Sampling Rare or Elusive Species

Concepts, Designs, and Techniques for Estimating Population Parameters

William Thompson

• Publisher: Island Press
• Year: 2013
• Print ISBN: 9781559634519, 9781559634502
• eBook ISBN: 9781610911061
• Subtopic: Biology

Information regarding population status and abundance of rare species plays a key role in resource management decisions. Ideally, data should be collected using statistically sound sampling methods, but by their very nature, rare or elusive species pose a difficult sampling challenge.

Sampling Rare or Elusive Species describes the latest sampling designs and survey methods for reliably estimating occupancy, abundance, and other population parameters of rare, elusive, or otherwise hard-to-detect plants and animals. It offers a mixture of theory and application, with actual examples from terrestrial, aquatic, and marine habitats around the world.

Sampling Rare or Elusive Species is the first volume devoted entirely to this topic and provides natural resource professionals with a suite of innovative approaches to gathering population status and trend data. It represents an invaluable reference for natural resource professionals around the world, including fish and wildlife biologists, ecologists, biometricians, natural resource managers, and all others whose work or research involves rare or elusive species.

1

Introduction

William L. Thompson

Natural resource professionals are typically faced with multiple sources of uncertainty when attempting to manage plant and animal populations in ways that ensure their persistence at acceptable levels into the foreseeable future. These sources of uncertainty can be broadly categorized as environmental variation, structural uncertainty, partial observability, and partial controllability (Nichols et al. 1995; Williams 1997). Environmental variation refers to inherent differences in the physical setting, such as those induced by habitat and weather, that may influence population status and trend. Structural uncertainty arises from the lack of knowledge about the underlying processes influencing population dynamics and how these processes relate to management practices. An example is the uncertainty related to whether hunting has an additive or a compensatory effect on waterfowl populations (Williams 1997). Partial observability concerns the incomplete knowledge of population status and trend, whereas partial controllability is the inability to strictly regulate the implementation of management actions.

Sources of uncertainty affecting management of plant and animal populations are not mutually exclusive. For instance, complex habitats (environmental variation) could lead to an incomplete and variable assessment of population status or trend (partial observability), which in turn could mask demographic responses to management practices (structural uncertainty). Therefore, effective management of plant and animal populations is predicated upon both recognizing and properly accounting for (hopefully, minimizing when possible) these various sources of uncertainty (Williams 2001). The optimum approach is to incorporate management options within an adaptive resource management framework (Holling 1978; Walters 1986). This is sometimes referred to as management by experiment (MacNab 1983), which recognizes uncertainty inherent within natural systems and incorporates learning as a fundamental objective so that management can improve with knowledge gained by previous actions. Competing models or hypotheses encapsulate this uncertainty and are evaluated as part of the learning process.

A critical element of adaptive resource management is monitoring responses of the population(s) under study to the management action(s) (Morrison et al. 2001). The monitoring program typically is based on obtaining an estimate of occupancy (spatial distribution or proportion of area occupied), abundance, density, or survival recorded at specified time intervals within the area of interest. However, estimation of these population parameters often is confounded by detection probability, that is, probability of correctly noting the presence of an individual (or species, for species-level surveys) within some area and time period. Failure to properly account for detection probability leads to biased estimators and, perhaps, misleading estimates of population status and trend (see Chapter 4). This could undermine the learning process in adaptive resource management and erode empirical support for management decisions and policies.

Development of methods and models to properly account for detection probability, especially for mobile species, continues to be one of the hotter areas in biometric research. This is reflected in the recent proliferation of books describing these approaches (e.g., Thompson et al. 1998; Elzinga et al. 2001; Borchers et al. 2002; Williams et al. 2002). However, sampling designs and counting methods described in these books are geared primarily toward moderately abundant to abundant species; many do not translate well to rare, elusive, or otherwise difficult-to-detect species (see Chapter 2). Unfortunately, these species are usually the ones of management concern because of their low numbers, limited geographical distribution, and/or our lack of reliable information on their population status and trend.

General Sampling Framework

The purpose of this volume is to describe the latest sampling designs and counting (estimation) techniques for reliably estimating occupancy, abundance, and other population parameters of rare or elusive plants and animals. Rare refers to low abundance or restricted geographical distribution (clustered or not) or both, whereas elusive refers to low probability of detection for whatever reason(s) (see Chapter 2). This demographic information in turn may be used as a basis for informed management and policymaking. Because mobile species typically present greater sampling challenges, the majority of this volume focuses on free-ranging animals. However, many of the designs and counting or estimation techniques also apply to sampling sessile organisms such as plants.

Throughout this volume, population or species surveys usually follow a sampling framework whereby a geographical area containing the population or species of interest is partitioned into sampling units, such as plots, quadrats, and transects. Then a subsample of these units is chosen based on some form of probability-based selection process, and some type of count or measurement is applied within chosen units. Such a scenario may be viewed as a two-stage sampling design (after Hankin 1984 and Skalski 1994), where the random selection of units is the first stage and the count or measurement is the second stage. This usage deviates from that in sample survey texts (e.g., Cochran 1977), which define a two-stage design as a random sample of subunits within randomly selected sampling units (e.g., subareas within a chosen plot). However, the traditional definition of a two-stage sample assumes a complete enumeration or measurement within selected subunits, which is usually not possible in plant or animal surveys. Therefore, this volume follows the definition of two-stage designs (or, generally, multistage designs) that treats the count as the last stage. Note that additional stages are possible, depending on the spatial extent of the area of interest.

This volume generally follows statistical notation suggested by Thompson et al. (1998) for plant and animal surveys, such that U is the total sampling units used for inference, u is the number of sampling units in a sample, and N is the true population size of individuals in the sampled population. However, notation in Chapter 13 deviates from this convention because it uses previously published notation in its equations; hence, in this chapter only, M is the total sampling units used for inference, m and n denote number of sampling units in a sample, and T is the true population size (total).

Scope of this Volume

Because of the extreme paucity of published books on sampling rare or elusive species, this volume contains a mixture of both theory and application, with a strong emphasis on application, This volume is organized into five sections: Overview and Basic Concepts; Sampling Designs for Rare Species and Populations; Estimating Occupancy; Estimating Abundance, Density, and Other Parameters; and The Future.

The first section provides an overview of sampling rare or elusive species and populations. In Chapter 2, McDonald discusses the results of his informal survey of biologists and biometricians in which he solicited their definitions of “rare” and asked them to recount their successes and failures when sampling rare species. This chapter serves as both an overview of designs commonly used to sample rare species and as a guide to their practicality and usefulness based on experiences of biologists and biometricians. The next two chapters in Part I deal with the issue of detection probability and its effects on parameter estimators. Pollock et al. (Chapter 3) discuss the components of detection probability and present applied examples of how to estimate these components. In Chapter 4, Conn et al. use simulation to rigorously evaluate implications of using counts (captures, in this case) uncorrected for incomplete detectability as surrogates for abundance in monitoring population trends of low-abundance species. Topics discussed in Chapters 3 and 4 are relevant to moderately abundant or abundant species as well as to rare ones.

Part II of this volume explores sampling designs for efficiently estimating abundance of rare species, with a greater emphasis on the first stage of a two-stage design, that is, probability-based selection of sampling units. Smith et al. provide a thorough review and practical evaluation of adaptive cluster sampling in Chapter 5. Manly follows with a description of an alternative adaptive approach to estimate abundance of rare species, based on a two-phase, stratified sampling regime. In Chapter 7, Christman reviews the sequential sampling design, another form of adaptive sampling, and applies it to estimating abundance of a waterfowl population.

The objective of a monitoring program may be to track changes in presence or occupancy of a species within the area of interest. Such information tends to be less costly to collect than data on abundance, density, or survival (MacKenzie et al. 2001, 2003). However, just like other population estimators, occupancy estimators are susceptible to bias from incomplete detectability. Therefore, Chapters 8 and 9 describe alternative approaches to estimating occupancy through correction of incomplete detection probability. MacKenzie et al. (Chapter 8) provide a review of the topic, present their latest work on developing a reliable estimator of occupancy, and apply their estimator to sampling rare populations. In the next chapter, Peterson and Bayley offer a Bayesian alternative for estimating probability of presence for rare or difficult-to-detect species. Poon and Margules complete this section by describing a practical sampling method for locating populations of rare plants in remote areas, which is essentially a form of species-presence survey.

Because abundance and density are commonly used in population monitoring, a section of this volume is devoted to methods for estimating these and other population parameters for rare or elusive species or populations. Chapters in Part IV generally have a greater emphasis on the second stage of a two-stage design (counting methods), as compared to chapters in Part II. Nonetheless, both stages are important to obtaining reliable parameter estimates.

The first four chapters in Part IV describe noninvasive methods for estimating abundance of rare or elusive species. Waits (Chapter 11) reviews the latest literature on genetic sampling methods for estimating abundance of rare animals and then critically appraises the current state-of-the-art for using these techniques. Karanth et al. follow with an overview of photographic sampling methods. They provide practical advice for applying this approach to sampling elusive mammals in tropical forests. At the other end of the climatic spectrum, Becker et al. (Chapter 13) describe their work with using tracks in the snow to estimate abundance of rare and elusive carnivores. We move to the marine environment in Chapter 14, where Hanselman and Quinn evaluate the use of adaptive sampling and double sampling, in conjunction with trawls (invasive) and hydroacoustics (noninvasive), to estimate abundance of rare and clustered, bottom-dwelling fish species.

In Chapter 15, O’Shea et al. tackle the difficult problem of sampling bats. They conclude that reliable survival estimates of bats are much more obtainable than meaningful abundance estimates. Therefore, they discuss the pros and cons of banding bats, provide an extensive review of survival studies of bats, critically evaluate these studies, and offer examples of their own research in estimating survival rates of bats. Ganey et al. conclude Part IV with a comprehensive evaluation of the effectiveness of a monitoring protocol for detecting trends in abundance and in finite rate of population growth (λ) of Mexican spotted owls (Strix occidentalis lucida). All other chapters could be viewed, and perhaps should be viewed, within this monitoring context. As with other chapters in this volume, concepts discussed by Ganey et al. are not limited in application to one species or taxon.

The final part and chapter discuss future avenues of research and methodological development for estimating abundance of rare or elusive species. Chapter 17 is presented within a two-stage sampling framework discussed earlier but could be applied to larger-stage contexts. Nonetheless, the focus is both on probability-based selection of plots and on counting or estimation techniques that properly correct for incomplete detectability of animals or plants within these plots.

Because relatively little has been published on sampling rare or elusive species, the scope of this volume is necessarily broad, with examples from plants and animals within terrestrial, aquatic, and marine environments. This breadth of application extends to geography as well; chapter authors hail from the United States, India, New Zealand, and Australia. I hope this volume will serve as a source of information for biologists and resource managers and as a source of motivation for biometricians to engage in further research and methodological development in this area of biological sampling. When possible, biologists and resource managers should try to incorporate designs and survey methods discussed in this volume to population monitoring programs, and in turn, conduct these programs within the adaptive management framework that was discussed earlier in this chapter.

REFERENCES

Borchers, D. L., S. T. Buckland, and W. Zucchini. 2002. Estimating Animal Abundance: Closed Populations. Springer-Verlag, London.

Cochran, W. G. 1977. Sampling Techniques. 3rd ed. Wiley, New York.

Elzinga, C. L., D. W. Salzer, J. W. Willough by, and J. P. Gibbs. 2001. Monitoring Plant and Animal Populations. Blackwell Science, Malden, Massachusetts.

Hankin, D. G. 1984. Multistage sampling designs in fisheries research: Applications in small streams. Canadian Journal of Fisheries and Aquatic Sciences 41:1575–1591.

Holling, C. S., ed. 1978. Adaptive Environmental Assessment and Management. Wiley, New York.

MacKenzie, D. I., J. D. Nichols, J. E. Hines, M. G. Knutson, and A. B. Franklin. 2003. Estimating site occupancy, colonization, and local extinction when a species is detected imperfectly. Ecology 84:2200–2207.

MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. A. Royle, and C. A. Langtimm. 2001. Estimating site occupancy rates when detection probabilities are less than one. Ecology 83:2248–2255.

MacNab, J....

Table of contents

• About Island Press
• Title Page
• Copyright Page
• Table of Contents
• Foreword
• Acknowledgments
• 1 - Introduction
• Part I - OVERVIEW AND BASIC CONCEPTS
• PART II - SAMPLING DESIGNS FOR RARE SPECIES AND POPULATIONS
• Part III - ESTIMATING OCCUPANCY
• Part IV - ESTIMATING ABUNDANCE, DENSITY, AND OTHER PARAMETERS
• Part V - THE FUTURE
• Contributors
• Reviewers
• About the Editor
• Index
• Island Press Board of Directors