Open Population Models Research Articles

Analysis of photo‐id data allowing for missed matches and individuals identified from opposite sides

Summary In many species, photo‐identification could be used as an alternative to artificial marking to provide data on demographic parameters. However, unless the population is very small or fragmented, software may be required to pre‐screen and reject most image pairs as potential matches. Depending on the species and method used to obtain images, currently available software may falsely reject some matches. We estimate the false rejection rate (FRR) of the ExtractCompare (EC) program when used to pre‐screen images of female grey seals. Filtering images manually to reduce the FRR involves subjective assessment of image quality, reduces the amount of data available and may bias the results in favour of relatively well‐marked individuals. The data may contain individuals identified only from the left side or the right side, as well as individuals identified from both sides. Missed matches resulting from false rejections by pre‐screening software and/or inclusion of individuals identified only from opposite sides cause some individuals to generate multiple encounter histories. We describe an open population model for data of this type which, given a measured risk of missing a match between a randomly selected pair of images of the same individual, provides maximum likelihood (ML) estimates of initial population size, survival/emigration and immigration/recruitment by calculating the expected frequency of any encounter history that could be generated. As a case study for the method, we used EC to pre‐screen photographs of female grey seals on a breeding colony and generate encounter histories over five successive seasons. Allowing for the measured FRR, we calculated ML estimates for comparison with estimates from previous studies. We also used the model with encounter histories simulated using the same FRR to give the same mixture of left side, right side and both sides histories and derived ML estimates for comparison with the values used to drive the simulation. With FRR set at up to 33%, the method gave estimates of the abundance and survival parameters used in the simulation model that were biased by at most 4·7% up and 3% down, respectively. The results of the grey seal case study were consistent with previous estimates of apparent survival and trends in abundance.

Evaluating the efficacy of non-invasive genetic methods and estimating wolf survival during a ten-year period

Genetic capture-recapture (CR) analysis is a highly promising tool to estimate population parameters and monitor populations through time. However, its level of accuracy has rarely been assessed and comparisons with traditional estimates in controlled settings have rarely been performed. We used CR analysis with long-term fecal genotyping data and applied open-population models to estimate survival rates and assess trend in abundance of wolves in Isle Royale over 10 years, while simultaneously estimating those parameters with traditional aerial-based techniques that are believed to be reasonably accurate. Comparison of the techniques indicated that there is a good correspondence of estimates only when the effort in genetic sampling is high, which guarantees a high recapture probability. Juvenile wolves had higher annual CR survival rates than adult wolves; kill rates most affected wolf sur- vival in this natural ecosystem, and higher annual kill rates were correlated with higher annual adult survival. Adult survival (but not juvenile survival) was an important pre- dictor of population growth rate. Hence, we show that kill rates indirectly affected population growth rate and directly affected adult survival rate. These reliable estimates of survival have unique value because the Isle Royale wolf population is not exposed to any human-caused mortality. Therefore, knowing long-term patterns of annual survival and its relationship to population growth rate for a not hunted wolf population represents a critical baseline for wolf conservation throughout its worldwide distribution.

Estimation of the density of the Near Threatened jaguarPanthera oncain Sonora, Mexico, using camera trapping and an open population model

AbstractOur objective in this study was to determine the density of the jaguarPanthera oncafrom camera-trap data, using an open population model, in a private protected natural area, the Northern Jaguar Reserve, and 10 adjoining cattle ranches in the state of Sonora, Mexico. The region is considered a long–term jaguar conservation unit. As well as being the most northerly recorded reproductive population of the jaguar, the arid habitat of this region is atypical for the species. During 16 months of sampling we identified 10 individual jaguars and the data met the three main assumptions of open population models. The estimated mean density was 1.05±SE 0.4 individuals per 100 km2, with a constant survival probability of 0.94 and capture probability of 0.23. This estimate of density is lower than reported in studies of the jaguar from more southerly locations in Mexico, Belize, Costa Rica, Bolivia and Brazil but cannot be attributed to a single factor even though in general there is an apparent relationship between jaguar density and precipitation. The main objectives of the management of the Northern Jaguar Reserve are to reduce the impact of cattle and restore jaguar habitat, with strategies focused on water retention, removal of invasive grass, reforestation and environmental education. Livestock have been gradually excluded since 2003 and, combined with the protection provided under the agreements with the surrounding ranches, the area is now a suitable place for long-term studies of the jaguar.

A computer‐assisted system for photographic mark–recapture analysis

Summary1. Photographic mark–recapture is a cost‐effective, non‐invasive way to study populations. However, to efficiently apply photographic mark–recapture to large populations, computer software is needed for image manipulation and pattern matching.2. We created an open‐source application for the storage, pattern extraction and pattern matching of digital images for the purposes of mark–recapture analysis. The resulting software package is a stand‐alone, multiplatform application implemented in Java. Our program employs the Scale Invariant Feature Transform (SIFT) operator that extracts distinctive features invariant to image scale and rotation.3. We applied this system to a population of Masai giraffe (Giraffa camelopardalis tippelskirchi) in the Tarangire Ecosystem in northern Tanzania. Over 1200 images were acquired in the field during three primary sampling periods between September 2008 and December 2009. The pattern information in these images was extracted and matched resulting in capture histories for over 600 unique individuals.4. Estimated error rates of the matching system were low based on a subset of test images that were independently matched by eye.5. Encounter histories were subsequently analysed with open population models to estimate apparent survival rates and population size.6. This new open‐access tool allowed photographic mark–recapture to be applied successfully to this relatively large population.

PRM32 One Day Money Will Only Be Spent on Effective Drugs…from Payers' Aspirations to Performance-Based Risk-Sharing Operations

To define an operational modelling framework intended to help the design of Performance-Based Risk-Sharing (PBRS) schemes. A time-to-event endpoint is used as a performance criterion. Such survival endpoints are commonly used in clinical studies, notably in oncology where PBRS schemes are gaining momentum. The framework is based on an open population model with a monthly cycle and 3-year time horizon from launch (i.e. when enrolment into the PBRS scheme starts). Entry into the model (i.e. the progressive arrival of new patients into the PBRS scheme) is determined by market diffusion assumptions and is modelled using a Logistic function. Exit from the model (i.e. patients experiencing the event or dying from any cause) is determined by survival curves from clinical/epidemiological studies and is modelled using a Weibull function. The model accommodates different treatment dosing schedules and performance levels (i.e. minimum survival times guaranteed). Multiple PBRS scenarios can be run and compared in terms of their operational and financial implications. Additionally, the effect of potential revisions of a PBRS scheme terms and conditions can also be examined as real-life information becomes available following scheme implementation (i.e. Bayesian updating). For example, assuming 1,000 patients enrolled in a PBRS scheme, with a monthly dosing schedule and given diffusion (Logistic α=5.0; β=0.4) and survival (Weibull λ=0.7; k=27.0) assumptions, the model predicts that 1937 (6970), 4050 (7861) and 9282 (4420) doses will be given to non-responding (responding) patients with 12, 18 and 24 months minimum survival time guaranteed scenarios, respectively. This framework provides both payer and manufacturer with valuable insight into the operational and financial dimensions of the potential PBRS schemes they may contemplate as they negotiate patient access conditions. Both parties can better anticipate the implications of the schemes and better plan resources, logistics and financial arrangements accordingly.

Surveying an endangered saproxylic beetle, Osmoderma eremita, in Mediterranean woodlands: a comparison between different capture methods

Measuring population size is riddled with difficulties for wildlife biologists and managers, and in the case of rare species, it is sometimes practically impossible to estimate abundance, whereas estimation of occupancy is possible. Furthermore, obtaining reliable population size estimates is not straightforward, as different sampling techniques can give misleading results. A mark-recapture study of the endangered saproxylic beetle Osmoderma eremita was performed in central Italy by applying four independent capture methods within a study area where 116 hollow trees were randomly selected to set traps. Detection probability and population size estimates were drawn from each of these four capture methods. There were strong differences in detection probability among methods. Despite using pheromone and beetle manipulation, capture histories were not affected by trap-happiness or trap-shyness. Population size estimates varied considerably in both abundance and precision by capture method. A number of 0.5 and 0.2 adult beetles per tree was estimated using the whole data set by closed and open population models, respectively. Pitfall trap appeared the optimal method to detect the occurrence of this species. Since in the southern part of its distribution range, a single population of O. eremita is widespread in the landscape, and includes beetles from more than one hollow tree, conservation efforts should focus not only on preserving few and isolated monumental hollow trees, but should be extended to large stands.

Estimating grizzly and black bear population abundance and trend in Banff National Park using noninvasive genetic sampling.

We evaluated the potential of two noninvasive genetic sampling methods, hair traps and bear rub surveys, to estimate population abundance and trend of grizzly (Ursus arctos) and black bear (U. americanus) populations in Banff National Park, Alberta, Canada. Using Huggins closed population mark-recapture models, we obtained the first precise abundance estimates for grizzly bears ( = 73.5, 95% CI = 64–94 in 2006; = 50.4, 95% CI = 49–59 in 2008) and black bears ( = 62.6, 95% CI = 51–89 in 2006; = 81.8, 95% CI = 72–102 in 2008) in the Bow Valley. Hair traps had high detection rates for female grizzlies, and male and female black bears, but extremely low detection rates for male grizzlies. Conversely, bear rubs had high detection rates for male and female grizzlies, but low rates for black bears. We estimated realized population growth rates, lambda, for grizzly bear males ( = 0.93, 95% CI = 0.74–1.17) and females ( = 0.90, 95% CI = 0.67–1.20) using Pradel open population models with three years of bear rub data. Lambda estimates are supported by abundance estimates from combined hair trap/bear rub closed population models and are consistent with a system that is likely driven by high levels of human-caused mortality. Our results suggest that bear rub surveys would provide an efficient and powerful means to inventory and monitor grizzly bear populations in the Central Canadian Rocky Mountains.

Modeling Trap-Awareness and Related Phenomena in Capture-Recapture Studies

Trap-awareness and related phenomena whereby successive capture events are not independent is a feature of the majority of capture-recapture studies. This phenomenon was up to now difficult to incorporate in open population models and most authors have chosen to neglect it although this may have damaging consequences. Focusing on the situation where animals exhibit a trap response at the occasion immediately following one where they have been trapped but revert to their original naïve state if they are missed once, we show that trap-dependence is more naturally viewed as a state transition and is amenable to the current models of capture-recapture. This approach has the potential to accommodate lasting or progressively waning trap effects.

Estimating Abundance of the Endangered Houston Toad on a Primary Recovery Site

Abstract The Griffith League Ranch is one of the primary recovery sites for the endangered Houston toad Bufo (Anaxyrus) houstonensis. New recovery initiatives have recently been implemented to increase Houston toad abundance; however, no robust estimate of population size has been conducted in the last decade of study, nor from this recovery site. To assist with inferences regarding efficacy of current and future management actions, we estimated adult Houston toad abundance on the Griffith League Ranch. Houston toads were sampled at breeding ponds during the 2010 breeding season using a mark–recapture methodology. We analyzed the data using a modified Jolly–Seber open population model in Program MARK. Models were built whereby the probability of capture remained constant, the apparent survival varied with time or was constant, and the probability of entry varied with time. Model averaging was used to account for uncertainty and the derived adult male Houston toad abundance ranged from 201 to 307 individuals. Using a previously determined Griffith League Ranch Houston toad functional sex ratio of 5∶1, we estimated the abundance of the total adult Houston toad population on this primary recovery site to be from 241 to 368 individuals. This study is the first to report a robust abundance estimate of a Houston toad population and provides a foundation for further research to quantify the impact of current and future management actions.

Transit station or destination? Attendance patterns, movements and abundance estimate of humpback whales off west South Africa from photographic and genotypic matching

Humpback whales Megaptera novaeangliae found off west South Africa (WSA) are known to display an atypical migration that may include temporary residency and feeding during spring and summer. At a regional scale there is uncertainty about how these whales relate to the greater West African Breeding Stock B as a whole, with evidence both for and against its division into two substocks. A database containing sighting information of humpback whales intercepted by boat in the WSA region from 1983 to 2008 was compiled. It included a total of 1 820 identification images of ventral tail flukes and lateral views of dorsal fins. After systematic within- and between-year matching of images of usable quality, it yielded 154 different individuals identified by tail flukes (TF), 230 by left dorsal fins (LDF), and 237 by right dorsal fins (RDF). Microsatellite (MS) matching of 216 skin biopsies yielded 156 individuals. By linking all possible sightings of the same individuals using all available identification features, the periodicity and seasonality of 281 individual whales were examined. In all, 60 whales were resighted on different days of which 44 were between different calendar years. The most resightings for one individual was 11 times, seen in six different years, and the longest interval between first and last sightings was about 18 years. A resighting rate of 15.6% of whales at intervals of a year or more indicates long-term fidelity to the region. Shorter intervals of 1–6 months between sequential sightings in the same year may suggest temporary residency. The TF image collection from WSA was compared to TF collections from four other regions, namely Gabon, Cabinda (Angola), Namibia and the Antarctic Humpback Whale Catalogue (AHWC). Three matches were detected between WSA (in late spring or summer) and Gabon (in winter), confirming direct movement between these regions. The capture–recapture data of four different identification features (TF, RDF, LDF and MS) from six successive subsets of data from periods with the highest collection effort (2001–2007) were used to calculate the number of whales that utilise the region, using both closed- and open-population models. Dorsal fins have never been used to estimate abundance for humpback whales, so the different identification features were evaluated for potential biases. This revealed 9–14% incidence of missed matches (false negatives) when using dorsal fins that would result in an overestimate, whereas variation in individual fluke-up behaviour may lower estimates by as much as 57–66% due to heterogeneity of individual capture probability. Taking into consideration the small dataset and low number of recaptures, the most consistent and precise results were obtained from a fully time-dependent version of the Jolly-Seber open-population model, with annual survival fixed at 0.96, using the MS dataset. This suggests that the WSA feeding assemblage during the months of spring and summer (September–March) of the study period numbered about 500 animals. The relationship of these whales to those (perhaps strictly migratory) that may occur in other seasons of the year, and their links to possible migratory routes and other feeding or breeding areas, remain uncertain.

Inference on Partially Observed Quasi-stationary Markov Chains With Applications to Multistate Population Models

If the full capture histories of captured individuals are available, inferences on multistate open population models may be conducted using the well known Arnason–Schwarz model. However, data of this detail is not always available. It is well known that inference on the transition probabilities of a Markov chain may be conducted using aggregate data and we extend this approach to aggregate data on multistate open population models. We show that for parameters to be identifiable we need to augment the aggregate data and we achieve this by batch marking a cohort of individuals according to their initial state, so that the batch marking augments the aggregate data. Model performance is examined by conducting several simulation studies and the model is applied to a real data set where full capture histories are available so it may be compared with the Arnason–Schwarz estimates. This article has supplementary material online.

An improved procedure to estimate wolf abundance using non-invasive genetic sampling and capture–recapture mixture models

Non-invasive genetic sampling (NGS) is increasingly used to estimate the abundance of rare or elusive species such as the wolf (Canis lupus), which cannot be directly counted in forested mountain habitats. Wolf individual and familial home ranges are wide, potentially connected by long-range dispersers, and their populations are intrinsically open. Appropriate demo- graphic estimators are needed, because the assumptions of homogeneous detection probability and demographic closeness are violated. We compiled the capture-recapture record of 418 individual wolf genotypes identified from ca. 4,900 non-invasive samples, collected in the northern Italian Apennines from January 2002 to June 2009. We analysed this dataset using novel capture-recapture multi- event models for open populations that explicitly account for individual detection heterogeneity (IDH). Overall, the detection probability of the weakly detectable individuals, probably pups, juveniles and migrants (P = 0.08), was ca. six times lower than that of the highly detectable wolves (P = 0.44), probably adults and dominants. The apparent annual survival rate of weakly detectable individuals was lower (U = 0.66) than those of highly detectable wolves (U = 0.75). The population mean annual finite rate of increase was k = 1.05 ± 0.11, and the mean annual size ranged from N = 117 wolves in 2003 to N = 233 wolves in 2007. This procedure, combining large-scale NGS and multievent IDH demographic models, provides the first estimates of abundance, multi-annual trend and survival rates for an open large wolf population in the Apennines. These results contribute to deepen our understanding of wolf population ecology and dynamics, and provide new information to implement sound long-term conservation plans.

Population size of neonate and juvenile dusky sharks Carcharhinus obscurus in the Port of Ngqura, South Africa

Identifying nearshore waters that serve as important habitats for neonate and juvenile sharks is crucial for effective fisheries management. This study examined the abundance of neonate and juvenile dusky sharks Carcharhinus obscurus within the Port of Ngqura, on the south-east coast of South Africa, between September 2006 and August 2007 using a combination of shore-angling catches and mark-recapture data. During this period, the port was non-operational and still undergoing construction. A total of 480 dusky sharks, ranging in size from 50 to 123 cm (precaudal length), was captured. Catch per unit effort was greatest between October and February, peaking in November at 0.51 sharks angler–1 h–1. Of the 219 C. obscurus tagged and released during the study period, 37 (16.9%) were recaptured. Sharks were at liberty from 0 to 409 days with the majority (89.2%) recaptured within the port. Using a Jolly-Seber open population model, the abundance of sharks within the port was estimated at 552 (95% CI: 422–765). Annual apparent survival probability was 0.38 (95% CI: 0.30–0.46), with an average annual recapture probability of 0.28 (95% CI: 0.19–0.39). This study demonstrates that the Port of Ngqura is an important summer habitat and core activity zone for both neonate and juvenile dusky sharks.

Habitat quality and habitat selection of golden-winged warblers in Costa Rica: an application of hierarchical models for open populations

Summary 1. Identifying high-quality habitat is integral to effective species conservation efforts and requires information about habitat-specific abundance and demographics. This information is lacking for many species of conservation concern because of the inherent difficulties associated with implementing mark–recapture studies at large spatial scales. 2. The golden-winged warbler Vermivora chrysoptera is a Neotropical–Nearctic migratory bird experiencing a long-term population decline; yet no information about non-breeding season habitat quality or habitat selection exists to inform conservation efforts. We used a novel hierarchical model that requires only spatially and temporally replicated count data to estimate habitat-specific abundance, apparent survival, recruitment and detection probability of golden-winged warblers wintering in Costa Rica. We quantified habitat selection at the home-range level using radiotelemetry. 3. Golden-winged warblers were absent from tropical dry forest and were most abundant in premontane evergreen forest. Within their home ranges, golden-winged warblers selected microhabitat features associated with intermediate disturbance that reflected their preference for foraging in hanging dead leaves. 4. Consistent with other evidence of a declining population, local population size decreased over the duration of the study. The rate of decrease was higher during the non-breeding season than among seasons. We found no differences in apparent survival or recruitment among habitat types; however, our estimates of these parameters were imprecise. 5. Synthesis and applications. Golden-winged warblers are forest-dependent species during the non-breeding season and have specialized microhabitat requirements that make them vulnerable to ongoing tropical deforestation. However, advanced secondary forests can provide the requisite microhabitat features, and because we found no evidence of reduced survival in this habitat type, regenerating forest on degraded lands may be an effective component of a conservation strategy for this species. Our study also demonstrates that information on population dynamics and habitat quality can be obtained using repeated counts instead of mark–recapture methods.

Abundance and site fidelity of Indo‐Pacific Humpback dolphins in the Great Sandy Strait, Queensland, Australia

AbstractThis study examined the population of Indo‐Pacific humpback dolphins, Sousa chinensis, inhabiting the Great Sandy Strait Marine Park, Queensland, Australia. A total of 106 dolphins were identified during 228 boat‐based surveys, completed between April 2004 and April 2007. Based on the distribution of resighted individuals and the pattern of associations, it was established that this population consists of two largely geographically distinct communities, referred to as the Northern Community (NC) and the Southern Community (SC). The only recorded interaction between the two groups was a single pod composed of one member of the NC and 11 dolphins from the SC. Abundance was estimated for the entire population and by geographical area using open population models. Estimates for the Great Sandy Strait indicate that about 150 dolphins (NGSS= 148.4, SE = 8.3, 95% CI: 132.5–165.2) used this area during the study. The NC and SC total population sizes was estimated to be 76 (NNGSS= 75.80, SE = 3.88, 95% CI = 71–86) and 75 (NSGSS= 74.98, SE = 4.43, 95% CI: 66–83), respectively. Analysis of residence patterns indicates that a majority of the identified dolphins are long‐term residents.

Full Open Population Capture–Recapture Models With Individual Covariates

Traditional analyses of capture–recapture data are based on likelihood functions that explicitly integrate out all missing data. We use a complete data likelihood (CDL) to show how a wide range of capture–recapture models can be easily fitted using readily available software JAGS/BUGS even when there are individual-specific time-varying covariates. The models we describe extend those that condition on first capture to include abundance parameters, or parameters related to abundance, such as population size, birth rates or lifetime. The use of a CDL means that any missing data, including uncertain individual covariates, can be included in models without the need for customized likelihood functions. This approach also facilitates modeling processes of demographic interest rather than the complexities caused by non-ignorable missing data. We illustrate using two examples, (i) open population modeling in the presence of a censored time-varying individual covariate in a full robust design, and (ii) full open population multi-state modeling in the presence of a partially observed categorical variable. Supplemental materials for this article are available online.

Local Weather, Regional Climate, and Annual Survival of The Northern Spotted Owl

Abstract. We used an information-theoretical approach and Cormack—Jolly—Seber models for open populations in program MARK to examine relationships between survival rates of Northern Spotted Owls and a variety of local weather variables and long-term climate variables. In four of the six populations examined, survival was positively associated with wetter than normal conditions during the growing season or high summer temperatures. At the three study areas located at the highest elevations, survival was positively associated with winter temperature but also had a negative or quadratic relation with the number of storms and winter precipitation. A meta-analysis of all six areas combined indicated that annual survival was most strongly associated with phase shifts in the Southern Oscillation and Pacific Decadal Oscillation, which reflect large-scale temperature and precipitation patterns in this region. Climate accounted for a variable amount (1–41%) of the total process variation in annual survival but for ...

Survivorship and Population Densities of Painted Turtles (Chrysemys picta) in Recently Modified Suburban Landscapes

Abstract Populations of long-lived animals, such as semi-aquatic turtles, that depend on high survivorship of reproductive adults, are particularly susceptible to the negative effects associated with habitat modification in suburban areas. Survivorship of semi-aquatic turtles in suburban landscapes may be reduced as a result of a number of factors, including the elimination of appropriate nesting habitat and the introduction of human-subsidized predators. Unfortunately, few studies on turtle populations in anthropogenically modified habitats estimate vital rates, and researchers are rarely able to study populations both before and after development. We studied painted turtle (Chrysemys picta) vital rates at 5 ponds in the Charlotte Metropolitan area of North Carolina; 2 ponds and their surrounding habitat underwent development after the first year of study, 1 pond was on a golf course, and 2 were farm ponds. We used Program MARK to generate open population models examining the effects of location and sex ...

Parameter-expanded data augmentation for Bayesian analysis of capture–recapture models

Data augmentation (DA) is a flexible tool for analyzing closed and open population models of capture–recapture data, especially models which include sources of hetereogeneity among individuals. The essential concept underlying DA, as we use the term, is based on adding “observations” to create a dataset composed of a known number of individuals. This new (augmented) dataset, which includes the unknown number of individuals N in the population, is then analyzed using a new model that includes a reformulation of the parameter N in the conventional model of the observed (unaugmented) data. In the context of capture–recapture models, we add a set of “all zero” encounter histories which are not, in practice, observable. The model of the augmented dataset is a zero-inflated version of either a binomial or a multinomial base model. Thus, our use of DA provides a general approach for analyzing both closed and open population models of all types. In doing so, this approach provides a unified framework for the analysis of a huge range of models that are treated as unrelated “black boxes” and named procedures in the classical literature. As a practical matter, analysis of the augmented dataset by MCMC is greatly simplified compared to other methods that require specialized algorithms. For example, complex capture–recapture models of an augmented dataset can be fitted with popular MCMC software packages (WinBUGS or JAGS) by providing a concise statement of the model’s assumptions that usually involves only a few lines of pseudocode. In this paper, we review the basic technical concepts of data augmentation, and we provide examples of analyses of closed-population models (M0, Mh, distance sampling, and spatial capture–recapture models) and open-population models (Jolly–Seber) with individual effects.

Effects of Exploitation on Black Bear Populations at White River National Wildlife Refuge

Abstract: We live‐trapped American black bears (Ursus americanus) and sampled DNA from hair at White River National Wildlife Refuge, Arkansas, USA, to estimate annual population size (N), growth (γ), and density. We estimated N and γ with open population models, based on live‐trapping data collected from 1998 through 2006, and robust design models for genotyped hair samples collected from 2004 through 2007. Population growth was weakly negative (i.e., 95% CI included 1.0) for males (0.901, 95% CI = 0.645–1.156) and strongly negative (i.e., 95% CI excluded 1.0) for females (0.846, 95% CI = 0.711–0.981), based on live‐trapping data, with N from 1999 to 2006 ranging from 94.1 (95% CI = 70.3–137.1) to 45.2 (95% CI = 27.1–109.3), respectively, for males and from 151.4 (95% CI = 127.6–185.8) to 47.1 (95% CI = 24.4–140.4), respectively, for females. Likewise, mean annual γ based on hair‐sampling data was weakly negative for males (0.742, 95% CI = 0.043–1.441) and strongly negative for females (0.782, 95% CI = 0.661–0.903), with abundance estimates from 2004 to 2007 ranging from 29.1 (95% CI = 21.2–65.8) to 11.9 (95% CI = 11.0–26.9), respectively, for males and from 54.4 (95% CI = 44.3–77.1) to 27.4 (95% CI =24.9–36.6), respectively, for females. We attribute the decline in the number of females in this isolated population to a decrease in survival caused by a past translocation program and by hunting adjacent to the refuge. We suggest that managers restructure the quota‐based harvest limits until these growth rates recover.