Search

Bird colonies on islands sustain elevated productivity and biomass on adjacent reefs, through nutrient subsidies. However, the implications of this localized enhancement on higher and often more mobile trophic levels (such as sharks and rays) are unclear, as spatial trends in mobile fauna are often poorly captured by traditional underwater visual surveys. Here, we explore whether the presence of seabird colonies is associated with enhanced abundances of sharks and rays on adjacent coral reefs. We used a novel long‐range water‐landing fixed‐wing unoccupied aerial vehicle (UAV) to survey the distribution and density of sharks, rays and any additional megafauna, on and around tropical coral islands (n = 14) in the Chagos Archipelago Marine Protected Area. We developed a computer‐vision algorithm to distinguish greenery (trees and shrubs), sand and sea glitter from visible ocean to yield accurate marine megafauna density estimation. We detected elevated seabird densities over rat‐free islands, with the commonest species, sooty tern, reaching densities of 932 ± 199 per km−2 while none were observed over former coconut plantation islands. Elasmobranch density around rat‐free islands with seabird colonies was 6.7 times higher than around islands without seabird colonies (1.3 ± 0.63 vs. 0.2 ± SE 0.1 per km2). Our results are evidence that shark and ray distribution is sensitive to natural and localized nutrient subsidies. Correcting for non‐sampled regions of images increased estimated elasmobranch density by 14%, and our openly accessible computer vision algorithm makes this correction easy to implement to generate shark and ray and other wildlife densities from any aerial imagery. The water‐landing fixed‐wing long‐range UAV technology used in this study may provide cost effective monitoring opportunities in remote ocean locations. [ABSTRACT FROM AUTHOR]

Camera trap surveys are a popular ecological monitoring tool that produce vast numbers of images making their annotation extremely time‐consuming. Advances in machine learning, in the form of convolutional neural networks, have demonstrated potential for automated image classification, reducing processing time. These networks often have a poor ability to generalise, however, which could impact assessments of species in habitats undergoing change.Here, we (i) compare the performance of three network architectures in identifying species in camera trap images taken from tropical forest of varying disturbance intensities; (ii) explore the impacts of training dataset configuration; (iii) use habitat disturbance categories to investigate network generalisability and (iv) test whether classification performance and generalisability improve when using images cropped to bounding boxes.Overall accuracy (72.8%) was improved by excluding the rarest species and by adding extra training images (76.3% and 82.8%, respectively). Generalisability to new camera locations within a disturbance level was poor (mean F1‐score: 0.32). Performance across unseen habitat disturbance levels was worse (mean F1‐score: 0.27). Training the network on multiple disturbance levels improved generalisability (mean F1‐score on unseen disturbance levels: 0.41). Cropping images to bounding boxes improved overall performance (F1‐score: 0.77 vs. 0.47) and generalisability (mean F1‐score on unseen disturbance levels: 0.73), but at a cost of losing images that contained animals which the detector failed to detect.These results suggest researchers should consider using an object detector before passing images to a classifier, and an improvement in classification might be seen if labelled images from other studies are added to their training data. Composition of training data was shown to be influential, but including rarer classes did not compromise performance on common classes, providing support for the inclusion of rare species to inform conservation efforts. These findings have important implications for use of these methods for long‐term monitoring of habitats undergoing change, as they highlight the potential for misclassifications due to poor generalisability to impact subsequent ecological analyses. These methods therefore need to be considered as dynamic, in that changes to the study site would need to be reflected in the updated training of the network. [ABSTRACT FROM AUTHOR]

Human activity in African tropical rainforests continues to threaten wild mammals. Many rural communities are dependent on hunting, yet there is a widespread lack of baseline data on ecology and the sustainability of hunting. We investigated the impacts of human activity on mammal species composition and distributions within a community forest surrounding a village in the buffer zone of the Dja Biosphere Reserve in south-east Cameroon. We conducted a camera-trap survey in August–November 2017 and detected 24 mammal species, including Critically Endangered western lowland gorilla Gorilla gorilla gorilla , Endangered central African chimpanzee Pan troglodytes troglodytes and Endangered tree pangolin Phataginus tricuspis. We used occupancy analysis to explore relationships between indicators of human activity (distance to a road and the Reserve), habitat quality (distance to the river and tree cover) and the distributions of species. We found that the local distribution of threatened mammals was not apparently limited by human activity, and proximity to the road did not negatively influence occupancy for any species. However, most of the Reserve's large species were not detected, including the African forest elephant Loxodonta cyclotis and the largest ungulates, and the occupancy of two species commonly hunted for wild meat was positively correlated with distance from the village, indicating hunting may be unsustainable. Our results show that the community forest provides habitat for threatened species outside the Reserve and in close proximity to people. However, effective conservation management will require continued monitoring and research to determine whether current rates of hunting are sustainable. [ABSTRACT FROM AUTHOR]

Animal density is a fundamental parameter in ecology and conservation, and yet it has remained difficult to measure. For terrestrial mammals and birds, camera‐traps have dramatically improved our ability to collect systematic data across a large number of species, but density estimation (except for species with natural marks) is still faced with statistical and logistical hurdles, including the requirement for auxiliary data and large sample sizes, and an inability to incorporate covariates.To fill this gap in the camera‐trapper's statistical toolbox, we extended the existing Random Encounter Model (REM) to the multi‐species case in a Bayesian framework. This multi‐species REM can incorporate covariates and provides parameter estimates for even the rarest species. As input to the model, we used information directly available in the camera‐trap data. The model outputs posterior distributions for the REM parameters—movement speed, activity level, the effective angle and radius of the camera‐trap detection zone, and density—for each species. We applied this model to an existing dataset for 35 species in Borneo, collected across old‐growth and logged forest. Here, we added animal position data derived from the image sequences in order to estimate the speed and detection zone parameters.The model revealed a decrease in movement speeds, and therefore day‐range, across the species community in logged compared to old‐growth forest, whilst activity levels showed no consistent trend. Detection zones were shorter, but of similar width, in logged compared to old‐growth forest. Overall, animal density was lower in logged forest, even though most species individually occurred at higher density in logged forest. However, the biomass per unit area was substantially higher in logged compared to old‐growth forest, particularly among herbivores and omnivores, likely because of increased resource availability at ground level. We also included body mass as a variable in the model, revealing that larger‐bodied species were more active, had more variable speeds, and had larger detection zones.Caution is warranted when estimating density for semi‐arboreal and fossorial species using camera‐traps, and more extensive testing of assumptions is recommended. Nonetheless, we anticipate that multi‐species density estimation could have very broad application. [ABSTRACT FROM AUTHOR]

Ranger‐led law enforcement patrols are the primary, site‐level response to – and the most common source of data on – illegal activity threatening wildlife in protected areas. Yet evidence that patrols effectively deter rule‐breaking is limited, and common management metrics for evaluating deterrence, which use ranger‐collected data, are particularly vulnerable to bias. "Differenced plots" (of the association between change in patrol effort and subsequent change in illegal activity) were recently proposed as a simple, new metric for deterrence, which, in tests with simulated patrol data, were more robust than the common alternatives. Here, we trial application of differenced plots to real patrol data collected in four protected areas, and explore methods for applying the metric in practice, using two indicators of rule‐breaking: snares, and people. We find evidence which is consistent with deterrence in some but not all sites, over shorter timescales than observed hitherto: increases in patrol effort were associated with subsequent reductions in snaring in one site, and in the presence of people in two sites. However, whether pressure on wildlife had been reduced or merely displaced was unclear from differenced plots, nor could the metric confirm absence of deterrence, raising questions for future applications. Our findings suggest differenced plots can be a useful metric, particularly for exploring variation in deterrence within sites, but should be applied and interpreted with care, and further work is urgently needed to determine whether and how patrols deter illegal activity, and to evaluate the effect reliably. [ABSTRACT FROM AUTHOR]

Remote Sensing in Ecology and Conservation, 1, 1 - 3. 31 Pittman, S.J. (2018) Seascape ecology. Active acoustic remote sensing technologies have historically been associated with military uses and the field of hydrography rather than with the remote sensing community of practice; this has slowed the integration of data processing and analysis methods that have proven effective in the study of terrestrial environments. Acoustic remote sensing can be passive (i.e., using hydrophones to capture sounds in the environment) or active (i.e., using sonars that produce directional sound and listen for returns); both have their place in support of marine ecology and conservation. Since its inception, the editorial board of I Remote Sensing in Ecology and Conservation i has welcomed contributions to coastal and marine ecology and conservation that rely on remote sensing (Pettorelli et al., 2015). [Extracted from the article]