The strong negative relationships observed between aridity and stability, nutrient cycling indices demonstrated that increasing aridity can reinforce erosive processes and deteriorate nutrient cycling (i.e. aridity decreased soil carbon, nitrogen and enzymatic activities). Similarly, aridity was negatively related to vegetation structural attributes including patch cover and landscape organization index. This finding suggests that any increase in aridity can promote the loss of perennial vegetation cover.

With ongoing climate change, the probability of crossing environmental thresholds promoting abrupt changes in ecosystem structure and functioning is higher than ever. In drylands (areas where it rains <65% of what could be potentially evaporated), recent research has shown how the crossing of three aridity thresholds [at aridity (1-Aridity Index) values of 0.54, 0.70 and 0.80] leads to abrupt changes on ecosystem structural and functional attributes. Despite the importance of these findings and their implications to develop effective monitoring and adaptation actions to combat climate change and desertification, we lack a proper understanding of the mechanisms unleashing these abrupt shifts.

Concurring evidence from independent studies highlights the existence of systemic (i.e. attaining several ecosystem compartments), sequential (as aridity increases) and global ecosystem thresholds associated with increases in aridity across drylands world-wide. According to aridity forecasts, up to 22% of emerged lands may cross one or several of these aridity thresholds by the end of this century.
These changes initialize with a nonlinear decrease in photosynthetic activity at aridity values of 0.54, followed by different and abrupt changes in the soil system and abrupt increases in shrublands at aridity 0.7. Finally, an abrupt decline in plant cover matches abrupt declines in richness and changes in leaf strategies to stress avoidant aridity values of 0.82.

Aridity thresholds spotted generally match previous divisions of drylands separating dry–subhumid from semi-arid (aridity ~0.5) and semi-arid from arid (aridity ~0.8). Indeed, soil disruption phase matches with the idea that transitioning from semi-arid to arid systems involves a lack of suitable soil fertility to perform agriculture.
Similarly, the definition of semi-arid areas as those in which vegetation indices are linearly correlated with aridity coincides with the vegetation decline phase. What aridity thresholds add to previous classification frameworks is matching the exact value of aridity with key ecosystem attributes that are affected by aridity in abrupt manners. This provides us the pieces of a puzzle to be assembled through mechanistic knowledge to understand abrupt changes in drylands.
The basic idea is to generate aridity maps capable of highlighting the areas in which the conditions are created for long-lasting changes to the vegetation, to the soil and more generally a systemic failure that leads to the loss of perennial vegetation
The Aridity Index (AI) is a simple but convenient numerical indicator of aridity based on long-term climatic water deficits and is calculated as the ratio P/PET. The AI is a widely used measure of dryness of the climate at a given location.

The basic idea is to generate aridity maps capable of highlighting the areas in which the conditions are created for long-lasting changes to the vegetation, to the soil and more generally a systemic failure that leads to the loss of perennial vegetation
The Aridity Index (AI) is a simple but convenient numerical indicator of aridity based on long-term climatic water deficits and is calculated as the ratio P/PET. The AI is a widely used measure of dryness of the climate at a given location.
This calculation uses the MODIS PET data generated every 8 days, and the daily CHIRPS precipitation data which is also aggregated as an 8-day sum. Both datasets are stored in Google Earth Engine and are queried, managed and analyzed using the Python API.
The visualization of the python code is carried out using a Jupyter Notebook, which using geemap makes it easy to check the analyses precisely and in real time.

The check of the generated aridity index is done through a comparison with the dataset generated by CGIAR which offers on a monthly basis high-resolution (30 arc-seconds) global raster climate data for the 1970-2000 period, related to evapotranspiration processes and rainfall deficit for potential vegetative growth, based on the implementation of a Penman-Montieth Reference Evapotranspiration (ET0) equation.