Introduction to the Special Issue on the

Estimation of Animal Abundance and Related

Parameters
For the Third Conference on Statistics in Ecology and Environmental Monitoring held
at the University of Otago, 6–10 December 1999, we chose the theme “The Estimation of
Animal Abundance and Related Parameters.” It was no accident that this is the title of a
book by George Seber because the conference intended to pay tribute to the singular con-
tribution made by Professor Seber to statistical ecology during his long career.
The breadth of Professor Seber’s contribution can be measured by the diversity of
papers presented at the conference by an international array of speakers who spoke on sub-
jects ranging from fisheries and wildlife management to plant ecology. The papers from
fisheries biologists were especially welcome, as there are obvious synergies between
wildlife and fisheries management that have not always been fully exploited.
Fittingly, the first paper in this special issue is a plenary presentation by Professor
Seber on some important recent advances in methods for estimating animal population
parameters. In addition to comments on the future of mark–recapture methods and multi-
stage sampling, his discussion on theoretical and practical aspects of adaptive and distance
sampling is of particular interest because Professor Seber has also made important contri-
butions to the theoretical foundations of these methods. To help put these contributions into
perspective, Ken Pollock pays tribute to Professor Seber in the second paper in this special
issue.
The main scientific program at the conference opened with a session led by Anne
Chao on closed population mark–recapture models. These models represent one of the ear-
liest marriages of demography and statistics with the application of what is now known as
the Petersen estimator by Pierre-Simon Laplace to estimate the population of 18th-centu-
ry France. Despite the relative longevity of closed-population models, they remain a fertile
area for theoretical development, as shown in the review by Professor Chao that forms the
third article in this special issue.
Closed-population models were featured in several of the papers presented at the con-
ference. The paper by Paul Yip, Emmy Wan, and K. S. Chan provides a unified mark–
recapture and removal analysis that allows for losses on capture. The paper by Darryl
MacKenzie and Bryan Manly reports the development of new methods for comparing
models based on randomization tests. Both of these papers represent important theoretical
developments. There were also some interesting applications of closed-population models.
Malcolm Faddy, Jennifer Brown, and Phil Commins describe a method for estimating
brushtail possum abundance on the basis of removals of animals by modeling the number
of possum captures in traps using a Poisson process. Klaus Follner and Klaus Henle used

137
138 R. BARKER AND D. FLETCHER

mark–recapture models to estimate species richness, a problem with a pedigree stretching

back to R. A. Fisher.
Carl Schwarz introduced the open-population, mark–recapture session and empha-
sized the need for statisticians to “think outside the square” in a lucid review of open-pop-
ulation models. In 1964, Richard Cormack developed what is now referred to as the
Cormack–Jolly–Seber model in which recaptures of marked animals are modeled to pro-
vide information on survival and capture probabilities. In this model, the number of ani-
mals released in each sample is regarded as uninformative. The advance provided by
George Jolly and George Seber in 1965 was to treat the numbers of unmarked animals cap-
tured at each sample as random variables generated from a distribution that has abundance
parameters. In recent years, much of the focus in the mark–recapture literature has been on
the Cormack–Jolly–Seber model, but Professor Schwarz stresses the need for an increased
focus on abundance and parameters related to abundance. He argues that parameterizing
the likelihood in terms of birth rates rather than the total number of unmarked animals in
the population is a more useful approach, allowing sensible constraints to be made on the
birth process.
The paper by Trent McDonald and Steven Amstrup and that by McDonald, Amstrup,
and I. Stirling both discuss the application of open-population models to the problem of
estimating the abundance of polar bears in the Beaufort Sea when capture probabilities are
heterogeneous. They model heterogeneity by use of individual covariates and use of a
Horvitz–Thompson-type estimator to estimate abundance.
The remainder of the papers in this special issue discuss interesting ecological appli-
cations. Gary Jackson and Yuk Cheng discuss stock assessment of a Western Australia
snapper fishery using egg surveys with adaptive sampling; Norm Good, Michelle Paterson,
Cris Brack, and Kerrie Mengersen present methods for estimating tree biomass; Paul
Scofield, David Fletcher, and Christopher Robertson reanalyze sooty shearwater data, col-
lected between 1940 and 1957, using a multi-state, mark–recapture model; and David
Gilbert, Jeremy McKenzie, and Nicholas Davies look at evidence for trap shyness in a
tagged snapper population.
The final session at the conference, led by Jim Nichols, was on the role of statistics in
advancing ecological theory. As statistical ecologists, our performance is measured by how
we have contributed to the advancement of statistical theory and the management of bio-
logical populations. This brings us back to our primary reason for the Statistics in Ecology
and Environmental Monitoring series of conferences, which is to foster the synergy
between ecology and statistics.
In Professor Seber’s opening remarks, he commented on the need for statistical theo-
ry not to outstrip ecological need. For many statisticians, the advancement of statistical the-
ory for its own sake can have a special allure. Ultimately, however, our contribution as sta-
tisticians will be measured by whether our science has been able to make a real contribu-
tion to the advancement of knowledge. In this, statisticians and ecologists each have their
own particular responsibilities. The statistician needs to listen carefully to what the ecolo-

138
 INTRODUCTION TO SPECIAL ISSUE 139

gists are saying about the problems they have and to the questions they want to answer. The
statistician should endeavor to solve these problems rather than work on peripheral ques-
tions that are mathematically easier or more attractive. The ecologists however, have a par-
ticular responsibility to think carefully about their needs and to communicate these needs
clearly to the statisticians.

Richard Barker and David Fletcher, Guest Editors

Department of Mathematics and Statistics
University of Otago, P.O. Box 56
Dunedin, New Zealand

139