This three-day workshop will introduce participants to spatial capture-recapture (SCR) methods and software, ranging from the basic to the most recently developed. Participants will be given a solid grounding in the models on which both maximum likelihood and Bayesian SCR methods are based, and get practice using the methods. The focus will be on maximum likelihood inference, but we will also deal with Bayesian inference for situations in which maximum likelihood is difficult or infeasible.
In addition to bringing participants up to date with recent developments in SCR, the workshop is an opportunity for those engaged in the analysis, design, and execution of SCR surveys to discuss common issues and problems, and suggest future research directions.
The workshop will be held at the Centre for Research into Ecological and Environmental Modelling, St Andrews.
The workshop is intended for scientists who want to use, or are using SCR methods for estimating abundance, density or distribution. We anticipate a mix of biologists and applied statisticians. To gain maximum benefit from the workshop, biologists and ecologists should have some quantitative training, including basic statistical methods.
Software and data
The workshop has a substantial practical component, with time scheduled for participants to analyse pre-prepared SCR datasets as well their own data, under the supervision of workshop presenters.
The majority of the practical sessions will use the R package secr, but we will also illustrate use of Bayesian software for some situations in which this is advantageous, and we will use the R package admbsecr for analyses of acoustic SCR surveys.
Participants are encouraged to bring their own data, which should be formatted for input to the R package secr (or admbsecr for acoustic surveys). Familiarity with R will be assumed. We recommend that R analyses are conducted using RStudio. In July, we will post instructions for organising simple datasets for analysis; we will also provide on-line video tutorials for use of secr for basic analyses to bring everyone up to an introductory level of proficiency.
The workshop covers all kinds of SCR surveys. This includes
- surveys that use proximity detectors (like camera traps, hair snares or acoustic detectors, which do not detain animals), multi-catch traps (like mist nets, which detain animals but can hold multiple animals at a time) and single-catch traps (which detain animals and can hold only one animal);
- surveys that involve detections at points (camera traps/hair snares/traps) as well as those that involve transect searches or area searches;
- surveys that generate binary data (caught/not on each occasion), count data (number of times detected per occasion), or time-to-detection data;
- surveys that involve detection of “cues” (things generated by animals rather than the animals themselves, e.g. acoustic surveys, scat surveys).
We deal with
- methods for estimating detection or capture probability and overall density or abundance,
- methods for fitting density surfaces as functions of habitat or other spatially-referenced variables to produce prediction maps of animal density across the landscape,
- model selection and interval estimation,
- survey design issues.
More advanced topics (for some of which no general user-friendly software is currently available) include:
- methods of dealing with individual heterogeneity and partially-observed covariates,
- detection functions with non-Euclidian distance, varying effort and density dependence, and effective survey area parameterisation,
- modelling change over time and open populations,
- inferring animal movement and activity patterns from SCR data,
- recapture uncertainty.
The workshop will be held in a purpose-built facility at CREEM, with a combined lecture and computing laboratory. Participants are encouraged bring their own laptop computers, although a number of desktop PCs will also be available for use at the workshop.
Enquiries can be addressed to David Borchers at firstname.lastname@example.org.
Registration and Payment
 Also called “spatially explicit capture-recapture” (SECR); we use “SCR” here for brevity.