• approximate Bayesian computation;
  • hybridization;
  • Neotoma ;
  • San Joaquin Valley


Identifying historic patterns of population genetic diversity and connectivity is a primary challenge in efforts to re-establish the processes that have generated and maintained genetic variation across natural landscapes. The challenge of reconstructing pattern and process is even greater in highly altered landscapes where population extinctions and dramatic demographic fluctuations in remnant populations may have substantially altered, if not eliminated, historic patterns. Here, we seek to reconstruct historic patterns of diversity and connectivity in an endangered subspecies of woodrat that now occupies only 1–2 remnant locations within the highly altered landscape of the Great Central Valley of California. We examine patterns of diversity and connectivity using 14 microsatellite loci and sequence data from a mitochondrial locus and a nuclear intron. We reconstruct temporal change in habitat availability to establish several historical scenarios that could have led to contemporary patterns of diversity, and use an approximate Bayesian computation approach to test which of these scenarios is most consistent with our observed data. We find that the Central Valley populations harbour unique genetic variation coupled with a history of admixture between two well-differentiated species of woodrats that are currently restricted to the woodlands flanking the Valley. Our simulations also show that certain commonly used analytical approaches may fail to recover a history of admixture when populations experience severe bottlenecks subsequent to hybridization. Overall our study shows the strength of combining empirical and simulation analyses to recover the history of populations occupying highly altered landscapes.