We performed bird predation experiments (dummy experiments), using artificial prey and bird community data to investigate the importance of predator diversity vs. predator identity in cacao agroforestry landscapes. All sample sites were situated at the northern tip of Napu Valley in Central Sulawesi, Indonesia. After an initial mapping of the study area, we selected 15 smallholder cacao plantations as sites for our exclosure experiments in March 2010. For our predation experiment, we selected 10 (out of 15) study sites and 5 cacao trees per site for the application of artificial prey for birds (dummy caterpillars made of plasticine). Our study trees (numbered from 1 to 5 per site) were randomly chosen and we kept spacing of at least two unmanipulated cacao trees between two study trees to avoid clumped distribution. To quantify both daytime/diurnal predation and night-time/nocturnal predation (e.g. birds vs. bats), we applied 7 caterpillar dummies on all study trees and controlled them for predation marks in the early morning (05:00-06:00 am), in the evening (17:00-18:00 pm) and in the early morning on the next day (completing one survey round). In total, we performed four survey rounds per study site (in June and July 2011). The caterpillar dummies were always applied in the same order and on three different parts of each cacao study tree: One 'control dummy' (located on first branching of the cacao tree); 3 'branch dummies' (located on one main branch coming from first branching; 20-25 cm between single dummies) and 3 'leaf dummies' (3 medium aged cacao trees adjacent to main branch were selected and single dummies placed in the center of each cacao leaf). The different positions were chosen to control for different foraging modes of predators (e.g. branch gleaners versus leaf gleaners). During day- and nighttime surveys, we controlled if the dummy caterpillars were still present in their original position, if they were absent and could not be relocated on the ground or if they were fallen to the ground, but could still be recorded. Eaten dummies were counted as 1 mark usually, except for those dummies, where two or more different kind of arthropods had eaten parts of the dummy (2 marks or more). Other predation marks were added to this number. For each dummy, we counted the total number of different predation marks. We focused on predation marks that could be identified with certainty (based on preliminary observations and/or literature): marks of birds, rodents and snails. Finally, we analysed the relationship of bird predation marks and bird community parameters (abundance vs. diversity), as well as effects of local and landscape management on the avian predation success.