Current Zoology, 2021, 67(2), 157–163 doi: 10.1093/cz/zoaa038 Advance Access Publication Date: 15 July 2020 Article Article Measuring disturbance at swift breeding colonies due to the visual aspects of a drone: a quasi-experiment study Geison P. MESQUITAa,b,*, José D. RODRÍGUEZ-TEIJEIROb,c, Serge A. WICHd,e, d and Margarita MULERO-PAZM   ANY a Department of Animal Biology, Plant Biology and Ecology, Faculty of Bioscience, Autonomous University of Barcelona, Barcelona 08193, Spain, bInstitut de Recerca de la Biodiversitat, University of Barcelona, Barcelona 08193, Spain, cDepartment of Evolutionary Biology, Ecology and Environmental Sciences, Faculty of Biology, Biodiversity Research Institute (IRBio), University of Barcelona, Barcelona 08193, Spain, dSchool of Biological and Environmental Sciences, Liverpool John Moores University, Liverpool L3 5UG, UK, and eInstitute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Amsterdam 1012 WX, The Netherlands *Address correspondence to Geison P. Mesquita. E-mail: geison.pires@e-campus.uab.cat. Handling editor: Zhi-Yun Jia Received on 9 April 2020; accepted on 8 July 2020 Abstract There is a growing body of research indicating that drones can disturb animals. However, it is usu- ally unclear whether the disturbance is due to visual or auditory cues. Here, we examined the effect of drone flights on the behavior of great dusky swifts Cypseloides senex and white-collared swifts Streptoprocne zonaris in 2 breeding sites where drone noise was obscured by environmental noise from waterfalls and any disturbance must be largely visual. We performed 12 experimental flights with a multirotor drone at different vertical, horizontal, and diagonal distances from the colonies. From all flights, 17% caused <1% of birds to temporarily abandon the breeding site, 50% caused half to abandon, and 33% caused more than half to abandon. We found that the diagonal distance explained 98.9% of the variability of the disturbance percentage and while at distances >50 m the disturbance percentage does not exceed 20%, at <40 m the disturbance percentage increase to > 60%. We recommend that flights with a multirotor drone during the breeding period should be con- ducted at a distance of >50 m and that recreational flights should be discouraged or conducted at larger distances (e.g. 100 m) in nesting birds areas such as waterfalls, canyons, and caves. Key words: Cypseloides senex, disturbance, drones, multirotors, Streptoprocne zonaris, unmanned aircraft systems Multirotor drones are one of the most widely used drone platforms affordability of commercial models, they are currently the most in the civilian environment and with the greatest commercial growth popular choice for recreational flyers (Rebolo-Ifrán et al. 2019), in recent years (Droneii 2019). The main growth factors for scientif- commercial services (Droneii 2019), and scientists (Chabot and Bird ic, commercial, and recreational drone use are associated with a di- 2015; Jiménez López and Mulero-Pázmány 2019). versity of models relatively easy-to-use, vertical take-off/landing, Within the scientific environment, the integration of drones as and easy transport. The high maneuverability of multirotor drones data-collection platforms has significantly facilitated vertebrate stud- and their ability to hover in the air make them the preferred ies, mainly focused on birds and mammals (Wich and Koh 2018) to option for filming and data collection in hard-to-access places (Bakó address a wide variety of topics, such as species monitoring (Rey et al. 2014; Chabot et al. 2015). For these reasons, along with the et al. 2017; Hodgson et al. 2018); behavioral analysis (Canal et al. C The Author(s) (2020). Published by Oxford University Press on behalf of Editorial Office, Current Zoology. V 157 This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com 158 Current Zoology, 2021, Vol. 67, No. 2 2016; Mulero-Pázmány et al. 2017; Cliff et al. 2018); management The Cachoeira do Prata is formed by a set of falls that reach up to (Mulero-Pázmány et al. 2014); habitat mapping (Castellanos-Galindo 18 m in height, and the Cachoeira de S~ ao Rom~ao has falls of up to et al. 2019); and spatial ecology and wildlife diseases (Barasona et al. 25 m in height (Figure 1). The region has a humid tropical climate 2014; Mulero-Pázmány et al. 2015; Laguna et al. 2018). Some of the characterized by 2 well-defined seasons: dry, which runs from May main advantages of using drones to study wildlife are the reduction of to October and wet from November to April, with an annual tem- logistical difficulties; costs; risks; and disturbance on wildlife when perature varying between 24 C and 26 C and an annual rainfall compared with conventional methods such as manned aircraft sur- varying between 1,200 and 1,600 mm (IMESC 2008). The water- veys or researchers on the ground (Dulava et al. 2015; Christie et al. falls are accessible to tourists but the number of visitors is low be- 2016). cause the access is currently limited to 50 km of dirt road that can The increase in drone use has raised concerns about the potential only be accessed by 4  4 vehicles. disturbance these systems can cause on wildlife (Bevan et al. 2018; The 2 study species were the great dusky and white-collared Bennitt et al. 2019; Weston et al. 2020). There are a number of fac- swifts. These are globally considered of least concern according to tors associated with drone characteristics (drone size, motor type, the Red List (IUCN 2020) with stable population for the great dusky and flight pattern) and animals (species, life-history stage, and level swift and declining population for the white-collared swift popula- of aggregation) that can be related to the level of disturbance caused tion. The great dusky swift distribution is restricted to Argentina, by these systems (Mulero-Pázmány et al. 2017). The threshold of Bolivia, Brazil, and Paraguay (Stopiglia and Raposo 2007) and the disturbance caused by a drone in a given species is often formed by a white-collared swift is distributed from the United States to set of interconnected factors: the sound signature of the drone, the Argentina (Chantler 1999). In Brazil, data for both species are environmental noise level, the visual ability of the species, and the sparse, leading to an inaccurate distribution map. Both species are association degree of the drone with a threatening stimulus of the strongly associated to areas with wet rocks walls next or behind species (Bevan et al. 2018). Birds have acute visual perception, and waterfalls, canyons, and caves. These sites are used with great fidel- therefore the visual stimuli generated by the drone can have a ity for breeding and nesting that occurs between October and greater effect than the noise. Even though some studies that assessed November (Whitacre 1989; Stopiglia and Raposo 2007). The 2 spe- drone disturbance in birds relating flight patterns and distances to cies often share nesting sites (Pearman et al. 2010). In this study, the sound and visual aspects of the drone (McEvoy et al. 2016; most of the individuals identified in the nesting sites were the great Rümmler et al. 2016; Brisson-Curadeau et al. 2017; Reintsma et al. dusky swift and few individuals of the white-collared swift. 2018), so far it has not been possible to analyze separately the dis- turbances caused by the visual stimuli of the sound stimuli coming Drone and experimental flights from the drones. The drone model used was a DJI Mavic Pro quad-copter, black Here, we describe an experiment in which we investigate color, with a diagonal size of 335 mm, 743 g weight, 677 dB responses from 2 species of swifts, great dusky swift Cypseloides (decibel) noise level, maximum flight speed of 65 km/h, and 20 min senex and white-collared swift Streptoprocne zonaris, to drone average flight autonomy, that carried a camera with a 1/2.300 CMOS flights in a scenario where noise is mainly masked by the back- (complementary metal-oxide semiconductor) and sensor with 12.35 ground noise of waterfalls and the visual stimulus the main disturb- effective megapixels. In each of the 2 swift breeding sites, we per- ance factor. We measured the disturbance caused by a multirotor formed 6 experimental flights at varying heights above the ground drone at varying distances from swift colonies located in wet rocks and distances to the breeding rocks walls (Table 1). walls next or behind waterfalls where the environmental noise is All the swift nests were located in the rock wall at 10 6 1 m louder than the drone noise. Our aims were to 1) bring a new per- above the ground in the Cachoeiras do Prata and 15 6 1 m in the spective of visual disturbance analysis caused by multirotor drones Cachoeiras de S~ ao Rom~ ao (Figure 2). Flights were conducted be- disassociated from the drone noise and 2) facilitate establishing tween 15 and 18 h local time. The drone was launched at a min- guidelines that allow minimizing disturbance to bird colonies that imum distance of 100 m from the breeding site. During a pilot study use places such as rocks walls next or behind waterfalls, canyons, conducted a week before the actual experiments, we checked that at and caves around the world as resting and nesting sites, places with this distance the drone did not lead to any noticeable reaction from high probability of drone–bird interaction due to the increased rec- the birds. Between the launch sites and the breeding areas, there was reational drone use and the tourist interest of such places. vegetation that prevented birds from viewing the drone’s take-off. We approached the nesting sites horizontally at a speed between 14 and 21 km/h which in a previous study did not seem to influence Material and Methods bird behavior (Vas et al. 2015) and allows for good control of the Study area and species drone. Once the drone reached the set point, which corresponds to This study was conducted in Chapada das Mesas National Park, the diagonal distance of each flight according to Table 1, it remained Maranh~ao, Brazil, in October, 2018. The park covers a total area of hovering stationary for a maximum time of 10 min or until we 1,600 km2 within the Cerrado biome, that has various vegetation detected any swifts’ behavioral reaction (flying away or mobbing). types, from “cerrad~ao,” which is a type of seasonal forest with dense Once we detected any reaction, we kept the flight time no >5 min to tree vegetation to “campos limpos” that are open fields as savannas minimize negative effects on the species. An experienced observer with few trees (Marques and Amorim 2014). The 2 breeding areas using a binocular (10  50) counted the number of birds that were of the study species were: Cachoeira do Prata (6 590 3600 S, present at the breeding site 5 min before the take-off of each flight 47 90 5500 W) and Cachoeira de S~ ao Rom~ ao (7 10 1100 S, 47 20 2600 W). and after the drone was landed. At both field sites, the observer was Both are located in the North of the park along different stretches of positioned between the nesting rocks walls and the drone, with free the “Farinha” river, a tributary of Araguaia/Tocantins basin, and view to both. Due to the difficulty of approaching the nesting rocks are 14 km away from each other in a straight line. The breeding walls and to avoid possible disturbance to the colony, the observer areas are the 2 most voluminous waterfalls present within the park. was positioned at a horizontal distance of 15–20 m from the base of Mesquita et al.  Drones can disturb animals 159 ~o Figure 1. Location of the studied swift breeding sites in Chapada das Mesas National Park, Brazil. Cachoeira do Prata (white square) and Cachoeira de Sa Rom~ao (white circle). Table 1. Experimental flights parameters Flight Date Time Study site Height nests Flight Vertical Horizontal Diagonal altitude distance distance distance 1 22 October 2018 16:00 Cachoeira do Prata 10 50 40 50 64.03 2 22 October 2018 17:30 Cachoeira do Prata 10 25 15 50 52.20 3 23 October 2018 16:00 Cachoeira de S~ ao Rom~ ao 15 50 35 50 61.03 4 23 October 2018 17:30 Cachoeira de S~ ao Rom~ ao 15 25 10 50 50.99 5 24 October 2018 16:00 Cachoeira do Prata 10 10 0 50 50.00 6 24 October 2018 17:30 Cachoeira do Prata 10 50 40 25 47.17 7 25 October 2018 16:00 Cachoeira de S~ ao Rom~ ao 15 10 5 50 50.25 8 25 October 2018 17:30 Cachoeira de S~ ao Rom~ ao 15 50 35 25 43.01 9 26 October 2018 16:00 Cachoeira do Prata 10 25 15 25 29.15 10 26 October 2018 17:30 Cachoeira de S~ ao Rom~ ao 15 25 10 25 26.93 11 27 October 2018 16:00 Cachoeira do Prata 10.00 10 0 25 25.00 12 27 October 2018 17:30 Cachoeira de S~ ao Rom~ ao 15.00 10.00 5 25 25.50 Note: Distances are in meters. the rocks walls, hidden from the colony’s line of sight. Because of The visual analysis included an assessment of the spots size on the large number of individuals of the 2-species agglomerated and the walls, which were agglomerations of the birds, and were used to the low luminosity at the waterfalls, we could not determine the define whether the birds had regrouped. This is, if the spot size number of individuals of each of the 2 species at the breeding sites returned to its original size, we assumed that the individuals had and therefore recorded the total number of birds. We established a returned. For the visual analysis of spot sizes, we compare the spot minimum interval of 30 min after landing of each flight or until the sizes with rock wall features as atypical marks, deformations, or birds regrouped in the breeding sites, and a maximum of 2 daily some plants. Due to the high environmental noise caused by the flights, to avoid major disturbances during the same day. waterfalls, in all the experimental flights in the 2 studied places it 160 Current Zoology, 2021, Vol. 67, No. 2 Figure 2. Design of experimental flights. Breeding group from “Cachoeira do Prata” and “Cachoeira de Sa ~o Roma~o.” Classification (circle, noticeable disturb- ance; triangle, moderate disturbance; and square, high disturbance), Diagonal distance (meters) and disturbance (%) for each drone flight. was not possible to hear the drone noise by the observer who was (Greenwell and Schubert 2014) packages in R 3.6.2 with RStudio positioned between the drone and the rock walls at a horizontal dis- 1.2.5033 (R Core Team 2019). tance of 15–20 m from the base of the rocks walls. Ethics Statement This project was the authorized No. 64630-1 (scientific purpose) by Statistical analysis the System of Authorization and Information on Biodiversity As drone disturbance we considered the change in swifts’ behavior (SISBIO) in Brazil (art. 28 of IN 03/2014) from the Chico Mendes (flying away or mobbing). We calculated this disturbance for each Institute for Biodiversity Conservation (ICMBio), and the flight experimental flight as the percentage of birds present in the breeding drone was register certificate No. PP-019272726 by the National colony 5 min before drone exposure minus the percentage of birds Civil Aviation Agency (ANAC). present after drone landing. Following Chabot et al. (2015), we clas- sified the drone disturbance level in 3 categories based on the per- centage of birds reacting: 1) noticeable disturbance, when the Results percentage does not exceed 1%; 2) moderate disturbance, when the Twelve drone flights were performed at different distances from 2 percentage does not exceed 50%; and 3) high disturbance, when the swift breeding colonies. A maximum disturbance of 93.3% was percentage is >50%. For vertical distance, we considered the differ- recorded when the drone flew at 25.5 m distance from a bird’s col- ence in height between the nest and the drone on each flight. The ony, and a minimum of 0.7% disturbance when the flight was con- horizontal was measured from the projection of the drone to the ducted at 64.0 m distance (Table 2). During the 6 flights that ground to the colony and the diagonal distance (hereafter distance) produced moderate disturbance initially, a few swifts, ranging from was obtained through the Pythagorean theorem. We also calculated 5 to 40 individuals, showed a mobbing behavior against the drone. the return time of the individuals to the breeding sites after the drone However, the majority of other individuals who showed reactions had landed on each flight and the average time for each of the 3 cat- just left the breeding sites and began to perform circular flights at a egories of disturbance. distance 20 6 5 m above the drone. Flights performed at less than A previous descriptive scatter plot showed the possibility of a 29 m produced high disturbance, causing the departure of most of nonlinear association between variables in the 2 ran models. The the colony of the breeding sites with just an average of 15.8% of the first model with diagonal distance as a predictor variable and the individuals remaining. In flights with high disturbance, we also disturbance percentage as a dependent variable, and the second recorded a larger number of individuals performing mobbing behav- model with the disturbance percentage as a predictor variable and ior toward the drone. In each of these flights, we landed as fast as the return time as a dependent variable. To choose the best models, possible. we initially consider the nature of the variables and Akaike’s infor- The nonlinear Gompertz model is the one that presents a lower mation criterion (AIC). For model validation, we tested for normal- AIC, 80.16, and the distance from the drone to the colony explained ity test (Shapiro–Wilk), heteroscedasticity (Breusch–Pagan) and set 98.9% of the variability of disturbance percentage. Thus, while at the significance level at 0.05. All analyses and charts were made distances >50 m the percentage of disturbances does not exceed using “car” (Fox 2016), “drc” (Ritz et al. 2015), and “investr” 20%, at <40 m the disturbance percentage increase to >60% Mesquita et al.  Drones can disturb animals 161 Table 2. Percentage disturbed and classification of experimental flights Classification Flight Date Time Study site Diagonal Total Disturbed Return distance (m) swifts (%) time (min) 1 1 22 October 2018 16:00 Cachoeira do Prata 64.03 3,000 0.7 1 1 3 23 October 2018 16:00 Cachoeira de S~ ao Rom~ ao 61.03 1,000 1.0 1 2 2 22 October 2018 17:30 Cachoeira do Prata 52.20 1,000 5.0 9 2 4 23 October 2018 17:30 Cachoeira de S~ ao Rom~ ao 50.99 3,000 10.0 9 2 5 24 October 2018 16:00 Cachoeira do Prata 50.00 1,000 15.0 12 2 6 24 October 2018 17:30 Cachoeira do Prata 47.17 2,500 32.0 15 2 7 25 October 2018 16:00 Cachoeira de S~ ao Rom~ ao 50.25 2,500 20.0 12 2 8 25 October 2018 17:30 Cachoeira de S~ ao Rom~ ao 43.01 1,500 46.7 16 3 9 26 October 2018 16:00 Cachoeira do Prata 29.15 1,000 70.0 20 3 10 26 October 2018 17:30 Cachoeira de S~ ao Rom~ ao 26.93 3,000 83.3 22 3 11 27 October 2018 16:00 Cachoeira do Prata 25.00 1,000 90.0 25 3 12 27 October 2018 17:30 Cachoeira de S~ ao Rom~ ao 25.50 3,000 93.3 25 Note: 1, noticeable disturbance; 2, moderate disturbance; and 3, high disturbance. disturbance the individuals returned almost immediately after the drone landing. Discussion In this study, we measured the drone visual disturbance separate from the drone noise disturbance in birds breeding colonies from a quasi-experiment where the drone’s noise is masked by environment noise, and we found that the response of birds to drone use follows a sigmoidal distribution with the diagonal distance from the drone to the colonies. Although our results are similar to studies that indi- cate that drone disturbance on birds increases as flight height decreases under different conditions and with different bird species Figure 3. Nonlinear Gompertz regression between diagonal distance and % (Rümmler et al. 2016; Mulero-Pázmány et al. 2017; van der Vliet disturbed of swifts. Blue, 95% confidence band; green, prediction band. et al. 2019), we found that the recommended minimum distance must be >50 m to avoid moderate and high disturbance in breeding sites, which is different from other studies, that were 15 m by com- mon gulls and other species in the bird reserve island Langenwerder in the Baltic Sea (Grenzdörffer 2013) and at least 20 m with drones to survey cliff-nesting seabirds as murres (Brisson-Curadeau et al. 2017). However, unlike all the studies mentioned above, our results show that this reaction to the drone at a greater distance from the colony could be due to the idiosyncrasy of these species but it could also be a consequence of the fact that the drone, without any appar- ent sound is more similar to a natural situation of approach of a winged predator to the colony and trigger the defensive reaction ear- lier. The drone’s sound could initially prevent the colony’s reaction by being an artificial stimulus not associated with a winged preda- tor, and only when the drone is close enough then triggers this de- Figure 4. Nonlinear power regression between % disturbed of swifts and re- fensive reaction. turn time. Blue, 95% confidence band; green, prediction band. The median bird hearing thresholds from 49 bird species suggest that the birds hear best at frequency between about 2 and 3 kHz, while humans generally have better auditory sensitivity with lower (Figure 3). The relationship between the disturbance percentage and auditory thresholds and with wider bandwidth than typical birds the return time, that is, the time it takes for the swifts to return to (Dooling and Popper 2007). Therefore, if an observer was unable to the colonies is better fitted to a nonlinear power model that explains hear the drone at 15 m, suppressed or muffled by waterfalls in this 97.3% of the variability of return time, and it was the one that experiment, it is assumed that the swifts could not hear the drone at presents a lower AIC, 54.5 (Figure 4). On the 4 flights classified as 25 m in the flight closest to the colony. This suggests that the drone high disturbance it took an average of 23.5 6 2.4 min for all individ- noise may lose importance for the disturbance, while the visual uals in the colony to return to the breeding sites after the drone had aspects such as the shape or the flight pattern can be determinant for landed. On flights classified as moderate disturbance this time was the swift’s behavior change. Indeed, the drone visual stimulation reduced to 12 6 2.9 min, whereas on flights with just noticeable was one of the possible causes of disturbances in colonies of greater 162 Current Zoology, 2021, Vol. 67, No. 2 crested tern Thalasseus bergii in a study that suggested that the noise regulations for the recreational use of drones in areas where these emitted by multirotor drones may not be audible to colonies of this species occur. species (Bevan et al. 2018). However, the drone shape of our study eschews the classic “hawk/goose” rule (Schleidt et al. 2011) because a multirotor does not look like any potential swift predator. The Acknowledgments new multirotor shape was one of the explanations for the lack of The authors are grateful to Neotropical Grassland Conservancy for donating flight response in waterfowl at low flight altitudes in other studies the drone used in this study; Fundaç~ ao de Amparo a Pesquisa e (McEvoy et al. 2016). In contrast, we found that swifts showed Desenvolvimento Cientı́fico (FAPEMA) for funding part of the study; and the mobbing behavior in flights near the nesting sites and may have rec- Chico Mendes Institute for Biodiversity Conservation (ICMBio) for providing ognized the multirotor drone as a potential predator. In the case of with permission and access to the park, in particular the chief of the park Mr. Deijacy Rego for the availability and assistance in the field. This study has the great dusky swift and white-collared swift, the only known aer- been carried out within the framework of biodiversity Ph.D. program at the ial predator is the peregrine falcon Falco peregrinus which has been Autonomous University of Barcelona. observed near the others colony sites awaiting to catch swifts as they enter or leave the colony to feed and collect nest materials (Whitacre 1989). So even though the multirotor does not have a “hawk” Conflict of Interest statement shape, it is possible that the mobbing behavior of the swifts facing The authors declare that they have no competing interests. the drone can be elicited due to the drone being perceived as an un- known potential predator. The time that swifts took to return to the colony after multirotor References flights classified of high disturbance was about 2 times longer than Bakó G, Tolnai M, Takács Á, 2014. Introduction and testing of a monitoring those classified of moderate disturbance and about 20 times longer and colony-mapping method for waterbird populations that uses high-speed than those of low disturbance. This time between departure and re- and ultra-detailed aerial remote sensing. Sensors 14:12828–12846. turn to the original location after the disturbance is also considered a Barasona JA, Mulero-Pázmány M, Acevedo P, Negro JJ, Vicente J et al., 2014. way to measure an animal’s response to a disturbance (van der Vliet Unmanned aircraft systems for studying spatial abundance of ungulates: et al. 2019). These types of responses can have a negative impact on relevance to spatial epidemiology. PLoS ONE 12:1–17. the reproductive process in the case of birds in their breeding season, Bennitt E, Bartlam-Brooks HLA, Hubel TY, Wilson AM, 2019. Terrestrial since it causes the individual to spend more energy, alters the incuba- mammalian wildlife responses to unmanned aerial systems approaches. Sci tion cycle and the care of altricial nestlings, and exposes them to pos- Rep 9:2142. sible predators. This negative impact caused by the return time to the Bevan E, Whiting S, Tucker T, Guinea M, Raith A et al., 2018. Measuring be- havioral responses of sea turtles, saltwater crocodiles, and crested terns to nests was different from others bird studies that measured this time drone disturbance to define ethical operating thresholds. PloS ONE 13: after drone disturbance in breeding colonies: ranging from 1 min to e0194460. common terns Sterna hirundo (Reintsma et al. 2018), 1–3 min for Brisson-Curadeau É, Bird D, Burke C, Fifield DA, Pace P et al., 2017. Seabird Iceland gulls Larus glaucoides, and 5–10 min for thick-billed murres species vary in behavioural response to drone census. Sci Rep 7:17884. Uria lomvia (Brisson-Curadeau et al. 2017), while our experiment Canal D, Mulero-Pázmány M, Negro JJ, Sergio F, 2016. Decoration increases demonstrated much longer return time, whether on high disturbance the conspicuousness of raptor nests. PLoS ONE 11:e0157440. flights, ranging from 20 to 25 min, or moderate disturbance flights, 9 Castellanos-Galindo GA, Casella E, Mejı́a-Renterı́a JC, Rovere A, 2019. to 16 min. This variability in return time suggests the need to carry Habitat mapping of remote coasts: evaluating the usefulness of lightweight out specific tests to know this effect in different species. Our experi- unmanned aerial vehicles for conservation and monitoring. Biol Conserv ment shows that this delay time in returning to the nesting site can 239:108282. cause very negative impacts on the reproductive process if the pres- Chabot D, Craik SR, Bird DM, 2015. Population census of a large common ence of these drones is intense over time. tern colony with a small unmanned aircraft. PLoS ONE 10:e0122588. Chabot D, Bird DM, 2015. Wildlife research and management methods in the Understanding the minimum operating distance at which drones 21st century: where do unmanned aircraft fit in? J Unmanned Veh Syst 3: can cause disturbance, which factors can cause them, and for which 137–155. species each distance can be tolerated is critical, whether for the Chantler P, 1999. Family Apodidae (swifts). In: del Hoyo J, Elliot A, Sargatal preparation of flight missions in scientific studies or to regulate the J, editors. Handbook of the Birds of the World: Barn-Owls to growing recreational use of drones in such environments. Despite Hummingbirds. Barcelona: Lynx Edicions, 387–466. the great diversity of responses to the drone use from different bird Christie KS, Gilbert SL, Brown CL, Hatfield M, Hanson L, 2016. Unmanned species due to the different types of ecological contexts in which aircraft systems in wildlife research: current and future applications of a they are found, almost always the greater the frequency and inten- transformative technology. Front Ecol Environ 14: 241–251. sity of the disturbance, the greater the negative impacts on breeding Cliff OM, Saunders DL, Fitch R, 2018. Robotic ecology: tracking small dy- bird populations. In this sense, the drone use, which is expanding in namic animals with an autonomous aerial vehicle. Sci Robot 3:eaat8409. sites as bird nesting areas, such as this study, should be considered Dooling RJ, Popper AN, 2007. The Effects of Highway Noise on Birds. as a possible source of negative effects in certain colony birds. California: California Department of Transportation Division of Environmental Analysis. Therefore, we suggest the flight distance with multirotor drone to Droneii, 2019. Report: Drone Manufacturer Ranking 2019. Available from: avoid high disturbance in the great dusky and white-collared swifts https://www.droneii.com/project/drone-manufacturer-ranking-2019 during the breeding period in nesting areas should be >50 m. We (accessed 13 December 2019). also recommend that recreational flights are generally discouraged Dulava S, Bean WT, Richmond OMW, 2015. Environmental reviews and case or conducted at larger distances (e.g., 100 m) in areas where swifts studies: applications of unmanned aircraft systems (UAS) for waterbird sur- occur such as waterfalls, canyons, and caves. This study serves as a veys. Environ Pract 17:201–210. basis both for the elaboration of new protocols for the use of drones Fox J, 2016. Applied Regression Analysis and Generalized Linear Models. 3rd over birds by researchers in conservation studies and for possible edn. California: SAGE Publication. Mesquita et al.  Drones can disturb animals 163 Greenwell BM, Schubert KCM, 2014. investr: an r package for inverse estima- R Core Team, 2019. R: A Language and Environment for Statistical tion. R J 6:90–100. Computing. Vienna, Austria: R Foundation for Statistical Grenzdörffer GJ, 2013. UAS-based automatic bird count of a common gull Computing. Available from: https://www.R-project.org/ (accessed 22 colony. ISPRS - International Archives of the Photogrammetry, Remote December 2019). Sensing and Spatial Information Sciences. Germany: Rostock. Vol. Rebolo-Ifrán N, Grilli MG, Lambertucci SA, 2019. Drones as a threat to wild- XL-1/W2, 169–174. life: youtube complements science in providing evidence about their effect. Hodgson JC, Mott R, Baylis SM, Pham TT, Wotherspoon S et al., 2018. Environ Conserv 46:205–210. Drones count wildlife more accurately and precisely than humans. Methods Reintsma KM, McGowan PC, Callahan C, Collier T, Gray D et al., 2018. Ecol Evol 9:1160–1167. Preliminary evaluation of behavioral response of nesting waterbirds to small IMESC, 2008. Perfil do Maranh~ ao 2006/2007. Available from: http://imesc. unmanned aircraft flight. Waterbirds 41:326–331. ma.gov.br/portal/Post/view/outras-publicacoes/38 (accessed 8 January Rey N, Volpi M, Joost S, Tuia D, 2017. Detecting animals in African Savanna 2019). with UAVs and the crowds. Remote Sens Environ 200:341–351. IUCN, 2020. The IUCN Red List of Threatened Species. Version 2020-1. Ritz C, Baty F, Streibig JC, Gerhard D, 2015. Dose-response analysis using R. Available from: https://www.iucnredlist.org (accessed 15 April 2020). PLoS ONE 10:e0146021. doi: 10.1371/journal.pone.0146021. Jiménez López J, Mulero-Pázmány M, 2019. Drones for conservation in pro- Rümmler M-C, Mustafa O, Maercker J, Peter H-U, Esefeld J, 2016. tected areas: present and future. Drones 3:1–23. Measuring the influence of unmanned aerial vehicles on Adélie penguins. Laguna E, Barasona JA, Triguero-Oca~ na R, Mulero-Pázmány M, Negro JJ et Polar Biol 39:1329–1334. al., 2018. The relevance of host overcrowding in wildlife epidemiology: a Schleidt W, Shalter MD, Moura-Neto H, 2011. The hawk/goose story: the new spatially explicit aggregation index. Ecol Indic 84:695–700. classical ethological experiments of Lorenz and Tinbergen, revisited. J Marques AR, Amorim M, 2014. Saberes geográficos integrados aos estudos Comp Psychol 125:121–133. territoriais sob a ótica da implantaç~ ao do Parque Nacional da Chapada das Stopiglia R, Raposo MA, 2007. Distribuiç~ ao e biologia do andor- Mesas, Sert~ao de Carolina/MA. Geogr Quest 7:100–117. inh~ao-preto-da-cascata Cypseloides fumigatus e do andor- McEvoy JF, Hall GP, Mcdonald PG, 2016. Evaluation of unmanned aerial ve- inh~ao-velho-da-cascata C. senex no Brasil: uma sı́ntese. Cotinga 28:49–57. hicle shape, flight path and camera type for waterfowl surveys: disturbance Vas E, Lescroel A, Duriez O, Boguszewski G, Grémillet D, 2015. Approaching effects and species recognition. PeerJ 4:e1831. birds with drones: first experiments and ethical guidelines. Biol Lett 11: Mulero-Pázmány M, Stolper R, Essen LD, Negro JJ, Sassen T, 2014. Remotely 20140754. piloted aircraft systems as a rhinoceros anti-poaching tool in Africa. PLoS van der Vliet R, Jeninga L, van den Burg A, 2019. RPAS over Natura 2000 ONE 9:1–10. areas: disturbance responses of wildlife and opportunities for research. Mulero-Pázmány M, Barasona JÁ, Acevedo P, Vicente J, Negro JJ, 2015. RPAS Civil Operators & Operations Forum 8th Annual International Unmanned aircraft systems complement biologging in spatial ecology stud- Conference, December. Netherlands: Culemborg. ies. Ecol Evol 5:4808–4818. Weston MA, O’Brien C, Kostoglou K, Symonds MRE, 2020. Escape responses Mulero-Pázmány M, Jenni-Eiermann S, Strebel N, Sattler T, Negro JJ et al., of terrestrial and aquatic birds to drones: towards a code of practice to min- 2017. Unmanned aircraft systems as a new source of disturbance for wild- imise disturbance. J Appl Ecol 57:777–785. life: a systematic review. PLoS ONE 12:e0178448. Whitacre DF, 1989. Conditional use of nest structures by white-naped and Pearman M, Areta JI, Roesler I, Bodrati A, 2010. Confirmation of the sooty white-collared swift. Condor 91:813–825. swift Cypseloides fumigatus in Argentina with notes on its nest placement, Wich SA, Koh LP, 2018. Conservation Drones: Mapping and Monitoring seasonality, and distribution. Ornitol Neotrop 21:351–359. Biodiversity. Oxford: Oxford University Press.