Satellites allow large‐scale surveys to be conducted in short time periods with repeat surveys possible at intervals of <24 h. Very‐high‐resolution satellite imagery has been successfully used to detect and count a number of wildlife species in open, homogeneous landscapes and seascapes where target animals have a strong contrast with their environment. However, no research to date has detected animals in complex heterogeneous environments or detected elephants from space using very‐high‐resolution satellite imagery and deep learning. In this study, we apply a Convolution Neural Network (CNN) model to automatically detect and count African elephants in a woodland savanna ecosystem in South Africa. We use WorldView‐3 and 4 satellite data –the highest resolution satellite imagery commercially available. We train and test the model on 11 images from 2014 to 2019. We compare the performance accuracy of the CNN against human accuracy. Additionally, we apply the model on a coarser resolution satellite image (GeoEye‐1) captured in Kenya, without any additional training data, to test if the algorithm can generalize to an elephant population outside of the training area. Our results show that the CNN performs with high accuracy, comparable to human detection capabilities. The detection accuracy (i.e., F2 score) of the CNN models was 0.78 in heterogeneous areas and 0.73 in homogenous areas. This compares with the detection accuracy of the human labels with an averaged F2 score 0.77 in heterogeneous areas and 0.80 in homogenous areas. The CNN model can generalize to detect elephants in a different geographical location and from a lower resolution satellite. Our study demonstrates the feasibility of applying state‐of‐the‐art satellite remote sensing and deep learning technologies for detecting and counting African elephants in heterogeneous landscapes. The study showcases the feasibility of using high resolution satellite imagery as a promising new wildlife surveying technique. Through creation of a customized training dataset and application of a Convolutional Neural Network, we have automated the detection of elephants in satellite imagery with accuracy as high as human detection capabilities. The success of the model to detect elephants outside of the training data site demonstrates the generalizability of the technique.

Methods in Ecology and EvolutionThis article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as

Conflicts between the interests of biodiversity conservation and other human activities pose a major threat to natural ecosystems and human well-being, yet few methods exist to quantify their intensity and model their dynamics. We develop a categorization of conflict intensity based on the curve of conflict, a model originally used to track the escalation and deescalation of armed conflicts. Our categorization assigns six intensity levels reflecting the discourse and actions of stakeholders involved in a given conflict, from coexistence or collaboration to physical violence. Using a range of case studies, we demonstrate the value of our approach in quantifying conflict trends, estimating transition probabilities between conflict stages, and modeling conflict intensity as a function of relevant covariates. By taking an evidence-based approach to quantifying stakeholder behavior, the proposed framework allows for a better understanding of the drivers of conservation conflict development across a diverse range of socioecological scenarios.

Satellites allow large-scale surveys to be conducted in short time periods with repeat surveys possible <24hrs. Very high-resolution satellite imagery has been successfully used to detect and count a number of wildlife species in open, homogeneous landscapes and seascapes where target animals have a strong contrast with their environment. However, no research to date has detected animals in complex heterogeneous environments or detected elephants from space using very high-resolution satellite imagery and deep learning.In this study we apply a Convolution Neural Network (CNN) model to automatically detect and count African elephants in a woodland savanna ecosystem in South Africa. We use WorldView-3 and 4 satellite data – the highest resolution satellite imagery commercially available. We train and test the model on eleven images from 2014-2019. We compare the performance accuracy of the CNN against human accuracy. Additionally, we apply the model on a coarser resolution satellite image (GeoEye-1) captured in Kenya to test if the algorithm can generalise to an elephant population outside of the training area.Our results show the CNN performs with high accuracy, comparable to human detection capabilities. The detection accuracy (i.e., F2 score) of the CNN models was 0.78 in heterogeneous areas and 0.73 in homogenous areas. This compares with the detection accuracy of the human labels with an averaged F2 score 0.77 in heterogeneous areas and 0.80 in homogenous areas. The CNN model can generalise to detect elephants in a different geographical location and from a lower resolution satellite.Our study demonstrates the feasibility of applying state-of-the-art satellite remote sensing and deep learning technologies for detecting and counting African elephants in heterogeneous landscapes. The study showcases the feasibility of using high resolution satellite imagery as a promising new wildlife surveying technique. Through creation of a customised training dataset and application of a Convolutional Neural Network, we have automated the detection of elephants in satellite imagery with as high accuracy as human detection capabilities. The success of the model to detect elephants outside of the training data site demonstrates the generalisability of the technique.

Background There is a rich body of literature addressing the topic of illegal hunting of wild terrestrial mammals. Studies on this topic have risen over the last decade as species are under increasing risk from anthropogenic threats. Sub-Saharan Africa contains the highest number of terrestrial mammals listed as vulnerable, endangered or critically endangered. However, the spatial distribution of illegal hunting incidences is not well documented. To address this knowledge gap, the systematic map presented here aims to answer three research questions: (1) What data are available on the spatial distribution of illegal hunting of terrestrial mammals in Sub-Saharan Africa in relation to environmental and anthropogenic correlates i.e. proximity to roads, water bodies, human settlement areas, different land tenure arrangements and anti-poaching ranger patrol bases? (2) Which research methodologies have primarily been used to collect quantitative data and how comparable are these data? (3) Is there a bias in the research body toward particular taxa and geographical areas? Methods Systematic searches were carried out across eight bibliographic databases; articles were screened against pre-defined criteria. Only wild terrestrial mammals listed as vulnerable, endangered or critically endangered by the International Union for Conservation of Nature (IUCN) whose geographical range falls in Sub-Saharan Africa and whose threat assessment includes hunting and trapping were included. To meet our criteria, studies were required to include quantitative, spatially explicit data. In total 14,325 articles were screened at the level of title and abstract and 206 articles were screened at full text. Forty-seven of these articles met the pre-defined inclusion criteria. Results Spatially explicit data on illegal hunting are available for 29 species in 19 of the 46 countries that constitute Sub-Saharan Africa. Data collection methods include GPS and radio tracking, bushmeat household and market surveys, data from anti-poaching patrols, hunting follows and first-hand monitoring of poaching signs via line transects, audio and aerial surveys. Most studies have been conducted in a single protected area exploring spatial patterns in illegal hunting with respect to the surrounding land. Several spatial biases were detected. Conclusions There is a considerable lack of systematically collected quantitative data showing the distribution of illegal hunting incidences and few comparative studies between different tenure areas. The majority of studies have been conducted in a single protected area looking at hunting on a gradient to surrounding village land. From the studies included in the map it is evident there are spatial patterns regarding environmental and anthropogenic correlates. For example, hunting increases in proximity to transport networks (roads and railway lines), to water sources, to the border of protected areas and to village land. The influence of these spatial features could be further investigated through meta-analysis. There is a diverse range of methods in use to collect data on illicit hunting mainly drawing on pre-existing law enforcement data or researcher led surveys detecting signs of poaching. There are few longitudinal studies with most studies representing just one season of data collection and there is a geographical research bias toward Tanzania and a lack of studies in Central Africa.