Fear of the dark? Contrasting impacts of humans versus lynx on diel activity of roe deer across Europe.

Humans, as super predators, can have strong effects on wildlife behaviour, including profound modifications of diel activity patterns. Subsequent to the return of large carnivores to human-modified ecosystems, many prey species have adjusted their spatial behaviour to the contrasting landscapes of fear generated by both their natural predators and anthropogenic pressures. The effects of predation risk on temporal shifts in diel activity of prey, however, remain largely unexplored in human-dominated landscapes. We investigated the influence of the density of lynx Lynx lynx, a nocturnal predator, on the diel activity patterns of their main prey, the roe deer Capreolus capreolus, across a gradient of human disturbance and hunting at the European scale. Based on 11 million activity records from 431 individually GPS-monitored roe deer in 12 populations within the EURODEER network (http://eurodeer.org), we investigated how lynx predation risk in combination with both lethal and non-lethal human activities affected the diurnality of deer. We demonstrated marked plasticity in roe deer diel activity patterns in response to spatio-temporal variations in risk, mostly due to human activities. In particular, roe deer decreased their level of diurnality by a factor of 1.37 when the background level of general human disturbance was high. Hunting exacerbated this effect, as during the hunting season deer switched most of their activity to night-time and, to a lesser extent, to dawn, although this pattern varied noticeably in relation to lynx density. Indeed, in the presence of lynx, their main natural predator, roe deer were relatively more diurnal. Overall, our results revealed a strong influence of human activities and the presence of lynx on diel shifts in roe deer activity. In the context of the recovery of large carnivores across Europe, we provide important insights about the effects of predators on the behavioural responses of their prey in human-dominated ecosystems. Modifications in the temporal partitioning of ungulate activity as a response to human activities may facilitate human-wildlife coexistence, but likely also have knock-on effects for predator-prey interactions, with cascading effects on ecosystem functioning.

Résumé Les humains, en tant que 'super-prédateurs', peuvent avoir des effets importants sur le comportement de la faune sauvage, y compris des modifications profondes de leurs rythmes circadiens d'activité. A la suite du retour des grands carnivores dans les écosystèmes anthropisés, de nombreuses espèces proies ont ajusté leur comportement spatial à ces paysages de la peur contrastés, générés à la fois par les pressions liées aux risques anthropiques et à la présence de leurs prédateurs naturels. Les effets du risque de prédation sur les modifications temporelles des rythmes circadiens d'activité des proies restent cependant largement inconnus dans les écosystèmes dominés par l'homme. Ici, nous avons étudié l'influence de la densité de lynx Lynx lynx, un prédateur nocturne, sur les rythmes circadiens d'activité de leur proie principale, le chevreuil Capreolus capreolus, à travers un gradient de pressions anthropiques à l'échelle Européenne. Sur la base de plus de 11 million de données d'activité issues de 431 suivis individuels de chevreuils équipés de colliers GPS provenant de 12 populations au sein du réseau EURODEER (http://eurodeer.org), nous avons analysé comment le risque de prédation par le lynx, associé aux risques létaux et non-létaux des activités humaines, influence la diurnalité des chevreuils. Nous avons démontré une forte plasticité des rythmes circadiens d'activité des chevreuils en réponse aux variations spatio-temporelles du risque, et notamment face aux activités humaines. Plus particulièrement, les chevreuils diminuent leur degré de diurnalité d'un facteur de 1.37 lorsque le dérangement humain est important. La chasse accentue cet effet, puisque durant la saison de chasse les chevreuils basculent la plupart de leur activité de nuit, et dans une moindre mesure, durant l'aube également, bien que ce patron soit essentiellement variable en fonction de la densité de lynx. En effet, en présence de lynx, leur principal prédateur, les chevreuils sont relativement plus diurnes. Globalement, nos résultats révèlent une forte influence des activités humaines et de la présence de lynx sur l'ajustement des rythmes circadiens d'activité des chevreuils. Dans le contexte du retour des grands carnivores en Europe, notre étude apporte de nouvelles connaissances sur les effets des prédateurs sur la réponse comportementale de leur proie dans des écosystèmes anthropisés. La modification de la répartition temporelle de l'activité des ongulés en réponse aux activités humaines pourrait être un facteur facilitant la coexistence homme-faune sauvage, avec toutefois des conséquences autres sur les interactions prédateurs-proies et leurs effets en cascade sur le fonctionnement des écosystèmes.

An evaluation of machine learning classifiers for next-generation, continuous-ethogram smart trackers.

BACKGROUND: Our understanding of movement patterns and behaviours of wildlife has advanced greatly through the use of improved tracking technologies, including application of accelerometry (ACC) across a wide range of taxa. However, most ACC studies either use intermittent sampling that hinders continuity or continuous data logging relying on tracker retrieval for data downloading which is not applicable for long term study. To allow long-term, fine-scale behavioural research, we evaluated a range of machine learning methods for their suitability for continuous on-board classification of ACC data into behaviour categories prior to data transmission. METHODS: We tested six supervised machine learning methods, including linear discriminant analysis (LDA), decision tree (DT), support vector machine (SVM), artificial neural network (ANN), random forest (RF) and extreme gradient boosting (XGBoost) to classify behaviour using ACC data from three bird species (white stork Ciconia ciconia, griffon vulture Gyps fulvus and common crane Grus grus) and two mammals (dairy cow Bos taurus and roe deer Capreolus capreolus). RESULTS: Using a range of quality criteria, SVM, ANN, RF and XGBoost performed well in determining behaviour from ACC data and their good performance appeared little affected when greatly reducing the number of input features for model training. On-board runtime and storage-requirement tests showed that notably ANN, RF and XGBoost would make suitable on-board classifiers. CONCLUSIONS: Our identification of using feature reduction in combination with ANN, RF and XGBoost as suitable methods for on-board behavioural classification of continuous ACC data has considerable potential to benefit movement ecology and behavioural research, wildlife conservation and livestock husbandry.

Settle Down! Ranging Behaviour Responses of Roe Deer to Different Capture and Release Methods.

The fitting of tracking devices to wild animals requires capture and handling which causes stress and can potentially cause injury, behavioural modifications that can affect animal welfare and the output of research. We evaluated post capture and release ranging behaviour responses of roe deer (Capreolus capreolus) for five different capture methods. We analysed the distance from the centre of gravity and between successive locations, using data from 14 different study sites within the EURODEER collaborative project. Independently of the capture method, we observed a shorter distance between successive locations and contextual shift away from the home range centre of gravity after the capture and release event. However, individuals converged towards the average behaviour within a relatively short space of time (between 10 days and one month). If researchers investigate questions based on the distance between successive locations of the home range, we recommend (1) initial investigation to establish when the animals start to behave normally again or (2) not using the first two to three weeks of data for their analysis. We also encourage researchers to continually adapt methods to minimize stress and prioritize animal welfare wherever possible, according to the Refinement of the Three R's.

Wave-like Patterns of Plant Phenology Determine Ungulate Movement Tactics.

Animals exhibit a diversity of movement tactics [1]. Tracking resources that change across space and time is predicted to be a fundamental driver of animal movement [2]. For example, some migratory ungulates (i.e., hooved mammals) closely track the progression of highly nutritious plant green-up, a phenomenon called "green-wave surfing" [3-5]. Yet general principles describing how the dynamic nature of resources determine movement tactics are lacking [6]. We tested an emerging theory that predicts surfing and the existence of migratory behavior will be favored in environments where green-up is fleeting and moves sequentially across large landscapes (i.e., wave-like green-up) [7]. Landscapes exhibiting wave-like patterns of green-up facilitated surfing and explained the existence of migratory behavior across 61 populations of four ungulate species on two continents (n = 1,696 individuals). At the species level, foraging benefits were equivalent between tactics, suggesting that each movement tactic is fine-tuned to local patterns of plant phenology. For decades, ecologists have sought to understand how animals move to select habitat, commonly defining habitat as a set of static patches [8, 9]. Our findings indicate that animal movement tactics emerge as a function of the flux of resources across space and time, underscoring the need to redefine habitat to include its dynamic attributes. As global habitats continue to be modified by anthropogenic disturbance and climate change [10], our synthesis provides a generalizable framework to understand how animal movement will be influenced by altered patterns of resource phenology.

Temporal patterns in road crossing behaviour in roe deer (Capreolus capreolus) at sites with wildlife warning reflectors.

Every year, there are millions of documented vehicle collisions involving cervids across Europe and North America. While temporal patterns in collision occurrence are relatively well described, few studies have targeted deer behaviour as a critical component of collision prevention. In this study, we investigated weekly and daily patterns in road crossing behaviour in roe deer. Using road crossing events and movement data obtained from GPS telemetry, we employed mixed-effect models to explain frequency and timing of crossings at five road segments by a number of predictors including traffic volume, deer movement activity and the presence of wildlife warning reflectors. We analysed 13,689 road crossing events by 32 study animals. Individual variation in crossing frequency was high but daily patterns in crossing events were highly consistent among animals. Variation in the intensity of movement activity on a daily and seasonal scale was the main driver of road crossing behaviour. The seasonal variation in crossing frequency reflected differences in movement activity throughout the reproductive cycle, while daily variation in the probability to cross exhibited a clear nocturnal emphasis and reflected crepuscular activity peaks. The frequency of road crossings increased as a function of road density in the home-range, while traffic volume only exerted marginal effects. Movement activity of roe deer in our study coincided with commuter traffic mainly in the early morning and late afternoon during winter and during periods of high spatial activity such as the rut. Both timing and frequency of crossing events remained unchanged in the presence of reflectors. Our results emphasise the importance of behavioural studies for understanding roe deer vehicle-collision patterns and thus provide important information for collision prevention. We suggest that mitigation of collision risk should focus on strategic seasonal measures and animal warning systems targeting drivers.

Synchrony in hunting bags: reaction on climatic and human induced changes?

Human induced land use changes negatively impact the viability of many wildlife species through habitat modifications and mortality, while some species seem to benefit from it. Roe deer (Capreolus capreolus), a wide spread ungulate increased both its abundance and range throughout Europe. This pattern is also reflected in the increasing hunting bags over the last 40 years. Such a development raises questions about the relationship between human hunting and population dynamics and, in particular, about the potential of human hunting to control related populations. We analysed and reconstructed annual hunting bags of roe deer for three federal states of northern Germany, Brandenburg, Lower Saxony and Mecklenburg West Pomerania for the years 1972 to 2011. Since 1992 the hunting bags from these three states are significantly higher than those reported for the years 1972-1991. Our reconstruction takes into consideration effects of climate variability, expressed by inter-annual changes in the North Atlantic Oscillation and impacts from rapeseed and wheat cultivation. We found that severe winters, which are indicated by negative values of the North Atlantic Oscillation during the months December-March, directly, or with a time lag of two years affect the number of deer shot. In contrast, an increase in the area used for rapeseed cultivation coincides with higher numbers of roe deer shot, with respect to the overall mean value. Consequently, we recommend that wildlife management addresses changes in large scale processes including land use pattern and climate variability.