framework to Inform Species Interactions with Offshore Wind Energy Developments.

Ingram, Evan¹*, Keith Dunton², Michael Frisk¹, Liam Butler³, ¹School of Marine and Atmospheric Sciences, Stony Brook University, Stony Brook, NY, ²Monmouth University, West Long Branch, NJ, ³Wake Forest University, Wake Forest, NC

The development of renewable wind energy on the outer continental shelf of the US has led to growing concerns for marine wildlife. However, there remains significant uncertainty regarding the technology’s impact on marine ecosystems and fauna. The noted scarcity of fisheries-independent data on habitat use for species of interest that may occupy planned development sites is further compounded by the novelty of the technology and the inherent difficulty of offshore monitoring. This makes the practical assessment of site- or species-specific threats that may require additional management intervention particularly problematic, and can result in regulatory incongruities. Species distribution models (SDMs) are correlative predictive models that are important decision-making tools for conservation planning because of their ability to quantify species-biogeographic relationships at biologically significant scales. SDMs can be used to describe and measure the importance of specific factors to distribution and predict habitat suitability or species distribution outside of sampled areas. Here we present a generalized framework for the use of SDMs to inform species interactions in marine habitats allocated for offshore wind energy development, using the federally protected Atlantic Sturgeon (Acipenser oxyrinchus oxyrinchus) as a case study. Acoustic-telemetry occurrence records of Atlantic Sturgeon in marine waters were collected over an 11-year monitoring period and used to create pooled-monthly SDMs. Presence-only data were derived from telemetry detections and background pseudo-absences were randomly generated. Environmental data were adopted from publicly available datasets to allow for general applicability. Preliminary results are promising and provide critical spatiotemporal information on the probability of occurrence for Atlantic Sturgeon over a broad swath of previously-unsampled marine habitats. Furthermore, monthly habitat suitability maps created from SDMs, when overlaid with offshore wind energy lease areas, can directly feed into management and inform best practices for potential habitat influences of Atlantic Sturgeon, as well as other species of commercial or conservation interest.