* Habitat conversion for agriculture is a major driver of biodiversity loss, but our understanding of the demographic processes involved remains poor. We typically investigate the impacts of agriculture in isolation even though populations are likely to experience multiple, concurrent changes in the environment (e.g. land and climate change). Drivers of environmental change may interact to affect demography, but the mechanisms have yet to be explored fully in wild populations. * Here, we investigate the mechanisms linking agricultural land use with breeding success using long-term data for the formerly Critically Endangered Mauritius kestrel Falco punctatus, a tropical forest specialist that also occupies agricultural habitats. We specifically focused on the relationship between breeding success, agriculture and the timing of breeding because the latter is sensitive to changes in climatic conditions (spring rainfall) and enables us to explore the interactive effects of different (land and climate) drivers of environmental change. * Breeding success, measured as egg survival to fledging, declines seasonally in this population, but we found that the rate of this decline became increasingly rapid as the area of agriculture around a nest site increased. If the relationship between breeding success and agriculture was used in isolation to estimate the demographic impact of agriculture, it would significantly under-estimate breeding success in dry (early) springs and over-estimate breeding success in wet (late) springs. * Analysis of prey delivered to nests suggests that the relationship between breeding success and agriculture might be due, in part, to spatial variation in the availability of native, arboreal geckos. * Synthesis and applications. Agriculture modifies the seasonal decline in breeding success in this population. As springs are becoming wetter in our study area and since the kestrels breed later in wetter springs, the impact of agriculture on breeding success will become worse over time. Our results suggest that forest restoration designed to reduce the detrimental impacts of agriculture on breeding may also help reduce the detrimental effects of breeding late due to wetter springs. Our results therefore highlight the importance of considering the interactive effects of environmental change when managing wild populations.
