Food webs can deliver win–win strategies for tropical agroforestry and biodiversity conservation

High biodiversity and agricultural productivity are commonly regarded as mutually exclusive. However, functionally diverse communities may provide valuable services to agroecosystems and therefore offer the possibility of win-win strategies. We developed a dynamic mechanistic community model of the bird-arthropod food web associated with African cocoa agroforestry, structurally informed by metabarcoding data on bird diets, and fitted to trapping data on species abundances. Our novel framework models rates of change and uses space-for-time substitution, thus providing insights into community dynamics without the need for long time-series data. We used our fitted model to predict equilibrium community composition under varying intensities of shade management and pesticide use. Our results indicate that low-intensity farming favors forest bird species, with at least two times the biomass of this bird group compared to any other at high shade cover. Low-intensity farms also favored potential pollinator abundance, while biomass of the main pest species of cocoa, brown capsid, was 25% lower at high shade than at low shade. Our model quantified the net effect of each taxon on the other taxa in the food web: most bird to arthropod interactions were negative, indicating important pest control services provided by birds. Furthermore, our simulations of pesticide application revealed that the long-term effect of pesticide use on biomass of taxa varied according to shade cover. Importantly, pesticide application resulted in the decline of non-pest taxa through trophic cascades: forest birds were the taxa that declined the fastest, and this trend was exacerbated in low shade farms. To achieve a decline of less than 50% in non-pest taxa, pesticide application could only reach 10% in the sunniest farms and 20% in shady farms, which results in a maximal reduction of 19% in pest biomass. By looking at the efficacy of agricultural management through the lens of community interactions, our holistic, quantitative approach demonstrates that low-intensity agriculture may provide a win-win for biodiversity and ecosystem services.