Methods in Ecology and Evolution

Animal social network analysis using GPS telemetry datasets provides insights into group dynamics, social structure, and interactions of the animal communities. It aids conservation by characterizing key aspects of animal sociality - including spatially explicit information on where sociality occurs (e.g., habitats, migratory corridors), contributing to informed management strategies for wildlife populations. The aniSNA package provides functions to assess and leverage data collected by sampling a subset of an animal population to perform social network analysis. The methodologies offered in this package are compatible with a variety of location and grouping data, collected through various means (e.g., direct observations, biologgers), however, they are particularly well suited to autocorrelated data streams such as data collected through GPS telemetry radio collars. The techniques assess the datas suitability to extract reliable statistical inferences from social networks and compute uncertainty estimates around the network metrics in the scenario where a fraction of the population is monitored. The package functions are user-friendly and allow for the implementation of pre-network data permutations for auto-correlated data streams, sensitivity analysis under downsampling, bootstrapping to establish confidence intervals for global and node-level network metrics, and correlation and regression analysis to assess the robustness of node-level network metrics. Using this package, animal ecologists will be able to compute social network metrics, both at the population and individual level, assess their reliability, and use such metrics in further analyses, e.g., to study social network variation within and across populations or link individual sociality to life history. This software also has plotting features that allow for visual interpretation of the findings.

O_LISuccessful animal tracking projects depend on well-informed sampling strategies and robust methods to yield biologically meaningful inferences. Considering financial and logistical constraints, the reliability of research outputs is shaped by key decisions regarding study duration (how long should each individual be tracked?), sampling frequency (how often should new locations be collected?), and how many individuals should be tracked. To maximize their conservation value, studies must avoid biased estimates of key parameters related to movement behavior and space use, as this can lead to wasted resources and misguided management actions. C_LIO_LITo address these challenges, we propose a workflow for determining the optimal sample sizes for population-level inferences in movement ecology, explicitly addressing the trade-offs between sampling duration (T), sampling interval (reciprocal of frequency; {Delta}t), and population sample size (m). While a priori study design is considered best practice, this workflow can be applied at multiple stages, including concurrent with data collection, or as a post hoc evaluation. C_LIO_LIBy selecting robust methods that are sampling-insensitive, and by quantifying and propagating uncertainty through downstream analyses, we can determine whether our sample sizes (both at the individual- and population-level) are sufficient to yield robust population-level inferences, such as mean home range area or mean movement speed. Furthermore, researchers can integrate additional logistical constraints such as fix success rate, location error, and potential device malfunctions, while also accounting for individual variation. We illustrate potential applications of this workflow through empirically-guided simulations. C_LIO_LITo facilitate its use and implementation, we incorporated this workflow into the user-friendly movedesign R Shiny application. This application enables researchers to easily test different sampling strategies, and as of version 0.3.2, integrates population-level analytical targets. This workflow has the potential to improve the rigor and reliability of animal tracking projects conducted under logistical and financial constraints, and thereby support more effective scientific research, wildlife management, and conservation efforts. C_LI

Behavioural ecologists aim to understand the causes of animal social structure. Connecting theoretical models of social structure with empirical observations remains, however, a for-midable challenge. While most of the current statistical methods for animal social network analysis rely on data that are aggregated over time and summarised as one behavioural dimension (e.g., an adjacency-matrix), common behavioural sampling techniques (e.g., focal-animal sampling) produce data in continuous time, and involve different behaviours. Furthermore, empiricists in the field are generally interested in causal inference, but lack a framework to rigorously analyse focal-animal sampling data in light of transparent causal assumptions. As a consequence, common methods are often inappropriate, and can lead to wrong biological conclusions. Here, we introduce a causal Bayesian modelling framework to empirically study the causes of social network structure from focal-animal sampling data. We start by outlining a generative model that encodes how biological and measurement processes jointly produce social network data in continuous time; namely, as a temporal sequence of dyadic behavioural states (e.g., no body contact, social resting, grooming). Building upon the generative model, we develop a statistical model: a multilevel, multiplex Bayesian model that takes raw focal observations as input, and produces a posterior probability distribution for the generative parameters as output. After validating the statistical models performance with sparse data-- common in real-world settings--we illustrate its application with an empirical data set collected in wild Assamese macaques. We notably showcase how researchers can compute probabilistic estimates for well-defined causal hypotheses about the drivers of social structure. With this work, we not only contribute novel theoretical and statistical tools to the field, but also illustrate a workflow that allows researchers to iteratively translate their domain expertise into a formal analytical strategy--bridging theoretical and empirical research in behavioural ecology.

O_LIPassive trapping techniques such as pitfall and malaise traps probably constitute the most widely used methods for standardised surveys of invertebrate populations worldwide. These methods typically yield aggregated count data over multi-day trapping periods, often spanning several weeks, during which species activity (i.e. phenology) can vary. The analysis of trapping data collected over temporally misaligned sampling intervals is challenging, hampering the integrated analysis of historically available trapping datasets. C_LIO_LIWe introduce a temporal disaggregation approach using interval-integrated B-splines to analyse data collected over misaligned sampling intervals while accounting for phenological influences. We present computationally efficient Taylor series approximations for integrating exponentiated B-splines over sampling intervals. We further tailor our approach to typical trapping datasets by providing several extensions, including joint species distribution modelling. C_LIO_LIThrough simulations and cross-validation, we demonstrate that our approach of temporal disaggregation outperforms naive approaches and provides improved inference on phenology and other parameters of interest, such as inter-annual trends. The first-order Taylor approximation, which can be fit using regular software routines, properly accounts for heterogeneity in sampling duration and timing, while the second-order Taylor approximation and the exact model additionally allow for improved estimation of phenological patterns. C_LIO_LIBy applying this model to a large pitfall trapping dataset, spanning almost 50 years and over 10,000 trapping events in the Belgian province of Limburg, we illustrate how this approach can be used to reveal phenological, spatiotemporal and co-distributional patterns for 331 spider species. C_LIO_LIThe interval-integrated B-splines approach we present provides a convenient way to infer phenology and other ecological parameters from temporally aggregated count data obtained over misaligned sampling intervals, facilitating the integrated analysis of heterogeneously collected datasets to infer biodiversity trends. C_LI

Species abundance data is commonly used to study biodiversity patterns. In this context, estimating - and {beta}-diversity based on incomplete samples can lead to undersampling biases. It is therefore essential to employ methods that enable accurate comparisons of - and {beta}-diversity across varying sample sizes. This involves relying on biodiversity measures that are focused on accurately estimating the total number of species within a community, as well as the total number of species shared by two communities. Rarefaction offers such a method, where -diversity is estimated for standardized sample sizes. Rarefaction methods can also be used as a basis for {beta}-diversity calculations for standardized sample sizes. In this application note, we introduce a new R package, rarestR, designed to estimate abundance-based - and {beta}-diversity measures for inconsistent samples using rarefaction metrics. Additionally, the package offers parametric extrapolations to estimate the total expected number of species within a single community and the total expected number of species shared between two communities. Furthermore, it provides visualization for the curve fitting associated with these estimators. Overall, the rarestR package is useful in estimating - and {beta}-diversity values for incomplete samples, for example in studies involving highly mobile or species-rich taxa. These species estimators offer a complementary approach to non-parametric methods, such as the Chao series of estimators.

O_LISocial network analysis of animal societies allows scientists to test hypotheses about social evolution, behaviour, dynamical processes, and transmission events such as the spread of disease. However, the accuracy of estimated social network metrics depends on the proportion of individuals sampled, actual sample size, and frequency of observations. Robustness of network metrics derived from a sample has thus far been examined through various simulation studies. However, simulated data do not necessarily reflect the nuances of real empirical data. C_LIO_LIWe used some of the largest available GPS telemetry relocation datasets from five species of ungulates characterised by different behavioural and ecological traits and living in distinct environmental contexts to study the bias and robustness of social network metrics. We introduced novel statistical methods to quantify the uncertainty in network metrics obtained from a partial population suited to autocorrelated data such as telemetry relocations. We analysed how social network metrics respond to down-sampling from the observed data and applied pre-network data permutation techniques, a bootstrapping approach, correlation, and regression analyses to assess the stability of network metrics when based on samples of a population. C_LIO_LIWe found that global network metrics like density remain robust when the sample size is lowered, whereas some local network metrics, such as eigenvector centrality, are entirely unreliable when a large proportion of the population is not monitored. We show how to construct confidence intervals around the point estimates of these metrics representing the uncertainty as a function of the number of nodes in the network. C_LIO_LIOur uncertainty estimates enable the statistical comparison of social network metrics under different conditions, such as analysing daily and seasonal changes in the density of a network. Despite the striking differences in the ecology and sociality among the five different ungulate species, the various social network metrics behave similarly under downsampling, suggesting that our approach can be applied to a wider range of species across vertebrates. Our methods can guide methodological decisions about animal social network research (e.g., sampling design and sample sizes) and allow more accurate ecological inferences from the available data. C_LI

Low-density social networks can be common in animal societies, even among species generally considered to be highly social. Social network analysis is commonly used to analyse animal societal structure, but edge weight (strength of association between two individuals) estimation methods designed for dense networks can produce biased measures when applied to low-density networks. Frequentist methods suffer when data availability is low, because they contain an inherent flat prior that will accept any possible edge weight value, and contain no uncertainty in their output. Bayesian methods can accept alternative priors, so can provide more reliable edge weights that include a measure of uncertainty, but they can only reduce bias when sensible prior values are selected. Currently, neither accounts for zero-inflation, so they produce edge weight estimates biased towards stronger associations than the true social network, which can be seen through diagnostic plots of data quality against output estimate. We address this by adding zero-inflation to the model, and demonstrate the process using group-based data from a population of male African savannah elephants. We show that the Bayesian approach performs better than the frequentist to reduce the bias caused by these problems, though the Bayesian requires careful consideration of the priors. We recommend the use of a Bayesian framework, but with a conditional prior that allows the modelling of zero-inflation. This reflects the fact that edge weight derivation is a two-step process: i) probability of ever interacting, and ii) frequency of interaction for those who do. Additional conditional priors could be added where the biology requires it, for example in a society with strong community structure, such as female elephants in which kin structure would create additional levels of social clustering. Although this approach was inspired by reducing bias observed in sparse networks, it could have value for networks of all densities.

Paleobiologists are increasingly employing network-based methods to analyze the complex data retrieved from geohistorical records, including stratigraphic sections, sediments, and fossil collections. However, the lack of a common framework for designing, performing, evaluating, and communicating these studies, leads to issues of reproducibility and communicability. The high-dimensional geohistorical data also raises questions about the limitations of standard network approaches, which assume independent interactions between pairs of components. Higher-order network models better suited for the complex relational structure of the geohistorical data provide an opportunity to overcome these challenges. These models can represent temporal and spatial constraints inherent to the biosedimentary record and describe higher-order interactions, capturing more accurate biogeographical, biostratigraphic, and macroevolutionary patterns. Here we describe how to use the Map Equation framework for designing higher-order network models of geohistorical data, address some practical decisions involved in modeling complex dependencies, and discuss critical methodological and conceptual issues that currently make it difficult to compare results across studies in the growing body of network-based paleobiology research. We illustrate different higher-order network representations and models, including multilayers, hypergraphs, and varying Markov times models, using case studies on gradient analysis, bioregionalization, and macroevolution, and delineate future research directions for current challenges in the emerging field of network paleobiology.

O_LIThe ability to monitor associations between wild animals is essential for understanding the processes governing gene transfer, information transfer, competition, predation and disease transmission. C_LIO_LIUntil recently, such insights have been confined to large, visible or captive animals. However, the rapid development of miniature sensors for consumer electronics is allowing ecologists to monitor the natural world in ways previously considered impossible. C_LIO_LIHere we describe miniature (<1g) proximity loggers we have developed that use Bluetooth Low Energy transmission to register contacts between individuals. Our loggers are open source, low cost, rechargeable, able to store up to 2000 contacts, can be programmed in situ and can download data remotely or through a mobile phone application, increasing their utility in remote areas or with species which are challenging to recapture. C_LIO_LIWe successfully trialled our loggers in a range of field realistic conditions, demonstrating that Bluetooth Low Energy is capable of logging associations in structurally complex habitats, and that changes in received signal strength can be equated to short range changes in distance between loggers. Furthermore, we tested the system on starlings (Sturnidae vulgaris). C_LIO_LIThe ability to include other sensors is retained in our prototypes, allowing for the potential integration of physiological and behavioural inference into social networks derived from our approach. Due to its open source nature, small size, flexibility of use and the active research currently being undertaken with Bluetooth Low Energy, we believe that our approach is a valuable addition to the biologging toolkit. C_LI

O_LIWith small-bodied species, it is difficult to directly track individual movements, leaving mark-recapture as the most feasible method for collecting movement data. Mark-recapture data are challenging to analyse because they are indirect: many individuals are never seen after release, and for recaptured individuals there is no information on the movements between release and recapture locations. This makes it difficult to apply many statistical approaches that have been developed for continuous movement data. Among the statistical methods targeted specifically to mark-recapture data, most are focused on the estimation of population sizes or vital parameters rather than the estimation of movement behaviours. C_LIO_LIWe present the R-package Jsmm that expands and implements the earlier published Joint Species Movement Modelling (JSMM) framework with Bayesian inference. Jsmm estimates parameters related to habitat selection (behaviour at edges between habitat types), diffusion (random component of movement), advection (directional component of movement) and reaction (mortality rate), and their dependence on spatial, temporal or spatiotemporal covariates. Jsmm implements both instantaneous capture process and cumulative capture process, enabling its applications to a broad range of studies. If applying Jsmm to data on multiple species, it can estimate how species-specific parameters depend on species traits and/or phylogenetic relationships. C_LIO_LIWe use real and simulated case studies to demonstrate the workflow of Jsmm: (1) defining the model through importing the spatial domain, the spatiotemporal covariates, and the capture-recapture data; (2) fitting the model with Bayesian inference and evaluating model fit through posterior predictive checks; and (3) using the fitted model for inference and/or prediction. The simulated example validates the technical implementation by showing that the estimated parameters match with the assumed values. The real data example on moth light-trapping illustrates the practical utility of the package. C_LIO_LIThe R-package Jsmm offers a flexible resource for analysing capture-recapture data in a model-based framework that explicitly accounts for the spatiotemporal study design of where and when captures are attempted. By analysing data jointly on multiple species, the approach facilitates analyses of sparse datasets where the low number of recaptures would not allow fitting species-specific models separately for each species. C_LI

O_LINetwork analysis is a powerful framework for investigating complex systems across disciplines, including ecology, where it helps uncover patterns in predator-prey, host- parasite, or plant-pollinator interactions. However, ecological network data are often incomplete or error-prone due to sampling limitations, detection failures, and taxonomic uncertainty--leading to missing (false negative) and spurious (false positive) links that obscure structure and hinder inference. The hierarchical stochastic block model (HSBM), particularly in its degree-corrected form, is among the most effective tools for reconstructing networks under such uncertainty. Despite its robustness, the primary implementation of HSBM in the Python-based graph-tool library has remained largely inaccessible to ecologists. C_LIO_LIHere, we introduce sabinaHSBM, the first R package that makes degree-corrected HSBM broadly available through a user-friendly, flexible workflow. By bridging a gap between advanced network modeling and widely used ecological analysis platforms, sabinaHSBM facilitates network reconstruction and link prediction from binary bipartite data. The workflow involves three main steps: (1) preparing input data, (2) estimating posterior link probabilities, and (3) reconstructing the network. The package supports detection of undocumented and spurious links, exploration of hierarchical structure, and propagation of uncertainty throughout. Key features include cross-validation, flexible thresholding, probabilistic evaluation metrics, and two link prediction modes: estimating all link probabilities or identifying undocumented ones. C_LIO_LIWe illustrate the packages functionality through a case study using a published global dataset of carnivore-parasite associations, showing that inferred groupings are phylogenetically clustered. To assess predictive accuracy, we examined the top 10 highest-probability links identified by the model and found published evidence for 8, despite their absence from the original dataset. This highlights the models ability to recover biologically meaningful but underreported interactions. C_LIO_LIBy integrating all components of HSBM-based reconstruction into an accessible R package, sabinaHSBM empowers researchers to improve relational data quality and uncover overlooked patterns in complex ecological networks and beyond. C_LI

O_LIMethods that estimate the niche of a species by calculating a convex hull or an elliptical envelope have become popular due to their simplicity and interpretation, given Hutchinsons conception of the niche as an n-dimensional hypervolume. C_LIO_LIIt is well known that convex hulls are sensitive to outliers and do not have the ability to differentiate between regions of low and high concentration of presences, while the elliptical envelopes may contain large regions of niche space that are not relevant for the species. Thus, when the goal is to estimate the realized niche of the species, both methods may overestimate the niche. C_LIO_LIWe present a methodology that combines both the convex hull and the elliptical envelope methods producing an n-dimensional hypervolume that better fits the observed density of species presences, making it a better candidate to model the realized niche. Our method, called the CHE approach, allows defining regions of iso-suitability as a function of the significance levels inherited from the method (Mahalanobis distance model, minimum covariance determinant, or minimum volume ellipsoid) used to fit an initial elliptical envelope from which we then discard regions not relevant for the species by calculating a convex hull. C_LIO_LIWe applied the CHE approach to a case study of twenty-five species of bats present in the Iberian Peninsula, fitting a hypervolume for each species and comparing them to both the convex hulls and elliptical envelopes obtained with the same data and different values of n. We show that as the number of variables used to define the niche space increases, both the convex hull and elliptical envelope models produce overly large hypervolumes, while the size of the hypervolume fitted with the CHE approach remains stable. As a consequence, similarity measures that account for the niche overlap among different species may be inflated when using convex hulls or elliptical envelopes to model the niche; something that does not occur under the CHE approach. C_LI

Many complex systems are modelled using modular models, where individual sub-models are estimated separately and then combined. While this simplifies inference, it fails to account for interactions between components. A natural solution is to estimate all components jointly, but this is often impractical due to intractable likelihoods. Approximate Bayesian Computation (ABC) provides a likelihood-free alternative, but its standard implementations are computationally inefficient, particularly when applied to high-dimensional modular models, or when sub-models involve costly machine learning methods, like Gaussian Process (GP) models. The ABC-Population Monte Carlo (ABC-PMC) framework improves on vanilla ABC by using sequential Monte Carlo sampling with adaptive tolerances and proposal kernels, yielding much higher acceptance rates and more efficient exploration of parameter space. Existing ABC-PMC algorithms are not, however, especially efficient in the high-dimensional parameter setting typical of modular models. We introduce a novel modification of the ABC-PMC method that leverages model modularity. Our approach refines the prior distribution and perturbation kernel by using precomputed Markov Chain Monte Carlo (MCMC) samples from individual sub-models, making parameter updates more efficient. Additionally, we employ an adaptive summary statistic weighting strategy that dynamically adjusts the contribution of different statistics, reducing the influence of less informative statistics. These modifications greatly reduce overall computational cost. In our case studies, the runtime for 10,000 simulation attempts drops from over 20 days to under 1 minute, following a one-off preprocessing step that consists of standard MCMC sampling for each sub-model (typically 3-10 hours, depending on model complexity). We apply our method to an ecological case study using an Integral Projection Model (IPM) for Cryptantha flava, where survival, growth, and reproduction processes are modelled using GP models. The results of the simulated and the real case studies demonstrate greatly improved computational efficiency while preserving inference quality. While the case study focuses on ecology, the method is applicable to a broad range of modular models where capturing interactions among sub-models is essential.

O_LISoundscape studies are increasingly common to capture landscape-scale ecological patterns. Yet, several aspects of soundscape diversity quantification remain unexplored. Although some processes influencing acoustic niche usage may operate in the 24h domain, most acoustic indices only capture the diversity of sounds co-occurring in sound files at a specific time of day. Moreover, many indices do not consider the relationship between the spectral and temporal traits of sounds simultaneously. To provide novel insights into landscape-scale patterns of acoustic niche usage at broader temporal scales, we present a workflow to quantify soundscape diversity through the lens of functional ecology. C_LIO_LIOur workflow quantifies the functional diversity of sound in the 24-hour acoustic trait space. We put forward an entity, the Operational Sound Unit (OSU), which groups sounds by their shared functional properties. Using OSUs as our unit of diversity measurement, and building on the framework of Hill numbers, we propose three metrics that capture different aspects of acoustic trait space usage: (i) soundscape richness; (ii) soundscape diversity; (iii) soundscape evenness. We demonstrate the use of these metrics by (a) simulating soundscapes to assess if the indices possess a set of desirable behaviours; and (b) quantifying the soundscape richness and evenness along a gradient in species richness to illustrate how these metrics can be used to shed unique insights into patterns of acoustic niche usage. C_LIO_LIWe demonstrate that: (a) the indices outlined herein have desirable behaviours; and (b) the soundscape richness and evenness are positively correlated with the richness of soniferous species. This suggests that the acoustic niche space is more filled where taxonomic richness is higher. Moreover, species-poor acoustic communities have a higher proportion of rare sounds and use the acoustic space less effectively. As the correlation between the soundscape and taxonomic richness is strong (>0.8) and holds at low sampling intensities, soundscape richness could serve as a proxy for taxonomic richness. C_LIO_LIQuantifying the soundscape diversity through the lens of functional ecology using the analytical framework of Hill numbers generates novel insights into acoustic niche usage at a landscape scale and provides a useful proxy for taxonomic richness measurement. C_LI

O_LIMeasuring time-activity budgets over the complete individual lifespan is now possible for many animals with the recent advances of life-long individual monitoring devices. Although analyses of changes in the patterns of time-activity budgets have revealed ontogenetic shifts in birds or mammals, no such technique has been applied to date on insects. C_LIO_LIWe tested an automated breakpoint-based procedure to detect, assess and quantify shifts in the temporal pattern of the flight activities in honey bees. We assumed that the learning and foraging stages of honey bees will differ in several respects, to detect the age at onset of foraging (AOF). C_LIO_LIUsing an extensive dataset covering the life-long monitoring of 2,100 individuals, we compared the AOF outputs with the more conventional approaches based on arbitrary thresholds. We further evaluated the robustness of the different methods comparing the foraging time-activity budget allocations between the presumed foragers and confirmed foragers. C_LIO_LIWe revealed a clear-cut learning-foraging ontogenetic shift that differs in duration, frequency, and time of occurrence of flights. Although AOF appeared to be highly plastic among bees, the breakpoint-based procedure seems better able to detect it than arbitrary threshold-based methods that are unable to deal with inter-individual variation. C_LIO_LIWe developed the aof R-package including a broad range of examples with both simulated and empirical dataset to illustrate the simplicity of use of the procedure. This simple procedure is generic enough to be derived from any individual life-long monitoring devices recording the time-activity budgets of honey bees, and could propose new ecological applications of bio-logging to detect ontogenetic shifts in the behaviour of central-place foraging insects. C_LI

The R package yaps was introduced in 2017 as a transparent open source alternative to closed source manufacturer-provided solutions to estimate positions of fish (and other aquatic animals) tagged with acoustic transmitters. Although yaps is open source and transparent, the process from raw detections to final tracks has proved to be challenging for many potential users, effectively preventing most users from accessing the benefits of using yaps. Especially, the very important process of synchronizing the hydrophone arrays have proven to be an obstacle for many potential users. To make yaps more approachable to the wider fish tracking community, we have developed and added user-friendly functions assisting users in the complex process of synchronizing the data. Here, we introduce these functions and a six-step protocol intended to provide users with an example workflow that can be used as a template enabling users to apply yaps to their own data. Using example data collected by an array of Vemco VR2 hydrophones, the protocol walks the user through the entire process from raw data to final tracks. Example data sets and complete code for reproducing results are provided.

O_LIAlthough the use of accelerometer-based behavioural classification to quantify animal activity budgets is gaining widespread traction, the interactions between key preprocessing decisions and modern classification algorithms remain poorly understood. Moreover, classification pipelines are commonly assessed using global performance metrics, despite increasing evidence that such metrics poorly reflect behaviour-specific patterns and ecological reliability. C_LIO_LIUsing a free-ranging primate (Chlorocebus pygerythrus) as a case study, we benchmarked how temporal segmentation (burst length), collar orientation correction, and model architecture jointly shape behavioural inference. We compared nine supervised algorithms spanning classical machine learning, feature-based deep learning including a tabular foundation model (TabPFN), and state of the art time-series architectures (HydraMultiROCKET). Beyond conventional metrics, performance was further evaluated using ecological validation against independent focal observations to assess model stability and biological plausibility. C_LIO_LIModel architecture exerted the strongest influence on classification outcomes. Modern deep-learning approaches substantially outperformed classical models, doubling recall for rare behaviours (e.g., grooming, self-scratching) without compromising precision. In contrast, burst length and collar orientation correction had little effect on global metrics but produced substantial, behaviour-specific trade-offs. Shorter bursts improved the detection of rare events by increasing training instances, while orientation correction suppressed dataset-specific artifacts at the cost of degrading common behaviours. Crucially, models with similar global and behaviour-level validation metrics produced divergent predictions when applied outside the annotated context. C_LIO_LIOur findings reveal that global metrics are insufficient for optimizing behavioural inference in complex wild systems. We demonstrate that modern deep-learning architectures, such as the ROCKET family, provide a robust, accessible baseline that handles class imbalance more effectively than traditional methods. We propose that reliable inference requires behaviour-aware evaluation frameworks that integrate ecological validation, and advocate for ensemble or hierarchical strategies to leverage the complementary strengths of different preprocessing and modelling configurations. C_LI

O_LIMechanistic models of resource competition underpin numerous foundational concepts and theories in ecology, and continue to be employed widely to address diverse research questions. Nevertheless, current software tools present a comparatively steep barrier to entry. C_LIO_LII introduce the R package rescomp to support the specification, simulation and visualisaton of a broad spectrum of consumer-resource interactions. rescomp is compatible with diverse model specifications, including an unlimited number of consumers and resources, different consumer functional responses (type I, II and III), different resource types (essential or substitutable) and supply dynamics (chemostats, logistic and/or pulsed), delayed consumer introductions, time dependent growth and consumption parameters, and instantaneous changes to consumer and/or resource densities. C_LIO_LISeveral examples on implementing rescomp are provided. In addition, a wide variety of additional examples can be found in the package vignettes, including using rescomp to reproduce the results of several well known studies from the literature. C_LIO_LIrescomp provides users with an accessible tool to reproduce classic models in ecology, to specify models resembling a wide range of experimental designs, and to explore diverse novel model formulations. C_LI

Process-based ecosystem models help us understand and predict ecosystem processes, but using them has long involved a difficult choice between performing data- and labor-intensive site-level calibrations or relying on general parameters that may not reflect local conditions. Hierarchical Bayesian (HB) calibration provides a third option that frees modelers from assuming model parameters to be completely generic or completely site-specific and allows a formal distinction between prediction at known calibration sites and "out-of-sample" prediction to new sites. Here, we compare calibrations of a process-based dynamic vegetation model to eddy-covariance data across 12 temperate deciduous Ameriflux sites fit using either site-specific, joint cross-site, or HB approaches. To be able to apply HB to computationally demanding process-based models we introduce a novel emulator-based HB calibration tool, which we make available through the PEcAn community cyberinfrastructure. Using these calibrations to make predictions at held-out tower sites, we show that the joint cross-site calibration is falsely over-confident because it neglects parameter variability across sites and therefore underestimates variance in parameter distributions. By showing which parameters show high site-to-site variability, HB calibration also formally gives us a structure that can detect which process representations are missing from the models and prioritize errors based on the magnitude of the associated uncertainty. For example, in our case-study, we were able to identify large site-to-site variability in the parameters related to the temperature responses of respiration and photosynthesis, associated with a lack of thermal acclimation and adaptation in the model. Moving forward, HB approaches present important new opportunities for statistical modeling of the spatiotemporal variability in modeled parameters and processes that yields both new insights and improved predictions.

Joint species distribution models (JSDMs) have become an increasingly utilised tool for modelling and predicting change within communities of species across environmental gradients. Here we present a new R package for the fitting of JSDMs using the Bayesian probabilistic programming language Stan. The jsdmstan package can model species responses to environmental covariates and also species interactions through either full specification of the species covariance matrix or through a latent variable formulation. It also provides tools for simulating joint species distribution data according to those models. It supports specification of prior distributions of all parameters in the model, as well as access to the full suite of Stan diagnostics. The ability of this package to fit these models is demonstrated upon two real data sets, one on tree species in a survey of broadleaved woodlands and another on dune spider populations. The models are able to successfully recover population characteristics such as richness and show ecologically interesting results regarding residual species correlations and responses to environmental factors. The jsdmstan package provides a user-friendly interface for fitting JSDMs, with tools available to better understand both the consequences of the model assumptions and how well the model is performing.