arXiv subjects

Olivier Gimenez

Publications and source records attributed to Olivier Gimenez.

11 recordsLinked to original sources

arXiv2026

Bayesian decision theory for wildlife management under uncertainty: from inference to action

Ecologists are increasingly expected to inform management decisions under uncertainty, yet most analytical workflows stop at statistical inference. Bayesian decision theory provides a coherent framework to bridge this gap by propagating posterior uncertainty to evaluate alternative actions through utility functions, but remains underused in ecology. Here, we present a practical workflow for implementing Bayesian decision theory using standard Bayesian tools, illustrated with two case studies: wolf management in France, where the decision is the number of wolves to remove under uncertain population dynamics, and invasive muskrat management in the Netherlands, where control effort is allocated across space. In both cases, expected utility integrates posterior uncertainty and management trade-offs. Optimal decisions emerge as compromises between competing objectives. For wolves, optimal harvest balances removal benefits and population risk. For muskrats, optimal effort increases with the importance of population reduction and is unevenly allocated across provinces. Bayesian decision theory provides a formal interface between scientists, who characterize ecological systems and uncertainty, and decision-makers, who define objectives, values and trade-offs. By making these trade-offs explicit, it enhances transparency and relevance for management. It also provides a common framework for bringing together Bayesian statistics, decision analysis and risk analysis, strengthening the link between ecological inference and action.

stat.AP↗

arXiv2026

A process-based dynamic occupancy model to study range dynamics under non-equilibrium conditions

Failing to account for ecological processes such as dispersal and connectivity when modeling distributions can lead to biased inference about environmental drivers and reduced predictive performance. Spatial dynamic occupancy models are promising to study range dynamics while accounting for dispersal and connectivity, but they currently rely on restrictive formulations of the colonization process, and computational constraints prevent their application at large spatial scales. Here, we propose a process-based dynamic occupancy model to study the distribution of range-expanding species while accounting for connectivity and effects of the environment. We introduce a formulation based on dispersal-pressure that provides a flexible and ecologically interpretable representation of the colonization process, and develop a computational approach based on sparse distance matrices that enables its application to national and transnational scales. We conducted a simulation study that showed unbiased parameter estimation across various ecological scenarios. We also applied our model to two range-expanding carnivores offering complementary insights: the grey wolf and the Eurasian otter. Our model revealed contrasting colonization dynamic, with wolves primarily constrained by altitude and forest cover while otters where only marginally affected by the environment, suggesting that their distribution is limited by dispersal history rather than habitat preferences. By explicitly disentangling the influence of dispersal and environment on distributions, our model provides better insight into occupancy-environment relationships under non-equilibrium conditions, and help identifies what limits species distributions. In light of the increasing availability of large-scale biodiversity data, our framework offers opportunities to study range dynamics using mechanistic approaches across entire landscapes.

q-bio.PE↗

arXiv2026

Estimating wolf population size in France using non-invasive genetic sampling and spatial capture recapture models

Population size is a key metric for management and conservation. This is especially true for large carnivore populations for which management decisions are often based on population size estimates. In France, gray wolves (Canis lupus) have been monitored for more than two decades using non-invasive genetic sampling and capture-recapture models. Population size estimates directly inform the annual number of wolves that can be killed legally. It is therefore key to use appropriate methods to obtain robust population size estimates. To track the recent numerical and geographical expansion of the population, a substantial increase in sample collection was performed during the winter 2023/24 within the entire wolf distribution range in France. A total of 1964 samples were genotyped and assigned to 576 different individuals using microsatellites genetic markers. During the winter 2023/24, spatial capture-recapture models estimated the wolf population size in France to be likely between 920 and 1125 individuals (95% credible interval). Detection probability varied spatially and was positively influenced by snow cover and accessibility. Wolf density was strongly associated with the recent presence of the species, reflecting the ongoing recolonization process from the Alps. This work illustrates the usefulness of non-invasive genetic data and spatial capture-recapture for large-scale population assessment. It also lays the ground for future improvements in monitoring to fully exploit the potential of spatial capture-recapture models.

q-bio.PE↗

arXiv2025

Estimating invasive rodent abundance using removal data and hierarchical models

Invasive rodents pose significant ecological, economic, and public health challenges. Robust methods are needed for estimating population abundance to guide effective management. Traditional methods such as capture-recapture are often impractical for invasive species due to ethical, legal and logistical constraints. Here, I showcase the application of hierarchical multinomial N-mixture models for estimating the abundance of invasive rodents using removal data. First, I perform a simulation study which demonstrates minimal bias, as well as good precision and reliable coverage of confidence intervals across a range of sampling scenarios. I also illustrate the consequences of violating the population closure assumption, showing how between-occasion dynamics can bias inference. Second, I analyze removal data for two invasive rodent species, namely coypus (Myocastor coypus) in France and muskrats (Ondatra zibethicus) in the Netherlands. Using hierarchical multinomial N-mixture models, I examine the effects of temperature on abundance while accounting for imperfect and time-varying capture probabilities. I also show how to accommodate spatial variability using random effects, quantify uncertainty in parameter estimates, and account for violations of closure by fitting an open-population model to multi-year data. Overall, I hope to demonstrate the flexibility and utility of hierarchical models in invasive species management.

q-bio.PE↗

arXiv2024

Spatial occupancy models for data collected on stream networks

To effectively monitor biodiversity in streams and rivers, we need to quantify species distribution accurately. Occupancy models are useful for distinguishing between the non-detection of a species and its actual absence. While these models can account for spatial autocorrelation, they are not suited for streams and rivers due to their unique network spatial structure. Here, I propose spatial occupancy models specifically designed for data collected on stream and river networks. I present the statistical developments and illustrate their application using data on a semi-aquatic mammal. Overall, spatial stream network occupancy models offer a robust method for assessing biodiversity in freshwater ecosystems.

stat.AP↗

arXiv2023

Nine tips for ecologists using machine learning

Due to their high predictive performance and flexibility, machine learning models are an appropriate and efficient tool for ecologists. However, implementing a machine learning model is not yet a trivial task and may seem intimidating to ecologists with no previous experience in this area. Here we provide a series of tips to help ecologists in implementing machine learning models. We focus on classification problems as many ecological studies aim to assign data into predefined classes such as ecological states or biological entities. Each of the nine tips identifies a common error, trap or challenge in developing machine learning models and provides recommendations to facilitate their use in ecological studies.

q-bio.PE↗

arXiv2021

Trade-off between deep learning for species identification and inference about predator-prey co-occurrence: Reproducible R workflow integrating models in computer vision and ecological statistics

Deep learning is used in computer vision problems with important applications in several scientific fields. In ecology for example, there is a growing interest in deep learning for automatizing repetitive analyses on large amounts of images, such as animal species identification. However, there are challenging issues toward the wide adoption of deep learning by the community of ecologists. First, there is a programming barrier as most algorithms are written in Python while most ecologists are versed in R. Second, recent applications of deep learning in ecology have focused on computational aspects and simple tasks without addressing the underlying ecological questions or carrying out the statistical data analysis to answer these questions. Here, we showcase a reproducible R workflow integrating both deep learning and statistical models using predator-prey relationships as a case study. We illustrate deep learning for the identification of animal species on images collected with camera traps, and quantify spatial co-occurrence using multispecies occupancy models. Despite average model classification performances, ecological inference was similar whether we analysed the ground truth dataset or the classified dataset. This result calls for further work on the trade-offs between time and resources allocated to train models with deep learning and our ability to properly address key ecological questions with biodiversity monitoring. We hope that our reproducible workflow will be useful to ecologists and applied statisticians. All material (source of the Rmarkdown notebook and auxiliary files) is available from https://github.com/oliviergimenez/computo-deeplearning-occupany-lynx.

stat.AP↗

arXiv2021

Fitting stochastic predator-prey models using both population density and kill rate data

Most mechanistic predator-prey modelling has involved either parameterization from process rate data or inverse modelling. Here, we take a median road: we aim at identifying the potential benefits of combining datasets, when both population growth and predation processes are viewed as stochastic. We fit a discrete-time, stochastic predator-prey model of the Leslie type to simulated time series of densities and kill rate data. Our model has both environmental stochasticity in the growth rates and interaction stochasticity, i.e., a stochastic functional response. We examine what the kill rate data brings to the quality of the estimates, and whether estimation is possible (for various time series lengths) solely with time series of population counts or biomass data. Both Bayesian and frequentist estimation are performed, providing multiple ways to check model identifiability. The Fisher Information Matrix suggests that models with and without kill rate data are all identifiable, although correlations remain between parameters that belong to the same functional form. However, our results show that if the attractor is a fixed point in the absence of stochasticity, identifying parameters in practice requires kill rate data as a complement to the time series of population densities, due to the relatively flat likelihood. Only noisy limit cycle attractors can be identified directly from population count data (as in inverse modelling), although even in this case, adding kill rate data - including in small amounts - can make the estimates much more precise. Overall, we show that under process stochasticity in interaction rates, interaction data might be essential to obtain identifiable dynamical models for multiple species. These results may extend to other biotic interactions than predation, for which similar models combining interaction rates and population counts could be developed.

q-bio.PE↗

arXiv2020

Bayesian Non-Parametric Detection Heterogeneity in Ecological Models

Detection heterogeneity is inherent to ecological data, arising from factors such as varied terrain or weather conditions, inconsistent sampling effort, or heterogeneity of individuals themselves. Incorporating additional covariates into a statistical model is one approach for addressing heterogeneity, but is no guarantee that any set of measurable covariates will adequately address the heterogeneity, and the presence of unmodelled heterogeneity has been shown to produce biases in the resulting inferences. Other approaches for addressing heterogeneity include the use of random effects, or finite mixtures of homogeneous subgroups. Here, we present a non-parametric approach for modelling detection heterogeneity for use in a Bayesian hierarchical framework. We employ a Dirichlet process mixture which allows a flexible number of population subgroups without the need to pre-specify this number of subgroups as in a finite mixture. We describe this non-parametric approach, then consider its use for modelling detection heterogeneity in two common ecological motifs: capture-recapture and occupancy modelling. For each, we consider a homogeneous model, finite mixture models, and the non-parametric approach. We compare these approaches using two simulation studies, and observe the non-parametric approach as the most reliable method for addressing varying degrees of heterogeneity. We also present two real-data examples, and compare the inferences resulting from each modelling approach. Analyses are carried out using the \texttt{nimble} package for \texttt{R}, which provides facilities for Bayesian non-parametric models.

stat.AP↗

arXiv2020

Uncovering ecological state dynamics with hidden Markov models

Ecological systems can often be characterised by changes among a finite set of underlying states pertaining to individuals, populations, communities, or entire ecosystems through time. Owing to the inherent difficulty of empirical field studies, ecological state dynamics operating at any level of this hierarchy can often be unobservable or "hidden". Ecologists must therefore often contend with incomplete or indirect observations that are somehow related to these underlying processes. By formally disentangling state and observation processes based on simple yet powerful mathematical properties that can be used to describe many ecological phenomena, hidden Markov models (HMMs) can facilitate inferences about complex system state dynamics that might otherwise be intractable. However, while HMMs are routinely applied in other disciplines, they have only recently begun to gain traction within the broader ecological community. We provide a gentle introduction to HMMs, establish some common terminology, and review the immense scope of HMMs for applied ecological research. We also provide a supplemental tutorial on some of the more technical aspects of HMM implementation and interpretation. By illustrating how practitioners can use a simple conceptual template to customise HMMs for their specific systems of interest, revealing methodological links between existing applications, and highlighting some practical considerations and limitations of these approaches, our goal is to help establish HMMs as a fundamental inferential tool for ecologists.

q-bio.QM↗

arXiv2020

A mechanistic-statistical species distribution model to explain and forecast wolf (Canis lupus) colonization in South-Eastern France

Species distribution models (SDMs) are important statistical tools for ecologists to understand and predict species range. However, standard SDMs do not explicitly incorporate dynamic processes like dispersal. This limitation may lead to bias in inference about species distribution. Here, we adopt the theory of ecological diffusion that has recently been introduced in statistical ecology to incorporate spatio-temporal processes in ecological models. As a case study, we considered the wolf (Canis lupus) that has been recolonizing Eastern France naturally through dispersal from the Apennines since the early 90's. Using partial differential equations for modelling species diffusion and growth in a fragmented landscape, we develop a mechanistic-statistical spatio-temporal model accounting for ecological diffusion, logistic growth and imperfect species detection. We conduct a simulation study and show the ability of our model to i) estimate ecological parameters in various situations with contrasted species detection probability and number of surveyed sites and ii) forecast the distribution into the future. We found that the growth rate of the wolf population in France was explained by the proportion of forest cover, that diffusion was influenced by human density and that species detectability increased with increasing survey effort. Using the parameters estimated from the 2007-2015 period, we then forecasted wolf distribution in 2016 and found good agreement with the actual detections made that year. Our approach may be useful for managing species that interact with human activities to anticipate potential conflicts.

q-bio.PE↗