A new conservation assessment for the Ecuadorian spiny pocket mouse informed by optimized ecological niche models and updated land cover and protected area data

Background/Question/Methods

Ecological niche models (ENMs) can be used to estimate the distribution of poorly known species from even a small sample of occurrence records. The spiny pocket mouse Heteromys teleus is known from only eight localities in the seasonal evergreen forests of western Ecuador. A study ten years ago modeled the species’ niche using the GARP algorithm and coarse climatic data. That prediction was used to make a conservation assessment, and abiotically suitable, current suitable, and protected suitable area totals were estimated. We reproduced this analysis with a newer algorithm, MaxEnt, and improved climatic data (finer resolution and greater variety of climatic variables). We applied evaluation criteria to estimate optimal model settings (by minimizing omission rate and maximizing evaluation AUC) over two climatic combinations (temperature + precipitation alone vs. those combined with solar radiation + soil moisture). We report a new conservation assessment for the species and suitable area totals for Ecuador using updated layers for 1) land cover, and 2) public and private protected areas.

Results/Conclusions

Optimal model settings for both climatic combinations were identical (linear features, regularization multiplier 1.5), and the suitability surfaces differed only slightly. Temperature + precipitation alone resulted in approximately 5% more suitable area than models made with the addition of solar radiation + soil moisture. The total abiotically suitable area was 1.5 times higher than the GARP prediction. Current (after taking into account land cover) and protected estimates were 3 and 6 times higher, respectively. Tentatively, we conclude that H. teleus may have more current suitable area (and much more protected suitable area) than the previous study estimated. However, various methodological differences may have led to these discrepancies. Therefore, future steps will include considering biotic interactions with a parapatric sister species, incorporating remote sensing deforestation datasets, and conducting field studies to confirm presence in newly predicted areas.