Discussion
In the GEA analysis, over half of the SNPs were associated with April 1st snowpack (PCK4). In this Mediterranean climate region, almost all of the annual precipitation occurs during the winter, and the melting of winter snow accumulation at high elevations feeds spring and summer streamflow (Serreze et al. 1999). Lack of snow can limit seedling establishment (Andrus et al. 2018). A “blanket” of snow can also insulate seedlings from extremely cold temperatures, but may also delay the start of their growing season (Ettinger & HilleRisLambers 2013; Renard et al. 2016). Consistent with this latter possibility, one of the associated SNPs was in a gene involved in light responses. Winter minimum temperature (TMIN), which has frequently been found to limit growth in tree-ring studies (Harvey et al. 2020), shows the next highest number of associations. The number of SNPs associated with more than one climatic variable was low (Fig. 2), which may indicate that we successfully selected semi-independent climatic variables that require different genetic adaptations. The highest overlap was between PCK4 and TMIN (64 SNPs) and between CWD and TMIN (17 SNPs). The former SNP set may be related to adaptation to cold and snow depth, while the latter SNP set may be related to how quickly the site warms up in spring, drying out the soil. A similar GEA we conducted for the co-occurring speciesPinus lambertiana also identified April snowpack as a key environmental variable that may have shaped local adaptation, and found low overlap in loci associated with different climate variables (Moranet al. In review).
In the GPA analysis, most SNPs associated with control phenotypic traits were linked with root-to-shoot ratio (R2S) and the number of abaxial stomatal rows (NR_AB). In contrast, most SNPs associated with phenotypic responses to drought were linked with shoot weight (SW), root length (RL), and R2S. Drought-stressed ponderosa pine seedlings allocated more to their root system, with longer root length, higher root-to-shoot dry mass ratio, less dry shoot mass, and less height growth. Other studies in pines have found similar patterns (Seiler & Johnson 1988; Irvine et al. 1998; Cregg & Zhang 2001; Taegeret al. 2015). This may indicate investment in greater water harvesting capacity at the cost of the overall low growth of aboveground structures – though low shoot growth can have the benefit of further reducing transpirational water loss (Moran et al. 2017b). We found that dry treatment root-to-shoot ratio was positively associated with survival in that treatment (Wu et al. 2023). Many of the SNPs associated with phenotypic drought responses were in genes associated with cell division & differentiation and with root growth, both of which make sense in light of the observed changes in allocation to root vs. shoot growth. The number of SNPs associated with more than one trait was low in both GPA analyses. The highest degree of overlap was in drought responsiveness of RL and R2S and of R2S and NR_AB (Fig. 6).
Non-synonymous (AKA missense) variants that may directly affect phenotype by changing protein form and function included 195 of the climate-associated, 93 of the control environment phenotype-associated, and 140 of the phenotype drought-response-associated SNPs (Tables 1, 2, & 3). Intragenic or synonymous variants are assumed to be neutral with respect to fitness but might be in linkage disequilibrium with a nearby causal variant. While linkage disequilibrium is usually low in conifers (Neale & Savolainen 2004), the GBS sequence fragments were relatively short (90-100 bp or less) and were trimmed further before SNP calling, so a linked non-synonymous variant could have been missed. We also found quite a few upstream and downstream SNPs in both GEA and GPA analysis that might directly affect gene expression or be linked to a protein-altering variant.
While we found no overlaps in specific SNPs between our GEA and GPA, identified several SNP-containing genes that were the same across the analyses (Tables 4 & 5). Most of these genes have been linked to stress responses in other studies. For example, gene wsc1 is involved in cell wall biosynthesis and gene PAT14 involved in leaf senescence, both in response to stress (Zu et al. 2001; Maddi et al. 2012; Laiet al. 2015; Zeng et al. 2018). Moreover, there was substantial overlap in functional categories found to be directly related to drought tolerance or other environmental responses in previous studies (Fig. 3, 4, 5). The prevalence of genetic associations related to abscisic acid (ABA)-signaling pathways and ubiquitination in GEA and GPA analyses is consistent with prior observations (Moranet al. 2017b) and with results of the P. lambertiana analysis (Moran et al. In review). Increasing ABA concentrations are used as a signal to keep stomata closed during dry conditions, reducing water loss (Brodribb et al. 2014). In addition, ABA signaling can also affect shoot growth and water uptake (Buckley 2005; Hamanishi & Campbell 2011). Ubiquitination is involved in drought responses in model species by playing a role in ABA-mediated dehydration stress responses (Ryu et al. 2010; Kim et al. 2012) or through the downregulation of plasma membrane aquaporin levels (Leeet al. 2009). Understanding of the role of ubiquitin in conifer drought response is still somewhat limited. A study in black spruce (Picea mariana ) identified 16 candidate genes correlated with precipitation, including the genes in the ubiquitin protein handling pathway (Prunier et al. 2011). The association between ubiquitin protein and roots and stomatal density may indicate previously unidentified roles in drought response.
Moreover, genes associated with seeds and seed dormancy can also be directly involved in drought tolerance; for instance, dehydrins can protect proteins from desiccation in both seeds and other plant tissues (Moran et al. 2017b). However, reproduction-related genes might also show associations with environmental gradients if they are involved in reproductive timing. Genes involved in xylem & phloem differentiation or cell wall formation could shape the hydraulic safety of water-transporting cells, which can be quite plastic in pines (Lauder et al. 2019). Other than these functions directly related to drought tolerance or different environmental responses, the other overlapping functions among GEA and GPA analysis are involved in gene expression (RNA or DNA binding, transcription factors, helicase activity, ribosome components, methylation) or ATP binding (motifs found in membrane transporters, microtubule subunits, enzymes, and other cell components that require energy). Our findings suggest the efficiency of combining GEA and GPA analysis with GBS to uncover potentially important adaptive genetic variation.
In conclusion, by investigating adaptive genetic variation in ponderosa pine with GEA and GPA association analysis, our study found thousands of genomic variants associated with response to climate and physiological traits. Some of these have previously-identified functions associated with drought responses, but for others, the gene function – or how that function is relevant for environmental responses – is still unknown. Molecular tools based on the associated genetic markers could be developed to assist breeders and land managers speed up selection for drought tolerance or selecting appropriate seed sources for a changing climate. In addition, our results should open new opportunities for functional studies to determine the molecular roles of the genes underlying these associated genetic makers in influencing trees’ adaptation.
The two environmental variables with the most genetic associations – snowpack and winter temperatures – are among those that have already undergone significant shifts in recent decades, with further substantial shifts being projected due to anthropogenic climate change (Rapacciuoloet al. 2014; Fyfe et al. 2017). This suggests that tree populations in the Western US will be under rapidly shifting selective pressures, making exploring the potential of genomic selection for seed selection of pressing concern. We found considerable heritable variation in drought-responsive traits (Wu et al. 2023), suggesting adaptive potential exists if change is not too rapid. We are also following up on this study by testing the ability of the SNP associations detected here to predict performance in post-fire restoration plantings.