Fig. 2: Potential for farm- and field-level causal inference.

a In a biodiversity-focused survey, species observations (bird icon, in blue) are detected with passive acoustic monitoring. The timing of the survey is aligned with the main breeding season of the target species (blue outline). Species detections are paired with categorical information on crop type (in dark green), common practice in state-of-the-art agroecological research. b Passive acoustic sensors are used to detect crop-damaging birds, while drones help monitor crop growth, damages, and health, but could also enable time-series on species-level bird abundances (c) based on detections recorded outside the breeding season (bird icons, in black). d With time-series on agriculture and biodiversity, together with their uncertainties (shaded ribbons), we can formally measure the presence and magnitude of causal effects, such as relating to biodiversity-yield or predator-prey interactions.