IBS: Estimating Abundance and Beyond

12.30-13.15 – Presidential Address – Ruth King (University of Edinburgh)

Estimating Abundance: Past, Present and Future

Estimating population sizes is important across many different areas from conservation management of wild animal populations to assessing government policies in relation to the hidden populations (such as problem drug users, homeless human trafficking etc.) and the associated socio-economic impact. Thus, unsurprisingly, there is also a long history of estimating different populations. I will focus on the particular approach of capture-recapture techniques that dates back to at least the 1600s and describe an (abridged and incomplete) history of the development of associated statistical approaches which have been applied primarily within the areas of ecology and epidemiology. Many of the statistical ideas have been developed in parallel in some sense within the different areas, including even the terminology of the techniques: “capture-recapture” in ecology; and “multiple systems estimation” in epidemiology. The talk will conclude with some discussion of the statistical challenges that remain – and which the following speakers will develop further with particular applications.

13.15-14.00 – Antony Overstall (University of Southampton)

Challenges of estimating human population sizes

Capture-recapture (multiple systems estimation) methods have been used to estimate elusive human population sizes including those of victims of human trafficking, people who misuse drugs and Covid-19 cases. Typically, members of the population are observed by several different administrative sources. Information from the different sources is combined using capture-recapture statistical modelling (usually log-linear models) to estimate the total population size (possibly stratified by age-group, sex, etc).

However, there are a number of challenges to using capture-recapture for estimating human population sizes including humans acting in a more heterogenous manner (compared to other animals) invalidating the assumptions of the statistical methods employed, different definitions of the target population by different sources, and privacy issues. The talk will discuss some of these issues and possible solutions in the context of estimating the number of people who inject drugs in the United Kingdom and the number of victims of human trafficking.

14.00-14.45 – David Borchers (University of St Andrews)

Addressing spatial questions in ecology using capture-recapture

Abstract: Spatial Capture-Recapture (SCR) methods adds “spatial” to “capture-recapture” in more than just name. SCR methods allow capture-recapture methods to be used to address questions of a fundamentally spatial nature. Aside from providing rigorous estimates of population density (in addition to abundance), SCR methods provide a means of modelling wildlife distribution in space, a means of investigating the drivers of this distribution, drivers of habitat use, and habitat connectivity.

I will illustrate these things using a class of SCR model that can be viewed as Poisson regression models in which animal activity centres are random (spatial) effects, and in which the thinning probabilities are unknown and depend on these random effects. Interest usually focusses on the random effect distribution (which governs animal density and abundance) and/or on the thinning probabilities, which in certain applications are informative about animal habitat use and habitat connectivity.