On the early warning signal of degradation in drylands: Patches or plants?

Abstract

Global drylands are threatened by grazing pressure intensification and climate change, which act as major drivers of land degradation. Detecting this process at an early stage is essential for predicting losses of ecological functions and for restoration management. Vegetation patch-size distribution is an indicator of dryland multifunctionality and has been proposed as a warning signal for the onset of degradation processes. However, we proposed and tested a general model that stresses that patchiness may fail to detect degradation of the forage provision, depending on plant community species composition. This is a key aspect since forage provision is strongly associated with human well-being in drylands. We hypothesized that grazing-induced changes in patchiness and forage provision converge in drylands dominated by forage species but are decoupled in those dominated or co-dominated by non-forage species. We tested the conceptual model in a unique regional-scale gradient with strong ecological differences but a common biogeographical and human impact history to reduce local contingencies effects. We compared datasets of grazing intensification impacts on (i) plant cover and patch-size distribution and (ii) plant density and plant-size distribution of dominant forage grasses (a proxy of forage provisioning). We showed that there is a decoupling between grazing-induced changes in vegetation patchiness and forage provisioning, particularly in drylands where non-forage species are dominant. In these drylands, plant cover and patch-size distribution were slightly affected by grazing intensification, whereas plant density of forage species was decimated and their plant-size distributions were strongly skewed towards small sizes. Synthesis. Our dryland conceptual model suggests that global change impacts on forage species populations can be detected even before changes in patch-size distribution and plant cover. Our findings support the model and indicate that the population status (plant density and plant-size distribution) of forage species allows for predicting forage dynamics and is useful to the early detection of losses of ecosystem services linked to human well-being in drylands.