We conducted the study in Reserva Florestal Adolpho Ducke (02°55'-03°01' S; 59°53'-59°59' W; Ducke Reserve hereafter), located on the outskirts of Manaus (Km 26 on the AM-110 Highway, Amazonas State, Brazil). The reserve is included in the Long-term Ecological Research Program of the Brazilian National Research Council (Programa de Pesquisas Ecológicas de Longa Duração – PELD/CNPq) and harbors 10,000 ha of mature terra firme forest. We captured bats in 49 plots, 24 of which are in riparian areas close to watercourses, and 25 in non-riparian areas. We captured bats using eight ground-level mist-nets (12×3 m, 19 mm mesh, Ecotone®, Poland) per plot between October 2013 and February 2014. Each plot was sampled on three non-consecutive nights, totaling 7056 net-hours (1 net-hour = one net open for one hour). Nets remained open between 1800 and 0000 h, and were checked every 15 minutes. Each captured bat was identified and allocated to a foraging guild: gleaning animalivores, frugivores, aerial insectivores, nectarivores, and sanguivores. We extracted the terrain elevation data for each plot from Shuttle Radar Topographic Mission (SRTM) in 90 m resolution rasters images provided by The Global Land Cover Facility (http://www.landcover.org). The geographical coordinates used to obtain terrain elevations were measured at the midpoint of each plot, and we used the ‘Point sampling tool’ from 2.2.0 Valmiera Quantum Geographic Information System software (QGIS) to extract the elevation values. We actively search for plants with mature fruits in an area of 600 m2 located along each plot. We visited each plot three times during daylight hours on the days preceding the nocturnal netting sessions in order to find plants of the given genera with available fruits on the same night as bat captures. Plants with fruits were identified to the genus level and fruit availability was quantified as the number of plants with fruits per plot. We quantified insect availability as the mass of insects collected by light traps, which consisted of 20 cm diameter cones inserted in plastic pots of 100 ml containing a 70% alcohol solution and detergent. A flashlight with 10 white LED bulbs pointed at the cone was set above the cone to attract insects, which were retained in the plastic pot. We installed two light traps in each plot, positioned at 65 and 130 m distant from the closest mist-net and along the central line of the plot and placed at a height of 1.5 m from the ground. The distances between the nets and the insect traps minimized the capture of insectivorous bats attracted to the light traps. We turned on the light traps at 1800 h on bat-capture nights and turned them off after 48 hours. Light traps functioned during the day and night, but probably attracted insects only at night. Each plot was sampled during two days three times, totaling 28224 trap-hours (1 trap-hour = one trap lit for one hour). Collected insects were dried on filter paper to remove excess alcohol and weighed individually on a precision balance (limit of reading 0.0001 g; Ohaus Discovery, Pine Brook, New Jersey). To estimate the total mass of insects collected in each plot, we combined the six insect-sampling nights of each plot. We identified the insects to the level of order.