ABSTRACT: Background:Drones are reliable tools for estimating colonial seabird numbers. Although most research has focused on methods of improving the accuracy of bird counts, few studies have evaluated the impacts of these methods on bird behavior. In this study, we examined the effects of the DJI Phantom 3 drone approach (altitude, horizontal and vertical descent speeds) on changes in the intensity of behavioral response of guano birds: guanay cormorants (Phalacrocorax bougainvilli), Peruvian boobies (Sula variegata) and Peruvian pelicans (Pelecanus thagus). The breeding and non-breeding condition was also evaluated. Methods:Eleven locations along the Peruvian coast were visited in 2016-2017. Drone flight tests considered an altitude range from 5 to 80 m from the colony level, a horizontal speed range from 0.5 to 15 m/s, and a vertical descent speed range from 0.5 to 3 m/s. The intensity of the behavioral response of birds was scored and categorized as: 0-no reacting, 1-head pointing to the drone (HP), 2-wing flapping (WF), 3-walking/running (WR) and 4-taking-off/flying (TK). Drone noise at specific altitudes was recorded with a sound meter close to the colony to discriminate visual from auditory effects of the drone. Results:In 74% of all test flights (N = 507), guano birds did not react to the presence of the drone, whereas in the remaining flights, birds showed a sign of discomfort: HP (47.7%, N = 130), WF (18.5%), WR (16.9%) and TK (16.9%). For the drone approach tests, only flight altitude had a significant effect in the intensity of the behavioral response of guano birds (intensity behavioral response <2). No birds reacted at drone altitudes above 50 m from the colony. Birds, for all species either in breeding or non-breeding condition, reacted more often at altitudes of 5 and 10 m. Chick-rearing cormorants and pelicans were less sensitive than their non-breeding counterparts in the range of 5-30 m of drone altitude, but boobies reacted similarly irrespective of their condition. At 5 m above the colony, cormorants were more sensitive to the drone presence than the other two species. Horizontal and vertical flights at different speeds had negligible effects (intensity behavioral response <1). At 2 m above the ground, the noise of the cormorant colony was in average 71.34 ± 4.05 dB (N = 420). No significant differences were observed in the drone noise at different flight altitudes because the background noise of the colony was as loud as the drone. Conclusions:It is feasible to use the drone DJI Phantom 3 for surveys on the guano islands of Peru. We recommend performing drone flights at altitudes greater than 50 m from guano bird colonies and to select take-off spots far from gulls. Likewise, this study provides a first step to develop guidelines and protocols of drone use for other potential activities on the Peruvian guano islands and headlands such as surveys of other seabirds and pinnipeds, filming and surveillance.