Loïc Brun

Lethal wildlife management operations are typically evaluated through their direct effects on target species, yet their disturbance footprint may extend to non-target animals sharing the same landscape. We opportunistically used a large-scale helicopter-assisted culling event targeting ungulates at Mawana Game Reserve, South Africa, as a natural experiment to examine how such an operation affected the behaviour and space use of non-target vervet monkeys ( Chlorocebus pygerythrus ). Using GPS and tri-axial accelerometry from 21 collared individuals across seven habituated social groups, we quantified behavioural and spatial responses around the culling event, including changes in activity budgets, circadian activity patterns, short-term walking and running responses, daily space use, and habitat composition. Vervets showed fear-consistent responses that differed among behaviours, social groups, and temporal scales. Foraging declined at the population level, with the strongest reduction occurring in the morning, when helicopter and shooting activity were the most intense. Inactivity increased during culling and remained elevated afterwards, suggesting both disturbance-related change and seasonal background effects, while social behaviour showed a transient increase consistent with a possible social-buffering response. Walking and running showed little consistent change in daily averages across the culling period, but finer-scale analyses revealed short-lived deviations concentrated in specific groups and time windows. Spatially, groups remained within existing ranges but showed divergent responses, from contraction and increased use of dense vegetation to greater movement in more open areas, highlighting how population-level inference can obscure context-dependent responses when heterogeneity among groups, individuals, or habitats is not explicitly considered. Together, our findings show that lethal management can generate multiscale, context-dependent responses in non-target wildlife, and that overlooking these effects may lead to systematic underestimation of their ecological consequences.

Although the use of accelerometer-based behavioural classification to quantify animal activity budgets is gaining widespread traction, the interactions between key preprocessing decisions and modern classification algorithms remain poorly understood. Moreover, classification pipelines are commonly assessed using global performance metrics, despite increasing evidence that such metrics poorly reflect behaviour-specific patterns and ecological reliability. Using a free-ranging primate ( Chlorocebus pygerythrus ) as a case study, we benchmarked how temporal segmentation (burst length), collar orientation correction, and model architecture jointly shape behavioural inference. We compared nine supervised algorithms spanning classical machine learning, feature-based deep learning including a tabular foundation model (TabPFN), and state of the art time-series architectures (HydraMultiROCKET). Beyond conventional metrics, performance was further evaluated using ecological validation against independent focal observations to assess model stability and biological plausibility. Model architecture exerted the strongest influence on classification outcomes. Modern deep-learning approaches substantially outperformed classical models, doubling recall for rare behaviours (e.g., grooming, self-scratching) without compromising precision. In contrast, burst length and collar orientation correction had little effect on global metrics but produced substantial, behaviour-specific trade-offs. Shorter bursts improved the detection of rare events by increasing training instances, while orientation correction suppressed dataset-specific artifacts at the cost of degrading common behaviours. Crucially, models with similar global and behaviour-level validation metrics produced divergent predictions when applied outside the annotated context. Our findings reveal that global metrics are insufficient for optimizing behavioural inference in complex wild systems. We demonstrate that modern deep-learning architectures, such as the ROCKET family, provide a robust, accessible baseline that handles class imbalance more effectively than traditional methods. We propose that reliable inference requires behaviour-aware evaluation frameworks that integrate ecological validation, and advocate for ensemble or hierarchical strategies to leverage the complementary strengths of different preprocessing and modelling configurations.

Facing rapid environmental changes and anthropogenic habitat destruction, animal behavioural plasticity becomes an adaptive potential that needs to be considered in conservation strategies along with, for example, genetic diversity. Here, we evaluate to what extent non‐invasive environmental DNA (eDNA) methods may contribute to the assessment of intraspecies behavioural plasticity in terms of foraging behaviour. We analysed DNA metabarcoding data for plant components in the diet of four neighbouring groups of wild vervet monkeys Chlorocebus pygerythrus to identify intergroup variation (IGV). The faecal samples considered for the analyses were limited to the summer season to minimise the impact of seasonality. Each sample was attributed by observation to individuals with known life history data. A plant survey was conducted in each group home range during the study period to account for environmental variation. We observed mixed results when testing whether IGV in plant consumption was greater than intragroup variation, indicating that the influence of social dynamics must be considered. Intragroup variation was positively correlated with group size. We observed IGV in diet composition among all groups as well as in some pairwise comparisons. We found significant dietary differences between two group pairs when considering only adult females. Lastly, we observed IGV in foraging of specific plants that were not explained by their distribution, suggesting behavioural differences in selectivity between groups. Our study system and organism, being a highly social and non‐threatened primate species, with constant gene flow and overlapping territories between groups, provides an ideal model to evaluate the usage of eDNA‐based methods to better understand the impact of social factors on IGV. Our results highlight the need to consider social and demographic factors, the impact of which remains complicated to disentangle from environmental factors. However, we emphasise the great potential for studying social groups using eDNA and that such studies are needed to better understand intraspecific behavioural plasticity in wild populations.

Assessing the diet of wild animals reveals valuable information about their ecology and trophic relationships that may help elucidate dynamic interactions in ecosystems and forecast responses to environmental changes. Advances in molecular biology provide valuable research tools in this field. However, comparative empirical research is still required to highlight strengths and potential biases of different approaches. Therefore, this study compares environmental DNA and observational methods for the same study population and sampling duration. We employed DNA metabarcoding assays targeting plant and arthropod diet items in 823 fecal samples collected over 12 months in a wild population of an omnivorous primate, the vervet monkey (Chlorocebus pygerythrus). DNA metabarcoding data were subsequently compared to direct observations. We observed the same seasonal patterns of plant consumption with both methods; however, DNA metabarcoding showed considerably greater taxonomic coverage and resolution compared to observations, mostly due to the construction of a local plant DNA database. We found a strong effect of season on variation in plant consumption largely shaped by the dry and wet seasons. The seasonal effect on arthropod consumption was weaker, but feeding on arthropods was more frequent in spring and summer, showing overall that vervets adapt their diet according to available resources. The DNA metabarcoding assay outperformed also direct observations of arthropod consumption in both taxonomic coverage and resolution. Combining traditional techniques and DNA metabarcoding data can therefore not only provide enhanced assessments of complex diets and trophic interactions to the benefit of wildlife conservationists and managers but also opens new perspectives for behavioral ecologists studying whether diet variation in social species is induced by environmental differences or might reflect selective foraging behaviors.

Most vertebrates host a wide variety of haematophagous parasites, which may play an important role in the transmission of vector-borne microorganisms to hosts. Surveillance is usually performed by collecting blood and/or tissue samples from vertebrate hosts. There are multiple methods to obtain samples, which can be stored for decades if properly kept. However, blood sampling is considered an invasive method and may possibly be harmful to the sampled individual. In this study, we investigated the use of ectoparasites as a tool to acquire molecular information about the presence and diversity of infectious microorganism in host populations. We tested the presence of three distinct vector-borne microorganisms in both bat blood and bat flies: Bartonella bacteria, malaria-like Polychromophilus sp. (Apicomplexa), and Trypanosoma sp. (Kinetoplastea). We detected the presence of these microorganisms both in bats and in their bat flies, with the exception of Trypanosoma sp. in South African bat flies. Additionally, we found Bartonella sp. in bat flies from one population in Spain, suggesting its presence in the host population even if not detected in bats. Bartonella and Polychromophilus infection showed the highest prevalence in both bat and bat fly populations. Single, co- and triple infections were also frequently present in both. We highlight the use of haematophagous ectoparasites to study the presence of infectious microorganism in host blood and its use as an alternative, less invasive sampling method.