Application of null models to evaluate the incorporation of intraspecific variation in ecological niche models for European beech (Fagus sylvatica L.)

Ecological niche models (ENMs) are vastly applied to predict species range shifts under climate change scenarios. However, classical ENMs assume species homogeneity, disregarding the effects of gene flow limitations and resulting intraspecific variation through local adaptation. Compared to classical ENMs, incorporating intraspecific variation (e.g., by using genetic data) can improve model performance and provide more realistic predictions of climate change impacts. Genetically informed ENMs are still rarely used and may suffer from biased estimates of model performance or overfitting. We propose the application of a null model approach to assess the model performance of genetically informed ENMs and provide further information about the meaning of intraspecific variation in ENMs.

We use publicly available range-wide genetic structure and observations from forest inventory plots of European beech (Fagus sylvatica L.) to create individual genetically informed models for the distinct genetic groups representing different ecotypes. The obtained individual models are combined into one overall genetically informed model. Model performance is assessed with measures of discriminatory ability, precisely area under the receiver operating characteristic curve (AUC), sensitivity, specificity and TSS. Null models based on random sets of spatial points are computed for each of the models. This allows us to estimate null distributions of the performance measures and compare them to the measures of the real model. By doing so, the null models will help us to check if indeed genetic adaptation or rather just statistical artifacts explain the improved habitat suitability found in such approaches. Finally, we compare model performance of the individual models and the overall model against a classical ENM approach. Thus, our approach will allow to quantify the importance of local adaptation for future habitat suitability.