Larissa L. Bailey

SAMPLING DESIGN TRADE‐OFFS IN OCCUPANCY STUDIES WITH IMPERFECT DETECTION: EXAMPLES AND SOFTWARE

Researchers have used occupancy, or probability of occupancy, as a response or state variable in a variety of studies (e.g., habitat modeling), and occupancy is increasingly favored by numerous state, federal, and international agencies engaged in monitoring programs. Recent advances in estimation methods have emphasized that reliable inferences can be made from these types of studies if detection and occupancy probabilities are simultaneously estimated. The need for temporal replication at sampled sites to estimate detection probability creates a trade-off between spatial replication (number of sample sites distributed within the area of interest/inference) and temporal replication (number of repeated surveys at each site). Here, we discuss a suite of questions commonly encountered during the design phase of occupancy studies, and we describe software (program GENPRES) developed to allow investigators to easily explore design trade-offs focused on particularities of their study system and sampling limitations. We illustrate the utility of program GENPRES using an amphibian example from Greater Yellowstone National Park, U.S.A.

Estimating Site Occupancy and Detection Probability Parameters for Meso- And Large Mammals in a Coastal Ecosystem

Abstract Large-scale, multispecies monitoring programs are widely used to assess changes in wildlife populations but they often assume constant detectability when documenting species occurrence. This assumption is rarely met in practice because animal populations vary across time and space. As a result, detectability of a species can be influenced by a number of physical, biological, or anthropogenic factors (e.g., weather, seasonality, topography, biological rhythms, sampling methods). To evaluate some of these influences, we estimated site occupancy rates using species-specific detection probabilities for meso- and large terrestrial mammal species on Cape Cod, Massachusetts, USA. We used model selection to assess the influence of different sampling methods and major environmental factors on our ability to detect individual species. Remote cameras detected the most species (9), followed by cubby boxes (7) and hair traps (4) over a 13-month period. Estimated site occupancy rates were similar among sampling methods for most species when detection probabilities exceeded 0.15, but we question estimates obtained from methods with detection probabilities between 0.05 and 0.15, and we consider methods with lower probabilities unacceptable for occupancy estimation and inference. Estimated detection probabilities can be used to accommodate variation in sampling methods, which allows for comparison of monitoring programs using different protocols. Vegetation and seasonality produced species-specific differences in detectability and occupancy, but differences were not consistent within or among species, which suggests that our results should be considered in the context of local habitat features and life history traits for the target species. We believe that site occupancy is a useful state variable and suggest that monitoring programs for mammals using occupancy data consider detectability prior to making inferences about species distributions or population change.

ITEROPARITY IN THE VARIABLE ENVIRONMENT OF THE SALAMANDER AMBYSTOMA TIGRINUM

Simultaneous estimation of survival, reproduction, and movement is essential to understanding how species maximize lifetime reproduction in environments that vary across space and time. We conducted a four-year, capture-recapture study of three populations of eastern tiger salamanders (Ambystoma tigrinum tigrinum) and used multistate mark-recapture statistical methods to estimate the manner in which movement, survival, and breeding probabilities vary under different environmental conditions across years and among populations and habitats. We inferred how individuals may mitigate risks of mortality and reproductive failure by deferring breeding or by moving among populations. Movement probabilities among populations were extremely low despite high spatiotemporal variation in reproductive success and survival, suggesting possible costs to movements among breeding ponds. Breeding probabilities varied between wet and dry years and according to whether or not breeding was attempted in the previous year. Estimates of survival in the nonbreeding, forest habitat varied among populations but were consistent across time. Survival in breeding ponds was generally high in years with average or high precipitation, except for males in an especially ephemeral pond. A drought year incurred severe survival costs in all ponds to animals that attempted breeding. Female salamanders appear to defer these episodic survival costs of breeding by choosing not to breed in years when the risk of adult mortality is high. Using stochastic simulations of survival and breeding under historical climate conditions, we found that an interaction between breeding probabilities and mortality limits the probability of multiple breeding attempts differently between the sexes and among populations.

Uncovering a latent multinomial: analysis of mark-recapture data with misidentification.

Natural tags based on DNA fingerprints or natural features of animals are now becoming very widely used in wildlife population biology. However, classic capture-recapture models do not allow for misidentification of animals which is a potentially very serious problem with natural tags. Statistical analysis of misidentification processes is extremely difficult using traditional likelihood methods but is easily handled using Bayesian methods. We present a general framework for Bayesian analysis of categorical data arising from a latent multinomial distribution. Although our work is motivated by a specific model for misidentification in closed population capture-recapture analyses, with crucial assumptions which may not always be appropriate, the methods we develop extend naturally to a variety of other models with similar structure. Suppose that observed frequencies f are a known linear transformation f=Ax of a latent multinomial variable x with cell probability vector pi=pi(theta). Given that full conditional distributions [theta | x] can be sampled, implementation of Gibbs sampling requires only that we can sample from the full conditional distribution [x | f, theta], which is made possible by knowledge of the null space of A. We illustrate the approach using two data sets with individual misidentification, one simulated, the other summarizing recapture data for salamanders based on natural marks.

Using occupancy models to understand the distribution of an amphibian pathogen, Batrachochytrium dendrobatidis

Batrachochytrium dendrobatidis is a fungal pathogen that is receiving attention around the world for its role in amphibian declines. Study of its occurrence patterns is hampered by false negatives: the failure to detect the pathogen when it is present. Occupancy models are a useful but currently underutilized tool for analyzing detection data when the probability of detecting a species is <1. We use occupancy models to evaluate hypotheses concerning the occurrence and prevalence of B. dendrobatidis and discuss how this application differs from a conventional occupancy approach. We found that the probability of detecting the pathogen, conditional on presence of the pathogen in the anuran population, was related to amphibian development stage, day of the year, elevation, and human activities. Batrachochytrium dendrobatidis was found throughout our study area but was only estimated to occur in 53.4% of 78 populations of native amphibians and 66.4% of 40 populations of nonnative Rana catesbeiana tested. We found little evidence to support any spatial hypotheses concerning the probability that the pathogen occurs in a population, but did find evidence of some taxonomic variation. We discuss the interpretation of occupancy model parameters, when, unlike a conventional occupancy application, the number of potential samples or observations is finite.

ESTIMATING THE ABUNDANCE OF MOUSE POPULATIONS OF KNOWN SIZE: PROMISES AND PITFALLS OF NEW METHODS

Knowledge of animal abundance is fundamental to many ecological studies. Frequently, researchers cannot determine true abundance, and so must estimate it using a method such as mark-recapture or distance sampling. Recent advances in abundance estimation allow one to model heterogeneity with individual covariates or mixture distributions and to derive multimodel abundance estimators that explicitly address uncertainty about which model parameterization best represents truth. Further, it is possible to borrow information on detection probability across several populations when data are sparse. While promising, these methods have not been evaluated using mark-recapture data from populations of known abundance, and thus far have largely been overlooked by ecologists. In this paper, we explored the utility of newly developed mark-recapture methods for estimating the abundance of 12 captive populations of wild house mice (Mus musculus). We found that mark-recapture methods employing individual covariates yielded satisfactory abundance estimates for most populations. In contrast, model sets with heterogeneity formulations consisting solely of mixture distributions did not perform well for several of the populations. We show through simulation that a higher number of trapping occasions would have been necessary to achieve good estimator performance in this case. Finally, we show that simultaneous analysis of data from low abundance populations can yield viable abundance estimates.

Bias, precision, and parameter redundancy in complex multistate models with unobservable states

Multistate mark-recapture models with unobservable states can yield unbiased estimators of survival probabilities in the presence of temporary emigration (i.e., in cases where some individuals are temporarily unavailable for capture). In addition, these models permit the estimation of transition probabilities between states, which may themselves be of interest; for example, when only breeding animals are available for capture. However, parameter redundancy is frequently a problem in these models, yielding biased parameter estimates and influencing model selection. Using numerical methods, we examine complex multistate mark-recapture models involving two observable and two unobservable states. This model structure was motivated by two different biological systems: one involving island-nesting albatross, and another involving pond-breeding amphibians. We found that, while many models are theoretically identifiable given appropriate constraints, obtaining accurate and precise parameter estimates in practice can be difficult. Practitioners should consider ways to increase detection probabilities or adopt robust design sampling in order to improve the properties of estimates obtained from these models. We suggest that investigators interested in using these models explore both theoretical identifiability and possible near-singularity for likely parameter values using a combination of available methods.

Using multilevel spatial models to understand salamander site occupancy patterns after wildfire

Studies of the distribution of elusive forest wildlife have suffered from the confounding of true presence with the uncertainty of detection. Occupancy modeling, which incorporates probabilities of species detection conditional on presence, is an emerging approach for reducing observation bias. However, the current likelihood modeling framework is restrictive for handling unexplained sources of variation in the response that may occur when there are dependence structures such as smaller sampling units that are nested within larger sampling units. We used multilevel Bayesian occupancy modeling to handle dependence structures and to partition sources of variation in occupancy of sites by terrestrial salamanders (family Plethodontidae) within and surrounding an earlier wildfire in western Oregon, USA. Comparison of model fit favored a spatial N-mixture model that accounted for variation in salamander abundance over models that were based on binary detection/non-detection data. Though catch per unit effort was higher in burned areas than unburned, there was strong support that this pattern was due to a higher probability of capture for individuals in burned plots. Within the burn, the odds of capturing an individual given it was present were 2.06 times the odds outside the burn, reflecting reduced complexity of ground cover in the burn. Ther was weak support that true occupancy was lower within the burned area. While the odds of occupancy in the burn were 0.49 times the odds outside the burn among the five species, the magnitude of variation attributed to the burn was small in comparison to variation attributed to other landscape variables and to unexplained, spatially autocorrelated random variation. While ordinary occupancy models may separate the biological pattern of interest from variation in detection probability when all sources of variation are known, the addition of random effects structures for unexplained sources of variation in occupancy and detection probability may often more appropriately represent levels of uncertainty.

Amphibian Research and Monitoring Initiative (ARMI): a successful start to a national program in the United States

Most research to assess amphibian declines has focused on local-scale projects on one or a few species. The Amphibian Research and Monitoring Initiative (ARMI) is a national program in the United States mandated by congressional directive and implemented by the U.S. Department of the Interior (specifically the U.S. Geological Survey, USGS). Program goals are to monitor changes in populations of amphibians across U.S. Department of the Interior lands and to address research questions related to amphibian declines using a hierarchical framework of base-, mid- and apex-level monitoring sites. ARMI is currently monitoring 83 amphibian species (29% of species in the U.S.) at mid- and apex-level areas. We chart the progress of this 5-year-old program and provide an example of mid-level monitoring from 1 of the 7 ARMI regions.

Methods for Estimating the Amount of Vernal Pool Habitat in the Northeastern United States

The loss of small, seasonal wetlands is a major concern for a variety of state, local, and federal organizations in the northeastern U.S. Identifying and estimating the number of vernal pools within a given region is critical to developing long-term conservation and management strategies for these unique habitats and their faunal communities. We use three probabilistic sampling methods (simple random sampling, adaptive cluster sampling, and the dual frame method) to estimate the number of vernal pools on protected, forested lands. Overall, these methods yielded similar values of vernal pool abundance for each study area, and suggest that photographic interpretation alone may grossly underestimate the number of vernal pools in forested habitats. We compare the relative efficiency of each method and discuss ways of improving precision. Acknowledging that the objectives of a study or monitoring program ultimately determine which sampling designs are most appropriate, we recommend that some type of probabilistic sampling method be applied. We view the dual-frame method as an especially useful way of combining incomplete remote sensing methods, such as aerial photograph interpretation, with a probabilistic sample of the entire area of interest to provide more robust estimates of the number of vernal pools and a more representative sample of existing vernal pool habitats.