Quantile Regression

Species Distribution Models predict abundance and its temporal variation in a steppe bird population.

Habitat Suitability Index (HSI) derived from Species Distribution Model (SDM) has been used to infer or predict local demographic properties such as abundance for many species. Across species studied, HSI has either been presented as a poor predictor of abundance or as a predictor of potential rather than realized abundance. The main explanation of the lack of relationship between HSI and abundance is that the local abundance of a species varies in time due to various ecological processes that are not integrated into correlative SDM. To better understand the HSI-abundance relationship, in addition to the study of the association between HSI and mean abundance, we explored its variation over time. We used data from 10-years monitoring of a Houbara bustard (Chlamydotis undulata undulata) population in Morocco. From various occurrence data we modelled the HSI. From (independent) count data we calculated four local abundance indices: mean abundance, maximum abundance, the temporal trend of abundance and the coefficient of variation of abundance over the study period. We explored the relationship between HSI and abundance indices using linear, polynomial and quantile regressions. We found a triangular relationship between local abundance (mean and maximum) and HSI, indicating that the upper limit of mean and maximum abundance increased with HSI. Our results also indicate that sites with the highest HSI were associated with least variation in local abundance, the highest variation being observed at intermediate HSI. Our results provide new empirical evidence supporting the generalization of the triangular relationship between HSI and abundance. Overall, our results support the hypothesis that HSI obtained from SDMs can reflect the local abundance potentialities of a species and emphasize the importance of investigating this relationship using temporal variation in abundance.