Sparse capture histories and an open population model?

Forum for discussion of general questions related to study design and/or analysis of existing data - software neutral.

Sparse capture histories and an open population model?

Postby ajdennhardt » Sat Oct 14, 2017 10:18 am


I am consulting a client who has secondarily acquired what I’m calling some “messy data.” The client is primarily interested in estimating population size from a sample of marked individuals.

The data are capture-mark-recaptures (by relocation and resight) for individuals of a particular mammal species in the Upper Midwest, and the client has capture histories for collared individuals spanning March 2013 to December 2016. From my perspective, the biggest problem with these data is that they appear to be conveniently sampled in the same study area over the time series. In brief, biologists simply went out on any number of days each year (e.g., whenever they had the time and/or money) to capture new or recapture previously marked animals. Furthermore, the target species is poorly known, and thus no strict time period within any one year was chosen for sampling the population in a state that can be considered closed.

Sampling, in terms of animal relocation (resight/recapture) timing, also appears haphazard and inconsistent from year to year.

What this leaves the client with is a capture history dataset, for approximately 40 individual animals, comprising many zeros and only some ones. The data are sparse, but the study area at least remained constant over the course of the surveys.

Today, I’m interested in gaining some advice on how to recommend that the client proceeds with analyzing these data in an open population mark-recapture framework, if at all.

-Are there any capture-mark-recapture models that can easily handle such sparse capture histories?

-Are any of these models supported in Program MARK and R (e.g., RMark)? References?

-Should we be gravely concerned with (almost certain and rampant) violations of key model assumptions and/or are there ways that we can correct these issues?

-Can we reasonably implement the following solution (and are there any known references for such an idea?) and apply traditional estimation methods like POPAN Jolly-Seber: replace zeros with ones in sampling occasions during a prolonged period of missingness for animals (a) initially captured and marked, (b) missing for a prolonged period (e.g., weeks to months), and then (3) recaptured later on in the time series (i.e. the animal survived from occasion i to i + n)? See example below (replacements in bold under “To:”).

ID Date_1 Date_2 Date_3 Date_4 Date_5 ... Date_n
1214m 0 1 0 0 1 0

ID Date_1 Date_2 Date_3 Date_4 Date_5 ... Date_n
1214m 0 1 1 1 1 0

Any suggestions or advice would be much appreciated. Thanks in advance!

Posts: 1
Joined: Wed Oct 04, 2017 10:42 am

Re: Sparse capture histories and an open population model?

Postby bacollier » Sun Oct 15, 2017 12:31 pm

What you are describing below is the case study of how not to do a mark-recapture study. Attempting to ram a block of wood of undetermined size and structure into a undefinable hole because a client has some data they want analyzed is (IMO) a useless scientific endeavor.

That said, you should at least rtfm and the 100s (if not more) of CMR papers that have been published on similar topics. When you ask explicitly if there are references or models for sparse data, or if things like what you want exist in MARK or if you should worry about assumptions or replacing zero's with one's all show that you have not done your homework as all you list are answered in 'the book' (see

The folks here will help all those that do the background work, but no effort on your part, no effort on ours.

Posts: 227
Joined: Fri Nov 26, 2004 10:33 am
Location: Louisiana State University

Return to analysis & design questions

Who is online

Users browsing this forum: No registered users and 1 guest