Biology subjects

Milleret, C.

Publications and source records attributed to Milleret, C..

13 recordsLinked to original sources

bioRxiv2024

Unravelling the effects of individual heterogeneity on connectivity estimates: Insights from Spatial Capture-Recapture modelling

Spatial Capture-Recapture (SCR) models using least-cost path distance offer an unified framework to estimate landscape connectivity and population size from individual detection data, while accounting for individual and spatial heterogeneity in space use. In our case study on the Pyrenean brown bear population, two "outliers" individuals were detected more often over very large spatial extent compared to other individuals. The integration of such individuals made SCR modelling challenging, especially since it remains unclear how unmodelled heterogeneity in space use may bias connectivity and population size estimates. To address this gap, we used simulations reflecting the Pyrenean brown bear population, with two groups of individuals differing in their space use due to individual characteristics but also in their spatial responses to landscape structure. We compared six SCR model formulations that varied in whether individual and spatial heterogeneity in space use were (1) ignored, (2) explicitly modelled, or (3) handled by removing outliers. We then applied the same models to the empirical bear dataset. By combining simulations and sensitivity analyses, we highlight the challenges of choosing the appropriate modelling approach when multiple sources of heterogeneity in space use occur simultaneously. Our results show that the treatment of heterogeneity in space use should match the research objective: removing outliers supports accurate population size estimation, while explicitly modelling heterogeneity is essential for reliable connectivity assessment. Overall, we provide a practical framework for identifying and addressing heterogeneity in SCR applications, to guide practitioners in deciding when additional model complexity is warranted.

ecology↗

bioRxiv2022

Wolverine density distribution reflects past persecution and current management in Scandinavia

After centuries of intense persecution, several large carnivore species in Europe and North America have experienced a rebound. Todays spatial configuration of large carnivore populations has likely arisen from the interplay between their ecological traits and current environmental conditions, but also from their history of persecution and protection. Yet, due to the challenge of studying population-level phenomena, we are rarely able to disentangle and quantify the influence of past and present factors driving the spatial distribution and density of these controversial species. Using spatial capture-recapture models and a data set of 742 genetically identified wolverines Gulo gulo collected over [1/2] million km2 across their entire range in Norway and Sweden, we identify landscape-level factors explaining the current population density of wolverines in the Scandinavian Peninsula. Distance from the relic range along the Swedish-Norwegian border, where the wolverine population survived a long history of persecution, remains a key determinant of wolverine density today. However, regional differences in management and environmental conditions also played an important role in shaping spatial patterns in present-day wolverine density. Specifically, we found evidence of slower recolonization in areas that had set lower wolverine population goals in terms of the desired number of annual reproductions. Management of transboundary large carnivore populations at biologically relevant scales may be inhibited by administrative fragmentation. Yet, as our study shows, population-level monitoring is an achievable prerequisite for a comprehensive understanding of the distribution and density of large carnivores across an increasingly anthropogenic landscape.

ecology↗

bioRxiv2022

Spatial variation in red deer density in relation to forest disturbance and ungulate management in a transboundary forest ecosystem

1. ContextForests in Europe are exposed to increasingly frequent and severe disturbances. The resulting changes in the structure and composition of forests can have profound consequences for the wildlife inhabiting them. Moreover, wildlife populations in Europe are often subjected to differential management regimes as they regularly extend across multiple national and administrative borders. The red deer (Cervus elaphus) population in the Bohemian Forest Ecosystem, straddling the Czech-German border, has experienced forest disturbances, primarily caused by windfalls and bark beetle (Ips typographus) outbreaks during the past decades. To adapt local management strategies to the changing environmental conditions and to coordinate them across the international border, reliable estimates of red deer density and abundance are highly sought-after by policymakers, wildlife managers, and stakeholders. 2. Approach and MethodsCovering a 1081-km2 study area, we conducted a transnational non-invasive DNA sampling study in 2018 that yielded 1578 genotyped DNA samples from 1120 individual red deer. Using spatial capture-recapture models, we estimated total and jurisdiction-specific abundance of red deer throughout the ecosystem and quantified the role of forest disturbance and differential management strategies in shaping spatial heterogeneity in red deer density. We hypothesized that (a) forest disturbances provide favourable habitat conditions (e.g., forage and cover), and (b) contrasting red deer management regimes in different jurisdictions create a differential risk landscape, ultimately shaping density distributions. 3. ResultsOverall, we estimated that 2851 red deer (95% Credible Intervals = 2609 - 3119) resided in the study area, with a relatively even overall sex ratio (1406 females, 1229 - 1612 and 1445 males, 1288 - 1626). The average red deer density was higher in Czechia (3.5 km-2, 1.2 - 12.3) compared to Germany (2 km-2, 0.2 - 11). The effect of forest disturbances on red deer density was context dependent. Forest disturbances had a positive effect on red deer density at higher elevations and a negative effect at lower elevations, which could be explained by partial migration and its drivers in this population. Density of red deer was generally higher in management units where hunting is prohibited. In addition, we found that sex ratios differed between administrative units and were more balanced in the non-intervention zones. 4. Synthesis and applicationsOur results show that the effect of forest disturbances on wild ungulates is modulated by additional factors, such as elevation and ungulate management practices. Overall density patterns and sex ratios suggested strong gradients in density between administrative units. With climate change increasing the severity and frequency of forest disturbances, population-level monitoring and management are becoming increasingly important especially for wide-ranging species as both wildlife and global change transcend administrative boundaries.

ecology↗

bioRxiv2022

Transnational non-invasive monitoring and spatial capture-recapture models reveal strong effects of humans on connectivity for the Pyrenean brown bear

Connectivity, in the sense of the persistence of movements between habitat patches, is key to maintain endangered populations and has to be evaluated in management plans. In practice, connectivity is difficult to quantify especially for rare and elusive species. Here, we use spatial capture-recapture (SCR) models with an ecological detection distance to identify barriers to movement. We focused on the transnational critically endangered Pyrenean brown bear (Ursus arctos) population, which is distributed over Spain, France and Andorra and is divided into two main cores areas following translocations. We integrate structured monitoring from camera traps and hair snags with opportunistic data gathered after depredation events. While structured monitoring focuses on areas of regular bear presence, the integration of opportunistic data allows us to obtain information in a wider range of habitat, which is especially important for ecological inference. By estimating a resistance parameter from encounter data, we show that the road network impedes movements, leading to smaller home ranges with increasing road density. Although the quantitative effect of roads is context-dependent (i.e. varying according to landscape configuration), our model predicts that a brown bear with a home range located in an area with relatively high road density (8.29km/km2) has a home range size reduced by 1.4-fold for males and 1.6-fold for females compared to a brown bear with a home range located in an area with low road density (1.38km/km2). When assessing connectivity, spatial capture-recapture modeling offers an alternative to the use of experts opinion when telemetry data are not available.

ecology↗

bioRxiv2022

Modelling spatially autocorrelated detection probabilities in spatial capture-recapture using random effects

Spatial capture-recapture (SCR) models are now widely used for estimating density from repeated individual spatial encounters. SCR accounts for the inherent spatial autocorrelation in individual detections by modelling detection probabilities as a function of distance between the detectors and individual activity centres. However, additional spatial heterogeneity in detection probability may still creep in due to environmental or sampling characteristics. if unaccounted for, such variation can lead to pronounced bias in population size estimates. Using simulations, we describe and test three Bayesian SCR models that use generalized linear mixed models (GLMM) to account for latent heterogeneity in baseline detection probability across detectors using: independent random effects (RE), spatially autocorrelated random effects (SARE), and a twogroup finite mixture model (FM). Overall, SARE provided the least biased population size estimates (median RB: -9 - 6%). When spatial autocorrelation was high, SARE also performed best at predicting the spatial pattern of heterogeneity in detection probability. At intermediate levels of autocorrelation, spatially-explicit estimates of detection probability obtained with FM where more accurate than those generated by SARE and RE. In cases where the number of detections per detector is realistically low (at most 1), all GLMMs considered here may require dimension reduction of the random effects by pooling baseline detection probability parameters across neighboring detectors ("aggregation") to avoid over-parameterization. The added complexity and computational overhead associated with SCR-GLMMs may only be justified in extreme cases of spatial heterogeneity. However, even in less extreme cases, detecting and estimating spatially heterogeneous detection probability may assist in planning or adjusting monitoring schemes.

ecology↗

bioRxiv2022

Estimating spatially variable and density-dependent survival using open-population spatial capture-recapture models

Open-population spatial capture-recapture (OPSCR) models use the spatial information contained in individual detections collected over multiple consecutive occasions to estimate occasion-specific density, but also demographic parameters. OPSCR models can also estimate spatial variation in vital rates, but such models are neither widely used nor thoroughly tested. We developed a Bayesian OSPCR model that not only accounts for spatial variation in survival using spatial covariates, but also estimates local density-dependent effects on survival within a unified framework. Using simulations, we show that OPSCR models provide sound inferences on the effect of spatial covariates on survival, including multiple competing sources of mortality, each with potentially different spatial determinants. Estimation of local density-dependent survival was possible but required more data due to the greater complexity of the model. Not accounting for spatial heterogeneity in survival led to positive bias in abundance estimates (up to 10% relative bias). We provide a set of features in R package nimbleSCR that allow computationally efficient fitting of Bayesian OPSCR models with spatially varying survival. The ability to make population-level inferences of spatial variation in survival is an essential step towards a fully spatially-explicit OPSCR model that can disentangle the role of multiple spatial drivers on population dynamics. Open Research statementcode to reproduce the analysis is available on github; https://github.com/Cyril-Milleret/Public/tree/master/SpatialSurvivalOPSCR

ecology↗

bioRxiv2021

Does the punishment fit the crime? Consequences and diagnosis of misspecified detection functions in Bayesian spatial capture-recapture modelling

O_LISpatial capture-recapture (SCR) is now used widely to estimate wildlife densities. At the core of SCR models lies the detection function, linking individual detection probability to the distance from its latent activity center. The most common function (half-normal) assumes a bivariate normal space use and consequently detection pattern. This is likely an oversimplification and misrepresentation of real-life animal space use patterns, but studies have reported that density estimates are relatively robust to misspecified detection functions. However, information about consequences of such misspecification on space use parameters (e.g. home range area), as well as diagnostic tools to reveal it are lacking. C_LIO_LIWe simulated SCR data under six different detection functions, including the half-normal, to represent a wide range of space use patterns. We then fit three different SCR models, with the three simplest detection functions (half-normal, exponential and half-normal plateau) to each simulated data set. We evaluated the consequences of misspecification in terms of bias, precision and coverage probability of density and home range area estimates. We also calculated Bayesian p-values with respect to different discrepancy metrics to assess whether these can help identify misspecifications of the detection function. C_LIO_LIWe corroborate previous findings that density estimates are robust to misspecifications of the detection function. However, estimates of home range area are prone to bias when the detection function is misspecified. When fitted with the half-normal model, average relative bias of 95% kernel home range area estimates ranged between -25% and 26% depending on the misspecification. In contrast, the half-normal plateau model (an extension of the half-normal) returned average relative bias that ranged between -26% and -4%. Additionally, we found useful heuristic patterns in Bayesian p-values to diagnose the misspecification in detection function. C_LIO_LIOur analytical framework and diagnostic tools may help users select a detection function when analyzing empirical data, especially when space use parameters (such as home range area) are of interest. We urge development of additional custom goodness of fit diagnostics for Bayesian SCR models to help practitioners identify a wider range of model misspecifications. C_LI

ecology↗

bioRxiv2020

Consequences of ignoring variable and spatially-autocorrelated detection probability in spatial capture-recapture

ContextSpatial capture-recapture (SCR) models are increasingly popular for analyzing wildlife monitoring data. SCR can account for spatial heterogeneity in detection that arises from individual space use (detection kernel), variation in the sampling process, and the distribution of individuals (density). However, unexplained and unmodeled spatial heterogeneity in detectability may remain due to cryptic factors, intrinsic and extrinsic to the study system. ObjectivesWe identify how the magnitude and configuration of unmodeled, spatially variable detection probability influence SCR parameter estimates. MethodsWe simulated realistic SCR data with spatially variable and autocorrelated detection probability. We then fitted a single-session SCR model ignoring this variation to the simulated data and assessed the impact of model misspecification on inferences. ResultsHighly autocorrelated spatial heterogeneity in detection probability (Morans I = 0.85 - 0.96), modulated by the magnitude of that variation, can lead to pronounced negative bias (up to 75%), reduction in precision (249%), and decreasing coverage probability of the 95% credible intervals associated with abundance estimates to 0. Conversely, at low levels of spatial autocorrelation (median Morans I = 0), even severe unmodeled heterogeneity in detection probability did not lead to pronounced bias and only caused slight reductions in precision and coverage of abundance estimates. ConclusionsUnknown and unmodeled variation in detection probability is liable to be the norm, rather than the exception, in SCR studies. We encourage practitioners to consider the impact that spatial autocorrelation in detectability has on their inferences and urge the development of SCR methods that can take structured unknown or partially unknown spatial variability in detection probability into account.

ecology↗

bioRxiv2020

A hierarchical point process model for spatial capture-recapture data

O_LISpatial capture-recapture (SCR) is a popular method for estimating the abundance and density of wildlife populations. A standard SCR model consists of two sub-models: one for the activity centers of individuals and the other for the detections of each individual conditional on its activity center. So far, the detection sub-model of most SCR models is designed for sampling situations where fixed trap arrays are used to detect individuals. C_LIO_LINon-invasive genetic sampling (NGS) is widely applied in SCR studies. Using NGS methods, one often searches the study area for potential sources of DNA such as hairs and faeces, and records the locations of these samples. To analyse such data with SCR models, investigators usually impose an artificial detector grid and project detections to the nearest detector. However, there is a trade-off between the computational efficiency (fewer detectors) and the spatial accuracy (more detectors) when using this method. C_LIO_LIHere, we propose a point process model for the detection process of SCR studies using NGS. The model better reflects the spatially continuous detection process and allows all spatial information in the data to be used without approximation error. As in many SCR models, we also use a point process model for the activity centers of individuals. The resulting hierarchical point process model enables estimation of total population size without imputing unobserved individuals via data augmentation, which can be computationally cumbersome. We write custom distributions for those spatial point processes and fit the SCR model in a Bayesian framework using Markov chain Monte Carlo in the R package nimble. C_LIO_LISimulations indicate good performance of the proposed model for parameter estimation. We demonstrate the application of the model in a real-life scenario by fitting it to NGS data of female wolverines (Gulo gulo) collected in three counties of Norway during the winter of 2018/19. Our model estimates that the density of female wolverines is 9.53 (95% CI: 8-11) per 10,000km2 in the study area. C_LI

ecology↗

bioRxiv2020

Astrocytes close the critical period for visual plasticity

Summary paragraphBrain postnatal development is characterized by critical periods of experience-dependent remodeling1,2. Termination of these periods of intense plasticity is associated with settling of neuronal circuits, allowing for efficient information processing3. Failure to end critical periods thus results in neurodevelopmental disorders4,5. Yet, the cellular processes defining the timing of these developmental periods remain unclear. Here we show in the mouse visual cortex that astrocytes control the closure of the critical period. We uncover a novel underlying pathway involving regulation of the extracellular matrix that allows interneurons maturation via an unconventional astroglial connexin signaling. We find that timing of the critical period closure is controlled by a marked developmental upregulation of the astroglial protein connexin 30 that inhibits expression of the matrix degrading enzyme MMP9 through the RhoA-GTPase signaling pathway. Our results thus demonstrate that astrocytes not only influence activity and plasticity of single synapses, but are also key elements in the experience-dependent wiring of brain developing circuits. This work, by revealing that astrocytes promote the maturation of inhibitory circuits, hence provide a new cellular target to alleviate malfunctions associated to impaired closure of critical periods.

neuroscience↗

bioRxiv2020

Instrumented individuals are not representative of the population. Survival of wolverines with and without GPS collars

Results from biologging studies are often scaled up to population-level inferences and this begs the question: Are instrumented animals representative of the population given the potential bias in individual selectivity, the influence of capture, handling and wearing bio-loggers? The answer is elusive due to the challenges of obtaining comparable data from individuals without bio-loggers. Using non-invasive genetic data of a large carnivore, the wolverine (Gulo gulo) in Scandinavia, and an open-population spatial capture-recapture model, we were able to estimate and compare survival rates of individuals fitted with GPS collars versus individuals that had never been captured and collared. We found that mortality of GPS collared individuals was 47% lower compared to individuals without GPS collars. While instrumented and non-instrumented wolverines had similar risks of dying from legal culling, GPS collared individuals displayed lower mortality rates due to causes other than legal culling. Reports of positive effects of biologging on the fitness of individuals are rather uncommon and we argue that the most likely explanation is that GPS-collars could shield animals from poaching. Our results highlight the challenges of drawing population level inferences for large carnivores when using data from an instrumented sample of the population.

ecology↗

bioRxiv2020

Efficient Estimation of Large-Scale Spatial Capture-Recapture Models

Capture-recapture methods are a common tool in ecological statistics, which have been extended to spatial capture-recapture models for data accompanied by location information. However, standard formulations of these models can be unwieldy and computationally intractable for large spatial scales, many individuals, and/or activity center movement. We provide a cumulative series of methods that yield dramatic improvements in Markov chain Monte Carlo (MCMC) estimation for two examples. These include removing unnecessary computations, integrating out latent states, vectorizing declarations, and restricting calculations to the locality of individuals. Our approaches leverage the flexibility provided by the nimble R package. In our first example, we demonstrate an improvement in MCMC efficiency (the rate of generating effectively independent posterior samples) by a factor of 100. In our second example, we reduce the computing time required to generate 10,000 posterior samples from 4.5 hours down to five minutes, and realize an increase in MCMC efficiency by a factor of 25. We also explain how these approaches can be applied generally to other spatially-indexed hierarchical models. R code is provided for all examples, as well as an executable web-appendix.

ecology↗

bioRxiv2019

Estimating abundance with interruptions in data collection using open population spatial capture-recapture models

O_LIThe estimation of population size remains one of the primary goals and challenges in ecology and provides a basis for debate and policy in wildlife management. Despite the development of efficient non-invasive sampling methods and robust statistical tools to estimate abundance, maintenance of field sampling is still subject to economic and logistic constraints. These can result in intentional or unintentional interruptions in sampling and cause gaps in data time series, posing a challenge to abundance estimation, and ultimately conservation and management decisions.\nC_LIO_LIWe applied an open population spatial capture-recapture (OPSCR) model to simulations and a real case study to test the reliability of abundance inferences models to interruption in data collection. Using individual detections occurring over consecutive sampling occasions, OPSCR models allow the estimation of abundance from individual detection data while accounting for lack of demographic and geographic closure between occasions. First, we simulated sampling data with interruptions in field sampling of different lengths and timing. We checked the performance of an OPSCR model in deriving abundance for species with slow and intermediate life history strategies. Finally, we introduced artificial sampling interruptions of various magnitudes and timing to a five-year non-invasive monitoring data set of wolverines (Gulo gulo) in Norway and quantified the consequences for OPSCR model predictions.\nC_LIO_LIInferences from OPSCR models were reliable even with temporal interruptions in monitoring. Interruption did not cause any systematic bias, but increased uncertainty. Interruptions occurring at occasions towards the beginning and the end of the sampling caused higher uncertainty. The loss in precision was more severe for species with a faster life history strategy.\nC_LIO_LIWe provide a reliable framework to estimate abundance even in the presence of sampling interruptions. OPSCR allows monitoring studies to provide contiguous abundance estimates to managers, stakeholders, and policy makers even when data are non-contiguous. OPSCR models do not only help cope with unintentional interruptions during sampling but also offer opportunities for using intentional sampling interruptions during the design of cost-effective population surveys.\nC_LI

ecology↗