Surveying mixed-species waterbird colonies with unmanned aerial systems (UAS): Visibility bias, disturbance, and protocol recommendations

Date

2017-12

Authors

Barr, Jarred

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Surveys of colonial-nesting waterbirds are necessary for assessing population trends, gaining insight into wetland ecosystem health, and even determining the impact of natural disasters and other environmental concerns. The popularity of unmanned aerial systems (UAS) for use as a surveying tool has risen in the past decade, but little research has been conducted on the effectiveness of such technology. I investigated visibility bias and disturbance impacts associated with using UAS to survey waterbird colonies in Texas, specifically in cypress-tupelo watershed and coastal island habitats. I used a stratified random design to place four waterbird decoy types (black skimmers, terns, and white- and dark-plumaged herons) in each habitat and had six observers independently count decoys from aerial imagery taken with a consumer-grade UAS (DJI Phantom). I used generalized linear mixed-effects models to estimate detection probabilities of each decoy type. Black skimmers were the only decoy type at the dredge-spoil island to have a detection probability of significantly less than 100% (0.54 [0.44-0.63 CI], P ≤ 0.001). Detectability of both white- and dark-plumaged herons decreased considerably in the canopied cypress-tupelo habitat when compared to the dredge-spoil island (by 80 and 84% respectively). In addition, for surveys in cypress-tupelo habitat where cloud cover was above 50%, detectability of white heron decoys decreased significantly by another 20% (0.09 [0.03-0.34 CI], P = 0.007). Detection rates varied among observers, but only significantly for models of white-plumaged herons (P = 0.022) and black skimmers (P = 0.05). Use of the double-sample method yielded biased-low abundance estimates for white- and dark-plumaged herons in canopied sites, suggesting that habitat differences were a greater source of bias than observer error. I investigated disturbance to waterbirds by setting up video cameras at the periphery of active nesting colonies while surveying with unmanned aircraft. I tested the effects of two UAS platforms - and a range of altitudes flown between them - on the behavioral reactions exhibited in four active colonies in Texas. Reactions were tallied in 1-minute sampling periods at each nesting colony, which were used to estimate generalized linear mixed-effects models for vigilance and flush behavior. I found that the consumer-grade UAS (DJI Phantom) increased vigilance in mixed-species colonies for survey altitudes of 91, 61, and 46 m when compared to a baseline control. Vigilant reactions were increased in magnitude by 72, 119, and 118% for these altitudes, respectively. Flush reactions were not influenced by either platform or any altitude flown. Surveys with the fixed-wing UAS did not impact vigilance or flush behavior, likely because it was used at suggested altitudes of 300 and 200 m and was hardly detectable from the ground. My results suggest that managers should employ UAS surveys on clear days in high-visibility habitats, or otherwise use another survey method to supplement photographic counts obtained by UAS. In addition, surveys should be flown between 46-91 m only when high resolution imagery is needed (e.g. for abundance estimates) to mitigate disturbance. Even though the Phantom UAS caused increased vigilance, if surveys are done promptly and in back-and-forth transects, the impact of this increased behavior is likely negligible especially when considering the much more harmful effects of ground-based survey methods.

Description

Keywords

Waterbirds, Drones, UAS, Aerial, Survey, Disturbance, Detection

Citation

Barr, J. (2017). <i>Surveying mixed-species waterbird colonies with unmanned aerial systems (UAS): Visibility bias, disturbance, and protocol recommendations</i> (Unpublished thesis). Texas State University, San Marcos, Texas.

Rights

Rights Holder

Rights License

Rights URI