Optimally designing drone‐based surveys for wildlife abundance estimation with N‐mixture models

Abstract

1. Hierarchical N-mixture models have been suggested for abundance estimation from spatiotemporally replicated drone-based count surveys, since they allow modeling abundance of unmarked individuals while accounting for detection errors. However, it is still necessary to understand how these models perform in the wide variety of contexts and species in which drone surveys are being used. This knowledge is fundamental to plan study designs with optimal allocation of scarce resources in ecology and conservation.2. We conduct a simulation study to address N-mixture model (binomial and multinomial) performance and optimal survey effort allocation in different scenarios of local abundance and detectability of individuals, focusing on their application for drone-based surveys. We also investigate the benefits of using a double-observer protocol (either human or algorithm) in image review to decompose the detection process in availability and perception. Finally, we illustrate our simulation-based survey design considerations by applying them to abundance estimation of marsh deer in the Pantanal wetland (Brazil). Accuracy of abundance estimation with N-mixture models increases with localabundance in sites and especially with the availability of individuals. The optimal design requires more visits at fewer sites when the availability probability is lower, and the optimal design is more flexible as local abundance increases.Two observers checking images can increase the estimator performance even at very high perception probabilities. We quantified how much the use of a doubleobserver protocol in image review can reduce fieldwork effort while achieving the same accuracy. 4. N-mixture models can deliver accurate abundance estimates from spatiotemporally replicated drone surveys in a wide variety of contexts while accounting for imperfect detection. The improvements achieved by a consciously planned design, rearranging survey efforts among sites and visits, as well as using a second observer in image review, can be crucial to detect trends when monitoring a population or to categorize a species as threatened or not.

On the move: Influence of animal movements on count error during drone surveys

“…While our assumptions of 100% availability and detectability are highly unlikely in real-world applications (Gilbert et al , 2021), for example, due to visual obstructions above the animals or the ability of the animal to dive underwater or move under cover (Hodgson, Peel and Kelly, 2017;Brunton, Leon and Burnett, 2020), this assumption allowed us to simplify our scenarios and better understand how flight patterns and animal movements may create counting errors. Typically, surveyors are concerned with omission rates associated with conventional animal survey methods (i.e., occupied aircraft and ground surveys) due to detectability issues, and there are means of addressing some of these problems (Steinhorst and Samuel, 1989;Samuel et al , 1992;Hamilton et al , 2018;Brack, Kindel, de Oliveira, et al , 2023). For example, the inclusion of detection probabilities in statistical models has greatly improved our ability to estimate animal populations (Martin et al , 2012;Corcoran, Denman and Hamilton, 2021), and incorporating detection probabilities into drone-based estimates would be a helpful advancement (Hodgson, Peel and Kelly, 2017;Brack, Kindel, de Oliveira, et al , 2023;Hodgson, Kelly and Peel, 2023).…”

“…Typically, surveyors are concerned with omission rates associated with conventional animal survey methods (i.e., occupied aircraft and ground surveys) due to detectability issues, and there are means of addressing some of these problems (Steinhorst and Samuel, 1989;Samuel et al , 1992;Hamilton et al , 2018;Brack, Kindel, de Oliveira, et al , 2023). For example, the inclusion of detection probabilities in statistical models has greatly improved our ability to estimate animal populations (Martin et al , 2012;Corcoran, Denman and Hamilton, 2021), and incorporating detection probabilities into drone-based estimates would be a helpful advancement (Hodgson, Peel and Kelly, 2017;Brack, Kindel, de Oliveira, et al , 2023;Hodgson, Kelly and Peel, 2023). It is also notable that false positives (i.e., multiple counts) are less frequent during ground-based and occupied aircraft surveys, something that researchers using drones need to carefully consider moving forward (Brack, Kindel and Oliveira, 2018).…”

Utility of Spectral Filtering to Improve the Reliability of Marine Fauna Detections from Drone-Based Monitoring

“…Drone-based aerial surveys in the marine environment are perceived as being a relatively efficient and reliable method for detecting and identifying coastal fauna, and have been used to assess animal behaviour [11], abundance [12][13][14], population health [15,16], as well minimising the potential for human-wildlife conflict, such as human-shark interactions along coastal beaches [17,18]. Despite the utility, sightability errors that affect reliability of the detections and identifications of marine life, can be apparent and similar to that reported from aerial surveys using crewed aircraft [4,8,17]. This can be particularly problematic in marine fauna surveys, where the detection reliability is governed by factors including water clarity, depth, sea state, sun glare, and sea-surface reflection, as well as animal size, behaviour, and its position in the water column [17,19].…”

“…This is particularly the case with the relatively recent appearance and development of aerial drones, also referred to as 'UAV', 'UAS', 'RPAS' (see Chabot et al [5]). Drones are now a common tool in ecology [6][7][8]. Furthermore, with the continued advancement of drone technology, as well as associated digital capture technology, it is anticipated that the effective spatial scales that can be efficiently sampled using drone-based methods will expand [9].…”

“…Perception biases occur when an animal in the water column should have been detected (i.e., it was positioned in the 'available' upper section of the water column), but it was not detected due to an error in human spotting or machine-learning, rather than external factors [1,4,17]. In many cases, sampling effort can be constrained to favourable locations and conditions, and methods can be employed (particularly in post-processing) to estimate the errors and biases to adjust the detection data for inferring abundance [8,19,20]. Minimising the uncertainty in count data has been, and is, a consistent objective across ecology.…”