Rearranging agricultural landscapes towards habitat quality optimisation: In silico application to pest regulation

Abstract

Modern agriculture suffers from its dependence on chemical inputs and subsequent impacts on health and environment. Alternatively, protecting crops against pests can be achieved through the reinforcement of regulation ecological services. Our work propounds a data-driven methodological framework to derive relevant agricultural landscape rearrangements enhancing populations of beneficial organisms regulating pests.Building on spatialised entomological and geographic data, we developed a parsimonious reaction–diffusion model describing the population dynamics of beneficial organisms. Parameter estimation was carried out in a Bayesian framework accounting for uncertainty in the measurement.Thousands of agricultural landscapes were generated under agronomic specifications dealt with as constraint satisfaction problems. Population dynamics was simulated on each landscape with the fitted reaction-diffusion model mentioned above, and two metrics of abundances allowed the assessment of the regulation performance of the landscape spatial arrangements. One metric is a mean field performance criterion assessing the regulation performance from the landscape composition only, the other is a spatial performance metric assessing the performance resulting from the whole landscape spatial configuration. The former is computed with a non-spatialised form of the population dynamics model, the latter results from the reaction-diffusion model of the population dynamics. Comparing these metrics enabled to quantify the impact of spatial arrangements, hence allowing arrangements proposals.This framework was applied to the case study of a ground beetle species involved in the biological regulation of weeds. The arrangement proposals abides by the productive agronomic constraint that is the landscape composition, while they allow for significant habitat quality enhancement (or deterioration) for the beneficial organism (or a pest). Minor adaptations of our integrated data-driven approach would suit numerous situations ranging from the provision of enhanced ecosystem services to land management for conservation.