Comparison of photo‐matching algorithms commonly used for photographic capture–recapture studies

Photographic capture–recapture is a valuable tool for obtaining demographic information on wildlife populations due to its noninvasive nature and cost‐effectiveness. Recently, several computer‐aided photo‐matching algorithms have been developed to more efficiently match images of unique individuals in databases with thousands of images. However, the identification accuracy of these algorithms can severely bias estimates of vital rates and population size. Therefore, it is important to understand the performance and limitations of state‐of‐the‐art photo‐matching algorithms prior to implementation in capture–recapture studies involving possibly thousands of images. Here, we compared the performance of four photo‐matching algorithms; Wild‐ID, I3S Pattern+, APHIS, and AmphIdent using multiple amphibian databases of varying image quality. We measured the performance of each algorithm and evaluated the performance in relation to database size and the number of matching images in the database. We found that algorithm performance differed greatly by algorithm and image database, with recognition rates ranging from 100% to 22.6% when limiting the review to the 10 highest ranking images. We found that recognition rate degraded marginally with increased database size and could be improved considerably with a higher number of matching images in the database. In our study, the pixel‐based algorithm of AmphIdent exhibited superior recognition rates compared to the other approaches. We recommend carefully evaluating algorithm performance prior to using it to match a complete database. By choosing a suitable matching algorithm, databases of sizes that are unfeasible to match “by eye” can be easily translated to accurate individual capture histories necessary for robust demographic estimates.     

A spatially explicit capture–recapture estimator for single‐catch traps

Single‐catch traps are frequently used in live‐trapping studies of small mammals. Thus far, a likelihood for single‐catch traps has proven elusive and usually the likelihood for multicatch traps is used for spatially explicit capture–recapture (SECR) analyses of such data. Previous work found the multicatch likelihood to provide a robust estimator of average density. We build on a recently developed continuous‐time model for SECR to derive a likelihood for single‐catch traps. We use this to develop an estimator based on observed capture times and compare its performance by simulation to that of the multicatch estimator for various scenarios with nonconstant density surfaces. While the multicatch estimator is found to be a surprisingly robust estimator of average density, its performance deteriorates with high trap saturation and increasing density gradients. Moreover, it is found to be a poor estimator of the height of the detection function. By contrast, the single‐catch estimators of density, distribution, and detection function parameters are found to be unbiased or nearly unbiased in all scenarios considered. This gain comes at the cost of higher variance. If there is no interest in interpreting the detection function parameters themselves, and if density is expected to be fairly constant over the survey region, then the multicatch estimator performs well with single‐catch traps. However if accurate estimation of the detection function is of interest, or if density is expected to vary substantially in space, then there is merit in using the single‐catch estimator when trap saturation is above about 60%. The estimator's performance is improved if care is taken to place traps so as to span the range of variables that affect animal distribution. As a single‐catch likelihood with unknown capture times remains intractable for now, researchers using single‐catch traps should aim to incorporate timing devices with their traps.     

Integrating high‐speed videos in capture‐mark‐recapture studies of insects

Capture–mark–recapture (CMR) studies have been used extensively in ecology and evolution. While it is feasible to apply CMR in some animals, it is considerably more challenging in small fast‐moving species such as insects. In these groups, low recapture rates can bias estimates of demographic parameters, thereby handicapping effective analysis and management of wild populations. Here, we use high‐speed videos (HSV) to capture two large dragonfly species, Anax junius and Rhionaeschna multicolor, that rarely land and, thus, are particularly challenging for CMR studies. We test whether HSV, compared to conventional “eye” observations, increases the “resighting” rates and, consequently, improves estimates of both survival rates and the effects of demographic covariates on survival. We show that the use of HSV increases the number of resights by 64% in A. junius and 48% in R. multicolor. HSV improved our estimates of resighting and survival probability which were either under‐ or overestimated with the conventional observations. Including HSV improved credible intervals for resighting rate and survival probability by 190% and 130% in A. junius and R. multicolor, respectively. Hence, it has the potential to open the door to a wide range of research possibilities on species that are traditionally difficult to monitor with distance sampling, including within insects and birds.     

A local evaluation of the individual state‐space to scale up Bayesian spatial capture–recapture

Spatial capture–recapture models (SCR) are used to estimate animal density and to investigate a range of problems in spatial ecology that cannot be addressed with traditional nonspatial methods. Bayesian approaches in particular offer tremendous flexibility for SCR modeling. Increasingly, SCR data are being collected over very large spatial extents making analysis computational intensive, sometimes prohibitively so. To mitigate the computational burden of large‐scale SCR models, we developed an improved formulation of the Bayesian SCR model that uses local evaluation of the individual state‐space (LESS). Based on prior knowledge about a species’ home range size, we created square evaluation windows that restrict the spatial domain in which an individual's detection probability (detector window) and activity center location (AC window) are estimated. We used simulations and empirical data analyses to assess the performance and bias of SCR with LESS. LESS produced unbiased estimates of SCR parameters when the AC window width was ≥5σ (σ: the scale parameter of the half‐normal detection function), and when the detector window extended beyond the edge of the AC window by 2σ. Importantly, LESS considerably decreased the computation time needed for fitting SCR models. In our simulations, LESS increased the computation speed of SCR models up to 57‐fold. We demonstrate the power of this new approach by mapping the density of an elusive large carnivore—the wolverine (Gulo gulo)—with an unprecedented resolution and across the species’ entire range in Norway (> 200,000 km²). Our approach helps overcome a major computational obstacle to population and landscape‐level SCR analyses. The LESS implementation in a Bayesian framework makes the customization and fitting of SCR accessible for practitioners working at scales that are relevant for conservation and management.     

multimark: an R package for analysis of capture–recapture data consisting of multiple “noninvasive” marks

I describe an open‐source R package, multimark , for estimation of survival and abundance from capture–mark–recapture data consisting of multiple “noninvasive” marks. Noninvasive marks include natural pelt or skin patterns, scars, and genetic markers that enable individual identification in lieu of physical capture. multimark provides a means for combining and jointly analyzing encounter histories from multiple noninvasive sources that otherwise cannot be reliably matched (e.g., left‐ and right‐sided photographs of bilaterally asymmetrical individuals). The package is currently capable of fitting open population Cormack–Jolly–Seber (CJS) and closed population abundance models with up to two mark types using Bayesian Markov chain Monte Carlo (MCMC) methods. multimark can also be used for Bayesian analyses of conventional capture–recapture data consisting of a single‐mark type. Some package features include (1) general model specification using formulas already familiar to most R users, (2) ability to include temporal, behavioral, age, cohort, and individual heterogeneity effects in detection and survival probabilities, (3) improved MCMC algorithm that is computationally faster and more efficient than previously proposed methods, (4) Bayesian multimodel inference using reversible jump MCMC, and (5) data simulation capabilities for power analyses and assessing model performance. I demonstrate use of multimark using left‐ and right‐sided encounter histories for bobcats (Lynx rufus) collected from remote single‐camera stations in southern California. In this example, there is evidence of a behavioral effect (i.e., trap “happy” response) that is otherwise indiscernible using conventional single‐sided analyses. The package will be most useful to ecologists seeking stronger inferences by combining different sources of mark–recapture data that are difficult (or impossible) to reliably reconcile, particularly with the sparse datasets typical of rare or elusive species for which noninvasive sampling techniques are most commonly employed. Addressing deficiencies in currently available software, multimark also provides a user‐friendly interface for performing Bayesian multimodel inference using capture–recapture data consisting of a single conventional mark or multiple noninvasive marks.     

Spatial capture–recapture with random thinning for unidentified encounters

1. 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.
2. 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.
3. 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).
4. 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.

     

The use of a non‐invasive tool for capture–recapture studies on a seahorse Hippocampus guttulatus population

In this study, the spot pattern in Hippocampus guttulatus was analysed using a computer programme algorithm that allowed individual comparison. This methodology was first tested in a controlled environment using 51 adult and 55 juvenile H. guttulatus. Positive matches were obtained in 86·3 and 83·6% of the adults and juveniles, respectively. In a second experiment, monthly surveys were carried out in five selected locations in the Ria Formosa Lagoon, south Portugal, over the course of a year and a total of 980 photographs were analysed. Photographed H. guttulatus were re‐sighted one to nine times during the course of the survey period with an overall re‐sight record of over 30%. Photo‐identification was therefore shown to be a useful tool for non‐invasive mark–recapture studies that can be successfully used to survey the population abundance of H. guttulatus aged 6 months or older in consecutive years. This could be of great value when considering the assessment of H. guttulatus populations and understanding changes over time.     

Complete tag loss in capture–recapture studies affects abundance estimates: An elephant seal case study

1. In capture–recapture studies, recycled individuals occur when individuals lose all of their tags and are recaptured as though they were new individuals. Typically, the effect of these recycled individuals is assumed negligible.
2. Through a simulation‐based study of double‐tagging experiments, we examined the effect of recycled individuals on parameter estimates in the Jolly–Seber model with tag loss (Cowen & Schwarz, 2006). We validated the simulation framework using long‐term census data of elephant seals.
3. Including recycled individuals did not affect estimates of capture, survival, and tag‐retention probabilities. However, with low tag‐retention rates, high capture rates, and high survival rates, recycled individuals produced overestimates of population size. For the elephant seal case study, we found population size estimates to be between 8% and 53% larger when recycled individuals were ignored.
4. Ignoring the effects of recycled individuals can cause large biases in population size estimates. These results are particularly noticeable in longer studies.

     

State–space mark–recapture estimates reveal a recent decline in abundance of North Atlantic right whales

North Atlantic right whales (Eubalaena glacialis Müller 1776) present an interesting problem for abundance and trend estimation in marine wildlife conservation. They are long lived, individually identifiable, highly mobile, and one of the rarest of cetaceans. Individuals are annually resighted at different rates, primarily due to varying stay durations among several principal habitats within a large geographic range. To date, characterizations of abundance have been produced that use simple accounting procedures with differing assumptions about mortality. To better characterize changing abundance of North Atlantic right whales between 1990 and 2015, we adapted a state–space formulation with Jolly‐Seber assumptions about population entry (birth and immigration) to individual resighting histories and fit it using empirical Bayes methodology. This hierarchical model included accommodation for the effect of the substantial individual capture heterogeneity. Estimates from this approach were only slightly higher than published accounting procedures, except for the most recent years (when recapture rates had declined substantially). North Atlantic right whales' abundance increased at about 2.8% per annum from median point estimates of 270 individuals in 1990 to 483 in 2010, and then declined to 2015, when the final estimate was 458 individuals (95% credible intervals 444–471). The probability that the population's trajectory post‐2010 was a decline was estimated at 99.99%. Of special concern was the finding that reduced survival rates of adult females relative to adult males have produced diverging abundance trends between sexes. Despite constraints in recent years, both biological (whales' distribution changing) and logistical (fewer resources available to collect individual photo‐identifications), it is still possible to detect this relatively recent, small change in the population's trajectory. This is thanks to the massive dataset of individual North Atlantic right whale identifications accrued over the past three decades. Photo‐identification data provide biological information that allows more informed inference on the status of this species.     

Genotyping validates the efficacy of photographic identification in a capture‐mark‐recapture study based on the head scale patterns of the prairie lizard (Sceloporus consobrinus)

Population studies often incorporate capture‐mark‐recapture (CMR) techniques to gather information on long‐term biological and demographic characteristics. A fundamental requirement for CMR studies is that an individual must be uniquely and permanently marked to ensure reliable reidentification throughout its lifespan. Photographic identification involving automated photographic identification software has become a popular and efficient noninvasive method for identifying individuals based on natural markings. However, few studies have (a) robustly assessed the performance of automated programs by using a double‐marking system or (b) determined their efficacy for long‐term studies by incorporating multi‐year data. Here, we evaluated the performance of the program Interactive Individual Identification System (I³S) by cross‐validating photographic identifications based on the head scale pattern of the prairie lizard (Sceloporus consobrinus) with individual microsatellite genotyping (N = 863). Further, we assessed the efficacy of the program to identify individuals over time by comparing error rates between within‐year and between‐year recaptures. Recaptured lizards were correctly identified by I³S in 94.1% of cases. We estimated a false rejection rate (FRR) of 5.9% and a false acceptance rate (FAR) of 0%. By using I³S, we correctly identified 97.8% of within‐year recaptures (FRR = 2.2%; FAR = 0%) and 91.1% of between‐year recaptures (FRR = 8.9%; FAR = 0%). Misidentifications were primarily due to poor photograph quality (N = 4). However, two misidentifications were caused by indistinct scale configuration due to scale damage (N = 1) and ontogenetic changes in head scalation between capture events (N = 1). We conclude that automated photographic identification based on head scale patterns is a reliable and accurate method for identifying individuals over time. Because many lizard or reptilian species possess variable head squamation, this method has potential for successful application in many species.     

Assessment of individual and conspecific reproductive success as determinants of breeding dispersal of female tree swallows: A capture–recapture approach

Breeding dispersal is a key process of population structure and dynamics and is often triggered by an individual's breeding failure. In both colonial and territorial birds, reproductive success of conspecifics (RSc) can also lead individuals to change breeding sites after a failure on a site. Yet, few studies have simultaneously investigated the independent contribution of individual reproductive success (RSi) and of RSc on dispersal decision. Here, we develop a modeling framework to disentangle the effects of RSi and RSc on demographic parameters, while accounting for imperfect individual detection and other confounding factors such as age or dispersal behavior in the previous year. Using a 10‐year capture–recapture dataset composed of 1,595 banded tree swallows, we assessed the effects of nonmanipulated RSi and RSc on female breeding dispersal in this semicolonial passerine. Dispersal was strongly driven by RSi, but not by RSc. Unsuccessful females were 9.5–2.5 times more likely to disperse than successful ones, depending if they had dispersed or not in the previous year, respectively. Unsuccessful females were also three times less likely to be detected than successful ones. Contrary to theoretical and empirical studies, RSc did not drive the decision to disperse but influenced the selection of the following breeding site once dispersal had been initiated. Because detection of individuals was driven by RSi, which was positively correlated to RSc, assuming a perfect detection as in previous studies may have lead us to conclude that RSc affected dispersal patterns, yet our approach corrected for this bias. Overall, our results suggest that the value and use of RSc as public information to guide dispersal decisions are likely dictated by multiple ecological determinants, such as landscape structure and extent, if this cue is indeed used.     

Use of hidden Markov capture–recapture models to estimate abundance in the presence of uncertainty: Application to the estimation of prevalence of hybrids in animal populations

Estimating the relative abundance (prevalence) of different population segments is a key step in addressing fundamental research questions in ecology, evolution, and conservation. The raw percentage of individuals in the sample (naive prevalence) is generally used for this purpose, but it is likely to be subject to two main sources of bias. First, the detectability of individuals is ignored; second, classification errors may occur due to some inherent limits of the diagnostic methods. We developed a hidden Markov (also known as multievent) capture–recapture model to estimate prevalence in free‐ranging populations accounting for imperfect detectability and uncertainty in individual's classification. We carried out a simulation study to compare naive and model‐based estimates of prevalence and assess the performance of our model under different sampling scenarios. We then illustrate our method with a real‐world case study of estimating the prevalence of wolf (Canis lupus) and dog (Canis lupus familiaris) hybrids in a wolf population in northern Italy. We showed that the prevalence of hybrids could be estimated while accounting for both detectability and classification uncertainty. Model‐based prevalence consistently had better performance than naive prevalence in the presence of differential detectability and assignment probability and was unbiased for sampling scenarios with high detectability. We also showed that ignoring detectability and uncertainty in the wolf case study would lead to underestimating the prevalence of hybrids. Our results underline the importance of a model‐based approach to obtain unbiased estimates of prevalence of different population segments. Our model can be adapted to any taxa, and it can be used to estimate absolute abundance and prevalence in a variety of cases involving imperfect detection and uncertainty in classification of individuals (e.g., sex ratio, proportion of breeders, and prevalence of infected individuals).     

Known unknowns in an imperfect world: incorporating uncertainty in recruitment estimates using multi‐event capture–recapture models

Studying the demography of wild animals remains challenging as several of the critical parts of their life history may be difficult to observe in the field. In particular, determining with certainty when an individual breeds for the first time is not always obvious. This can be problematic because uncertainty about the transition from a prebreeder to a breeder state – recruitment – leads to uncertainty in vital rate estimates and in turn in population projection models. To avoid this issue, the common practice is to discard imperfect data from the analyses. However, this practice can generate a bias in vital rate estimates if uncertainty is related to a specific component of the population and reduces the sample size of the dataset and consequently the statistical power to detect effects of biological interest. Here, we compared the demographic parameters assessed from a standard multistate capture–recapture approach to the estimates obtained from the newly developed multi‐event framework that specifically accounts for uncertainty in state assessment. Using a comprehensive longitudinal dataset on southern elephant seals, we demonstrated that the multi‐event model enabled us to use all the data collected (6639 capture–recapture histories vs. 4179 with the multistate model) by accounting for uncertainty in breeding states, thereby increasing the precision and accuracy of the demographic parameter estimates. The multi‐event model allowed us to incorporate imperfect data into demographic analyses. The gain in precision obtained has important implications in the conservation and management of species because limiting uncertainty around vital rates will permit predicting population viability with greater accuracy.