Standard spatial capture-recapture (SCR) models routinely discard a large fraction of non-invasive genetic detections that fail genotyping, wasting costly field effort and undermining the precision of population estimates. In environmental impact assessments (EIA), where conclusions must withstand regulatory scrutiny under fixed monitoring budgets, this data loss is a problem that can be solved. We applied the random-thinning SCR framework (RT-SCR) to a three-year Iberian wolf (Canis lupus signatus) EIA monitoring programme in northern Portugal. We treat genotyping failure as a stochastic thinning process within a multisession Bayesian model with covariates on density and detection. Of 173 confirmed wolf scats, 56 (32%) lacked individual identity, and rather than discarding them, we analysed them alongside 117 identified samples. We benchmarked RT-SCR against standard SCR, evaluated model fit using posterior predictive checks and used simulation to test whether gains generalised beyond this specific dataset. Incorporating unidentified detections improved precision for the habitat effect on density (15% CV improvement), estimates of space use (8% CV improvement) and detection parameter (up to 5% CV improvement). Wolves had stable year-to-year density (2.89-5.07 wolves/100 km2). Simulations showed that RT-SCR yielded consistent precision gains and restored near-nominal coverage for detection parameters, whereas excluding unidentified samples produced overconfident credible intervals under standard SCR (0.920 vs. 0.505; nominal 0.89). Posterior predictive checks found no strong evidence of systematic lack of fit, even for a pack-living species. Synthesis and applications. RT-SCR provides a principled and cost-effective way to recover information from samples that are usually discarded, improving both precision and the reliability of parameter estimates without extra field or laboratory investment. Gains were strongest for shared ecological parameters, including detection, space use and habitat associations, and are expected to be even greater where genotyping success is lower. This makes the framework especially valuable for EIA and other conservation monitoring programmes where every hard-won sample matters.

Environmental DNA (eDNA) concentration varies through space and time, and measurements collected close together are often correlated. Ignoring this dependence can inflate the rate of incorrect ecological inferences (Type I error rate). Although spatial correlation in eDNA has received considerable attention, temporal correlation has been less well studied. Statistical models and study designs that account for temporal correlation are increasingly important to understand time-dependent effects in complex systems. We developed a hierarchical model that separates temporal ecological variation from variability stemming from sampling and laboratory processes and applied it to four single-site eDNA time series collected over 17-24 days, three of which provided sufficient information for parameter estimation. We then used the empirically estimated parameter magnitudes in a simulation study to evaluate alternative temporal sampling designs that considered (1) how a fixed number of samples are allocated across different numbers of sampling times with different levels of temporal replication and (2) equally spaced versus cluster-spaced sampling (short bursts separated by longer gaps). Across the three time series sufficient for analysis, we observed substantial sampling variability, temporal variability, and temporal correlation, although correlation was estimated imprecisely (large coefficient of variation). Simulations showed that when sampling intervals were shorter than the effective temporal correlation range, models that ignored temporal dependence produced inflated Type I error rates and frequently detected spurious temporal trends. Accounting for temporal correlation substantially reduced this inflated Type I error rate. Optimal sampling strategies depended on study objectives. Clustered sampling most effectively estimated temporal correlation. When temporal dependence was negligible, evenly spaced sampling maximized power to detect trends. Estimating sampling variability required concentrating effort into fewer sampling times with more replicates per time, whereas estimating temporal variance was most precise with intermediate levels of replication. Together, these results indicate that temporal dependence can strongly affect inference from quantitative eDNA time series when sampling intervals approach the correlation timescale. Designs that ignore this dependence risk inferring ecological change or difference where none exists. Our framework provides practical guidance for allocating sampling effort in temporally intensive eDNA monitoring and for interpreting trends from short time series.

Context Management of broadly distributed invasive species requires knowledge of population densities across multiple ecosystems. The feral cat (Felis catus) has a continental distribution in Australia and has caused many declines and extinctions; however, density estimates from several ecosystems are few or lacking. Camera trapping data coupled with spatially explicit capture–recapture analysis is a suitable approach for estimating cat densities. However, if a large portion of individuals cannot be identified, density estimation may be difficult and estimates may be too low. Aims We aimed to estimate the feral cat density and population size within the subtropical Gondwanan Rainforest of eastern Australia, an area with world heritage status and high biodiversity values, and in which cat density was not known. Methods We used feral cat data from a grid of 60 camera traps deployed over 305 days in Border Ranges National Park, totalling 18,300 camera trapping nights. We employed a ‘random thinning’ spatially explicit capture–recapture model that considered detections of both known and unknown feral cat identities. Our modelling included a primary detection history of identified individuals and a secondary detection history that included pelage type, which allowed us to account for individuals with homogeneous pelage types. Key results Feral cat density was estimated at 0.858 cats km−2 (95% HPDI 0.432, 1.385), which is much higher than the average feral cat density across the Australian continent of 0.27 cats km−2 (95% CI: 0.18–0.45). The probability of identifying unique individuals was low for black and tabby cats compared with other pelage types. Population sizes during the four survey periods were estimated to be similar, with posterior medians ranging from 197 to 202 individuals and 95% highest posterior density intervals ranging from 95 to 329 individuals within a ~234 km2 area. Conclusions Our study provides the first robust feral cat density estimate from a subtropical rainforest ecosystem. Implications Our results add to a growing body of literature that suggests feral cat densities are high in productive mesic reserves of eastern Australia. Management of feral cats in Gondwanan World Heritage Areas should be prioritised to limit impacts on narrow-range endemic species that are likely prey to feral cats.

Louisiana black bears ( Ursus americanus luteolus ) occur in semi‐isolated fragments of bottomland hardwood forest in the lower Mississippi Alluvial Valley and were listed as threatened under the United States Endangered Species Act in 1992. A population viability analysis based on radio‐telemetry and capture‐mark‐recapture (CMR) data collected from 2002 to 2012 revealed that the probability of bears persisting in at least 1 subpopulation was >0.999, which prompted the United States Fish and Wildlife Service to remove the Louisiana black bear from the threatened species list in 2016. A post‐delisting monitoring plan was developed, which included continued noninvasive CMR surveys to monitor subpopulation trends. We used genetic CMR data based on non‐invasively collected hair samples for a post‐delisting assessment of the demographic status of the black bear subpopulations in Louisiana, USA. The assessment included primary range in the Tensas River Basin (TRB; 2006–2020), the Upper Atchafalaya River Basin (UARB; 2007–2020), and the Three Rivers Complex (TRC; 2014 and 2021), where bears had been reintroduced beginning in 2001, and adjacent areas of possible range expansion (i.e., secondary range). We used spatially explicit closed‐population capture‐recapture models to estimate abundance ( N ), density ( D ), and the realized population growth rate ( λ ) for female bears at TRB, UARB, and TRC. Model‐averaged estimates of N for primary range at TRB ranged from 135 (95% CI = 121–149) female bears in 2006 to 140 (95% CI = 123–157) in 2020 with mean λ of 1.003 (95% CI = 0.991–1.015), which suggested a stable primary population. An estimated 56 (95% CI = 19–92) additional female bears occurred in secondary range at TRB in 2020. Model‐averaged estimates of N for primary range at UARB ranged from 35 (95% CI = 28–43) females in 2007 to 33 (95% CI = 28–43) in 2020, with mean λ = 0.996 (95% CI = 0.972–1.020), again suggesting a stable population. We estimated 3 (95% CI = 0–6) additional female bears occurred in secondary range at UARB during 2020. The estimate of N at TRC was 28 (95% CI = 18–44) females in 2014 and 42 (95% CI = 19–94) in 2021. The number of bears in the secondary ranges suggests some bears may have emigrated from the primary areas and colonized secondary areas, likely facilitated by increases in bottomland hardwood forests adjacent to the core populations. The stable primary populations, the reintroduced population at TRC, and the number of bears in secondary range adjacent to TRB indicate the overall number of Louisiana black bears has increased since monitoring began.

AbstractPrevention and early detection of invasive species are championed as the most cost‐effective and efficient strategies for reducing or preventing negative impacts on ecosystems. Spotted lanternfly (SLF), Lycorma delicatula, is a recently introduced invasive insect whose range in the United States has been expanding rapidly since it was first discovered in Pennsylvania in 2014. Feeding by this planthopper can cause severe impacts on agricultural production, particularly grapes (Vitis spp.). Human visual surveys are the most common search method employed for detection but can be ineffective due to the insect's cryptic egg masses and low density during early stages of infestation. Therefore, finding alternative early detection methods has become a priority for agencies tasked with addressing SLF management. This study experimentally tested whether trained detector dogs could improve the probability of detecting SLF in both agricultural and forest settings. We surveyed transects in 20 vineyards and their adjacent wooded areas in Pennsylvania and New Jersey, USA, and used a multiscale occupancy model to estimate detection probability achieved by human observers and detection dogs as a function of SLF infestation level, weather, and habitat covariates. We modeled transect‐level occupancy of SLF as a function of infestation level, habitat type, topographic position index, and distance to forests. Occupancy probability of SLF was higher on vines within vineyards than in forests, and occupancy declined with increasing distance from forests, which is informative for future search efforts. Detection probability of SLF was lower at forested sites but was higher at high infestation sites. Detection dogs had a lower detection probability than humans in the vineyards, but the detection probability of dogs was >3× greater than that of humans in forested sites. Our study suggests that detection dogs are more effective than human visual searches as an early detection method for SLF in forested areas, and utilizing detector dogs could strengthen SLF early detection efforts. This study demonstrates the potential applicability of using canine‐assisted search strategies combined with occupancy models to enhance the surveillance and prevention of other difficult‐to‐detect invasive species.

Unprecedented rates of biodiversity loss and ecosystem function necessitate the use of rapid, efficacious, and cost-effective biomonitoring tools. The combination of autonomous samplers and high throughput sequencing (i.e., “metabarcoding”) of environmental DNA (eDNA) samples enables characterization of entire communities at high frequency and can be an important tool for conservation and management, allowing researchers to track fluctuations in biodiversity. We deployed two autonomous samplers at two U.S. Geological Survey streamgage sites in the upper Snake River (Wyoming and Idaho, USA) to collect eDNA samples from July-September 2021 and 2022 to characterize fish diversity. We used a probabilistic approach to evaluate the effects of water temperature, water discharge, filter pore size, water volume filtered, number of samples collected, timing, and primers on the probability of detecting eDNA from fish species known to be present. We detected eDNA from 13/15 species present in these areas of the Snake River. Overall, we did not find evidence that filter pore size, water volume filtered, water discharge, and water temperature affected the probability of detecting fish species’ eDNA. By contrast, primers and sampling day affected fish detection probabilities, indicating that primer choice and sampling day can either over- or under- estimate species diversity. These results indicate that users would ideally consider sampling on non-consecutive days and which primer set will maximize species detections.

Species interactions shape biodiversity patterns, community assemblage, and the dynamics of wildlife populations. Ecological theory posits that the strength of interspecific interactions is fundamentally underpinned by the population sizes of the involved species. Nonetheless, prevalent approaches for modeling species interactions predominantly center around occupancy states. Here, we use simulations to illuminate the inadequacies of modeling species interactions solely as a function of occupancy, as is common practice in ecology. We demonstrate erroneous inference into species interactions due to error in parameter estimates when considering species occupancy alone. To address this critical issue, we propose, develop, and demonstrate an abundance-mediated interaction framework designed explicitly for modeling species interactions involving two or more species from detection/non-detection data. We present Markov chain Monte Carlo (MCMC) samplers tailored for diverse ecological scenarios, including intraguild predation, disease- or predator-mediated competition, and trophic cascades. Illustrating the practical implications of our approach, we compare inference from modeling the interactions in a three-species network involving coyotes (Canis latrans), fishers (Pekania pennanti), and American marten (Martes americana) in North America as a function of occupancy states and as a function of abundance. When modeling interactions as a function of abundance rather than occupancy, we uncover previously unidentified interactions. Our study emphasizes that accounting for abundance-mediated interactions rather than simple co-occurrence patterns can fundamentally alter our comprehension of system dynamics. Through an empirical case study and comprehensive simulations, we demonstrate the importance of accounting for abundance when modeling species interactions, and we present a statistical framework equipped with MCMC samplers to achieve this paradigm shift in ecological research.

Environmental DNA (eDNA) analysis has become a transformative technology, but sample collection methods lack standardization and sampling at effective frequencies requires considerable field effort. Autonomous eDNA samplers that can sample water at high frequencies offer potential solutions to these problems. We present results from four case studies using a prototype autonomous eDNA sampler as part of the U.S. Geological Survey's Rapid Environmental eDNA Assessment and Deployment Initiative & Network (READI-Net) project. These case studies involved short-term deployments of an eDNA autosampler (Smith-Root) across a range of riverine habitats with the objectives of (a) identifying what insights could be gained from high-frequency autosampling and (b) benchmarking these autosamples against manually collected samples. The high frequency autosampling revealed high temporal variability of eDNA concentrations and provided valuable insights about eDNA associations with environmental covariates, such as discharge and turbidity. Benchmarking assessments indicated autosamples had similar detection rates to manual samples and obtained similar or greater eDNA quantities. We did find minimal carryover contamination in autosampler field controls. We conclude that eDNA autosamplers have potential to improve freshwater biosurveillance by reducing logistical sampling barriers, standardizing collection methods, and clarifying the influence of environmental covariates on eDNA results.

The U.S. Fish and Wildlife Service monitors species-specific waterfowl (ducks, seaducks, geese, and brant) harvest through two hunter surveys, one that estimates the total harvest for each waterfowl group, and a second that estimates the species composition of each waterfowl group. Point estimates for species-specific harvest can be computed by multiplying the estimated total harvest by the estimated proportion of the total harvest of each species. However, to date, no uncertainty estimates have been available. Here, we combine these two data sources to provide species-specific harvest estimates at the state and flyway level while characterizing the uncertainty via Bayesian estimation. We take a similar approach to Smith et al. (2022), providing both estimates that treat yearly data as independent and estimates that share information across years via a random walk process. We then discuss the advantages and disadvantages of each approach. ### Competing Interest Statement The authors have declared no competing interest.

Environmental DNA (eDNA) sampling is an increasingly important tool for answering ecological questions and informing aquatic species management; however, several factors currently limit the reliability of ecological inference from eDNA sampling. Two particular challenges are (1) determining species source location(s) and (2) accurately and precisely measuring low concentration eDNA samples in the presence of multiple sources of ecological and measurement variability. The recently introduced eDNA Integrating Transport and Hydrology (eDITH) model provides a framework for relating eDNA measurements to source locations in riverine networks, but little empirical work has been done to test and refine model assumptions or accommodate low concentration samples, that can be systematically undermeasured. To better understand eDNA fate and transport dynamics and our ability to reliably quantify low concentration samples, we developed a hierarchical model and used it to evaluate a fate and transport experiment. Our model addresses several low concentration challenges by modeling the number of copies in each PCR replicate as a latent variable with a count distribution and conditioning detection and quantification on replicate copy number. We provide evidence that the eDNA removal rate declined through time, estimating that over 80

Free-roaming cats are a conservation concern in many areas but identifying their impacts and developing mitigation strategies requires a robust understanding of their distribution and density patterns. Urban and residential areas may be especially relevant in this process because free-roaming cats are abundant in these anthropogenic landscapes. Here, we estimate the occupancy and density of free-roaming cats in Washington D.C. and relate these metrics to known landscape and social factors. We conducted an extended camera trap survey of public and private spaces across D.C. and analyzed data collected from 1483 camera deployments from 2018 to 2020. We estimated citywide cat distribution by fitting hierarchical occupancy models and further estimated cat abundance using a novel random thinning spatial capture-recapture model that allows for the use of photos that can and cannot be identified to individual. Within this model, we utilized individual covariates that provided identity exclusions between photos of unidentifiable cats with inconsistent coat patterns, thus increasing the precision of abundance estimates. This combined model also allowed for unbiased estimation of density when animals cannot be identified to individual at the same rate as for free-roaming cats whose identifiability depended on their coat characteristics. Cat occupancy and abundance declined with increasing distance from residential areas, an effect that was more pronounced in wealthier neighborhoods. There was noteworthy absence of cats detected in larger public spaces and forests. Realized densities ranged from 0.02 to 1.75 cats/ha in sampled areas, resulting in a district-wide estimate of ~7296 free-roaming cats. Ninety percent of cat detections lacked collars and nearly 35% of known individuals were ear-tipped, indicative of district Trap-Neuter-Return (TNR) programs. These results suggest that we mainly sampled and estimated the unowned cat subpopulation, such that indoor/outdoor housecats were not well represented. The precise estimation of cat population densities is difficult due to the varied behavior of subpopulations within free-roaming cat populations (housecats, stray and feral cats), but our methods provide a first step in establishing citywide baselines to inform data-driven management plans for free-roaming cats in urban environments.

Aerial count surveys of wildlife populations are a prominent monitoring method for many wildlife species. Traditionally, these surveys utilize human observers to detect, count, and classify observations to species. However, given recent technological advances, many research groups are exploring the combined use of remote sensing and deep learning methods to replace human observers in order to improve data quality and reproducibility, reduce disturbance to wildlife, and increase aircrew safety. Given that deep learning detection and classification are not perfect and that statistical inference from ecological models is generally very sensitive to misclassification, we require study designs and statistical models to accommodate these observation errors. As part of an ongoing effort by the U.S. Fish and Wildlife Service, Bureau of Ocean Energy Management, and U.S. Geological Survey to survey marine birds and other marine wildlife using digital aerial imagery and deep learning object detection and classification, we developed a general hierarchical model for estimating species-specific abundance that accommodates object-level errors in classification. We consider hierarchical deep learning classification at multiple taxonomic levels subject to misclassification, hierarchically-structured human validation data subject to partial and erroneous misclassification, and an image censoring process leading to preferential sampling. We demonstrate that this model can estimate species-specific abundance and habitat relationships without bias when the assumptions are met, and we discuss the plausibility of these assumptions in practice for this study and others like it. Finally, we use this model to demonstrate the relevance of the features of the ecological systems under study to the classification task itself. In models that couple the ecological and classification processes into a single, hierarchical model, the true classes are treated as latent variables to be estimated and are informed by both the classification probability parameters and the ecological parameters that determine the expected frequencies of each class at the level the data are being modeled (e.g., site or site by occasion). We show that ignoring the expected frequencies of each class (when they are imbalanced) can cause correction for misclassification to produce biased parameter estimates, but coupling the ecological and classification models allows for the variability in relative class frequency across space and time due to ecological and sampling conditions to be accommodated with spatial or temporal covariates. As a result, bias is removed, classification is more accurate, and uncertainty is propagated between the ecological and classification models. We therefore argue that ability of deep learning classifiers, and classifiers more generally, to produce reliable ecological inference depends, in part, on the ecological system under study.

Designing a population monitoring program for Asian bears presents challenges associated with their low densities and detectability, generally large home ranges, and logistical or resource constraints. The use of an occupancy-based method to monitor bear populations can be appropriate under certain conditions given the mechanistic relationship between occupancy and abundance. The form of the occupancy–abundance relationship is dependent on species-specific characteristics such as home range size and population density, as well as study area size. To assess the statistical power of tests to detect population change of Asian bears, we conducted a study using a range of scenarios by simulating spatially explicit individual-based capture-recapture data from a demographically open model. Simulations assessed the power to detect changes in population density via changes in site-level occupancy or abundance through time, estimated using a standard occupancy model or a Royle-Nichols model, both with point detectors (representing camera traps). We used IUCN Red List criteria as a guide in selection of two population decline scenarios (20% and 50%), but we chose a shorter time horizon (10 years = 1 bear generation), meaning that declines were steeper than used for IUCN criteria (3 generations). Our simulations detected population declines of 50% with high power (>0.80) and low false positive rates (FPR: incorrectly detecting a decline) (<0.10) when detectors were spaced at > 0.67 times the home range diameter (home-range spacing ratio: HRSR, a measure of spatial correlation), such that bears would tend to overlap no more than two detectors. There was high (0.85) correlation between realized occupancy and N in these scenarios. The FPR increased as the HRSR decreased because of spatial correlation in the occupancy process induced when individual home ranges overlap multiple detectors. The mean statistical power to detect more gradual population declines (20% in 10 years) with HRSR > 0.67 was low for occupancy models 0.22 (maximum power 0.67) and Royle-Nichols models (0.24; maximum power 0.67), suggesting that declines of this magnitude may not be described reliably with 10 years of monitoring. Our results demonstrated that under many realistic scenarios that we explored, false positive rates were unacceptably high. We highlight that when designing occupancy studies, the spacing between point detectors be at least 0.67 times the diameter of the home range size of the larger sex (e.g., males) when the assumptions of the spatial capture-recapture model used for simulation are met.

Abstract Effective monitoring of wildlife populations forms the foundation of modern‐day conservation biology. Without reliable estimates of population size, it is not possible to determine population trends, a key requirement in determining species status under international legislation. Carnivores are one of the more difficult taxonomic groups to monitor due to low population densities and elusive behavior. Here, we compare conventional live trapping and two more modern, noninvasive field methods of population estimation: genetic fingerprinting from hair tube sampling and camera trapping for the pine marten (Martes martes). We apply marked spatial capture–recapture (SCR) models to the genetic and live‐trapping data where individuals were identifiable, and unmarked SCR (uSCR), camera‐trap distance sampling (CT‐DS), and random encounter models (REMs) to the camera‐trap data where individual ID was not possible. All five approaches produced plausible and relatively consistent point estimates (0.49–1.20 individuals/km2) despite differences in precision, cost, and effort being apparent. Genetic fingerprinting produced the most precise estimate out of the two approaches for marked animal populations and had the key benefit of being noninvasive but was the most expensive of all the methods. Live trapping produced the highest point estimate while being cheapest, but the most labor intensive and least precise. The camera‐trapping methods for unmarked animal populations were the most time efficient and precise except uSCR with a moderately informative prior (uSCRm), which produced the second least precise density estimate of all the methods compared. The CT‐DS produced the most precise estimate of all the methods, followed by REM and then uSCR with a strongly informative prior (uSCRs). While choice of method of density estimation depends on objectives and funding constraints, as well as the species of interest, we demonstrate the importance of using a priori knowledge of target species and consideration of planned statistical analysis to produce appropriate experimental designs with critical consideration required regarding trap spacing and spatial extent. Such considerations broaden the comparability and applicability of these methods and will serve to provide key reference estimates for researchers, wildlife managers, and non‐governmental organizations involved in monitoring wildlife populations.

Group living in species can have complex consequences for individuals, populations, and ecosystems. Therefore, estimating group density and size is often essential for understanding population dynamics, interspecific interactions, and conservation needs of group-living species. Spatial capture-recapture (SCR) has been used to model both individual and group density in group-living species, but modeling either individual-level or group-level detection results in different biases due to common characteristics of group-living species, such as highly cohesive movement or variation in group size. Furthermore, no SCR method currently estimates group density, individual density, and group size jointly. Using clustered point processes, we developed a cluster SCR model to estimate group density, individual density, and group size. We compared the model to standard SCR models using both a simulation study and a data set of detections of African wild dogs (Lycaon pictus), a group-living carnivore, on camera traps in northern Botswana. We then tested the model's performance under various scenarios of group movement in a separate simulation study. We found that the cluster SCR model outperformed a standard group-level SCR model when fitted to data generated with varying group sizes, and mostly recovered previous estimates of wild dog group density, individual density, and group size. We also found that the cluster SCR model performs better as individuals' movements become more correlated with their groups' movements. The cluster SCR model offers opportunities to investigate ecological hypotheses relating group size to population dynamics while accounting for cohesive movement behaviors in group-living species.

Abstract Spatial capture–recapture (SCR) models have increasingly been used as a basis for combining capture–recapture data types with variable levels of individual identity information to estimate population density and other demographic parameters. Recent examples are the unmarked SCR (or spatial count model), where no individual identities are available and spatial mark–resight (SMR) where individual identities are available for only a marked subset of the population. Currently lacking, though, is a model that allows unidentified samples to be combined with identified samples when there are no separate classes of “marked” and “unmarked” individuals and when the two sample types cannot be considered as arising from two independent observation models. This is a common scenario when using noninvasive sampling methods, for example, when analyzing data on identified and unidentified photographs or scats from the same sites. Here we describe a “random thinning” SCR model that utilizes encounters of both known and unknown identity samples using a natural mechanistic dependence between samples arising from a single observation model. Our model was fitted in a Bayesian framework using NIMBLE. We investigate the improvement in parameter estimates by including the unknown identity samples, which was notable (up to 79% more precise) in low‐density populations with a low rate of identified encounters. We then applied the random thinning SCR model to a noninvasive genetic sampling study of brown bear (Ursus arctos) density in Oriental Cantabrian Mountains (North Spain). Our model can improve density estimation for noninvasive sampling studies for low‐density populations with low rates of individual identification, by making use of available data that might otherwise be discarded.

Modeling the population dynamics of patchily distributed species is a challenge, particularly when inference must be based on incomplete and small data sets such as those from most species of conservation concern. Open population spatial capture-recapture (SCR) models are ideally suited to quantify population trends, but have seen only limited use since their introduction. To investigate population trend and sex-specific population dynamics, we applied an open SCR model to a capercaillie ( Tetrao urogallus ) population in Switzerland living in eight distinct forest patches totalling 22 km 2 within a region of 908 km 2 . The population was surveyed using genetic sampling of scat in 2009, 2012 and 2015. We fit an open SCR model with sex-specific detection and population dynamics parameters while accounting for the patchy distribution of habitat and the uncertainty introduced by observing the population in three years only. Between 2009 and 2015, a total of 143 males, 112 females and 4 individuals of uncertain sex were detected. The annual per capita recruitment rate was estimated at 0.115 (SE 0.0144) for males and 0.127 (0.0168) for females. The estimated annual survival probability for males was 0.758 (0.0241) and 0.707 (0.0356) for females. The population trajectory implied by these survival and recruitment estimates was a decline of 2% per year; however, the sex specificity of the model revealed a decline in the male population only, with no evidence of decline in the female population. Further, the population decline observed in males was explained by the demography of just two of the eight patches. Using a customized open population SCR model, we determined that the endangered capercaillie in our Swiss study area had a stable female population and a declining male population, with the male decline due to population dynamics in a subset of the study area. Our study highlights the flexibility of open population SCR models for assessing population trajectories through time and across space and emphasizes the desirability of estimating sex-stratified population trends especially in species of conservation concern.

Accelerating declines of an increasing number of animal populations worldwide necessitate methods to reliably and efficiently estimate demographic parameters such as population density and trajectory. Standard methods for estimating demographic parameters from noninvasive genetic samples are inefficient because lower-quality samples cannot be used, and they assume individuals are identified without error. We introduce the genotype spatial partial identity model (gSPIM), which integrates a genetic classification model with a spatial population model to combine both spatial and genetic information, thus reducing genotype uncertainty and increasing the precision of demographic parameter estimates. We apply this model to data from a study of fishers (Pekania pennanti) in which 37% of hair samples were originally discarded because of uncertainty in individual identity. The gSPIM density estimate using all collected samples was 25% more precise than the original density estimate, and the model identified and corrected three errors in the original individual identity assignments. A simulation study demonstrated that our model increased the accuracy and precision of density estimates 63 and 42%, respectively, using three replicated assignments (e.g., PCRs for microsatellites) per genetic sample. Further, the simulations showed that the gSPIM model parameters are identifiable with only one replicated assignment per sample and that accuracy and precision are relatively insensitive to the number of replicated assignments for high-quality samples. Current genotyping protocols devote the majority of resources to replicating and confirming high-quality samples, but when using the gSPIM, genotyping protocols could be more efficient by devoting more resources to low-quality samples.

Recently introduced unmarked spatial capture-recapture (SCR), spatial mark-resight (SMR), and 2-flank spatial partial identity models (SPIM) extend the domain of SCR to populations or observation systems that do not always allow for individual identity to be determined with certainty. For example, some species do not have natural marks that can reliably produce individual identities from photographs, and some methods of observation produce partial identity samples as is the case with remote cameras that sometimes produce single flank photographs. These models share the feature that they probabilistically resolve the uncertainty in individual identity using the spatial location where samples were collected. Spatial location is informative of individual identity in spatially structured populations with home range sizes smaller than the extent of the trapping array because a latent identity sample is more likely to have been produced by an individual living near the trap where it was recorded than an individual living further away from the trap. Further, the level of information about individual identity that a spatial location contains is determined by two key ecological concepts, population density and home range size. The number of individuals that could have produced a latent or partial identity sample increases as density and home range size increase because more individual home ranges will overlap any given trap. We show this uncertainty can be quantified using a metric describing the expected magnitude of uncertainty in individual identity for any given population density and home range size, the Identity Diversity Index (IDI). We then show that the performance of latent and partial identity SCR models varies as a function of this index and produces imprecise and biased estimates in many high IDI scenarios when data are sparse. We then extend the unmarked SCR model to incorporate partially identifying covariates which reduce the level of uncertainty in individual identity, increasing the reliability and precision of density estimates, and allowing reliable density estimation in scenarios with higher IDI values and with more sparse data. We illustrate the performance of this “categorical SPIM” via simulations and by applying it to a black bear data set using microsatellite loci as categorical covariates, where we reproduce the full data set estimates with only slightly less precision using fewer loci than necessary for confident individual identification. The categorical SPIM offers an alternative to using probability of identity criteria for classifying genotypes as unique, shifting the “shadow effect”, where more than one individual in the population has the same genotype, from a source of bias to a source of uncertainty. We discuss the difficulties that real world data sets pose for latent identity SCR methods, most importantly, individual heterogeneity in detection function parameters, and argue that the addition of partial identity information reduces these concerns. We then discuss how the categorical SPIM can be applied to other wildlife sampling scenarios such as remote camera surveys, where natural or researcher-applied partial marks can be observed in photographs. Finally, we discuss how the categorical SPIM can be added to SMR, 2-flank SPIM, or other future latent identity SCR models.

Habitat loss and overexploitation extirpated American black bears (Ursus americanus) from most of the Central Appalachians, USA, by the early 20th Century. To attempt to restore bears to the southwestern portion of this region, 2 reintroductions that used small founder groups (n = 27 and 55 bears), but different release methods (hard vs. soft), were conducted during the 1990s. We collected hair samples from black bears during 2004-2016 in the reintroduced Big South Fork (BSF) and Kentucky- Virginia populations (KVP), their respective Great Smoky Mountains (GSM) and Shenandoah National Park (SNP) source populations, and a neighboring population in southern West Virginia (SWV) to investigate the early genetic outcomes of bear reintroduction. Despite having undergone genetic bottlenecks, genetic diversity remained similar between reintroduced populations and their sources approximately 15 years after the founder events (ranges: A(R) = 4.86-5.61; H-O = 0.67-0.75; H-E = 0.65-0.71). Effective population sizes of the reintroduced KVP and BSF (N-E = 31 and 36, respectively) were substantially smaller than their respective SNP and GSM sources (N-E = 119 and 156, respectively), supporting founder effects. Genetic structure analysis indicated that the hard-released (i.e., no acclimation period) KVP founder group likely declined considerably, whereas the soft-released BSF founder group remained mostly intact, suggesting superior effectiveness of soft releases. Asymmetrical gene flow via immigration from the SWV has resulted in the KVP recovering from the initial founder group reduction. Sustained isolation, small N-E, and small population size of the BSF may warrant continued genetic monitoring to determine if gene flow from neighboring populations is established or N-E declines. For future bear reintroductions, we suggest managers consider sourcing founders from populations with high genetic diversity and soft-releasing bears to locales that are, if possible, within the dispersal capability of extant populations to mitigate the potential consequences of founder effects and isolation.