Living on the edge—Modelling habitat suitability for species at the edge of their fundamental niche

Abstract

Predictive species distribution models have become increasingly common in conservation management. Among them, envelope-based approaches like the Ecological Niche Factor Analysis (ENFA) are particularly advantageous, as they require only presence data. Based on the assumption that the absolute frequency of species presence is a direct indicator of habitat suitability (HS), habitat suitability indices (HSI) are computed. However, this assumption may be misleading when the scarcity of optimal habitat forces most of the individuals to live in suboptimal conditions. This often happens when the environmental conditions in the study area represent only a marginal part of the species fundamental niche. In this study we propose three new HS algorithms for ENFA models, which address such ‘edge of niche’ situations. The first algorithm (area-adjusted median, M a) takes the availability of environmental conditions in the study area into account, the second (median + extremum, M e) addresses situations where the species’ optimum is at or beyond the extremum of the investigated environmental gradient, and the third (area-adjusted median + extremum, M ae) combines both approaches. These algorithms were applied to two populations of capercaillie ( Tetrao urogallus), situated in different positions relative to the environmental gradient represented in the respective study area, and compared with the classical median algorithm (M). We evaluated the models using cross-validation and a comparison with an expert model based on external data. In both study areas, the HS maps obtained with the three new algorithms differed visibly from those calculated with the median algorithm. Cross-validation and comparison with external data showed that the new algorithms always provided better models, with the extremum-based algorithms (M e and M ae) performing best. We conclude that the new algorithms can extend the applicability of ENFA-models to a broader range of conservation-relevant species by improving HS calculations for skewed species–habitat relationships in marginal habitats.