Variation among Global Circulation Models for reconstructions of geographic distributions at the Last Glacial Maximum: relevance for phylogeography

Abstract

Guevara, L., León-Paniagua, L., Ríos, J., Anderson, R. P. 2018. Variation among Global Circulation Models for reconstructions of geographic distributions at the Last Glacial Maximum: relevance for phylogeography. Ecosistemas 27(1): 62-76. Doi.: 10.7818/ECOS.1443

Reconstructing geographic distributions during the Last Glacial Maximum (LGM) is an increasingly common practice in phylogeographic studies. It allows the inference of stable areas over time (potential refugia) or areas of postglacial expansion, which are hypotheses that can be tested by patterns of genetic variation. Usually, researchers estimate more than one paleodistribution by transferring the ecological niche model using the output of various Global Circulation Models (GCMs, which estimate the climate). Here, we focus on the variation between paleodistributions reconstructed for the LGM, which could be due to two related reasons: (1) the use of different GCMs to transfer an ecological niche model and (2) the degree of extrapolation that is necessary (according to the GCM used) to predict in conditions more extreme than those existing in the present (non-analog conditions). First, we quantified the differences of two GCMs (CCSM4 and MIROC-ESM) commonly used for transferring niche models into the climate of the LGM (in three regions widely studied in phylogeography). Second, we investigated whether these differences between GCMs did or did not lead to variation in paleoreconstructions reported in the 31 phylogeographic studies found for these regions. The two CGMs examined showed significant differences for almost all variables in the three study areas. Importantly, the differences between the two GCMs were manifested in different ways in each study area. These differences led to notable discrepancies in the paleoreconstructions reported by the articles. Despite the importance of detecting non-analog conditions and deciding how to extrapolate into them (and their effects on results and inferences), none of the examined studies considered these factors. The present study illustrates the value of assessing the variation among GCMs in each study area to create paleoreconstructions (made using ecological niche models) that should be tested with additional information, such as genetic and paleoecological data.