PS 32-187 - Effects of model complexity on ecological niche models for detecting Allocapnia species distribution patterns

Abstract: Stoneflies are crucial components of stream food webs and are important biological indicators of optimal stream habitats and water quality, however climate change has significantly impacted distribution range of individual species and biodiversity hotspots.

We compiled historical collection records for nine winter stoneflies to compare species richness across ecoregions in the southern U.S.A. and examine the effects of model complexity (Regularization Multiplier (RM) and the Feature Combinations (FCs)) on the predictive performance of machine learning maximum entropy (Maxent) models. In addition to identifying stonefly biodiversity hotspots, we quantified the importance of environmental variables based on a comparison of past and future patterns of stonefly species distributions under two Shared Socio-economic Pathways (SSP) climate change scenarios (SSP2-4.5 and SSP5-8.5).

Our results indicate both default and optimized Maxent models were effective in predicting species occurrence, with Average Area Under the Curve (AUC) values above 0.80 for all species except A. ozarkana. Allocapnia species were predominantly found in small- or medium-sized streams, and fewer in larger streams. Moreover, the Ozark Mountain ecoregion had the highest species richness, while the Western Corn Belt Plains had the lowest species richness. Increasing temperature scenarios resulted in a decrease in the probability of occurrence for all the target Allocapnia species. Greater deciduous forest coverage increased the likelihood of occurrence. Higher sediment concentration significantly reduced the probability of stonefly presence, but the effect on A. mystica was inconsistent with the predictions for the other species. Based on our findings, climate change will have a negative impact on the suitable habitat for Allocapnia species, with a reduction in available optimal environments for their survival and reproduction. Nevertheless, certain species may exhibit an increase in habitat suitability, possibly due to their varying temperature optima, facilitating spatial distribution changes. Future predictions indicate that A. ozarkana will have the largest growth in suitable area, with expanded habitat in both the Ozark Mountain and Boston Mountain ecoregions. The study's results also have significant implications for species-distribution modeling, especially when using Maxent.