Using convolutional neural networks to count parrot nest-entrances on photographs from the largest known colony of Psittaciformes

Counting animal populations is fundamental to understand ecological processes. Counts make it possible to estimate the size of an animal population at specific points in time, which is essential information for understanding demographic change. However, in the case of large populations, counts are time-consuming, particularly if carried out manually. Here, we took advantage of convolutional neural networks (CNN) to count the total number of nest-entrances in 222 photographs covering the largest known Psittaciformes (Aves) colony in the world. We conducted our study at the largest Burrowing Parrot Cyanoliseus patagonus colony, located on a cliff facing the Atlantic Ocean in the vicinity of El Cóndor village, in north-eastern Patagonia, Argentina. We also aimed to investigate the distribution of nest-entrances along the cliff with the colony. For this, we used three CNN architectures, U-Net, ResUnet, and DeepLabv3. The U-Net architecture showed the best performance, counting a mean of 59,842 Burrowing Parrot nest-entrances across the colony, with a mean absolute error of 2.7 nest-entrances over the testing patches, measured as the difference between actual and predicted counts per patch. Compared to a previous study conducted at El Cóndor colony more than 20 years ago, the CNN architectures also detected noteworthy differences in the distribution of the nest-entrances along the cliff. We show that the strong changes observed in the distribution of nest-entrances are a measurable effect of a long record of human-induced disturbance to the Burrowing Parrot colony at El Cóndor. Given the paramount importance of the Burrowing Parrot colony at El Cóndor, which concentrates 71% of the world's population of this species, we advocate that it is imperative to reduce such a degree of disturbance before the parrots reach the limit of their capacity of adaptation.

Coral reefs in transition: Temporal photoquadrat analyses and validation of underwater hyperspectral imaging for resource-efficient monitoring in Guam

The island of Guam in the west Pacific has seen a significant decrease in coral cover since 2013. Lafac Bay, a marine protected area in northeast Guam, served as a reference site for benthic communities typical of forereefs on the windward side of the island. The staghorn coral Acropora abrotanoides is a dominant and characteristic ecosystem engineer of forereef communities on exposed shorelines. Photoquadrat surveys were conducted in 2015, 2017, and 2019, and a diver-operated hyperspectral imager (i.e., DiveRay) was used to survey the same transects in 2019. Machine learning algorithms were used to develop an automated pipeline to assess the benthic cover of 10 biotic and abiotic categories in 2019 based on hyperspectral imagery. The cover of scleractinian corals did not differ between 2015 and 2017 despite being subjected to a series of environmental disturbances in these years. Surveys in 2019 documented the almost complete decline of the habitat-defining staghorn coral Acropora abrotanoides (a practically complete disappearance from about 10% cover), a significant decrease (~75%) in the cover of other scleractinian corals, and a significant increase (~55%) in the combined cover of bare substrate, turf algae, and cyanobacteria. The drastic change in community composition suggests that the reef at Lafac Bay is transitioning to a turf algae-dominated community. However, the capacity of this reef to recover from previous disturbances suggests that this transition could be reversed, making Lafac Bay an excellent candidate for long-term monitoring. Community analyses showed no significant difference between automatically classified benthic cover estimates derived from the hyperspectral scans in 2019 and those derived from photoquadrats. These findings suggest that underwater hyperspectral imagers can be efficient and effective tools for fast, frequent, and accurate monitoring of dynamic reef communities.

Assessment of the utility of underwater hyperspectral imaging for surveying and monitoring coral reef ecosystems

Technological innovations that improve the speed, scale, reproducibility, and accuracy of monitoring surveys will allow for a better understanding of the global decline in tropical reef health. The DiveRay, a diver-operated hyperspectral imager, and a complementary machine learning pipeline to automate the analysis of hyperspectral imagery were developed for this purpose. To evaluate the use of a hyperspectral imager underwater, the automated classification of benthic taxa in reef communities was tested. Eight reefs in Guam were surveyed and two approaches for benthic classification were employed: high taxonomic resolution categories and broad benthic categories. The results from the DiveRay surveys were validated against data from concurrently conducted photoquadrat surveys to determine their accuracy and utility as a proxy for reef surveys. The high taxonomic resolution classifications did not reliably predict benthic communities when compared to those obtained by standard photoquadrat analysis. At the level of broad benthic categories, however, the hyperspectral results were comparable to those of the photoquadrat analysis. This was particularly true when estimating scleractinian coral cover, which was accurately predicted for six out of the eight sites. The annotation libraries generated for this study were insufficient to train the model to fully account for the high biodiversity on Guam's reefs. As such, prediction accuracy is expected to improve with additional surveying and image annotation. This study is the first to directly compare the results from underwater hyperspectral scanning with those from traditional photoquadrat survey techniques across multiple sites with two levels of identification resolution and different degrees of certainty. Our findings show that dependent on a well-annotated library, underwater hyperspectral imaging can be used to quickly, repeatedly, and accurately monitor and map dynamic benthic communities on tropical reefs using broad benthic categories.

Mapping the regeneration potential of native vegetation in cultivated pastures of the Brazilian Cerrado

The Brazilian Cerrado is a hotspot of biodiversity conservation and an important global agricultural region. Cultivated pastures under different degradation levels are dominant in the landscape and are being targeted for sustainable agricultural intensification and restoration of native vegetation. In this study, we classified the cultivated pastures of the Brazilian Cerrado according to their potential for natural regeneration, based on field surveys and environmental predictors. We surveyed the native vegetation cover in 186 plots distributed along 93 cultivated pastures. The environmental predictors considered in this study were the proportion of sand in the soil, cation exchange capacity, climate water deficit, pasture age, slope, and pasture vigor index. We then applied the Random Forest regression algorithm to predict and map the cultivated pastures according to their potential for natural regeneration in the 19 Cerrado ecoregions. The potential for natural regeneration was classified into low (< 30% of native plant cover), medium (30-50%), and high (> 50%). Our prediction explained 75% of the data variability. Most of the cultivated pastures presented medium potential for natural regeneration (57%), while 31% and 12% presented high and low potentials, respectively. Cultivated pastures in ecoregions with high mechanization, due to their high water availability and extensive flat terrains, presented low potential for natural regeneration. This first attempt to map the potential for natural regeneration in the cultivated pastures of the Brazilian Cerrado can be used as a proxy for planning low-cost and predictable restoration or environmentally sustainable intensification in this major type of land use found in this biome.

Unlocking the Potential of Deep Learning for Migratory Waterbirds Monitoring Using Surveillance Video

Estimates of migratory waterbirds population provide the essential scientific basis to guide the conservation of coastal wetlands, which are heavily modified and threatened by economic development. New equipment and technology have been increasingly introduced in protected areas to expand the monitoring efforts, among which video surveillance and other unmanned devices are widely used in coastal wetlands. However, the massive amount of video records brings the dual challenge of storage and analysis. Manual analysis methods are time-consuming and error-prone, representing a significant bottleneck to rapid data processing and dissemination and application of results. Recently, video processing with deep learning has emerged as a solution, but its ability to accurately identify and count waterbirds across habitat types (e.g., mudflat, saltmarsh, and open water) is untested in coastal environments. In this study, we developed a two-step automatic waterbird monitoring framework. The first step involves automatic video segmentation, selection, processing, and mosaicking video footages into panorama images covering the entire monitoring area, which are subjected to the second step of counting and density estimation using a depth density estimation network (DDE). We tested the effectiveness and performance of the framework in Tiaozini, Jiangsu Province, China, which is a restored wetland, providing key high-tide roosting ground for migratory waterbirds in the East Asian–Australasian flyway. The results showed that our approach achieved an accuracy of 85.59%, outperforming many other popular deep learning algorithms. Furthermore, the standard error of our model was very small (se = 0.0004), suggesting the high stability of the method. The framework is computing effective—it takes about one minute to process a theme covering the entire site using a high-performance desktop computer. These results demonstrate that our framework can extract ecologically meaningful data and information from video surveillance footages accurately to assist biodiversity monitoring, fulfilling the gap in the efficient use of existing monitoring equipment deployed in protected areas.

Sea Turtles in the Cancer Risk Landscape: A Global Meta-Analysis of Fibropapillomatosis Prevalence and Associated Risk Factors

Several cancer risk factors (exposure to ultraviolet-B, pollution, toxins and pathogens) have been identified for wildlife, to form a “cancer risk landscape.” However, information remains limited on how the spatiotemporal variability of these factors impacts the prevalence of cancer in wildlife. Here, we evaluated the cancer risk landscape at 49 foraging sites of the globally distributed green turtle (Chelonia mydas), a species affected by fibropapillomatosis, by integrating data from a global meta-analysis of 31 publications (1994–2019). Evaluated risk factors included ultraviolet light exposure, eutrophication, toxic phytoplanktonic blooms, sea surface temperature, and the presence of mechanical vectors (parasites and symbiotic species). Prevalence was highest in areas where nutrient concentrations facilitated the emergence of toxic phytoplankton blooms. In contrast, ultraviolet light exposure and the presence of parasitic and/or symbiotic species did not appear to impact disease prevalence. Our results indicate that, to counter outbreaks of fibropapillomatosis, management actions that reduce eutrophication in foraging areas should be implemented.

COVID-19 disruption reveals mass-tourism pressure on nearshore sea turtle distributions and access to optimal breeding habitat

Quantifying the extent to which animals detect and respond to human presence allows us to identify pressure (disturbance) and inform conservation management objectively; however, obtaining baselines against which to compare human impact is hindered in areas where human activities are already well established. For example, Zakynthos Island (Greece, Mediterranean) receives around 850,000 visitors each summer, while supporting an important loggerhead sea turtle rookery (~300 individuals/season). The coronavirus (COVID-19)-driven absence of tourism in May-June 2020 provided an opportunity to evaluate the distribution dynamics of this population in the absence (2020) vs. presence (2018 and 2019) of visitors using programmed unmanned aerial system (UAS) surveys. Ambient sea temperature transitioned from suboptimal for breeding in May to optimal in late June, with turtle distribution appearing to shift from shallow (to benefit from waters 3-5°C above ambient) to deeper waters in 2018 and 2019, but not 2020. The 2020 data set demonstrated that increased tourism pressure, not temperature, drives turtles offshore. Specifically, >50% of turtles remained within 100 m of shore at densities of 25-50 visitors/km, even when sea temperature rose, with 2018 and 2019 data supporting this trend. Reduced access to warmer, nearshore waters by tourism could delay the onset of nesting and increase the length of the egg maturation period between nesting events (internesting interval) at this site. A coastal refuge zone could be delimited in May-June where touristic infrastructure is minimal, but also where turtles frequently aggregate. In conclusion, sea turtles appear capable of perceiving changes in the level of human pressure at fine spatial and temporal scales and adjusting their distribution accordingly.

Cancer risk landscapes: A framework to study cancer in ecosystems

Cancer is a family of diseases that has been documented in most metazoan species and ecosystems. Human induced environmental changes are increasingly exposing wildlife to carcinogenic risk factors, and negative repercussions on ecosystems and on the conservation of endangered species are already been observed. It is therefore of key importance to understand the spatiotemporal variability of those risk factors and how they interact with the biosphere to mitigate their effects. Here we introduce the concept of cancer risk landscape that can be applied to understand how species are exposed to, interact with, and modify cancer risk factors. With this publication we aim to provide a framework in order to stimulate a discussion on how to mitigate cancer-causing risk factors.

Going Batty: The Challenges and Opportunities of Using Drones to Monitor the Behaviour and Habitat Use of Rays

The way an animal behaves in its habitat provides insight into its ecological role. As such, collecting robust, accurate datasets in a time-efficient manner is an ever-present pressure for the field of behavioural ecology. Faced with the shortcomings and physical limitations of traditional ground-based data collection techniques, particularly in marine studies, drones offer a low-cost and efficient approach for collecting data in a range of coastal environments. Despite drones being widely used to monitor a range of marine animals, they currently remain underutilised in ray research. The innovative application of drones in environmental and ecological studies has presented novel opportunities in animal observation and habitat assessment, although this emerging field faces substantial challenges. As we consider the possibility to monitor rays using drones, we face challenges related to local aviation regulations, the weather and environment, as well as sensor and platform limitations. Promising solutions continue to be developed, however, growing the potential for drone-based monitoring of behaviour and habitat use of rays. While the barriers to enter this field may appear daunting for researchers with little experience with drones, the technology is becoming increasingly accessible, helping ray researchers obtain a wide range of highly useful data.

Drivers of Bornean Orangutan Distribution across a Multiple-Use Tropical Landscape

Logging and conversion of tropical forests in Southeast Asia have resulted in the expansion of landscapes containing a mosaic of habitats that may vary in their ability to sustain local biodiversity. However, the complexity of these landscapes makes it difficult to assess abundance and distribution of some species using ground-based surveys alone. Here, we deployed a combination of ground-transects and aerial surveys to determine drivers of the critically endangered Bornean Orangutan (Pongo pygmaeus morio) distribution across a large multiple-use landscape in Sabah, Malaysian Borneo. Ground-transects and aerial surveys using drones were conducted for orangutan nests and hemi-epiphytic strangler fig trees (Ficus spp.) (an important food resource) in 48 survey areas across 76 km2, within a study landscape of 261 km2. Orangutan nest count data were fitted to models accounting for variation in land use, above-ground carbon density (ACD, a surrogate for forest quality), strangler fig density, and elevation (between 117 and 675 m). Orangutan nest counts were significantly higher in all land uses possessing natural forest cover, regardless of degradation status, than in monoculture plantations. Within these natural forests, nest counts increased with higher ACD and strangler fig density, but not with elevation. In logged forest (ACD 14–150 Mg ha−1), strangler fig density had a significant, positive relationship with orangutan nest counts, but this relationship disappeared in a forest with higher carbon content (ACD 150–209 Mg ha−1). Based on an area-to-area comparison, orangutan nest counts from ground transects were higher than from counts derived from aerial surveys, but this did not constitute a statistically significant difference. Although the difference in nest counts was not significantly different, this analysis indicates that both methods under-sample the total number of nests present within a given area. Aerial surveys are, therefore, a useful method for assessing the orangutan habitat use over large areas. However, the under-estimation of nest counts by both methods suggests that a small number of ground surveys should be retained in future surveys using this technique, particularly in areas with dense understory vegetation. This study shows that even highly degraded forests may be a suitable orangutan habitat as long as strangler fig trees remain intact after areas of forest are logged. Enrichment planting of strangler figs may, therefore, be a valuable tool for orangutan conservation in these landscapes.

The Drone Revolution of Shark Science: A Review

Over the past decade, drones have become a popular tool for wildlife management and research. Drones have shown significant value for animals that were often difficult or dangerous to study using traditional survey methods. In the past five years drone technology has become commonplace for shark research with their use above, and more recently, below the water helping to minimise knowledge gaps about these cryptic species. Drones have enhanced our understanding of shark behaviour and are critically important tools, not only due to the importance and conservation of the animals in the ecosystem, but to also help minimise dangerous encounters with humans. To provide some guidance for their future use in relation to sharks, this review provides an overview of how drones are currently used with critical context for shark monitoring. We show how drones have been used to fill knowledge gaps around fundamental shark behaviours or movements, social interactions, and predation across multiple species and scenarios. We further detail the advancement in technology across sensors, automation, and artificial intelligence that are improving our abilities in data collection and analysis and opening opportunities for shark-related beach safety. An investigation of the shark-based research potential for underwater drones (ROV/AUV) is also provided. Finally, this review provides baseline observations that have been pioneered for shark research and recommendations for how drones might be used to enhance our knowledge in the future.

Expanding conservation culturomics and iEcology from terrestrial to aquatic realms

The ongoing digital revolution in the age of big data is opening new research opportunities. Culturomics and iEcology, two emerging research areas based on the analysis of online data resources, can provide novel scientific insights and inform conservation and management efforts. To date, culturomics and iEcology have been applied primarily in the terrestrial realm. Here, we advocate for expanding such applications to the aquatic realm by providing a brief overview of these new approaches and outlining key areas in which culturomics and iEcology are likely to have the highest impact, including the management of protected areas; fisheries; flagship species identification; detection and distribution of threatened, rare, and alien species; assessment of ecosystem status and anthropogenic impacts; and social impact assessment. When deployed in the right context with awareness of potential biases, culturomics and iEcology are ripe for rapid development as low-cost research approaches based on data available from digital sources, with increasingly diverse applications for aquatic ecosystems.

Operational Protocols for the Use of Drones in Marine Animal Research

The use of drones to study marine animals shows promise for the examination of numerous aspects of their ecology, behaviour, health and movement patterns. However, the responses of some marine phyla to the presence of drones varies broadly, as do the general operational protocols used to study them. Inconsistent methodological approaches could lead to difficulties comparing studies and can call into question the repeatability of research. This review draws on current literature and researchers with a wealth of practical experience to outline the idiosyncrasies of studying various marine taxa with drones. We also outline current best practice for drone operation in marine environments based on the literature and our practical experience in the field. The protocols outlined herein will be of use to researchers interested in incorporating drones as a tool into their research on marine animals and will help form consistent approaches for drone-based studies in the future.