Amanda Hodgson

Understanding the fine-scale behavioural and feeding ecology of marine megafauna is imperative for effective management of their habitat areas; however, obtaining the relevant data can be both costly and challenging. Here we integrate the use of small drones for dugong surveys with underwater benthic habitat assessment techniques at the local spatial scale (∼30 km2), to determine the drivers of dugong (Dugong dugon) distribution across three locations in the Pilbara, Western Australia. Paired assessment data was collected three times over two years. Benthic habitat (percent cover), seagrass nutritional quality and environmental parameters (temperature, water clarity, water current, water depth) were tested as predictor variables using generalised linear models, to examine drivers of both dugong presence/absence and abundance. We found that low cover (typical for this region; 2–10 %) of colonising seagrass is a key driver of the presence and abundance of dugongs. Halophila ovalis and Halodule uninervis were the main predictors of dugong presence and abundance across the three locations surveyed. Where both seagrass species simultaneously occurred, the likelihood of dugongs being present increased by over 60 times. The presence of H. uninervis alone was predicted to increase the abundance of dugongs by 1.4 times across all locations and by 6.8 times in one location, Exmouth Gulf, compared to when no seagrass was present. This study provided evidence of critical seagrass habitat, which is important knowledge for the protection and conservation of dugongs and their foraging habitat. The methods developed in this study could be employed in environmental impact assessments to predict and confirm potential seagrass forage habitat.
[Display omitted]

Drones have emerged as a powerful tool in animal detection, significantly advancing wildlife monitoring, conservation, and management by capturing high-resolution, real-time imagery over areas often inaccessible or challenging for human observers to reach. However, manual analysis of drone imagery for animal detection is labour-intensive and time-consuming. The application of deep learning methods, particularly convolutional neural networks, in automating animal detection from drone imagery has the potential to revolutionise wildlife monitoring, conservation, and management protocols.
This review provides a comprehensive overview of the increasing use and prospects of deep learning in animal detection using drone imagery. It explores successful applications of deep learning for animal detection, localisation, recognition, and their combinations. The paper also discusses the challenges, limitations, and future research directions of this field. A key message from this review is the need for representative training data covering the various scenarios in which target animals may appear, image annotation difficulties, and the comparability of DL model results across studies. Many studies have focused on single species, locations, or images with a high density of common target species. Assessments of models are potentially biased from using a single test set; many studies lack metrics to evaluate model efficiency, feasibility, and generalizability, and there are uncertainties regarding the optimal number of training images and required ground sample distance (GSD) for different animal detection tasks in drone imagery.
The potential applications of deep learning in wildlife monitoring, conservation, and ecological research using drone imagery are substantial. By enhancing the accuracy and efficiency of animal detection in imagery, this technology could contribute to the understanding and protecting animal populations. To expand the applicability of deep learning to diverse species, environments, and spatial scales, researchers should create standardised benchmark datasets and prioritise open collaboration and data sharing, which would aid in addressing the current challenges.

Can you picture cows grazing on a meadow of grass? Did you know that there are also “cows” under the sea that graze on seagrass meadows? Dugongs—a type of sea-cow—are threatened with extinction, mainly as a result of human activities and loss of their main food source, seagrass. Seagrasses are a group of flowering plants that grow in the ocean! Seagrasses are important not only as a food source for dugongs, but they provide a home for many animals, absorb carbon dioxide aiding in climate change mitigation, and so much more! However, seagrasses are declining globally, which is bad news not only for dugongs, but for humans as well. Luckily, dugong presence can aid scientists in understanding the health of seagrasses in an area, as well as help scientists locate and protect our important seagrass ecosystems.

There are many advantages to transitioning from conducting marine wildlife surveys via human observers onboard light-aircraft, to capturing aerial imagery using drones. However, it is important to maintain the validity of long-term data series whilst transitioning from observer to imagery surveys. We need to understand how the detection rates of target species in images compare to those collected from observers in piloted aircraft, and the factors influencing detection rates from each platform. We conducted trial ScanEagle drone surveys of dugongs in Shark Bay, Western Australia, covering the full extent of the drone’s range (∼100 km), concurrently with observer surveys, with the drone flying above or just behind the piloted aircraft. We aimed to test the assumption that drone imagery could provide comparable detection rates of dugongs to human observers when influenced by same environmental conditions. Overall, the dugong sighting rate (i.e., count of individual dugongs) was 1.3 (95% CI [0.98–1.84]) times higher from the drone images than from the observers. The group sighting rate was similar for the two platforms, however the group sizes detected within the drone images were significantly larger than those recorded by the observers, which explained the overall difference in sighting rates. Cloud cover appeared to be the only covariate affecting the two platforms differently; the incidence of cloud cover resulted in smaller group sizes being detected by both platforms, but the observer group sizes dropped much more dramatically (by 71% (95% CI [31–88]) compared to no cloud) than the group sizes detected in the drone images (14% (95% CI [−28–57])). Water visibility and the Beaufort sea state also affected dugong counts and group sizes, but in the same way for both platforms. This is the first direct simultaneous comparison between sightings from observers in piloted aircraft and a drone and demonstrates the potential for drone surveys over a large spatial-scale.

Aerial surveys are frequently used to estimate the abundance of marine mammals, but their accuracy is dependent upon obtaining a measure of the availability of animals to visual detection. Existing methods for characterizing availability have limitations and do not necessarily reflect true availability. Here, we present a method of using small, vessel-launched, multi-rotor Unoccupied Aerial Vehicles (UAVs, or drones) to collect video of dolphins to characterize availability and investigate error surrounding group size estimates. We collected over 20 h of aerial video of dive-surfacing behaviour across 32 encounters with Australian humpback dolphins Sousa sahulensis off north-western Australia. Mean surfacing and dive periods were 7.85 sec (se = 0.26) and 39.27 sec (se = 1.31) respectively. Dolphin encounters were split into 56 focal follows of consistent group composition to which example approaches to estimating availability were applied. Non-instantaneous availability estimates, assuming a 7 sec observation window, ranged between 0.22 and 0.88, with a mean availability of 0.46 (CV = 0.34). Availability tended to increase with increasing group size. We found a downward bias in group size estimation, with true group size typically one individual more than would have been estimated by a human observer during a standard aerial survey. The variability of availability estimates between focal follows highlights the importance of sampling across a variety of group sizes, compositions and environmental conditions. Through data re-sampling exercises, we explored the influence of sample size on availability estimates and their precision, with results providing an indication of target sample sizes to minimize bias in future research. We show that UAVs can provide an effective and relatively inexpensive method of characterizing dolphin availability with several advantages over existing approaches. The example estimates obtained for humpback dolphins are within the range of values obtained for other shallow-water, small cetaceans, and will directly inform a government-run program of aerial surveys in the region.

The Journal of Unmanned Vehicle Systems—led by its founding editor, David Bird—published its first issue in December 2013. It was the first peer-reviewed scientific journal entirely dedicated to research relating to all types of remotely piloted or autonomous robotic vehicles, including those that operate in the air, on the ground, or on or below the water’s surface. Although rare currently, it could also expand to include those that operate in outer space environments (Potter 2020). This is a uniquely eclectic field of research that encompasses multiple engineering and design aspects of the vehicles themselves in addition to a diverse and ever-growing array of practical applications of the technology...

The advent of unoccupied aerial vehicles (UAVs) has enhanced our capacity to survey wildlife abundance, yet new protocols are still required for collecting, processing, and analysing image-type observations. This paper presents a methodological approach to produce informative priors on species misidentification probabilities based on independent experiments. We performed focal follows of known dolphin species and distributed our imagery amongst 13 trained observers. Then, we investigated the effects of reviewer-related variables and image attributes on the accuracy of species identification and level of certainty in observations. In addition, we assessed the number of reviewers required to produce reliable identification using an agreement-based framework compared with the majority rule approach. Among-reviewer variation was an important predictor of identification accuracy, regardless of previous experience. Image resolution and sea state exhibited the most pronounced effects on the proportion of correct identifications and the reviewers’ mean level of confidence. Agreement-based identification resulted in substantial data losses but retained a broader range of image resolutions and sea states than the majority rule approach and produced considerably higher accuracy. Our findings suggest a strong dependency on reviewer-related variables and image attributes, which, unless considered, may compromise identification accuracy and produce unreliable estimators of abundance.

Understanding species’ distribution patterns and the environmental and ecological interactions that drive them is fundamental for biodiversity conservation. Data deficiency exists in areas that are difficult to access, or where resources are limited. We use a broad-scale, non-targeted dataset to describe dolphin distribution and habitat suitability in remote north Western Australia, where there is a paucity of data to adequately inform species management. From 1,169 opportunistic dolphin sightings obtained from 10 dugong aerial surveys conducted over a four-year period, there were 661 Indo-Pacific bottlenose dolphin (Tursiops aduncus), 191 Australian humpback dolphin (Sousa sahulensis), nine Australian snubfin dolphin (Orcaella heinsohni), 16 Stenella sp., one killer whale (Orcinus orca), one false killer whale (Pseudorca crassidens), and 290 unidentified dolphin species sightings. Maximum Entropy (MaxEnt) habitat suitability models identified shallow intertidal areas around mainland coast, islands and shoals as important areas for humpback dolphins. In contrast, bottlenose dolphins are more likely to occur further offshore and at greater depths, suggesting niche partitioning between these two sympatric species. Bottlenose dolphin response to sea surface temperature is markedly different between seasons (positive in May; negative in October) and probably influenced by the Leeuwin Current, a prominent oceanographic feature. Our findings support broad marine spatial planning, impact assessment and the design of future surveys, which would benefit from the collection of high-resolution digital images for species identification verification. A substantial proportion of data were removed due to uncertainties resulting from non-targeted observations and this is likely to have reduced model performance. We highlight the importance of considering climatic and seasonal fluctuations in interpreting distribution patterns and species interactions in assuming habitat suitability.

Simulated social facilitation techniques (e.g. decoys and call playbacks) are commonly used to attract seabirds to restored and artificially created nesting habitats. However, a lack of social stimuli and conspecific cueing at these habitats may limit the use of these sites, at least in the short term. Therefore, testing the effectiveness of simulated audio-visual cues for attracting gregarious birds is important for conservation planning. In this study, we (1) assessed whether call playback and decoys were associated with an increased likelihood of Australian fairy terns Sternula nereis nereis visiting potentially suitable nesting habitats; (2) tested their behavioral response to different cues; and (3) documented whether social facilitation had the potential to encourage colony establishment. A full cross-over study design consisting of all possible pairings of decoy and call playback treatments (control [no attractants], decoys, call playback, both decoys and playback), allocated as part of a random block design, was undertaken at 2 sites. Linear modeling suggested that call playback was important in explaining the time spent aerial prospecting as well as the maximum number of fairy terns aerial prospecting, although this only appeared to be the case for 1 of the 2 sites. Decoys, on the other hand, did not appear to have any effect on time spent aerial prospecting. The results from this study suggest that audio cues have the potential to encourage site selection by increasing social stimuli, but attractants may be required over several breeding seasons before colonies are established.

There is growing interest from research and conservation groups in the potential for using small unoccupied aerial vehicles (UAVs; <2 kg) to conduct wildlife surveys because they are affordable, easy to use, readily available and reliable. However, limitations such as short flight endurance, and in many situations, aviation regulations, have constrained the use of small UAVs in survey applications. Thus, there is a need to refine survey methods adapted to small UAVs that conform to standard operations within aviation law. We developed a novel survey approach based on a grid sampling design using two multirotor UAVs (Phantom 4 Pros) flying simultaneously, within visual line of sight, from our vessel base-station. We used this approach to assess the fine-scale distribution and abundance of dugongs (Dugong dugon) in the remote waters of the Pilbara, Western Australia during three field seasons across 2 years. We surveyed 64 non-overlapping survey cells in random order one or more times and obtained complete image coverage of each surveyed cell of our 31 km2 survey area. Our sampling design maximizes sampling effort while limiting survey time by surveying four cells, two at a time, from one location. Overall, we conducted 240 flights with up to 17 flights per day (mean = 14 flights per day) and could obtain complete coverage of up to 11.36 km2per day. A total of 149 dugongs were sighted within the 50,482 images which we manually reviewed. Spatially-explicit models of dugong density distribution (corrected for availability and perception bias) were produced using general additive models to identify areas more or less used by dugongs (range of corrected dugong densities across all field season = 0.002–1.79 dugongs per 0.04 km2). Dugong abundance estimates ranged from 47 individuals in June 2019 (CV = 0.17) to 103 individuals in May 2018 (CV = 0.36). Our method, which proved convincing in a real-word application by its feasibility, ease of implementation, and achievable surface coverage has the potential to be used in a wide range of applications from community-based local-scale surveys, to long-term repeated/intensive surveys, and impact assessments and environmental monitoring studies.