Hybrid zone populations exhibit adaptive potential to novel climatic conditions
Current patterns of genetic diversity in populations of forest trees are outcomes of various evolutionary processes structured by past events (Hampe & Petit, 2005; Mayol et al ., 2015), which often results in extant populations exhibiting adaptational lags to their current climate conditions. Although this has been documented for several tree taxa (Seliger et al ., 2021), it does not necessarily mean these populations are devoid of adaptive potential. Theory suggests that GEI should be prevalent in populations exhibiting adaptational lags and those occurring in heterogenous environments (Via & Lande, 1985; Ghalambor et al ., 2007). Sequential founder events, high levels of landscape fragmentation and the associated loss of genetic diversity, however, can restrict the evolution of GEI (Schmid et al ., 2019).
The populations sampled in this study naturally occur across the fragmented landscapes of the southwestern United States, where they experience markedly different seasonal and annual means and fluctuations in environmental conditions, as well as gene flow from a northern congener, P. flexilis (Menon et al ., 2018). By treating gene expression patterns and resulting co-expression modules as quantitative traits, we revealed strong signals of garden-specific adaptive trait differentiation (i.e., GEI) thus supporting our first and second hypotheses. In general, our estimates ofQ ST agree with previously published estimates from studies of forest trees (Lind et al ., 2018), and more specifically with the estimates obtained using expression datasets (Roberge et al ., 2007; Leder et al ., 2015). Although the presence of neutral population structure in the transcriptomic dataset used to construct the co-expression modules could generate false positives, eight of the strongly differentiated modules were also enriched for various QST categories (Fig. 4). We suggest these are likely true candidates for garden-specific signatures of adaptive evolution, given the overall weak population structure in this hybrid zone (Menon et al ., 2018).
The signals of local adaptation noted under the novel environmental conditions of our study suggest that populations of long-lived tree species, such as conifers, might not be limited in their ability to adapt to rapidly changing climatic conditions. Our results contrast with predictions of extensive future maladaptation suggested for other long-lived tree species such as oak (Browne et al ., 2019), poplar (Fitzpatrick et al ., 2020; Gougherty et al ., 2021) and spruce (Frank et al ., 2017). This contrast may relate to methodological differences among past studies and ours. Specifically, by using a space-for-time substitution study we allow for the architecture of adaptive evolution to reflect response to novel conditions. However, similarities between the space-for-time substitution design used in the present study and by Fitzpatrick et al . (2020) and Browneet al . (2019) suggest that our contrasting results might follow from the inclusion of hybrid populations, which matters because hybrid populations frequently show increased additive genetic variance (Reif et al ., 2007; Kulmuni, Wiley & Otto, 2023) (Table 2). Nevertheless, we advise caution in the interpretation that populations of long-lived tree species may be more adaptable to novel climates than expected because the presence of additive genetic variation underlying climatically relevant traits is only one of the important conditions needed for an adaptive response. Correlations between traits can be modulated by environmental conditions (Wood & Brodie, 2015) and correlations antagonistic to the direction of selection can impede adaptive responses (Walsh & Blows, 2009). While this does not seem to be the case in our study, as is evident throughQST enrichment at the core of the networks and strong co-expression module differentiation, we cannot conclude that evolution is occurring in the direction of the novel selective pressure. Such a conclusion would require a larger sample size and more thorough quantitative evaluation of the multivariate trait space.