These studies do not automatically provide information about which, if any, environmental variables are responsible for the pattern. One can test whether patterns of differentiation match an environmental gradient, but this is necessarily a post-hoc interpretation . Association studies use a regression approach to identify loci in which genetic variation is associated with variation in trait values or home environment. Such analyses can be carried out at the individual or population level. Genotype-to-environment association studies identify loci that vary along environmental gradients . An association between an SNP and aridity, for example, suggests that the gene or its regulatory region affects performance in wet vs dry environments. This does not reveal how the locus affects phenotype, and careful interpretation is needed as a result of correlation between climatic variables. Genotype-to-phenotype association studies identify loci correlated with a particular phenotype , but the phenotype may or may not be relevant for fitness in the field. Most association studies in conifers to date have used SNPs in a limited number of candidate genes . This ensures that genes suspected of involvement are surveyed, but limits the ability to identify additional loci. However, with the decreasing cost of sequencing, approaches that generate large numbers of SNPs are increasingly being used for genome-wide association studies . One set of approaches, including RAD-seq and geno typing-by-sequencing, flood and drain tray involves the use of restriction enzymes to cut and sequence a small subset of the genome . This can produce tens of thousands of SNPs with high coverage .

Many of these SNPs will be in noncoding regions, which is good for the potential discovery of regulatory regions, but can limit the number of gene associations detected. Another approach involves the creation of a transcriptome or full genome sequence for a species, and the development of probes for all or most of the putative genes to identify SNPs . This approach can also yield useful gene expression data if multiple tissue types or treatments are included in the development of the transcriptome .Most drought gene expression studies in conifers have focused on pine seedlings, with a few investigating other Pinaceae genera . The direction of expression responses to the environment, including dry conditions, is highly conserved between Pinus contorta and Picea glauca 9 engelmannii, even though average expression levels often differ . It is unclear whether this is true across conifer families. No expression studies have focused on adult drought responses. The methods used to induce drought stress vary. Studies have withheld water for a specified period , until soil moisture reached a threshold or needles wilted , or needle water content declined to a certain level . Some have used chemically induced water stress . Caution must therefore be used in interpreting differences across studies, as these could be methodological artifacts . Genes related to signaling and gene transcription are frequently upregulated in drought-stressed seedlings. Changes in signal cascades must precede changes in their targets, and such expression shifts often occur within the first week of drought stress. Those in the ABA pathway are well represented . In addition to being involved instomatal closure, ABA signaling can affect shoot growth and water uptake . However, there are also ABA-independent pathways in most taxa, which may use leaf water potential as a signal .

Upreg ulation of genes in the ethylene pathway could be related to reduced shoot growth or leaf area . Genes related to protective molecules are also frequently upregulated . Late-embryogenesis-abundant proteins, named for their role in seeds, appear to stabilize proteins and membranes and prevent protein aggregation . Dehydrins, a subgroup of LEAs, often protect against drought stress, although some are induced by other abiotic stresses . Heat shock proteins, detoxification enzymes and genes in the synthesis and transport pathways of osmoprotective carbohydrates and proline may also be upregulated. Genes involved in pathogen or biotic stress defenses are often upregulated during drought stress, but those involved in growth, including cell division and wall construction, are often downregulated . Up or down changes in carbohydrate and lipid metabolism and protein handling pathways are also evident, although these are more difficult to interpret. Aquaporins, which affect membrane water permeability, were found to be upregulated in two studies . When drought-stressed seedlings are re-watered, most gene expression quickly returns to normal. In Pinus taeda, only 76 of the 2445 genes with altered expression during drought were still different after 48 h of recovery . Lorenz et al. found 11 genes upregulated in ‘recovered’ P. taeda seedlings relative to either drought-stressed or well-watered seedlings, including probable cell wall proteins, an aquaporin and a gene involved in vacuole function. These may reflect recovery or repair processes that occur once drought stress is removed.Gene families illustrate the complexity of expression responses to drought. Pinus pinaster has at least eight dehydrin genes, based on expressed sequence tag analyses . Three of five were downregulated during drought, whereas the other two were upregulated . Most dehydrin induction occurred after 20 d of drought , which may be why a similar but shorter study did not reveal the upregulation of dehydrins. Expression can also vary by tissue.

Of seven dehydrins examined in P. abies, drought stress upregulated four in needles, but only two in bark, with one being down regulated in bark . To investigate the link between drought and defense gene expression, Fossdal et al. exposed P. abies seedlings to a pathogen , drought stress or both, and examined the transcription of 14 candidate defense genes. Genes were upregulated more slowly in drought-stressed seedlings than in pathogen inoculated seedlings. The combined treatment led to more rapid and/or higher expression of many defense genes than either alone, which may be related to the synergistic mortality risks posed by biotic and abiotic stressors. Pleiotropic effects for some drought/ defense-related genes are also possible, but none have been identified to date.Multiple provenance studies have identified patterns consistent with local adaptation to drought. Trees from drier climates often exhibit conservative growth strategies , such as slower height or needle growth , less above ground biomass or a shorter growing season . Seedlings from dry environments often also exhibit more root growth and higher drought survival . Provenance trials of Pinus halepensis have shown mixed responses, with low growth and high water use efficiency in dry-sourced populations , but high growth in populations from intermediate-aridity areas , which may be related to growth plasticity. Because of the importance of carbon resources for plants, WUE – the ratio of carbon fixed to water lost – has long been considered to be closely tied to drought tolerance . Measures that integrate over longer time periods, such as the carbon isotope ratio d13C ,nft hydroponics are most frequently used to represent changes in WUE in trees. However, although different measures of WUE are often correlated , they are not interchangeable. For example, carbon discrimination is sensitive to chloroplast carbon concentrations and mesophyll conductance, whereas WUE itself is heavily influenced by evaporative demand, which does not directly affect D . In addition, nitrogen fertilization can increase WUE and decrease D, but does not affect gs or transpiration . Thus, WUE and D do not always co-vary, and caution is needed in the interpretation of d13C as a measure of WUE. Additional caution is warranted when using WUE as an indication of drought tolerance. High WUE may not be adaptive in some dry environments if the use of less water per unit carbon fixed does not result in slower depletion of soil water , or if plants with higher WUE grow faster and thus use more total water. Although a few studies have shown higher d13C for populations from dry sites , others have shown the opposite . There was no difference between three populations of Pinus ponderosa seedlings from varying climates in d13C or instantaneous WUE; the drought adapted populations exhibited greater plasticity in water use . In P. halepensis, however, individuals from more mesic sources showed higher plasticity of WUE than those from drier sources , but dry sources may show higher average WUE . Highly plastic growth and water usage reduce apparent WUE over the whole season compared with consistently moderate to low water usage . Instantaneous measures of WUE can change over a day, whereas integrated measures can differ significantly for a source population grown under different conditions or for the same tree across years . Changes in WUE may thus be a useful indication of drought stress, but, in conifers, radial growth and WUE are often weakly or negatively correlated . In pines, higher WUE usually results from reduced gs and/or reduced leaf area , which can limit photosynthesis and growth . Low gs can also result in higher tissue temperatures, which can be damaging, particularly in seedlings . Drought length and severity can influence measures of relative drought tolerance between populations. In P. ponderosa seedlings,the relative growth rate under moist conditions was positively correlated with previously measured tolerance to severe drought, whereas, under 4-wk drought, the intermediate-drought-tolerant population grew faster .

When Silim et al. examined Picea sitchensis, P. glauca and their hybrids, they found that P. sitchensis and the hybrids had the highest WUE and growth in well-watered conditions, but P. glauca and the hybrids had higher WUE and growth in drought conditions. Similarly, the relative transpiration and photosynthetic rates, WUE and growth of P. halepensis tree provenances differed between near-desert and Mediterranean planting sites . Such shifts in ranking are often a result of plasticity differences between populations. Provenances of P. pinaster from across the species’ range did not vary in cavitation resistance, suggesting uniform selection or lack of genetic variation . In P. halepensis, however, the percentage loss of conductivity differed significantly between provenances, but not between environments . Although plasticity has been observed in xylem wall thickening, time to thickening and number of cells in Picea mariana in drought experiments , cell anatomy studies often focus on only one population, so that the extent of local adaptation is unknown.Genome scans have identified loci in conifers that may be under differential selection across environments . Of 13 candidate genes for drought response in P. pinaster, two showed signs of divergent selection, although only one exhibited a pattern correlated with a climatic gradient; three, including two dehydrins, showed evidence of balancing selection . Prunier et al. examined SNPs from 313 candidate genes in P. mariana and found 16 that exhibited differentiation correlated with precipitation, including a LEA protein and genes in the ubiquitin protein handling pathway. However, differentiation between populations can be driven by processes unrelated to climatic gradients. Conifer populations are likely to violate the assumptions of such tests because they rarely form discrete isolated populations and are often far from demographic equilibrium; mis-specification of population hierarchical structure can lead to high false positive rates . However, newer methods are being developed that avoid frequently violated assumptions and reduce false positives . The approach preferred by recent studies is to directly assess the association of loci with environmental gradients , whilst controlling for population structure . Jaramillo Correa et al. examined the correlation of P. pinaster candidate gene SNP allele frequencies with climate principal component axes, using transcriptome-wide SNPs to control for population structure and demographic history. They identified 18 environmentally associated SNPs, many of which were in genes relating to carbohydrate transport, cell wall construction and photosynthesis. Two surveys of P. taeda examined associations between candidate gene SNPs and environmental gradients. One examined the association of these loci with five climatic PC axes , whereas the other used an aridity index for each county . There was some overlap in function between the loci identified . However, the studies disagreed on whether SNPs associated with climate also tended to be Fst outliers. G2E associations have been detected even over short geograph ical distances, suggesting that selection can drive local adaptation in the presence of high gene flow. Eckert et al. examined Pinus lambertiana populations around Lake Tahoe, an area of c. 35 9 65 km2 , and found 11 genes associated with environmental PCs reflecting differences in water availability. These included genes involved in carbohydrate metabolism and transport and response to biotic stress .