Movement Ecology

Understanding how animals respond to vessels is a fundamental question in conservation, since interactions between marine animals and vessels can expose wildlife to threats such as fisheries bycatch and vessel collisions. Although animal-borne tags and vessel-tracking systems now allow both components to be monitored at fine spatiotemporal scales, distinguishing behavioural interactions from incidental co-occurrence remains challenging, particularly in areas of dense marine traffic. Here, we developed a simulation-based framework to determine whether an animal is attracted to or follows a vessel, or instead moves independently. We applied the framework to concurrent GPS data from 2,705 foraging trips by Scopolis shearwaters (Calonectris diomedea) and Automatic Identification System (AIS) trajectories from fishing and nonfishing vessels in the northwestern Mediterranean. Candidate association events (n = 1,567) were first identified using spatiotemporal proximity and subsequently tested against simulated seabird trajectories representing movement in the absence of a vessel response. The framework classified apparent associations arising when vessels approached stationary birds or when independently moving trajectories crossed by chance as independent movement. Overall, 53.5% of association events showed evidence of vessel attraction, following, or both, whereas the remainder were consistent with independent movement. Although only approximately 3% of the vessels recorded by AIS were fishing vessels, they accounted for 69% of attraction events and 74% of following events, whereas most associations with nonfishing vessels were consistent with incidental proximity. Attraction and following became less likely as birds spent more time stationary and were more likely during daytime, while following was less likely for nonfishing than fishing vessels. These ecologically coherent patterns support the biological relevance of the interactions identified by the simulation-based approach. Sensitivity analyses showed that the number of associations detected by conventional threshold methods increased markedly with broader proximity criteria, whereas inference of attraction and, particularly, following remained comparatively stable. Implemented in the R package intersimR, the framework provides an accessible, reproducible, and scalable approach for separating behavioural responses from incidental co-occurrence at the event level and identifying the animal-vessel interactions most relevant to conservation and management.

In Baltic grey seals (Halichoerus grypus), interannual variation in sea-ice availability can alter the proportion of females breeding on land, generating substantial variation in local breeding density among years. We investigated preweaning survival, post-weaning residence, and weaning body condition at a Baltic grey seal land-breeding colony over two years with contrasting colony densities. Drone imagery provided repeated counts of pups in distinct developmental stages and estimates of moulting pup mass as an index of weaning body condition. A Bayesian state-space population model estimated substantially greater pup production in the high- (282.5 [95% CrI = 252.8 to 320.9] pups) than the low-density (188.3 [95% CrI =168.9 to 210.8] pups) year, but similar preweaning survival (0.66 [95% CrI = 0.53 to 0.81] and 0.67 [95% CrI = 0.56 to 0.79], respectively). In contrast, post-weaning residence differed substantially between the high- (11.8 [95% CrI = 9.2 to 15.7] days) and low-density (30.3 [95% CrI = 23.1 to 39.1] days) years. Greater exposure to other pups during early life was associated with lower weaning body mass ({beta} = -0.45 [-0.89 to -0.01], P[{beta} < 0] = 0.98). Together, these results suggest that the consequences of higher breeding density may emerge through poorer weaning condition and altered post-weaning residence even when effects on preweaning survival are not detectable.

Movement models often assume that animal decisions depend on perceived environmental conditions. However, optimal foraging and cognition theory predict that accumulated experiences should drive changes in animal motivation and decision-making. Here, we propose a mechanistic framework that models state-switching habitat selection as a history-dependent process, where internal states, such as fear or energy, emerge as latent variables accumulating through environmental exposure. Simulations showed that our model accurately recovers memory timescales and cumulative effects of environmental exposure on behavioural switching that would be undetected by existing state-switching habitat selection models. Applied to woodland caribou, it reveals that an individual integrates predation risk experienced over the previous 15 days, but food intake over only 2 days, when deciding to remain or leave an area. Our model advances perspectives on movement ecology by quantifying how accumulated experiences shape changes in motivation driving animal movement decisions.

1. Understanding how animals acquire and use information from the environment is critical for linking movement to population dynamics, species distributions and conservation. Advances in tracking technologies and growing interest in learning processes have opened opportunities to study behaviours such as habitat exploration in translocated animals or ontogeny of migration and dispersal movements. However, accessible statistical methods for studying these behavioural processes are lacking. 2. We present a learning-explicit step-selection analysis (SSA) that integrates movement, habitat selection, and learning into a single evolving process. At the start of the movement trajectory, the animal is assumed to have no knowledge of the landscape, i.e. its internal habitat quality map is initialized to a constant. As the animal moves, this map is updated dynamically: with each step, only the habitat within its perceptual range becomes known and contributes to future decisions. 3. Through simulations, we show that this method separates true habitat preferences from learning effects and reveals when large-scale behaviours, such as attraction to resources or avoidance of risks, emerge as knowledge accumulates. Critically, we demonstrate that ignoring learning, as in traditional SSA, can lead to biased estimators of habitat selection. 4. Finally, we apply our approach to a real-world GPS dataset of naive individuals, consisting of 10 juvenile red kites dispersing in Switzerland. Using leave-one-individual-out cross-validation, we show that learner SSFs consistently outperform traditional SSFs in predicting movement decisions. 5. We conclude by discussing methodological considerations and future directions for integrating learning into movement ecology. By explicitly modelling the formation of memory that integrates both spatial information and habitat quality, our approach advances process-based movement ecology while remaining compatible with standard SSA workflows, offering practical tools for both theoretical research and applied wildlife management.

Space-for-time Cormack-Jolly-Seber (CJS) models have been developed to estimate survival of migrating animals that are imperfectly detected at spatially discrete sampling stations. However, space-for-time CJS models typically ignore survival over time and age-specific probability of movement, missing the diversity of life history strategies. To address these limitations, we extended space-for-time Cormack-Jolly-Seber models to explicitly incorporate time, age, and individual-level covariates to account for diverse life history strategies. We incorporate a sub-model that includes uncertainty in individuals age using a finite-mixture model. The detection and transition probabilities were parameterized using generalized linear models, facilitating flexible model specification and inclusion of spatiotemporal and individual covariates. We apply the model to detections of juvenile steelhead (Oncorhynchus mykiss) from two populations in the Snake River Basin in Idaho, USA, to estimate trends in survival with respect to population, age, time, and the length of individuals. Additionally, we generalized the model so that it can be applied to other systems and populations with different life history strategies and monitoring infrastructure.

Acoustic telemetry is a core methodology for collecting fine-scale movement data for aquatic animals. When telemetry receivers are set up in closely spaced arrays with overlapping detection ranges, the detection times of a tagged animal can be used to estimate its position and movement paths. In practice, estimating these paths can be challenging. Traditional time-difference-of-arrival methods generally provide positioning accuracy too poor for inferring fine-scale behaviours, while more robust state-space positioning models can be computationally intensive and practically infeasible to run on large datasets. Here, a novel telemetry positioning method is presented where time-of-arrival positioning is implemented as a state-space model within an extended Kalman filter. The resulting model, termed EK-TOA, provides closed-form solutions to track estimation. Simulated datasets of fish movement within a 2D telemetry array are used to verify the models performance and a real case study is provided where EK-TOA is utilized for the long-term tracking of a tagged fish. In comparison to currently available positioning models, EK-TOA provides a fast and accurate solution for tracking fine-scale movement behaviours of aquatic animals over long, continuous periods of time.

From the iconic V-formations of geese to the spectacular murmurations of starlings and communal roosting of millions of bramblings, bird flocks keep sparking wonder and scientific curiosity. In science, common themes include how flocking formation impacts the energy expenditure during migration, how it affects risks of predation, how it relates to information spread on food abundance, and how birds manoeuvre in dense flocks. Here, we report on observations of a very different type of flock formation: migrating terns over open sea flying side by side in a search party manner -- strikingly orthogonal to their heading -- with interbird distances up towards 300 m and flocks of around ten birds covering widths up to one kilometer. Such sparse formations of migrating birds stand in great contrast to the classes of bird flocks previously described in the literature. We present simultaneous field observations and ship-based radar measurements of flock characteristics and argue that this is a form of fly-and-forage migration that may be a common, but overlooked, network foraging strategy of surface feeding birds: while most migrants cannot combine migration and foraging, the situation is different for surface feeders like terns.

1. Common dolphins (Delphinus delphis) are the most abundant cetaceans off mainland Portugal and a species of conservation and bycatch concern, so reliable abundance estimates are a management priority. Yet no single survey provides adequate spatiotemporal coverage: dedicated programmes offer sparse snapshots, while opportunistic programmes are spatially constrained by their routes. Integrating them is methodologically demanding, and the existing methods that do so in a single model estimate the density of animal groups, not of the individual animals that management often requires. 2. We develop a spatiotemporal marked log-Gaussian Cox process that integrates four structurally distinct surveys through a shared intensity surface, comprising environmental covariates, a latent spatial field, and survey-specific detection functions. Group size enters as a zero-truncated negative binomial mark with its own spatiotemporal field. We apply it to a dedicated aerial survey and three opportunistic ship-based programmes off mainland Portugal, comprising 162,514 km of on-effort transects and 1,756 sightings of 17,000 animals. We fit the model in a Bayesian framework with integrated nested Laplace approximations, yielding seasonally resolved group and animal density surfaces across the Portuguese Exclusive Economic Zone, and abundance estimates with fully propagated uncertainty. 3. Predicted animal density concentrated over the shelf and upper slope in every season, peaking northwest, and increased in shallower, cooler, more productive waters and over steeper seabed. Abundance nearly doubled from summer (77,000 [95% credible interval: 60,000, 98,000]) to winter (146,000 [99,000, 208,000]). Seasons shared 56% of the spatial field's variance, so effort from any programme in any season informed all others, and the seasonal contrast remained estimable despite uneven coverage. 4. Synthesis and applications. Our framework allows managers to combine dedicated surveys with existing platforms of opportunity to estimate animal density by season. Because group size is modelled explicitly, densities are estimated in animals rather than groups, as required for bycatch risk assessment and marine spatial planning. Because each survey enters through its own observation model, the approach accommodates programmes that differ in platform and protocol and transfers to any species or region where several surveys each cover part of a population.

1. Migration is a widespread phenomenon across taxa, yet the ecological mechanisms underlying its evolution and maintenance, particularly whether migratory behaviors are primarily driven by access to spatially restricted breeding sites or by seasonal tracking of trophic resources, remain poorly documented outside birds and large mammals. Despite increasing evidence that reptiles perform seasonal migrations, the ecological mechanisms underlying these movements have rarely been formally tested. 2. The critically endangered Galapagos pink land iguana (Conolophus marthae), endemic to Wolf Volcano, Isabela Island, exhibits partial migration along a steep altitudinal gradient, providing an opportunity to disentangle the relative roles of breeding-site availability, trophic resource dynamics, and thermoregulatory conditions as drivers of migration. 3. We used GPS tracking data from 22 individuals (7 males, 15 females) monitored between 2019 and 2023, combined with high-resolution spatio-temporal models of vegetation productivity and air temperature across the species' altitudinal range, to characterize population-level movement patterns and evaluate competing hypotheses explaining the evolution of this migratory behavior. 4. Movement models revealed a clear pattern of partial migration: 16 out of 22 tracked individuals performed seasonal altitudinal movements between a restricted high-elevation mating area and a larger dispersal area at lower elevation, with males reaching the mating area approximately 48 days earlier than females. The dispersal area remained consistently more productive than the mating area throughout the year, rejecting the prediction that individuals should track the shifting trophic resource peaks. Instead, the mating season coincided with the local productivity peak within the mating area, whereas temperature differences between areas were small (ca. 2{degrees}C) and did not explain migration timing. 5. These results support a site-dependent hypothesis of partial migration over a resource-tracking hypothesis, indicating that access to spatially restricted breeding sites is the primary driver of migration in this species, with local trophic resource dynamics fine-tuning reproductive timing. Providing empirical evidence for the ecological mechanisms underlying migration in a large terrestrial reptile, our results extend site-dependent theories of migration beyond birds and mammals and identify breeding-site availability as a key ecological driver of migratory behaviors across taxa.

Ecologists remotely track movement steps of animals (e.g., via global positioning systems) and use step selection functions to study the effect of environmental factors upon their movement decisions. Constructing such functions requires pairing each observed step with some number of unobserved but feasible comparison steps. Larger numbers of comparison steps generally yield better estimates but also incur potentially challenging computational demands. Thus, it is important to determine an appropriate number of comparison steps. No established guidance exists for this decision. Here, we use simulated tracks to assess how many comparison steps are needed, fitting each set of steps to a conditional logistic regression model. We monitor errors in estimated effects for several classes of tracks, identifying the number of comparison steps for which mean relative absolute error in estimated effects is consistently low. By this criterion, 32 comparison steps per observed step are needed for our primary class of simulated tracks. Tracks in more homogeneous landscapes, tracks with shorter mean step lengths, or shorter tracks generally require more comparison steps (ranging from 64 to 128 per observed step) to achieve the same level of accuracy. Longer tracks generally require fewer comparison steps (16 per observed step). These results clearly demonstrate that the number of comparison steps influences how well step selection functions estimate covariate effects and provides initial direction in a research area that currently lacks quantitative guidance. Movement ecologists should take care when selecting the number of comparison steps paired with each observed step because those decisions matter.

For organisms with biparental care, successful reproduction hinges on coordination between partners. Seabirds face an extreme coordination challenge because parents must schedule nest attendance on land with long-distance foraging trips at sea. We present a computational model of incubation schedules for a vulnerable seabird, the Leachs Storm-Petrel (Hydrobates leucorhous). Using only simple energetic rules and parameters, the model recapitulates natural incubation rhythms, exposes a tradeoff between parent energy and egg attendance, and predicts severe reproductive failure in harsh environments. Incubation primarily fails through "schedule breakdown" -- a single point in the season when both parents spend too long foraging and the egg dies from cold. The resilience of the developing offspring to neglect is thus a fundamental adaptation to the uncertainties of biparental care. These results raise new alarms about the indirect causes of reproductive failure in sensitive marine species and provide theoretical foundations for the evolutionary ecology of scheduling behaviors.

Protecting deep-sea sessile communities as Marine Protected Areas (MPAs) requires accurate estimates of their spatial distribution and abundance. Existing computer-vision detection methods require careful calibration to avoid counting errors; multi-object tracking (MOT) theoretically offers a more robust alternative. In practice, however, poor lighting and turbidity limit the efficacy of purely kinematic MOT algorithms in deep-sea environments. At the same time, the intra-species visual homogeneity typical of deep-sea scenes means traditional per-frame appearance features are often too weak to maintain stable identities in deep-association MOT algorithms. In this paper we introduce JEPA-SORT, an MOT algorithm that injects V-JEPA 2.1 embeddings into detection-to-track association, to show that video world-model embeddings - which capture not only physical appearance but spatiotemporal trajectory across viewpoints - can help maintain stable identities on visually homogeneous specimens. We provide a benchmark of twelve ROV transects surveying deep-sea sessile communities and evaluate JEPA-SORT against standard MOT algorithms across a range of simulated optical environments. We also propose the Biodensity Utility Metric (BDU), which measures tracker performance in the context of downstream ecological modeling. JEPA-SORT outperforms all benchmarked trackers on IDF1, MOTA, and BDU and is competitive on specimen-count accuracy. Ablations indicate its advantage stems primarily from the discriminative power of world-model embeddings on visually homogeneous footage.

1. Natal dispersal is a key process in population ecology because it links local demographic processes to broader-scale population dynamics by redistributing individuals. When using capture-recapture data, multistate capture-recapture models using discrete spatial units as states are the gold standard for estimating natal dispersal among spatial units while accounting for spatial variation in survival, recruitment and imperfect detection. However, because their computational cost increases rapidly with the number of spatial units, applications have been limited to a small number of units. Therefore, in practice, these models cannot provide spatially detailed inference on natal dispersal across large landscapes. 2. We develop a computationally efficient Bayesian capture-recapture model, called the efficient natal dispersal (END) model, to estimate natal dispersal among discrete spatial units jointly with spatial variation in demographic parameters and detection probabilities. The END model relies on two key structural features: juveniles and breeders are separated into two arrays, and resightings outside the natal spatial unit are aggregated over time for individuals released as juveniles. 3. Using simulations, we show that the END model is considerably (up to 30 times) more computationally efficient than a conventional multistate model, while maintaining comparable parameter accuracy. We then apply the END model to white stork (Ciconia ciconia) capture-recapture data from Germany across 101 hexagonal spatial units, a spatial resolution at which a conventional multistate model is computationally infeasible. We estimate natal dispersal among units jointly with spatial variation in survival and recruitment. This allows us to identify areas of lower or higher survival, earlier or delayed recruitment, and dispersal probabilities among all units. By combining estimated dispersal probabilities with existing data on the number of juveniles born in each spatial unit, we estimate natal dispersal in terms of numbers of individuals and identify units with positive or negative net migration, sources and sinks. 4. Overall, our approach moves capture-recapture analyses from estimating natal dispersal among a few spatial units to inferring dispersal networks and assessing their demographic consequences across large domains. Our approach is applicable to many spatially structured capture-recapture datasets, opening new opportunities for studying spatial population dynamics.

Translocations are commonly used to reduce the risk of extinction of threatened species, and often to reintroduce animals to their former habitats. If extirpation occurred a long time ago, individuals may be placed into novel environments outside of conditions they have previously experienced, which can affect translocation success. To measure success of translocations, particularly those to novel habitats, it is important to monitor post-release spatial behaviour. In 2023, ten critically endangered k[a]k[a]p[o] (Strigops habroptilus) were released to form a new population at the fenced mainland island of Sanctuary Mountain Maungatautari (hereafter Maungatautari) in the North Island of New Zealand, an area from which they were extirpated approximately a century earlier. The spatial behaviour and habitat selection of these k[a]k[a]p[o] was analysed using dynamic space use and step selection functions with temporally dynamic responses applied to GPS logger data recorded over a seven month period. These showed three unexpected results: k[a]k[a]p[o] were highly mobile and used larger areas compared to other sites; there was synchronicity of movement, with most individuals simultaneously moving toward the perimeter fence in spring; and when near the fence k[a]k[a]p[o] selected for areas of pasture and exotic vegetation more than anticipated. While direct comparisons are difficult due to varying data availability, this unexpected spatial behaviour and habitat selection suggests that k[a]k[a]p[o] behavioural responses were different at Maungatautari compared to other sites. This may indicate that k[a]k[a]p[o] released at Maungatautari had a longer exploratory stage than elsewhere; that there were behavioural effects influenced by conspecific density or the availability of novel and exotic vegetation; or that there were specific breeding habitat requirements not met at the site during the study period. These results highlight the importance of high-resolution spatial monitoring to evaluate the success of translocations, particularly those to novel environments.

Developing robust methods to quantify how animals allocate time across behaviours is essential for understanding energy use, habitat requirements, and responses to environmental change. For cryptic, semi-aquatic mammals such as the platypus, direct observation is difficult, creating a reliance on remote biologging approaches that can reliably infer behaviour in the wild. However, aquatic environments can both smooth acceleration signals through hydrodynamic damping and introduce noise from water movement, turbulence, and drag, potentially obscuring behavioural differences of similar magnitudes. We tested whether progressively incorporating biomechanical and frequency-domain (FFT-derived) predictors improved behavioural classification in hydrodynamically challenging aquatic environments. Tri-axial accelerometers were deployed on four ex situ platypuses, with synchronised video observations used to validate behaviour. From the acceleration data, we derived three predictor classes of increasing complexity: summary statistics describing activity level, engineered biomechanical variables capturing posture and body orientation, and FFT-derived features describing movement rhythm. These predictors were progressively incorporated into Random Forest models to classify five behaviours: burrow resting, surface resting, grooming, travelling/foraging, and diving. Model performance improved with increasing predictor complexity, although gains were behaviour specific. FFT-derived features substantially improved classification of rhythmic behaviours such as diving and foraging, while engineered biomechanical predictors improved grooming detection. In contrast, resting behaviours, particularly surface resting, showed little improvement. Overall accuracy increased from [~]75% to [~]88% when frequency-domain features were included. Misclassification was greatest among behaviours with overlapping or low-amplitude signals, and cross-individual validation revealed reduced model generalisability, indicating that individual variation in movement patterns constrained transferability. Incorporating frequency-domain features substantially improved behavioural classification in platypuses, particularly for rhythmic behaviours such as diving and foraging. This study provides the first validated accelerometry-based behavioural classification framework for the species and highlights the importance of matching predictor selection to behavioural mechanics. More broadly, the approach offers a transferable framework for aquatic and semi-aquatic taxa.

Passive acoustic sensors and machine learning classifiers offer a scalable approach to monitoring wildlife populations. These classifiers typically detect whether a species is present in a short time-window of audio; ecologists, however, want ecologically meaningful measures such as indices of site-level abundance. Vocal density, the proportion of time-windows containing a vocalization, is a useful abundance proxy. Recovering vocal density from classifiers requires calibrating their outputs against expert-labeled data. Because classifier performance varies between sites, a single global calibration yields overconfident site-level estimates, yet calibrating every site independently requires prohibitive labeling effort. Estimating site-level vocal density at the scale of modern sensor networks demands a label-efficient alternative. We present a Bayesian hierarchical extension of Platt scaling, a commonly used calibration method, that calibrates classifiers at the site level while borrowing strength across sites. Where site-level covariates are available, a Gaussian-process prior lets the model learn how calibration varies across covariate space. Averaging calibrated per-clip probabilities across a site's recordings yields a posterior distribution over vocal density, and we show how to carry that uncertainty into downstream regressions of vocal density on environmental covariates. Across simulations and two fully annotated field datasets from Hawai'i and the Pacific Northwest, our model attained desired credible-interval coverage of vocal density estimates with narrower intervals than independent site-level calibration at equal labeling effort, and lower mean squared error. Global calibration, by contrast, failed to attain coverage except when simulated sites were genuinely homogeneous. Unlike site-level calibration, our model enables calibration at sites with zero labeled data, and incorporating covariates further improved precision when site heterogeneity was covariate-driven. Applied to a 283-site dataset in Pennsylvania with only two labeled clips per site, we recovered known habitat associations for the declining Wood Thrush (Hylocichla mustelina). Our method provides a principled, label-efficient route from bioacoustic classifier scores to site-level abundance indices with well-quantified uncertainty, making rigorous ecological inference feasible at the scale at which acoustic sensor networks are now deployed.

Animals often attempt to resolve the trade-off between foraging and predation by regulating their activity timings and patterns. Driven by individual energetic requirements and the local environment, onset of activity in a species varies across space and time. These differences are particularly enhanced on a latitudinal gradient, influenced by varying lengths of daylight, temperature and seasonality. For a nocturnal mammal like the bat, such patterns translate to shorter activity windows at higher latitudes, both on a nightly and a seasonal basis. This constructs the timing of activity onset as a crucial variable for a successful night of foraging. While roost exit timings and influence of local conditions on bat activity have been extensively studied across species inhabiting a variety of locations around the world, our understanding of bat emergence behaviour on a global scale remains limited. To investigate emergence patterns in bats across a latitudinal gradient, we conducted a meta-analysis spanning 91 species from across 13 Chiropteran families. We evaluated how timing and luminosity (i.e., sun altitude) at exit varied across latitudes and species. Results revealed that bats across the globe tended to emerge under conserved thresholds of brightness, a pattern that was facilitated by a temporal delay in roost exits at higher latitudes. By maintaining emergence within the confines of nautical twilight (sun altitude between 0{degrees} to -12{degrees}), bats across the phylogeny thus seemed to balance the dilemma of predation versus foraging. This pattern was mirrored across the two broad dietary types (frugivores and insectivores) as well as different social structures (maternal v. non-maternal colonies). Furthermore, our models indicated that while local conditions dictate final exit decisions, shared ancestry was potentially influencing the extent of plasticity seen in emergence patterns. Taken together, these results highlight how broad-scale latitudinal gradients of light can influence the foraging strategies in a globally distributed nocturnal mammal.

Recent work on the design of spatial capture-recapture (SCR) studies has largely focused on settings with a single set of movement and detection parameters. In many applied contexts, however, populations are naturally stratified-for example, when different sexes exhibit markedly different movement scales and encounter processes, or when more than one focal species is monitored. Designs that perform well for one group may be inefficient or suboptimal for others, despite the substantial logistical effort typically required to implement an SCR survey. We describe several potential SCR design strategies for surveying multiple groups, and compare these using simulation. Designs considered include regular and optimised grids, clustered and lacework designs, and designs generated using genetic algorithms, including new approaches that optimise information from both groups and a recently published two-stage design. Using proximity detectors such as camera traps, we evaluate design performance in terms of precision, bias, and coverage, and examine how performance changes with the number of available resources. Our results show that multi-group SCR surveys can perform well even when focal groups differ greatly in movement scale and encounter rate, provided that some simple guidelines are followed. Systematic layouts accommodating more than one characteristic spacing e.g. cluster designs, performed consistently well and are a reliable default option. Regular grid designs with a single spacing between detectors are feasible when differences between groups are moderate but performed poorly as differences increased. Establishing when these designs become problematic adds unnecessary risk to the design process. Algorithmic designs did not outperform systematic designs even in highly-skewed scenarios expected to favour them. In practice, when there is more than one focal group we recommend identifying the smallest and largest plausible movement scales and rejecting potentially pathological designs, primarily those that cannot generate recaptures at the smaller scale or cover a substantial area at the larger scale.

Long-distance migration has evolved repeatedly across the animal kingdom, yet the underlying processes giving rise to and maintaining these complex eco-behavioral phenotypes remain poorly understood. Here, we present the first evidence of genomic determinants of migratory phenotypes in sea turtles, using whole genome resequencing to demonstrate that complex genomic architecture underlies divergent migratory destinations and reproductive timing in the critically endangered western Pacific leatherback turtle (Dermochelys coriacea). Individuals from this admixed population that navigate to foraging grounds on opposite sides of the Pacific Ocean have a putative inversion on chromosome 2 encompassing one gene, potentially conferring pleiotropic physiological effects and supporting magnetoreception. Genomic architecture underlying divergent reproductive timing is more dispersed, aligned with reduced gene flow, and is associated with genes that may influence reproductive success. Genes underlying both traits suggest a role for neurodevelopment and memory. Our study adds to the increasing evidence of at least partial genomic control of migratory traits in wild populations, with important potential implications for conservation measures such as translocation and genetic rescue. Our results align with a growing body of work describing complex genomic architecture and structural variants underlying key eco-behavioral traits, advancing the understanding of evolution of long-distance migration across taxa.

Motion-activated camera traps are an effective tool for monitoring animals, but a single study can produce millions of images, which must be analyzed before they yield data. The burden of analyzing them has led many teams to automate the work with computer vision. These models lose accuracy on images unlike those they were trained on, however, and a general model may not use the categories a project needs. Adapting a model to a project usually requires at least one of the following: many labeled images per species, programming expertise, a graphics processing unit (GPU), or uploading images to a remote server. I describe WildObserve, a desktop application that adapts to each project without any of these requirements. SpeciesNet suggests a species for each animal that MegaDetector finds, and WildObserve keeps SpeciesNet's feature vector for each one. A project model, a logistic regression on those features, then retrains in about a second as the reviewer labels, learning the project's own species and names. From the reviewed records, and optionally the suggestions, WildObserve also fits occupancy and abundance models. Images stay on the user's computer, and every record says which model suggested its species and who confirmed it. WildObserve processed 8.1 images/s on an Apple M1 Pro laptop. On Oregon Critters, a dataset released after SpeciesNet was trained, the project model's accuracy passed that of SpeciesNet with ten labels per species (balanced accuracy 0.783 against 0.751). I also tested occupancy on 63 cameras in Malawi that SpeciesNet had not seen. With 5% of the images reviewed and the project model's suggestions counted for the rest, species' occupancy estimates differed from those of a full review by 0.11 on average. From the reviewed images alone, they differed by 0.41. WildObserve takes a project from images to species records in its own categories and on to occupancy estimates, without writing code, needing a GPU, or sending images elsewhere.