User Tools

Site Tools


The Expansion-Contraction Model of Pleistocene Refugia

Author: Bryan_P_White

Original Publication: 09/14/2012

Edited for grammar/syntax: 07/26/2019


The last glacial maximum (LGM) was a distinct geological period in which ice and glaciers covered all of modern Canada and extended down into some regions of the United States. It has been shown through fossil and pollen records that many terrestrial species underwent large shifts in distribution following the end of the LGM, 20,000 years ago. However, few studies have attempted to characterize the existence of, and/or extent of, shifts in the distributions of marine species following the end of the LGM.

The existence of LGM-caused patterns in distribution shifts can be tested using standard population genetics methods, where populations exhibiting re-colonization events following the end of the LGM should exhibit patterns consistent with large population expansions. Range expansions (and potentially contractions) should be expected following the end of the LGM as ice sheets retract and water temperatures rise. Empirical evidence for this expansion and contraction (EC) model has been demonstrated in both North America and western Europe wherein species survive in southern refugia and later re-colonize northern latitudes following the end of the LGM.

In order to test for the presence of patterns corresponding to the EC model, the authors conducted a community-level analysis of 14 rocky shore species spanning four phyla. To the author's best knowledge, no biogeographic study of these 14 species had been conducted spanning for more than 30 years, or with spatial scales ranging more than a few meters.

Methods A combination of new and old sequences were combined, resulting in roughly ~300 to ~600 bp of character data from both mtDNA and nuclear DNA, although sequence length and loci varied from species to species. Population genetic statistics were calculated using Arlequin, haplotype networks were constructed using PAUP*, and Bayesian skyline plots generated using BEAST.

Results Sequence Diversity Most species showed patterns consistent with either recent expansion, or selective sweeps.

Population Structure Nine out of 14 species showed no spatial structure according to population genetic statistics consistently for both mtDNA and nuclear loci. Other population structure seems unclear and difficult to understand.

mtDNA haplotype networks and mismatch analysis Mismatch distributions seem reflect haplotype diversity and coalescence time. Differences in population expansion times seem to be clearly associated with adult form – either benthic or sessile – with the confidence interval of benthic species expansion occurring before, during, or after the LGM.

Bayesian demographic reconstructions Some species show patterns of increases in NeT during and at the end of the LGM, but many show increases before, leading up to, and peaking at the start of the LGM.


Overall, seven species showed patterns consistent with LGM persistence, 5 with re-colonization, and 2 with spatial variability.

Critique It is unclear where, or if, combined data sets were used. Confidence intervals for most analyses are very large, suggesting high variability in data which could be related to differing coalescent times, which could be confounded by incomplete lineage sorting. Given the variability encountered in this data set, I do not think species can be confidently classified as either persistent or post-LGM expansionary. What I see is strong evidence for LGM persistence for all taxa, with the potential for some sub-populations exhibiting re-colonization and expansions – a pulse effect (Figure 6). Assuming a 2%/MY mutation rate across all taxa could be very problematic, particularly across different phyla, although most metazoan mitochondria (excluding Cnidaria) evolve at similar rates. In this case, I would have liked to see the minimum and maximum divergence times for each species, and looked for a pattern of different mutation rates resulting in concordant population expansion times. That is to say, if species A evolved at a 1%/MY and species B evolved at a 3%/MY, would they then have concordant expansion times? For future work I would suggest focusing on a single taxa and a combined data set in order to gain better estimates of coalescent times.

Article Reviewed: Marko, P. B., J. M. Hoffman, S. A. Emme, T. M. McGovern, C. C. Keever, and L. N. Cox. 2010. The 'Expansion-Contraction' model of Pleistocene biogeography: rocky shores suffer a sea change? Molecular Ecology 19:146-169. doi:

bpwhite/expansion_contraction_model.txt · Last modified: 2019/07/26 19:58 by bpwhite