354 www.newphytologist.org Research Blackwell Publishing Ltd Collinsia sparsiflora in serpentine and nonserpentine habitats: using F2 hybrids to detect the potential role of selection in ecotypic differentiation Jessica W. Wright 1 and Maureen L. Stanton Section of Evolution and Ecology and Center for Population Biology, One Shields Avenue, University of California, Davis, CA 95616, USA; 1 Present address: USDA Forest Service, Pacific South-west Research Station, Institute of Forest Genetics, 1100 West Chiles Road, Davis CA 95618, USA Summary • Here we document phenotypic differences between serpentine and nonserpentine ecotypes of Collinsia sparsiflora, as well as patterns of selection in these contrasting soil habitats. • We transplanted the two parental ecotypes and experimental F2 hybrids into six field sites, and collected morphological, phenological and fitness data on emergent plants. To focus on edaphically mediated selection, rather than on pollinator-mediated selection, we used pollinator-exclusion cages. • Transplanted parentals of the two ecotypes showed genetic differentiation for floral traits, but not for phenological traits or cotyledon size. For the F2 hybrids growing on serpentine soils, there was significant directional selection on cotyledon size, flower size and flower shape. However, the pattern of selection did not differ significantly between serpentine and nonserpentine sites. • Overall, we did not see evidence for divergent selection on the two soil types. We conclude that differences in floral traits between the ecotypes do not reflect adaptation to physical conditions associated with soil type, and that there are unmeasured traits that must be contributing to ecotypic differentiation. Key words: ecotypic differentiation, fitness components, floral traits, natural selection, principal components analysis, serpentine soils. New Phytologist (2007) 173: 354 –366 No claim to original US government works. Journal compilation © New Phytologist (2006) doi: 10.1111/j.1469-8137.2006.01925.x Author for correspondence: Jessica W. Wright Tel: +530 758 6350 Fax: +530 758 1070 Email: jessicawwright@fs.fed.us Received: 18 July 2006 Accepted: 20 September 2006 Introduction Ecotypic differences result from divergent selection in contrasting environments. Reciprocal transplant experiments are the most powerful tool for determining the presence of local adaptation and ecotypic differentiation within species (reviewed in Briggs & Walters, 1984; Galen et al., 1991; Jordan, 1992; Montalvo & Ellstrand, 2001; Waser & Price, 1985). Traditionally, seeds or plants from a series of focal sites are moved into each of the environments, and then are evaluated for survival and/or reproductive success. Although the reciprocal transplant experiment is a power- ful tool for demonstrating adaptive ecotypic differentiation, one cannot assume that all phenotypic traits distinguishing locally adapted ecotypes are under divergent selection. The gene complexes that give each ecotype a distinct phenotype are partially the result of divergent selection, but will also be influenced by genetic linkage, pleiotropic effects of genes under selection, and random drift (Kaplan et al., 1989; Lynch & Walsh, 1998). Because multiple traits are often highly cor- related within ecotypes, it may not be possible to evaluate selection on individual traits using transplanted ecotypes alone (Lande & Arnold, 1983; Jordan, 1991; Nagy, 1997; Lexer et al., 2003). Hybridization between ecotypically differentiated popula- tions can break up complexes of coadapted and linked genes, and consequently the use of experimental hybrids can allow investigators to measure selection on individual traits (reviewed