Int. J. Plant Sci. 175(6):688–701. 2014. Copyright is not claimed for this article. DOI: 10.1086/676305 EFFECT OF A GEOGRAPHIC BARRIER ON ADAPTATION IN THE DWARF SUNFLOWER (HELIANTHUS PUMILUS NUTT.) Patrick A. Reeves1,* and Christopher M. Richards* *USDA-ARS, National Center for Genetic Resources Preservation, 1111 South Mason Street, Fort Collins, Colorado 80521, USA Editor: Pamela Soltis Premise of research. Geographic isolation promotes local adaptation but may also prevent the movement of advantageous gene complexes throughout a species distribution. We identify a geographic barrier that inhibits gene flow and adaptation across the disjunct species range of the Rocky Mountain endemic sunflower Helianthus pumilus. Methodology. Habitat suitability modeling and least-cost path analysis were used to characterize envi- ronmental preferences across the species range and to identify barriers to gene flow. Simple sequence repeat data were used to assess genetic structure and estimate recent levels of gene flow across the distributional disjunction. A common garden was used to study phenotypic differentiation and local adaptation. Controlled crosses tested for intrinsic barriers to reproduction. Pivotal results. Despite significant differences in available habitat, individuals in northern and southern regions occupy similar sites, suggesting limited adaptive differentiation in environmental preference. A region of low-suitability habitat separates the two regions. Northern and southern plants are genetically distinct. There has been little recent gene flow. Southern plants had higher fitness (measured as seed set) in the northern region than northern plants. A suite of morphological characters, present in half of southern plants, conveyed a twofold fitness advantage in the north, but this adaptive phenotype is virtually nonexistent in the northern range. Between-region crosses were equal to or more productive than within-region crosses. Conclusions. The Pikes Peak massif, a high-altitude eastward projection of the Rocky Mountains, forms a geographic barrier to gene flow in H. pumilus, a lowland foothills species. No intrinsic barriers to gene exchange between regions have yet evolved. Therefore, the geographic barrier appears to impede the movement of the adaptive phenotype found in the southern region, resulting in a northern group that is less fit than it might be, contrary to the expectation of local adaptation. Keywords: fitness, gene flow, niche model, population structure, reproductive isolation, common garden. Online enhancement: appendix. Introduction mechanism (Wiens 2004). Given that dispersal distances for seeds and pollen are usually limited (by pollination vectors, Geographic isolation drives biological diversification. Re- dispersal agents, or physical processes), spatial disjunction re- strictions in gene flow between populations caused by simple duces the flow of genes between populations in proportion to spatial disjunction are ultimately responsible for many of the the magnitude of the distance between them. Geographic iso- species we observe today (Jordan 1905; Mayr 1963; Barra- lation, over time, leads to the serendipitous accumulation of clough and Vogler 2000; Coyne and Orr 2004, pp. 83–124, other—intrinsic—isolating mechanisms, which evolve as a by- 168–178; Fitzpatrick and Turelli 2006). Excluding polyploid product of adaptation to distinct environments or via genetic speciation, a few cases of sympatric speciation have been doc- drift, thereby promoting the formation of new species (Dob- umented in plants (Gottlieb 1973; Savolainen et al. 2006; zhansky 1951, pp. 206–211; Mayr 1959; Schluter 2001; Papadopulos et al. 2011), but most botanists would agree that Coyne and Orr 2004). While spatial discontinuity alone, in- allopatric speciation appears to be the dominant mode by dependent of any intrinsic mechanism, might cause complete which plants diversify. reproductive isolation, one must also consider the possibility Geographic isolation promotes diversification by function- that the spatial discontinuity itself is a consequence of genetic ing as a powerful, premating, extrinsic reproductive isolating differences—adaptations—that preclude a continuous distri- bution (Stebbins 1950, pp. 197–199; Schemske 2000). Sobel 1 Author for correspondence; e-mail: [email protected]. et al. (2010) recognized this conundrum and proposed two Manuscript received June 2013; revised manuscript received December 2013; definitions that echo Stebbins (1950, p. 196): “effective geo- electronically published May 27, 2014. graphic isolation” describes the contribution of both extrinsic 688 REEVES & RICHARDS—BARRIERS TO GENE FLOW IN HELIANTHUS PUMILUS 689 and intrinsic factors to reductions in gene flow between al- their location and describing their consequences in popula- lopatric populations, whereas “ecogeographic isolation,” on tions, as well as farther along the speciation continuum, is the other hand, describes the contribution of intrinsic factors critical for fully understanding plant diversification. For some alone, i.e., adaptations that promote geographic isolation. For well-collected species, the location of a potential barrier can convenience, we utilize a third term from Stebbins (1950), be found using maps, as a geographic gap in the species dis- “spatial isolation,” which refers to the reduction in gene flow tribution. If a geographic gap can be shown to contain suitable attributable solely to the physical distance between groups, habitat, either by transplant experiment or by modeling, it may i.e., to dispersal limitation. Spatial isolation plus ecogeographic form a contemporary barrier due to dispersal limitations but isolation equals effective geographic isolation. is unlikely to remain as such over evolutionary time periods, The contribution of geographic isolation relative to other as intervening suitable habitat is colonized and a conduit for isolating mechanisms has rarely been quantified, but where it gene flow is established. If, on the other hand, intervening has it appears to be important. When pollinator isolation was habitat is unsuitable, the gap may restrict gene flow by pre- suspected a priori to be the predominant mechanism main- venting colonization. The latter type of gap is more likely to taining species distinctness between sympatric species pairs, function as a barrier over evolutionary time because gene flow two studies found that geography (technically, effective geo- across it requires infrequent, long-distance dispersal events. graphic isolation) was, in fact, an equivalent or stronger iso- Finally, if a gap can be associated with a restriction in gene lating force, accounting for 35%–60% of total reproductive flow, it may then be considered a barrier. Two conditions are isolation (Ramsey et al. 2003; Kay 2006). In a diploid/tetra- necessary. First, genetic differentiation should be demonstrable ploid species pair, Husband and Sabara (2004) found that across the barrier. This can be tested using molecular marker geographic isolation was more important than triploid block, data (Pearse and Crandall 2004) or via traditional morpho- a strong isolating mechanism caused by ploidy incompatibil- logical studies. Second, the barrier itself must be responsible ities in the backcross generation (Ramsey and Schemske 1998). for the genetic differentiation, as opposed to migration from This finding is remarkable because polyploid speciation is often already-differentiated source populations (e.g., distinct glacial cited as the most plausible form of sympatric speciation in refugia). To satisfy this condition, one must demonstrate that plants (Otto and Whitton 2000), yet geographic isolation may habitat on both sides of the barrier has been continuously still function as the critical (and largely unappreciated) factor occupied and that the barrier region has been unoccupied, or allowing divergence. unsuitable, for an evolutionarily relevant period of time. This Effective geographic isolation can be estimated using dis- will generally be difficult, although for a minority of species persal distance and range overlap (Ramsey et al. 2003; Kay fossil or palynological data may be available to help prove the 2006). Best practices for teasing apart, in a quantitative sense, case. For most species, paleoclimatic habitat suitability mod- the contribution of spatial and ecogeographic components of eling seems to offer a reasonable alternative (Hugall et al. effective geographic isolation have yet to be developed, but 2002; Carstens and Richards 2007). Sobel et al. (2010) present experimental recommendations to Geographic barriers to gene flow are generally thought of help determine which form predominates. First, reciprocal as a creative force in evolution, one that promotes adaptive transplants can be used to estimate the relative fitness of two differentiation (Slatkin 1987; Garant et al 2007) by disrupting disjunct populations in sympatry. If local genotypes perform maladaptive gene flow from other, ecologically distinct regions better than foreign genotypes, local adaptation has occurred (Dobzhansky 1940; Kirkpatrick and Barton 1997; Lenormand and ecogeographic isolation is indicated. If local and foreign 2002; Coyne and Orr 2004). Localized, adaptively differen- genotypes perform equivalently or if foreign genotypes out- tiated populations result and are an important step toward perform locals, ecogeographic isolation should be rejected. Sec- new species formation. But barriers might also impede adap- ond, habitat suitability