Determining Occurrence Dynamics when False Positives Occur: Estimating the Range Dynamics of Wolves from Public Survey Data

et al. (2013) Determining Occurrence Dynamics when False Positives Occur: Estimating the Range Dynamics of Wolves from Public Survey Data. PLoS ONE 8(6): e65808. doi:10.1371/journal.pone.0065808 Determining Occurrence Dynamics when False Positives Occur: Estimating the Range Dynamics of Wolves from Public Survey Data David A. W. Miller 0 James D. Nichols 0 Justin A. Gude 0 Lindsey N. Rich 0 Kevin M. Podruzny 0 James E. Hines 0 Michael S. Mitchell 0 Evelyn Merrill, University of Alberta, Canada 0 1 United States Geological Survey, Patuxent Wildlife Research Center, Laurel, Maryland, United States of America, 2 Pennsylvania State University, Department of Ecosystem Science and Management, University Park, Pennsylvania, United States of America, 3 Montana Fish, Wildlife and Parks, Helena, Montana, United States of America, 4 United States Geological Survey, Montana Cooperative Wildlife Research Unit, University of Montana , Missoula, Montana , United States of America Large-scale presence-absence monitoring programs have great promise for many conservation applications. Their value can be limited by potential incorrect inferences owing to observational errors, especially when data are collected by the public. To combat this, previous analytical methods have focused on addressing non-detection from public survey data. Misclassification errors have received less attention but are also likely to be a common component of public surveys, as well as many other data types. We derive estimators for dynamic occupancy parameters (extinction and colonization), focusing on the case where certainty can be assumed for a subset of detections. We demonstrate how to simultaneously account for non-detection (false negatives) and misclassification (false positives) when estimating occurrence parameters for gray wolves in northern Montana from 2007-2010. Our primary data source for the analysis was observations by deer and elk hunters, reported as part of the state's annual hunter survey. This data was supplemented with data from known locations of radio-collared wolves. We found that occupancy was relatively stable during the years of the study and wolves were largely restricted to the highest quality habitats in the study area. Transitions in the occupancy status of sites were rare, as occupied sites almost always remained occupied and unoccupied sites remained unoccupied. Failing to account for false positives led to over estimation of both the area inhabited by wolves and the frequency of turnover. The ability to properly account for both false negatives and false positives is an important step to improve inferences for conservation from largescale public surveys. The approach we propose will improve our understanding of the status of wolf populations and is relevant to many other data types where false positives are a component of observations. - Funding: Funding for data collection was provided by the sale of hunting and fishing licenses in Montana, the Pittman-Robertson program administered by the United States Fish and Wildlife Service, and the United States Fish and Wildlife Service wolf recovery program. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. Presence-absence surveys have become increasingly prominent in large-scale ecological and conservation research [1,2]. Occurrence data have the advantage of being relatively easy to collect and can be related to many important ecological processes such as habitat use, range-dynamics, metapopulation dynamics, and occupancy-abundance relationships [1]. Particularly important from the standpoint of understanding ecological processes is the ability to use empirical data to estimate occupancy transition probabilities (i.e., colonization and extinction) and to investigate how occupancy dynamics are affected by dynamics of habitat and co-occurring species [1,3]. New methods for estimating species occurrence probabilities have opened the door to utilizing largescale occurrence data collections, many of which have engaged the public in the data collection process. Utilizing the public can expand the scope and scale of data collection by orders of magnitude as compared to typical research efforts [4,5]. However, observation error is likely to be an especially significant issue for these types of data, meaning they should be approached with proper caution [610] If inferences are to be reliable it is necessary to account for observation uncertainty, including both nondetection and misidentification [11]. Ecologists have long recognized the need to account for imperfect detection when estimating parameters for wildlife populations and have developed an extensive set of methods to deal with non-detection [12]. Recent effort has focused on the need to also account for misclassification and misidentification when estimating population parameters. For example, adaptations of traditional mark-recapture models have focused on various types of classification uncertainty [13]. Additionally, the availability of analytical techniques to deal with individual misclassification have increased the utility of techniques that identify individuals using genetic identifiers [14] and visual patterns [15]. Similarly, most attention for studies of species occurrence have focused on non-detection [1], although recent efforts have also considered misclassification errors. In occupancy studies, misclassification happens when sites that are unoccupied are recorded as being occupied. These false positive errors are common in many occurrence sampling methods [7,11,1618]. This is problematic because, when unaddressed, even small rates of false positive errors can result in substantial bias in single season estimators of occupancy [19,20] and estimators for colonization and extinction rates [11,21]. Detection errors are often ignored when public surveys are analyzed and, when addressed, effort has generally focused on false negative errors [6,10]. Previous attempts to address misclassification have largely focused on ad hoc methods to try to reduce their occurrence in data sets [7,22]. Two approaches have been suggested for estimating occupancy when false positives occur in single season occupancy analyses. The first is a simple modification of the standard occupancy estimator [23], which allows for false positive detections to occur at unoccupied sites [19]. Observed numbers of detections at sites are treated as a binomial mixture of the true positive detection probability at occupied sites and false positive detection probability at unoccupied sites. Miller et al. [20] extend this to deal with cases where detections can be divided into uncertain detections, which have some probability of being a false positive, and certain detections, which are assumed to have zero probability of being a false positive. Th (...truncated)