Molecular Ecology (2010) 19, 3421–3443 doi: 10.1111/j.1365-294X.2010.04754.x Multiple Pleistocene refugia and Holocene range expansion of an abundant southwestern American desert plant species (Melampodium leucanthum, Asteraceae) CAROLIN A. REBERNIG,* GERALD M. SCHNEEWEISS,†1 KATHARINA E. BARDY,†* PETER SCHO¨ NSWETTER,†‡ JOSE L. VILLASEN˜ OR,– RENATE OBERMAYER,* TOD F. STUESSY* and HANNA WEISS-SCHNEEWEISS* *Department of Systematic and Evolutionary Botany, University of Vienna, Rennweg 14, A-1030 Vienna, Austria; †Department of Biogeography and Botanical Garden, University of Vienna, Rennweg 14, A-1030 Vienna, Austria; ‡Department of Systematics, Palynology and Geobotany, Institute of Botany, University of Innsbruck, Sternwartestrasse 15, A-6020 Innsbruck, Austria; –Instituto de Biologı´a, Departamento de Bota´nica, Universidad Nacional Auto´noma de Me´xico, Tercer Circuito s ⁄ n, Ciudad Universitaria, Delegacio´n Coyoaca´n, MX-04510 Me´xico D. F., Me´xico Abstract Pleistocene climatic fluctuations had major impacts on desert biota in southwestern North America. During cooler and wetter periods, drought-adapted species were isolated into refugia, in contrast to expansion of their ranges during the massive aridification in the Holocene. Here, we use Melampodium leucanthum (Asteraceae), a species of the North American desert and semi-desert regions, to investigate the impact of major aridification in southwestern North America on phylogeography and evolution in a widespread and abundant drought-adapted plant species. The evidence for three separate Pleistocene refugia at different time levels suggests that this species responded to the Quaternary climatic oscillations in a cyclic manner. In the Holocene, once differentiated lineages came into secondary contact and intermixed, but these range expansions did not follow the eastwardly progressing aridification, but instead occurred independently out of separate Pleistocene refugia. As found in other desert biota, the Continental Divide has acted as a major migration barrier for M. leucanthum since the Pleistocene. Despite being geographically restricted to the eastern part of the species’ distribution, autotetraploids in M. leucanthum originated multiple times and do not form a genetically cohesive group. Keywords: desert biota, Holocene aridification, Melampodium, phylogeography, polyploidy, refugia Received 24 March 2010; revision received 28 May 2010; accepted 5 June 2010 Abbott & Brochmann 2003; Scho¨nswetter et al. 2005). Introduction The role of these climatic fluctuations in other regions, The impact of Pleistocene climatic fluctuations on however, remains less well understood. This is particu- directly affected areas, such as the Arctic or temperate larly the case for arid regions in northern Mexico and high mountain ranges, has been comparatively well adjacent southwestern United States. Paleoclimatic and investigated phylogeographically in both plants and paleovegetational evidence unambiguously suggests animals (Hewitt 1996, 2001; Brunsfeld et al. 2001; that desert vegetation was strongly restricted during the wetter and cooler pluvial periods (Wells 1966; Van Correspondence: Gerald M. Schneeweiss, Fax: +43 1 4277 9541; Devender & Spaulding 1979; Thompson & Anderson E-mail: [email protected] 1Present Address: Systematic Botany and Mycology, Ludwig- 2000) and confined to refugia in the west and south, Maximilians-University Munich, Menzingerstrasse 67, D-80638 such as the lower Colorado River Basin, the plains Munich, Germany. of Sonora, or the southern Chihuahuan Desert (Van Ó 2010 Blackwell Publishing Ltd 3422 C. A. REBERNIG ET AL. Devender 1990; Thompson & Anderson 2000; Hunter tinental Divide (Nason et al. 2002; Clark-Tapia & Moli- et al. 2001). Large-scale aridification of the whole region na-Freaner 2003; Fehlberg & Ranker 2009; Garrick et al. started only after the end of the last glacial maximum 2009; Sosa et al. 2009) or they do not employ molecular (Van Devender & Spaulding 1979; McClaran & Van methods (Hunter et al. 2001; Holmgren et al. 2007). Devender 1995; Bousman 1998; Metcalfe et al. 2000; By affecting the distribution of a species, environmen- Musgrove et al. 2001; Holmgren et al. 2007) and was tal changes will also shape its evolution via, for accompanied by a shift from xeric woodlands, abundant instance, enabling or interrupting gene flow in phases until 8000 years BP (Van Devender 1977), to semidesert of continuous distribution and range disruption, respec- grassland and eventually desert shrubland vegetation tively, or affecting the success of establishment of newly (Neilson 1986). Consequently, drought-adapted species formed polyploids (Husband 2004; Baack & Stanton are expected to have persisted in one or more distinct 2005; Ramsey et al. 2008). The latter is of particular rele- refugia (Nason et al. 2002; Fehlberg & Ranker 2009), vance, because polyploidy is recognized as an impor- from where they reached their current distribution after tant mode of speciation in general and one of the more range expansion within the last 10 000–6000 years (Van likely means of sympatric speciation in particular (Otto Devender & Spaulding 1979; Spaulding 1990; Van & Whitton 2000; Coyne & Orr 2004; Soltis et al. 2007). Devender 1990; Holmgren et al. 2007). While the role of allopolyploidy for speciation has long These range expansions into new arid regions are been recognized (Ramsey & Schemske 1998, 2002; Le- expected to have had major impacts on population itch & Leitch 2008), the rapidly mounting evidence of a structure and genetic diversity, for instance resulting in high frequency of autopolyploids, often in mixed popu- loss of alleles because of bottlenecks and founder lations with their diploid progenitors (Husband 2004; events, or in secondary contact of genetic lineages dif- Suda et al. 2007), has led to a more positive view con- ferentiated in allopatric refugia (Hewitt 2001, 2004). Pa- cerning the evolutionary significance of autopolyploidi- leoclimatic modelling indicates that the aridification zation (Soltis et al. 2007). Despite several recent studies progressed from the Sonoran Desert north- and east- dealing with the dynamics of diploid–autopolyploid wards (Holmgren et al. 2007), and it can be expected complexes (Baack & Stanton 2005; Scho¨nswetter et al. that range expansion of drought-adapted species fol- 2007; Ramsey et al. 2008; Hu¨ lber et al. 2009), their evo- lowed the same general direction (Fehlberg & Ranker lutionary significance and the factors involved in poly- 2009). Additionally, rapid expansion should also be ploid cytotype formation and establishment are still reflected in geographic patterns of genetic diversity, poorly understood (Baack & Stanton 2005). which is expected to be lower in more recently colo- Here we use Melampodium leucanthum (Asteraceae), nized areas because of founder effects (Hewitt 1996). A an abundant taxon of the North American desert and longitudinal migration pattern may, however, be modi- semi-desert regions, to investigate the impact of the fied by the Continental Divide, whose establishment in major aridification in southern North America within the late Tertiary is thought to have caused vicariant the last 10 000 years on phylogeography and evolution, diversification in a number of warm-desert animals including cytotype differentiation, in a drought- (Riddle & Hafner 2006; Castoe et al. 2007). Since then, adapted plant species. This phylogenetically distinct the divide has acted as a formidable migration barrier (Blo¨ch et al. 2009) and morphologically and taxonomi- for desert biota because of the lack of a spatially and cally homogeneous species is particularly well suited temporally continuous connection between the Sonoran to address these questions, because it is distributed and the Chihuahuan Deserts, which currently come over several major arid and semi-arid biogeographic closest at the Derning Plains near the border between regions ranging from the Sonoran and Chihuahuan Arizona and New Mexico (Morafka 1977; Riddle & Haf- Deserts to the Tamaulipan Plain and Southern Plain ner 2006; Castoe et al. 2007). This barrier is expected to region (Stuessy 1972), and it comprises diploid and tet- enhance founder effects in the course of eastward raploid cytotypes, the latter restricted to the eastern migration. Alternatively, if refugia of drought-adapted part of the distribution area (Fig. 1, Table 1; Stuessy species were also located east of the Continental Divide et al. 2004). Our first aim is to analyse the phylogeo- (Hunter et al. 2001; Castoe et al. 2007), this region prob- graphic patterns caused by post-Pleistocene aridifica- ably is the contact zone of western and eastern lineages tion and subsequent migration events. Specifically, we (Castoe et al. 2007). While several of these hypotheses want (i) to determine the locations of the refugia of have been tested in a number of animal groups (Jaeger M. leucanthum and test whether these are congruent et al. 2005; Riddle & Hafner 2006; Castoe et al. 2007; with those suggested by paleoclimatic and phylogeo- Haenel 2007; Fontanella et al. 2008), comparable studies graphic data (Hunter et al. 2001; Castoe et al. 2007; in plants are lacking. The few studies from desert plants Holmgren et al. 2007; Fehlberg & Ranker 2009); (ii) to either investigate species from only one side of the Con- infer the directionality of the range expansion, in Ó 2010 Blackwell Publishing Ltd PHYLOGEOGRAPHY OF NORTH AMERICAN DESERT PLANT 3423 (a) (b) (a) (b) Fig. 1 Physical map