Association Genetics of Coastal Douglas Fir ( Pseudotsuga menziesii var . menziesii , Pinaceae ) . I . Cold - Hardiness Related Traits Andrew

Adaptation to cold is one of the greatest challenges to forest trees. This process is highly synchronized with environmental cues relating to photoperiod and temperature. Here, we use a candidate gene-based approach to search for genetic associations between 384 single-nucleotide polymorphism (SNP) markers from 117 candidate genes and 21 cold-hardiness related traits. A general linear model approach, including population structure estimates as covariates, was implemented for each marker–trait pair. We discovered 30 highly significant genetic associations [false discovery rate (FDR) Q , 0.10] across 12 candidate genes and 10 of the 21 traits. We also detected a set of 7 markers that had elevated levels of differentiation between sampling sites situated across the Cascade crest in northeastern Washington. Marker effects were small (r , 0.05) and within the range of those published previously for forest trees. The derived SNP allele, as measured by a comparison to a recently diverged sister species, typically affected the phenotype in a way consistent with cold hardiness. The majority of markers were characterized as having largely nonadditive modes of gene action, especially underdominance in the case of coldtolerance related phenotypes. We place these results in the context of trade-offs between the abilities to grow longer and to avoid fall cold damage, as well as putative epigenetic effects. These associations provide insight into the genetic components of complex traits in coastal Douglas fir, as well as highlight the need for landscape genetic approaches to the detection of adaptive genetic diversity. A fundamental goal of molecular population and quantitative genetics is to discover polymorphisms that underlie adaptive phenotypic traits. Elucidation of the genetic components for ecologically relevant traits within natural populations has been slow, due mostly to the disconnect between organisms with detailed genomic resources and those that have phenotypes with ecological relevance (Stinchcombe and Hoekstra 2008). Rapid advances and applications of high-throughput marker technologies are beginning to amend this disconnect for forest trees. Several applications of association mapping approaches using functional marker data have been fruitful in identifying putatively causal single-nucleotide polymorphisms (SNPs) for an array of adaptive phenotypes across different forest tree species. The importance of these associations is clear, with putative applications ranging from marker-assisted breeding to gene conservation in the face of climate change (Walther et al. 2002; Aitken et al. 2008). Adaptation to cold is one of the greatest challenges to forest trees and is highly synchronized with environmental cues, primarily photoperiod and temperature (Saxe et al. 2001; Howe et al. 2003). The annual growth cycle of temperate forest trees involves a trade-off between the timing of initiation and cessation of growth that takes full advantage of favorable climatic conditions, while avoiding cold damage from late frosts in the spring and early frosts in the fall. Timing of bud flush is predominantly influenced by temperature following adequate chilling, while bud set is influenced by photoperiod (short days), as well as temperature, soil moisture, nutrition, and light quality (Sakai and Larcher 1987; Howe et al. 2003). The first stage of cold hardiness is also induced by short days, while low temperatures induce the second stage (Weiser 1970; Sakai and Larcher 1987). Here, we take an association genetic approach to the dissection of cold-hardiness related traits within natural populations of coastal Douglas fir [Pseudotsuga menziesii (Mirb.) Franco var. menziesii]. The range of this species extends from the Pacific Coast of North America to the eastern slope of the Rocky Mountains, with trees from the Pacific Coast classified as P. menziesii var. menziesii Supporting information is available online at http:/ www.genetics.org/ cgi/content/full/genetics.109.102350/DC1. Corresponding author: Department of Plant Sciences, Mail Stop 6, University of California, Davis, CA 95616. E-mail: [email protected] Genetics 182: 1289–1302 (August 2009) and those from the Rocky Mountains classified as P. menziesii var. glauca (Bessin.) Franco. The success of Douglas fir across this highly heterogeneous landscape is due largely to its ability to maximize growth during favorable climatic conditions, balanced with tolerance to low temperatures (Rehfeldt 1989; St. Clair et al. 2005; St. Clair 2006). Genetic variation for cold hardiness in coastal Douglas fir is well documented among geographic sources and among families within sources (Campbell and Sorensen 1973; White 1987; Loopstra and Adams 1989; Aitken and Adams 1996, 1997; O’Neill et al. 2001; St. Clair 2006). Most of these traits are also heritable, with h values ranging from 0.10 to 0.85. Population differences in cold adaptation across the range of Douglas fir are strongly influenced by geographic and climatic variables (Howe et al. 2003). Differences in cold season temperature and associated geographic variables (e.g., latitude, elevation, and distance from the ocean) are important selective forces driving local adaptation of populations (St. Clair et al. 2005). For example, population differentiation at quantitative traits (QST) related to fall cold hardiness is eightfold greater than differentiation at anonymous and presumably neutral markers (FST), suggesting the action of natural selection acting upon these traits (St. Clair 2006). The genes underlying cold hardiness, however, have remained elusive, despite numerous efforts to map quantitative trait loci (QTL) ( Jermstad et al. 2001a,b, 2003) and to analyze patterns of collocation between QTL and candidate genes (Wheeler et al. 2005). Expression studies support the hypothesis that similar types of genes to those identified in Arabidopsis are involved with cold adaptation in conifers (Guy et al. 1985; Thomashow 1999; Fowler and Thomashow 2002; Sekai et al. 2002; Lee et al. 2005; Yakovlev et al. 2006; Holliday et al. 2008). Population genetic investigations into patterns of diversity and divergence at candidate genes for cold adaptation, as well as a suite of other adaptive phenotypes, however, often find few loci consistent with the action of natural selection (Brown et al. 2004; Krutovsky and Neale 2005; GonzálezMartı́nez et al. 2006; Heuertz et al. 2006; Ingvarsson et al. 2006; Hall et al. 2007; Pyhäjärvi et al. 2007; Eveno et al. 2008) (reviewed by Neale 2007; Savolainen and Pyhäjärvi 2007; Neale and Ingvarsson 2008). Even the low power of the methods employed in these investigations (Zhai et al. 2009) and the theoretical expectations that selected loci may be unable to be detected using outlier approaches (Le Corre and Kremer 2003) are unlikely to account for the paucity of results. Larger sets of candidate genes are crucial, therefore, for the continued investigation and identification of major portions of the adaptive genetic diversity in forest trees. Similar patterns have been found in association genetic analyses, where only a small number of markers all of small effect are detected (Neale and Savolainen 2004; Thumma et al. 2005; González-Martı́nez et al. 2007, 2008; Ingvarsson et al. 2008) (reviewed by Neale 2007; Grattapaglia and Kirst 2008; Grattapaglia et al. 2009). Much of this work has focused on point mutations within coding regions, thus ignoring regulatory regions affecting gene expression. Seminal work has illuminated the possibility that many of the adaptive responses by forest trees to their environments, however, may stem from epigenetic effects ( Johnsen et al. 1996; Hänninen et al. 2001; Saxe et al. 2001; Johnsen et al. 2005a,b; Webber et al. 2005; Kvaalen and Johnsen 2008). The prevalence of such effects modifies the expectation of the quantity, type, and effect size of genes involved with adaptation by forest trees. The segregation of adaptive genetic diversity by coastal Douglas fir along environmental gradients is clearly established. Surveys of molecular diversity and divergence across 139 candidate genes have documented a set of those genes that deviate from the standard neutral model (Krutovsky and Neale 2005; Eckert et al. 2009b). These are prime candidates for the further dissection of cold-hardiness related traits using association mapping (cf. Wright and Gaut 2005). Here, we aim to bridge the gap between molecular population and quantitative genetics, using an association mapping approach. Our primary goal is to identify single-marker associations with 21 cold-hardiness traits. In doing so, we highlight the need for future investigations into landscape approaches to the description of adaptive genetic diversity, as well as studies of epigenetic effects in coastal Douglas fir. MATERIALS AND METHODS Association population and phenotypic data: Association population: The association population consisted of 700 of the 1338 unrelated families that were assessed in the genecology study of St. Clair et al. (2005). They represent an extensive rangewide sample covering 6.8 of latitude, 4.1 of longitude, and a diversity of environmental conditions (Figure 1; supporting information, Table S1). Wind-pollinated seed was collected from trees that originated from naturally regenerated stands throughout the range of Douglas fir in western Oregon and Washington. Twenty progeny were grown in raised nursery beds that were located in Corvallis, Oregon. Families were randomly assigned to five-tree row plots in each of the four raised beds, with each bed treated as a block. The term family is used to refer to source trees (i.e., mothers) because the phenotypic values we use are breeding values, and this is the terminology used in the original studies in which the phenotypes were measured (cf. St. Clair et al. 2005; St. Clair 2006). Phenotypes: Seedlings were grown for 2 years, during which they were measured for 21 traits related to cold injury, emergence, bud phenology, growth, and resource partitioning (Table 1). The data for cold-tolerance traits were obtained from St. Clair (2006). Emergence was determined following procedures described by Campbell and Sorensen (1979). Height and bud set were measured at the end of the first 1290 A. J. Eckert et al.