Camera-traps have revolutionized the way ecologists monitor biodiversity and population abundances. Their full potential is however only realized when the hundreds of thousands of images collected can be rapidly classified with minimal human intervention. Machine learning approaches, and in particular deep learning methods, have allowed extraordinary progress towards this end. Trained classification models remain rare however, and for instance are only emerging for the European fauna. This can be explained by the technical expertise they require but also by the limited availability of large datasets of annotated pictures, which are key to obtaining successful recognition models.In this context, we set-up the DeepFaune initiative (https://deepfaune.cnrs.fr), a large-scale collaboration between dozens of partners involved in research, conservation and management of wildlife in France. The aim of DeepFaune is to aggregate individual datasets of annotated pictures to train species classification models based on convolutional neural networks, an established deeplearning approach.Here we report on our first milestone, a two-step pipeline built upon the MegaDetector algorithm for detection (discarding empty pictures and cropping the animal) and a classification model for 18 species or higher-level taxa as well as people and vehicles. The classification model achieved 92% validation accuracy and showed > 90% sensitivity and specificity for many classes. Most importantly, these performances were generally conserved when tested on an independent out-of-sample dataset. In addition, we developed a cross-platform graphical-user-interface that allows running the pipeline on images stored locally on a personal computer.In conclusion, the DeepFaune initiative provides a freely available (for non-commercial purposes) toolbox with high performance to classify the French fauna in camera-trap images.

In a context of changing carnivore populations worldwide, it is crucial to understand the consequences of these changes for prey populations. The recolonization by wolves of the French Vercors mountain range and the long‐term monitoring (2001–2017) of roe deer in this area provided a unique opportunity to assess the effects of wolves on this prey. Roe deer was the main prey of wolves in the west Vercors mountain range during this recolonization. We compared roe deer abundance and fawn body mass in two contrasted areas of a wolf pack territory: a central area (core of the territory characterized by an intense use by wolves) and a peripheral area (used more occasionally). Roe deer population growth rates were lower in the central area between 2001 and 2006, resulting in a decline in roe deer abundance. Roe deer abundance substantially dropped in the two study areas after an extremely severe winter but the abundance of roe deer in the central area facing with wolves was slower to recover and remained at lower abundance levels for 6 years. Fawn body mass was consistently lower in the central area, varied similarly as roe deer abundance, and was not influenced by weather conditions or red deer population abundance. Altogether, the effects of wolves on roe deer in the central area occurred during a 10‐year period following the establishment of wolves, through the interplay between wolf predation (before wolves started preying on red deer), harsh winter conditions and possibly naivety of prey to this recolonizing predator.

22 started preying on red deer), harsh winter conditions and naïveté of prey to this recolonizing 38 predator. 39 40