G2P studies focusing on quantitative traits have generally been successful in identifying associated loci

A cross-species comparison of environmental associations suggests some similarities in the genetic mechanisms involved in climatic tolerances across conifer genera. For each of four European conifer species in the Italian Alps, 6–18 SNPs were associated with precipitation/temperature PC axes . There was some overlap between species in the genes represented, including heat shock proteins, and cell wall construction and carbohydrate metabolism genes .Gene expression studies have identified a range of genes that may be involved in drought responses, but these results are not easily connected to the results of physiological or provenance response studies. First, RNA transcripts reflect the genes being expressed at a particular instant, whereas morphological or physiological traits are the result of processes acting over a longer time. Second, most gene expression studies do not examine differences between populations. Although some evidence suggests that stronger gene expression changes during stress are associated with greater growth or survival, different genotypes and demographic stages can show significant differences in gene expression changes . A few studies have begun to address this. Provenances of P. pinaster differed in the expression response of two dehydrin genes, as well as in physiology and mortality rates . Similarly, three genotypes of P. taeda differed in their gene expression responses to drought and re-watering . More such studies are needed, but care must be taken to distinguish between drivers of expression differences. For instance, a more drought-sensitive tree might express higher levels of dehydrins at a given drought stage because the leaf water potential has dropped faster than in a drought-resistant tree, whereas the resistant tree might express higher levels of dehydrins than the sensitive tree at a given leaf water potential.

Genome scan and G2E association studies can be useful tools in the search for genes responsible for local adaptation. Although such studies can identify loci at which allele frequencies differ between environments,nft system it is not always clear how these differences are connected to phenotypic differences, and thus what traits are under selection in a given environment. This is where QTL and G2P association studies are useful.Most conifer QTL studies have focused on wood traits, growth or yield. Of the two that have examined drought tolerance, the first identified four significant and four suggestive QTLs for d13C in P. pinaster, none of which co-located with QTLs for ring width . The second examined a wider range of traits – photosynthesis , chlorophyll fluorescence, gs, d13C, intrinsic WUE and specific leaf area – in F1 cross seedlings of P. pinaster when well watered or after 1 or 2 wk without water, and identified 28 significant and 27 suggestive QTLs . Locations of the QTLs for each trait varied by time point. Candidate genes within the QTLs were identified : those for gs and WUEi included stomatal regulation, ABA signaling and cell wall construction genes; those for d13C included an aquaporin; and those for chlorophyll fluorescence included transcription factors and a histone chaperone.However, only a few studies have investigated drought tolerance in conifers , with less success. All such studies used d13C as the focal trait. As we argue in Section VI, other traits would probably yield results that are more helpful for the understanding of drought responses. Gonzalez-Martinez et al. examined 41 candidate stress response genes of P. taeda, using 61 tree families planted at two sites. However, drought stress was probably mild, and they only identified one strongly associated gene and one weakly associated gene at each site. A later study on the same species examining 3938 SNPs identified seven new associations with d13C . Four of the associations were with unknown proteins, with only a transcription factor probably involved in the ABA-mediated stress response having an obvious connection to drought responses. G2P and G2E association studies complement one another, with the first identifying loci linked to targeted traits, but not whether these loci are under selection in nature, and the second doing the opposite.

The combination of these approaches is useful for the identification of genes and traits under selection in natural settings, but so far few studies have taken this approach. Eckert et al. tested the association of SNPs with five phenotypic traits and 11 environmental variables across 10 P. lambertiana populations around Lake Tahoe. This study identified six genes associated with phenotypic traits , and 31 associated with environmental PCs. Two genes were associated with both a trait and an environmental axis, including a glucose transport protein associated with d13C and environmental variables linked to water availability. A study focusing on multiple drought response traits and a larger number of SNPs might be able to identify more genes that have variants associated with both environmental gradients and drought tolerance traits. Some traits and processes involved in drought response have been better studied at the genetic level than others . Provenance studies have indicated that differences in stomatal control and shoot growth are often involved in local adaptation to drought, and all other study types have identified the genes likely to be involved . However, although root growth has also been identified as important by provenance studies, root-growth-related genes have not been identified. Conversely, although genes related to resistance traits, such as changes in carbohydrate metabolism, and protective and pathogen defense molecule production, have been identified in expression or association studies, these traits have been largely ignored in provenance studies. Finally, xylem traits, including refilling ability, have not been the focus of any genetic study type.Tree improvement programs that aim to increase growth potential and stress resistance face the challenges of long generation times, the need for large-scale field experiments and the late expression of traits such as wood density . Genomic selection, already routinely used in livestock breeding, has been proposed as a method of speeding up this process by using marker-predicted breeding values for phenotypes of interest . This approach is suitable for species with low LD and for traits with complex genetic architectures as it uses thousands of markers with effects that are estimated simultaneously . As with traditional phenotypic selection, accuracy is likely to be greatest when tests are carried out in environments similar to the target environment, because of the high likelihood of geno type 9 environment interactions .

Several recent studies have demonstrated the potential of genomic selection approaches for traits of interest to forestry. Resende et al. carried out an early evaluation of genomic selection in P. taeda, making use of clonally replicated individuals grown on four sites and genotyped at 4825 SNPs. They found that the accuracy of prediction models within sites ranged from 0.63 to 0.75 for diameter and height,hydroponic gutter and estimated that the breeding cycle could be speeded up by 50% with this method. Gamal El-Dien et al. used GBS to genotype over 1000 interior spruce trees over three sites that had been pheno typed for yield and wood attributes, and found that the incorpo ration of genomic information produced more accurate heritability estimates. Genomic estimated breeding values were most accurate when data from multiple sites were used to fit the model. Of even more relevance to selection for drought tolerance, Jaramillo-Correa et al. identified 18 SNPs associated with climatic PC axes in P. pinaster, and found that the frequency of locally advantageous alleles at these loci correlated with population level survival rates in a common garden at the hot/dry end of the species range. Together with the growth trait analyses, these results suggest that association techniques could be applied to predict breeding values for overall drought tolerance or particular drought tolerance traits even though only some of the loci involved have been identified. There is evidence of significant potential for selection approaches to improve drought responses in conifers. Provenance studies have shown evidence of genetic differentiation between populations in drought responses, and genome scan and G2E associations are finding evidence of natural selection on within-species genetic variation. Second, heritabilities for drought tolerance traits, when these have been examined, tend to be moderate to high. The calculation of heritability requires pedigree information: parent– offspring or sibling and half-sib comparisons. Narrow-sense heritability is the fraction of the variance in a trait attributable to additive genetic variation, as opposed to environmental and non additive genetic variation. Because heritability depends on both genetic variation in the population assessed and the degree of variation caused by the environment, estimates are not transferable between situations. In P. pinaster, estimates of d13C narrow-sense heritability ranged from 0.17 to 0.41, depending on how many individuals of what populations were assessed in what sites; and ring width and height growth rates were also moderately heritable . In the same species, heritability of P50 was 0.44, but this was driven more strongly by low levels of other sources of variation rather than high additive genetic variation . Across species, measured heritabilities for d13C range from the very high 0.7 for Araucaria cunninghamii to < 0.1 for P. taeda . Managers of wild forests are often focused on ensuring the resilience and function of the ecosystem rather than productivity. G2E and G2P association studies may help to identify seed sources that could be ‘preadapted’ to projected conditions for replanting in wild lands. However, wild trees face a range of challenges, including disease and competition, as well as drought . Stand structure and soil properties may also directly affect how trees experience drought stress. Studies that integrate stand level processes with genetic testing can further bridge gaps between genetic experiments and forest-scale management. Restoration projects could be used as experiments to test genomic predictions of survival and growth in a given environment, as well as the effects of genetic composition and diversity of the planted population on restoration success.

Common garden, gene expression and genetic association studies all have different strengths and weaknesses, and none alone will answer the question of how genetic differences affect drought tolerance . As described previously, a combination of different types of association study may help to identify loci that are under selection in the wild and the traits they influence. Similarly, gene expression studies could easily be combined with common garden studies of adults or seedlings to address whether differences in drought responses between populations or genotypes are a result of differences in gene sequences, gene expression patterns or both.Many studies to date have focused on WUE, often using d13C as a proxy. As discussed above, however, WUE is a ratio of changes in photosynthesis and transpiration, which can both vary, and higher WUE may or may not be associated with greater survival or growth in dry conditions. Moreover, different measures of WUE are not entirely consistent. We therefore recommend that future studies use survival and/or growth during and following drought as the metric of overall ‘drought tolerance’, and measure photosynthesis and water loss separately if these are processes of interest. The time involved in the measurement of traits for hundreds or thousands of individuals has encouraged the focus on easily measured d13C, but much progress has been made in high-throughput phenotyping techniques . For instance, thermal and long wave infrared sensors can measure leaf temperature or stomatal conductance, near and short-wave infrared sensors can measure leaf water content, and fluorescence sensors can measure chlorophyll content and photosystem efficiency .There are several traits and processes that have been suggested to be important for drought response by physiological studies, but about which there is little genetic information . Genetic studies frequently identify genes related to carbohydrate metabolism and transport as having altered expression or allele frequencies depending on water availability. It is difficult to make sense of these patterns because the link between these metabolic changes and tree function and survival during drought is still unclear. We also know relatively little about which species can refill cavitated xylem, under what circumstances and by what mechanisms. Thus, it is difficult to determine whether any genes identified by expression or G2E studies are involved in this process. Similarly, how roots and root growth respond to changes in water availability, and what genes are involved in these responses, remain poorly understood. Although the measurement of root architecture can be complex, high-throughput methods are being developed for this as well .Most experimental studies, including those looking at gene expression, have focused on seedlings.