OOS 24-6 - Data integration for ecological forecasting: Elevational and physiological offsets improve species distribution models

Biodiversity is increasingly threatened by human disturbance and global warming causing changes in species distributions and projected extinctions within the century with implications for the functioning of ecosystems and ultimately human well-being. Spatially detailed information of species distributions is essential to forecasting and mitigating the biological impacts of ongoing climate change and human modifications. Yet, our information on biodiversity is often incomplete for most facets of biodiversity. Here, we discuss how model-based approaches which integrate point occurrences with additional trait based or physiological information can pose a significant step forward for species distribution modeling predictions. Specifically, we showcase how incorporating expert-knowledge based species elevational range limits and expert range maps with point occurrences as offsets in a Poisson point process species distribution model (SDM) consistently improves the distributions of 276 hummingbird Trochilidae species at 1 km resolution. Specifically, SDMs that included expert range information out-performed presence-only models based only on occurrence data in 92.5% of cases and had higher sensitivity, lower false positive rates and smaller predicted range sizes. Separately, we show how integrating laboratory-derived thermal physiological information as ‘offsets’ into an SDM framework for ten reptiles and amphibian species leads to small but consistent gains in model performance and smaller calculated species geographic range sizes. Further, when projecting species distributions under several climate change scenarios, we find that physiologically informed SDMs suggest 2-30% greater range loss than traditional correlational models due to the incorporation of upper physiological thermal limits. Generally, these integrated models can remove unsuitable areas from range estimates and decreased overestimates in geographic range size with implications for estimating biodiversity patterns and identifying regions for conservation actions. This work highlights the value gained from integrating multiple data types in a single framework.