2025/06/01 by Andrew Allyn, Stephanie Brodie, Katherine E. Mills +9 · 1 voice
Environmental Science · Biochemistry, Genetics and Molecular Biology · #Species Distribution and Climate Change #Wildlife Ecology and Conservation #Genetic diversity and population structure
paper · pdf · doi:10.1111/ddi.70036
openalex publication_date 2025/06/01 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
ABSTRACT Aim Despite the rapid development and application of species distribution models (SDMs) to predict species responses to climate‐driven changes, we have a limited understanding of model predictive performance under novel environmental conditions. Location California Current (CC) and Northeast U.S. Shelf (NES) Large Marine Ecosystems, USA. Methods We used a simulation experiment to evaluate how novel environmental conditions and species‐specific environmental tolerances influence SDM predictability. We leveraged sea surface temperatures in the CC and NES large marine ecosystems (LMEs) and simulated the distribution of a resident‐mobile and a seasonally‐migrating ectotherm in each LME. For each LME and species archetype, we fitted boosted regression tree SDMs using data from 1985 to 2004, then predicted the monthly probability of presence from 2005 to 2020 and calculated the novelty of prediction conditions. Results Ocean warming increased environmental novelty during the prediction period, with seasonal variations as novel conditions increased in summer and fall, and decreased in winter and spring as cool conditions became rarer. Predictive performance declined as novelty increased, and this decline occurred before prediction conditions became distinguishable from observation conditions and with temperature increases below those expected given continued climate change. However, unexpected increases in performance under novel environmental conditions arose when they occurred over optimum species‐response curve temperatures. Main Conclusions While environmental novelty reduces prediction performance, this isn't always the case. Prediction challenges depend on where novel conditions map onto species‐response curves, with performance increasing when novel conditions occur over species‐response curve optimums. As SDM applications expand, there will be an ongoing need to maximise data quantity and quality to better characterise species' range of suitable conditions, explore novelty relative to species‐response curves, and improve methods for quantifying and communicating model uncertainty. These efforts will open opportunities for model improvement and support stakeholders' capacity to integrate predictions into decision‐making.