ABSTRACT Modelling the distribution of microendemic species presents significant challenges due to limited occurrence records and the coarse resolution of available bioclimatic data. This is particularly true for montane regions, which harbour high levels of endemism and environmental heterogeneity. In this study, we modelled the potential distribution of Brachycephalus pernix group toadlets to assess their range and identify key environmental drivers of their ecological niches. We applied two correlative modelling approaches—model‐selection procedures using MaxEnt and Ensemble Small Models—incorporating a broad suite of environmental predictors beyond traditional bioclimatic variables. Our results highlight that Ensemble Small Models outperformed model‐selection procedures (MaxEnt) in predicting suitable habitats for these microendemic species, yielding more spatially precise predictions centred in highland, montane and submontane regions. Suitability was strongly associated with environmental variables related to precipitation and moisture, which play a critical role in shaping the realised niche of the B. pernix group. The species exhibited niche conservatism, likely reflecting the retention of ancestral ecological preferences that facilitate persistence in montane environments. This supports the hypothesis that mountain ranges act as long‐term refugia during climatic fluctuations. Importantly, models incorporating heterogeneous environmental data outperformed those using only bioclimatic variables, highlighting the value of accounting for topographic and climatic complexity when modelling narrow‐range taxa. Despite identifying additional suitable habitats, many of these areas remain unprotected and are increasingly threatened by deforestation and land‐use change. Our findings provide new insights into the ecological requirements and distribution dynamics of the B. pernix group and emphasise the urgent need for targeted conservation efforts to safeguard their specialised habitats and ensure long‐term persistence.
Rivadeneira et al. (Tue,) studied this question.