The number of buds from 30 inflorescences (on five branches) from the 10 trees at each agroforest was counted. Developing and mature fruit on the same inflorescences were counted on four subsequent occasions: early fruit set (mid-March to early May), pre-monsoon fruit set (June), post-monsoon fruit set (September), and final production (December to January). We also considered pre-monsoon fruit drop (early fruit set minus pre-monsoon fruit set) and monsoon fruit drop (post-monsoon fruit set minus pre-monsoon fruit set). Once collected, berries were dried in an oven at 60 °C for 24–36 h until moisture content was around 12%, as required by the coffee sector. For each of the 10 trees, we evaluated berry production per inflorescence in absolute terms (most relevant from a farmer’s perspective). The ratio of fruits to flowers was not used as this would have obscured the interaction effect of flower number and bee visitation with regard to fruit production. The initial number of flowers and previous counts were included as covariates in the corresponding models.