Biology subjects

Miele, V.

Publications and source records attributed to Miele, V..

11 recordsLinked to original sources

bioRxiv2026

Six decades of change across all North American bird interaction networks

Species interactions are essential for maintaining biodiversity in the face of global changes. Yet lack of interaction data has hindered progress in understanding spatiotemporal changes in ecological network structure and function. To address this shortfall, we perform network analysis on a continental-scale metanetwork of bird interactions (AvianMetaNetwork), and quantify shifts in trophic, commensalism, parasitism, competition, mobbing, and facilitation networks using 13762 interactions among 687 North American birds. Since 1970, changes in species abundance and human activity drove network shifts, especially in eastern North America. With this metanetwork, we uncovered more than a half century of previously undocumented shifts in the structure of each avian network at a continental scale, with large implications for the functioning of bird communities and the ecosystems that rely on them.

ecology↗

bioRxiv2025

Paying Attention to Other Animal Detections Improves Camera Trap Classification Models

O_LIIn ecological studies, automated species classification models are increasingly used to process large volumes of camera trap images. Most current classification models rely on a two-step pipeline: a detector first locates and crops animals, followed by a classifier that predicts species independently for each crop. While effective, these models still struggle under challenging conditions and ignore temporal context or information from nearby animals available in sequences of camera trap images, whereas human annotators often use it in difficult cases. C_LIO_LIWe propose to leverage self-attention, a core mechanism of Large Language Models and Vision Transformers, to enable the model to learn relationships between crops occurring in similar contexts. Our self-attention module operates directly on the set of crop embeddings, producing new representations enriched with information from other crops, improving species classification. The module fits into the two-step pipeline without requiring structural change and adds only minimal computational overhead. To address the lack of annotated multi-species sequences, we develop a training strategy that synthetically generates multi-species sequences from monospecific ones. C_LIO_LICompared to an independent classifier baseline, our multi-crop model achieves higher accuracy on mono-specific sequences, both real and synthetic. In multispecies settings, evaluated using synthetic test sets, we also observe a substantial improvement in accuracy. Using real but weakly annotated multi-species sequences, we reformulate the task as multi-label set classification, and conduct a visual analysis, to highlight the benefits of our approach. C_LIO_LIBy leveraging the information brought by all detections of animals in the image and others of the same sequence, our approach reduces species misclassifications and enables more accurate estimates for downstream ecological analyses focusing on, for instance, species richness, occupancy, and species interaction. C_LI

ecology↗

bioRxiv2025

Limits and Promises of Earth Observation Foundation Models in Predicting Multi-Trophic Soil Biodiversity

Soil biodiversity is essential for terrestrial ecosystems, influencing nutrient cycling, carbon sequestration, agricultural productivity, and resilience to environmental changes. Yet, it faces significant threats from land-use changes, pollution, agricultural intensification, and climate change. Effective predictive modeling tools are urgently needed to inform conservation and management strategies. Although species distribution models (SDMs) have been successful for aboveground biodiversity, their application to soil biodiversity is limited by scarce large-scale datasets with spatial mismatches between environmental data and soil habitats. Recent advances, including environmental DNA (eDNA) metabarcoding, now allow extensive multi-taxa assessments of soil biodiversity. Simultaneously, remote sensing technologies provide high-resolution spatial data, potentially overcoming traditional coarse-gridded environmental limitations. This study evaluates Earth Observation Foundation (EOF) models, deep learning models pretrained on massive remote sensing datasets to summarize earth observation images into embeddings, to predict multi-trophic soil biodiversity in the French Alps. We compare models using EOF-derived embeddings from orthophotos with coarse-gridded and high-quality in-situ variables. We modeled relative abundance for 51 trophic groups across seven taxa using Random Forest, Light Gradient Boosting Machine, and Artificial Neural Networks, evaluating four data configurations: coarse-gridded environmental data, high-quality in-situ data, EOF embeddings, and a hybrid embedding-tabular approach. High-quality in-situ climate and soil data consistently delivered the highest predictive accuracy, especially for microbial and fungal groups. EOF embeddings provided valuable spatial context but did not surpass in-situ data performance, showing partial redundancy. Integrating remote sensing data can enhance biodiversity modeling in areas lacking detailed in-situ measurements, underscoring their complementary role in ecological assessments.

ecology↗

bioRxiv2024

Zero-shot animal behavior classification with image-text foundation models

1. Understanding the behavior of animals in their natural habitats is critical to ecology and conservation. Camera traps are a powerful tool to collect such data with minimal disturbance. They however produce very a large quantity of images, which can make human-based annotation cumbersome or even impossible. While automated species identification with artificial intelligence has made impressive progress, automatic classification of animal behaviors in camera trap images remains a developing field. 2. Here, we explore the potential of foundation models, specifically Vision Language Models (VLMs), to perform this task without the need to first train a model, which would require some level of human-based annotation. Using an original dataset of alpine fauna with behaviors annotated by participatory science, we investigate the zero-shot capabilities of different kind of recent VLMs to predict behaviors and estimate behavior-specific diel activity patterns in three ungulate species. 3. Our results show that using these methods, it is possible to achieve accuracies over 91% in behavior classification and produce activity patterns that closely align with those derived from participatory science data (overlap indexes between 84% and 90%). 4. These findings demonstrate the potential of foundation models and vision-language models in ecological research. Ecologists are encouraged to adopt these new methods and leverage their full capabilities to facilitate ecological studies.

ecology↗

bioRxiv2023

Beyond accuracy : score calibration in deep learning models for camera trap image sequences

In ecological studies, machine learning models are increasingly being used for the automatic processing of camera trap images. Although this automation facilitates and accelerates the identification step, the results of these models may lack interpretability and their immediate applicability to ecological downstream tasks (e.g occupancy estimation) remain questionable. In particular, little is known about their calibration, a property that guarantees that confidence scores can be reliably interpreted as probabilities that a models predictions are true. Using a large and diverse European camera trap dataset, we investigate whether deep learning models for species classification in camera trap images are well calibrated, or in contrast over/under-confident. Additionally, as camera traps are often configured to take multiple photos of the same event, we also explore the calibration of predictions at the sequence level. Finally, we study the effect and the practicality of a post-hoc calibration method, i.e. temperature scaling, for predictions made at image and sequence levels. Based on five established models and three independent test sets, our findings show that, using the right methodology, it is possible to enhance the interpretability of the confidence scores, with clear implication for, for instance, the calculation of error rates or the selection of confidence score thresholds in ecological studies making use of artificial intelligence models.

ecology↗

bioRxiv2022

The DeepFaune initiative: a collaborative effort towards the automatic identification of the French fauna in camera-trap images

Camera traps have revolutionized how ecologists monitor wildlife, but their full potential is realized only when the hundreds of thousands of collected images can be readily classified with minimal human intervention. Deep-learning classification models have allowed extraordinary progress towards this end, but trained models remain rare and are only now emerging for European fauna. We report on the first milestone of the DeepFaune initiative (https://www.deepfaune.cnrs.fr), a large-scale collaboration between more than 50 partners involved in wildlife research, conservation and management in France. We developed a classification model trained to recognize 26 species or higher-level taxa. The classification model achieved 0.97 validation accuracy and often >0.95 precision and recall for many classes. These performances were generally higher than 0.90 when tested on independent out-of-sample datasets for which we used image redundancy contained in sequence of images. We implemented our model in a software to classify images stored locally on a personal computer, so as to provide a free, user-friendly and high-performance tool for wildlife practitioners to automatically classify camera-trap images.

ecology↗

bioRxiv2020

Deep learning for species identification of modern and fossil rodent molars

Reliable identification of species is a key step to assess biodiversity. In fossil and archaeological contexts, genetic identifications remain often difficult or even impossible and morphological criteria are the only window on past biodiversity. Methods of numerical taxonomy based on geometric morphometric provide reliable identifications at the specific and even intraspecific levels, but they remain relatively time consuming and require expertise on the group under study. Here, we explore an alternative based on computer vision and machine learning. The identification of three rodent species based on pictures of their molar tooth row constituted the case study. We focused on the first upper molar in order to transfer the model elaborated on modern, genetically identified specimens to isolated fossil teeth. A pipeline based on deep neural network automatically cropped the first molar from the pictures, and returned a prediction regarding species identification. The deep-learning approach performed equally good as geometric morphometrics and, provided an extensive reference dataset including fossil teeth, it was able to successfully identify teeth from an archaeological deposit that was not included in the training dataset. This is a proof-of-concept that such methods could allow fast and reliable identification of extensive amounts of fossil remains, often left unstudied in archaeological deposits for lack of time and expertise. Deep-learning methods may thus allow new insights on the biodiversity dynamics across the last 10.000 years, including the role of humans in extinction or recent evolution.

zoology↗

bioRxiv2020

Core-periphery structure in mutualistic networks: an epitaph for nestedness?

The calculation of nestedness has become a routine analysis in the study of ecological networks, as it is commonly associated with community resilience, robustness and species persistence. While meaningful in species distributional patterns, for an interaction matrix to be nested, specialist species must interact with ordered subsets of subsequently more generalized species -- not just with a lower number of species. However, after reviewing 419 papers on mutualistic networks published since nestedness was introduced for the study of species interactions in 2003, we have found that only two theoretical studies considered explicitly ordered subsets. Instead, most studies interpret nestedness as a core of densely connected generalist species, surrounded by a periphery of specialist species attached to this core -- a so-called core-periphery structure. Such a topological feature is generally perceived as a core-periphery structure in network science. Here, we argue that the concept of core-periphery may be more relevant for studies on mutualistic networks than the concept of nestedness, as ecologists are usually not interested in exploring in detail the ordered subsets that characterize nestedness but instead use nestedness to describe a topology with a core of densely linked generalist species surrounded by a sparsely linked periphery of specialists. To illustrate our arguments and the quantification of core-periphery structures, we calculate core-periphery and nestedness in a large publicly available dataset of mutualistic networks. We also describe the calculation of core-periphery structures, its relationship with nestedness, and provide the code inside the R package econetwork for its calculation in mutualistic networks. We hope that our review will help ecologists to move beyond nestedness towards a more explicit representation of the structure of ecological networks.

ecology↗

bioRxiv2020

Revisiting giraffe photo-identification using deep learning and network analysis

An increasing number of ecological monitoring programs rely on photographic capture-recapture of individuals to study distribution, demography and abundance of species. Photo-identification of individuals can sometimes be done using idiosyncratic coat or skin patterns, instead of using tags or loggers. However, when performed manually, the task of going through photographs is tedious and rapidly becomes too time consuming as the number of pictures grows. Computer vision techniques are an appealing and unavoidable help to tackle this apparently simple task in the big-data era. In this context, we propose to revisit animal re-identification using image similarity networks and metric learning with convolutional neural networks (CNNs), taking the giraffe as a working example. We first developed an end-to-end pipeline to retrieve a comprehensive set of re-identified giraffes from about 4, 000 raw photographs. To do so, we combined CNN-based object detection, SIFT pattern matching, and image similarity networks. We then quantified the performance of deep metric learning to retrieve the identity of known individuals and detect unknown individuals never seen in the previous years of monitoring. After a data augmentation procedure, the re-identification performance of the CNN reached a Top-1 accuracy of about 90%, despite the very small number of images per individual in the training data set. While the complete pipeline succeeded in re-identifying known individuals, it slightly under-performed with unknown individuals. Fully based on open-source software packages, our work paves the way for further attempts to build automatic pipelines for re-identification of individual animals, not only in giraffes but also in other species.

ecology↗

bioRxiv2019

Temporal switching of species roles in a plant-pollinator network

Mutualistic networks are highly dynamic, characterized by high temporal turnover of species and interactions. Yet, we have a limited understanding of how the internal structure of these networks and the roles species play in them vary through time. We used six years of observation data and a novel statistical method (dynamic stochastic block models) to assess how network structure and species roles change across time in a quantitative plant-pollinator network from a dryland ecosystem in Argentina. Our analyses revealed a core-periphery structure persistent through seasons and years. Yet, species roles as core or peripheral were highly dynamic: virtually all species that played a core role in some seasons were also peripheral in other seasons, while many other species remained always peripheral. Our results illuminate our understanding of the dynamics of ecological networks and have important implications for ecosystem management and conservation.

ecology↗

bioRxiv2018

The diversity of interaction types drives the functioning of ecological communities

Ecological communities are undeniably diverse, both in terms of the species that compose them as well as the type of interactions that link species to each other. Despite this long-recognition of the coexistence of multiple interaction types in nature, little is known about the consequences of this diversity for community functioning. In the ongoing context of global change and increasing species extinction rates, it seems crucial to improve our understanding of the drivers of the relationship between species diversity and ecosystem functioning.\n\nHere, using a multispecies dynamical model of ecological communities including various interaction types (e.g. competition for space, predator interference, recruitment facilitation), we studied the role of the presence and the intensity of these interactions for species diversity, community functioning (biomass and production) and the relationship between diversity and functioning.\n\nTaken jointly, the diverse interactions have significant effects on species diversity, whose amplitude and sign depend on the type of interactions involved and their relative abundance. They however consistently increase the slope of the relationship between diversity and functioning, suggesting that species losses might have stronger effects on community functioning than expected when ignoring the diversity of interaction types and focusing on feeding interactions only.

ecology↗