Analysis of extreme values of soil ecosystem services predicted from associated soil properties and weather conditions

Soils provide services that can increase food security and mitigate global warming. Simulation modelling of biophysical processes is one approach used to estimate values of soil ecosystem services (SES) based on soil properties and climate conditions. The SES are usually investigated using statistical approaches that assume that values of SES are normally distributed. This assumption, however, ignores atypical values of SES, which deviate strongly from average values. Classic approaches have, therefore, failed to identify such atypical values of SES. To address this issue, we applied extreme value theory (EVT) as a statistical tool to identify thresholds above which SES can be considered as 'extreme'. EVT was applied to a dataset of four SES-climate regulation, groundwater recharge, water-to-plant provision and plant biomass provision-estimated by biophysical soil and crop modelling based on a dataset of 64 cultivated soils located in northwestern France under a single climate observed over 30 years (1988-2018) and a single crop management of a maize and wheat rotation. Extreme values of the SES considered in our study were identified, and soil properties associated with these values were characterised. For instance, soils with extreme minimum values of climate regulation service had the highest pH, clay and organic carbon contents and maximum rooting depth. Applying EVT also enabled identification of the 'return period', which is the amount of time after which an extreme value of SES is likely to occur again. For instance, under the maize and wheat rotation simulated and weather conditions equivalent to those from 1988 to 2018, storage of organic carbon in the soil of 1000 kg C ha(-1) year(-1) is expected to be exceeded by at least one of the 64 of soils once every 2 years. This approach can thus be useful for decision making in natural resource management by identifying atypical values of SES and predicting how frequently they may occur.HighlightsExtreme value theory identified soils with extreme values of ecosystem services (SES).Some soil properties differed significantly among samples of extreme values of SES.Soils in extreme samples of SES were identified as a function of their parent material.The mean extreme maxima of the SES were expected to be exceeded every 2 years.