How to Use Reduced-Representation Genomic Methods to Estimate Contemporary Effective Population Size

Effective population size (N_e) is a fundamental concept in population genetics, but it is a hard parameter to estimate in wild populations. A common single-sample method to estimate contemporary N_e uses linkage disequilibrium (LD) observed across loci presumed to be neutral and unlinked. Reduced-representation genomic methods can be leveraged to produce data from many thousands of loci in non-model species. These partial characterizations of a genome facilitate analyses of physical linkage (see the proliferation of linkage maps) and selection (see the proliferation of genome scans). However, these same aspects of genomics data also generate LD and downwardly bias estimates of contemporary N_e. LD also creates problems of pseudo-replication because loci in LD do not act independently; this in turn means that as the number of loci increases, precision does not increase as rapidly as one might expect. Here we present results from simulation studies designed to evaluate the bias and precision of LD-based N_e estimates.  Based on these results, we provide guidelines for reducing bias and assigning confidence intervals to genetic estimates of contemporary N_e.