Human appropriation of net primary production as driver of change in landscape-scale vertebrate richness.

Aim: Land use is the most pervasive driver of biodiversity loss. Predicting its impact on species richness (SR) is often based on indicators of habitat loss. However, the degradation of habitats, especially through land-use intensification, also affects species. Here, we evaluate whether an integrative metric of land-use intensity, the human appropriation of net primary production, is correlated with the decline of SR in used landscapes across the globe.

Location: Global.

Time Period: Present.

Major Taxa Studied: Birds, mammals and amphibians.

Methods: Based on species range maps (spatial resolution: 20 km × 20 km) and an area-of-habitat approach, we calibrated a "species-energy model" by correlating the SR of three groups of vertebrates with net primary production and biogeographical covariables in "wilderness" areas (i.e., those where available energy is assumed to be still at pristine levels). We used this model to project the difference between pristine SR and the SR corresponding to the energy remaining in used landscapes (i.e., SR loss expected owing to human energy extraction outside wilderness areas). We validated the projected species loss by comparison with the realized and impending loss reconstructed from habitat conversion and documented by national Red Lists.

Results: Species-energy models largely explained landscape-scale variation of mapped SR in wilderness areas (adjusted -values: 0.79-0.93). Model-based projections of SR loss were lower, on average, than reconstructed and documented ones, but the spatial patterns were correlated significantly, with stronger correlation in mammals (Pearson's  = 0.68) than in amphibians ( = 0.60) and birds ( = 0.57).

Main Conclusions: Our results suggest that the human appropriation of net primary production is a useful indicator of heterotrophic species loss in used landscapes, hence we recommend its inclusion in models based on species-area relationships to improve predictions of land-use-driven biodiversity loss.