Monitoring large and complex wildlife aggregations with drones

Abstract

Recent advances in drone technology have rapidly led to their use for monitoring and managing wildlife populations but a broad and generalised framework for their application to complex wildlife aggregations is still lacking. We present a generalised semi‐automated approach where machine learning can map targets of interest in drone imagery, supported by predictive modelling for estimating wildlife aggregation populations. We demonstrated this application on four large spatially complex breeding waterbird colonies on floodplains, ranging from c. 20,000 to c. 250,000 birds, providing estimates of bird nests. Our mapping and modelling approach was applicable to all four colonies, without any modification, effectively dealing with variation in nest size, shape, colour and density and considerable background variation (vegetation, water, sand, soil, etc.). Our semi‐automated approach was between three and eight times faster than manually counting nests from imagery at the same level of accuracy. This approach is a significant improvement for monitoring large and complex aggregations of wildlife, offering an innovative solution where ground counts are costly, difficult or not possible. Our framework requires minimal technical ability, is open‐source (Google Earth Engine and R), and easy to apply to other surveys.

Counting Mixed Breeding Aggregations of Animal Species Using Drones: Lessons from Waterbirds on Semi-Automation

“…We collected imagery over the colony (15-20 ha) using a DJI Phantom 3 Professional multi-rotor drone, again with the stock standard camera and an additional neutral density filter (4000 × 3000 image size, lens FOV 94 • 20 mm). We flew at 5-10 ms −1 aiming to acquire imagery with~70% forward and lateral overlap, along parallel flight lines at 90 • [3,25]. We processed the imagery using the commercial software Pix4DMapper (v4.19,166 Pix4D SA), with a photogrammetry technique called 'structure from motion', which identified points in overlapping images, building a three-dimensional (3D) point cloud reconstruction of the landscape, and finally, generating a digital surface model and an orthorectified image mosaic ( Figure 2, Step 1).…”

“…There is an increasing need to estimate aggregations of animals around the world, including turtles, seals and birds [1][2][3][4][5][6]. Regular monitoring of these concentrations allows decision-makers to not only track changes to these colonies but also long-term environmental changes, given that large aggregations of some species can be used to monitor environmental change (e.g., waterbird breeding colonies) [7,8].…”

“…They provide advantages over in-situ observations, accessing physically inaccessible or dangerous areas in a relatively small amount of time [17][18][19]. Drones are also relatively cheap, safe and less disturbing, improving traditional wildlife surveys [3,18,20,21]. They can, however, disturb some animal populations, requiring careful consideration of appropriateness when surveying [22].…”

“…Such benefits allow for real-time monitoring and management decisions and could, for example, assist in the targeted delivery of environmental flows for waterbird breeding events [29].Generally, semi-automated counting methods are most effective for species where there are strong contrasts against the backgrounds, particularly when background colours and shapes are consistent [28]. They can distinguish large single species aggregations on relatively simple backgrounds [30][31][32], up to sixteen avian species (numbering in the hundreds) on simple single colour backgrounds, such as oceans [33,34], or single species aggregations of hundreds of thousands on complex backgrounds [3].Development of flexible, repeatable and efficient methods, using open source software, is important in ensuring methods are applicable across a range of datasets [35,36]. Further, there are potential cost implications of processing data, given that some processing software can be expensive (i.e., compulsory licence fees, called 'payware' in this paper) and so are often only accessible to large organisations in high-income countries [37].…”

“…Generally, semi-automated counting methods are most effective for species where there are strong contrasts against the backgrounds, particularly when background colours and shapes are consistent [28]. They can distinguish large single species aggregations on relatively simple backgrounds [30][31][32], up to sixteen avian species (numbering in the hundreds) on simple single colour backgrounds, such as oceans [33,34], or single species aggregations of hundreds of thousands on complex backgrounds [3].…”