When scale matters: Drivers of faunal diversity in cacao production landscapes in the Peruvian Amazon

Preserving biodiversity in tropical agricultural areas relies on a fine balance between crop management strategies and landscape characteristics. Moreover, the response of different species to these local and landscape metrics may greatly differ. Therefore, identifying strategies to maximize biodiversity in tropical agricultural landscapes relies on analyses that encompass multiple spatial scales and taxa simultaneously. To address this, we assessed a multidiversity index, comprising the diversity of birds, bats, ground and flying insects in four different land uses in a highly deforested landscape in the Peruvian amazon basin: forest fragments, cacao monocultures, cacao agroforests and degraded areas, mainly covered by pastures and early successional vegetation. We identified the main biophysical drivers influencing annual multidiversity across the four land uses and across the dry and wet seasons. We complemented random forest (RF) analyses with generalized linear mixed-effects models. Our RF models explained between 46% and 62% of data variability, for annual multidiversity and for the dry and wet seasons. We observed that cacao growing in agroforestry systems held similar multidiversity values to those observed for forest fragments. Further, our findings provide specific management recommendations to integrate agricultural production with biodiversity conservation in a highly intervened landscape: such as the relevance of maintaining and increasing forest cover within the landscape as well as increasing the connectivity of tree-dominated land use types (i.e. agroforestry systems, forests fragments, agricultural areas mainly covered by trees).