Drought and hybrid ancestry influence adaptive evolution and generate GEI
Results from our transcript categorisation (Fig. 2) and co-expression module enrichment analyses (Fig 4) support the second hypothesis of garden-specific trait differentiation. Despite the more arid conditions at the warm garden, the marginal effect of drought on population-level transcript differentiation was significant only at the cold garden. The strong impact of drought on population differentiation found here conforms with previous studies conducted in P. strobiformis (Goodrich et al ., 2016; Bucholz et al ., 2020; Menon et al ., 2021). While drought was an important selective pressure in both gardens, its effect may have been exaggerated at the cold garden because of freeze-related rupturing of cell wells during the previous winter season (Bachofen et al . 2015) leading to elevated susceptibility to drought stress. Given the overlap in molecular pathways leading to stress tolerance (Blödner et al. 2005) such carry-over effects or cross-susceptibility may be common across long-lived plants, although the opposite indicating higher tolerance to the second stressor has also been noted (Kong & Henry, 2019).
The absence of association with freezing stress could be due to global rather than local adaptation (Booker et al ., 2020) aided by expression of variants introgressed from P. flexilis , strong confounding between geography and freeze-related variables or compounded effects of drought and freeze stress in the cold garden. Ongoing work demonstrating higher post-winter survival in the cold garden when compared to the warm garden (Moler ERV. 2020) as well as a significant association between P. flexilis ancestry and survival in the cold garden provide some evidence for garden-specific effect of introgressed variants. At the network level, we noted very limited preservation across gardens using our joint threshold (Table S3; Fig. 4). The modules that were strongly preserved consisted of transcripts with low garden specific population differentiation and were generally related to metabolic processes that are likely essential for basic organismal functioning. Overall, this highlights strong and consistent patterns of GEI.
At the multivariate level, modules correlated with ancestry were often also strongly correlated with survival (Table S3; Table 2). Specifically, two QST enriched and weakly preserved modules – ME24 (warm garden) and ME13 (cold garden) – were strongly associated with P. flexilis ancestry, as well as with survival, making the traits and the loci underlying them a key candidate for further detailed studies of GEI (Fig. 4; Table 2). Furthermore, none of the transcripts encompassing these modules were shared across the two gardens. Since we sampled the transcriptomes of juvenile hybrid trees, it is possible that some of the strongly differentiated traits documented here are involved in post-zygotic isolating barriers (Lindtkeet al ., 2014; Zhao et al ., 2014). Overall associations between ancestry and expression levels of the threeQST categories, however, were not significant. There are at least two potential explanations for these results. First, ancestry estimates were obtained from a ddRADseq dataset, which although is representative of overall genomic ancestry, provides less coverage of genic regions in species with large genomes such as pines (Parchmanet al ., 2018). Second, ancestry was obtained for maternal trees and correlated with the maternal tree’s genetic values that were estimated using only the surviving seedlings. Given that postzygotic barriers are often expressed in seedling stage (Ogasawara & Okubo, 2009) resulting in high mortality, it is likely that our design could not fully capture the impact of ancestry on maternal tree’s genetic expression value as we only used the surviving seedlings without strong incompatibilities.
Interestingly, the proportion of P. flexilis ancestry was positively associated with survival across both gardens, but significant only at the cold garden. Survival and ancestry remained uncorrelated to population-level transcript abundances. These results thus provide only partial support for the third hypothesis of hybrid ancestry interacting with GEI and highlight the need for an in-depth evaluation mapping expression difference between parentals and hybrid trees in novel environments. Nevertheless, drawing from our previous work (Menonet al ., 2021) and the likely wider climatic tolerance of P. flexilis (Windmuller-Campione & Long, 2016), we speculate that the significant correlation between ancestry and survival only at the cold garden is indicative of P. flexilis -like variants enhancing survival in cooler environments, which aligns with the higher latitudinal distribution of the species range of P. flexilis thanP. strobiformis .