CKMRpop: Forward-in-time simulation and tabulation of pairwise kin relationships in age-structured populations

In the last five years, interest in close-kin mark-recapture (CKMR), a variant of mark-recapture that uses genetically inferred kin as 'recaptures', has grown dramatically. However, understanding the basis of CKMR, and properly implementing it, remains challenging. This paper describes an R package, CKMRpop, for simulating age-structured populations with user-specified demography, overdispersed variance in reproductive success (allowing for different ratios of effective to census size) and random sampling of individuals. Using compiled code for the simulation makes it feasible to simulate populations of millions of individuals. From the simulation output, pairs of sampled individuals related within a user-specified number of generations are found. Such pairs form the foundation for CKMR inference, and simulating them provides insight for understanding the statistical basis for CKMR and for assessing the feasibility of CKMR in different scenarios. We predict that CKMRpop will serve as an important tool for researchers contemplating CKMR estimation of population size. Furthermore, the methods presented here for identifying and categorizing relationships beyond half-siblings allow a more complete picture of the wide variety of kin pairs encountered in populations. This identifies the fraction of kin pairs that may not be the target of a CKMR experiment, but may be inadvertently mistaken for a more closely related 'target' kin pair. Additionally, as more distant kin categories will likely be accurately inferred from increasingly available and inexpensive whole genome resequencing, understanding the distributions of more distant relationships in populations is a first step towards broadening the scope of CKMR to include them.

Automated summary

The study introduces an R package, CKMRpop, for simulating age-structured populations to help understand the foundations of CKMR and assess its feasibility in different scenarios. By simulating output, sampled pairs of individuals related within a user-specified number of generations can be found, providing a basis for CKMR inference.