Transcriptome wide profiling illustrates the prevalence of
pleiotropic architectures
Strongly correlated traits within biological networks often experience
selective constraints, as variants occurring in the genes underlying
these traits could cause deleterious pleiotropic effects (Jordanet al ., 2004). On the other hand, traits with lower connectivity
will often be involved in adaptive evolution and in GEI, as selection
can fine tune population responses to local environments without
disrupting key functional components (Cork & Purugganan, 2004; Josephset al ., 2017). Despite this, our work demonstrates an
overrepresentation of GEI transcripts at the core of networks and of the
modules, thereby leading us to reject our fourth hypothesis of adaptive
evolution being dominated by weakly connected peripheral traits.
Following Fisher’s geometric model (Fisher, 1930), if populations are
further from their fitness optima, variants that impart pleiotropic
influences are advantageous during the initial stages of the adaptive
walk (Orr, 1998). We suggest that this is likely the case for the
expanding hybrid zone populations (Menon et al ., 2020) that are
exposed to novel selective pressures. Under this scenario of demographic
and climatic mismatch, pleiotropy speeds up evolution to fitness optima.
However, our study uses a narrow definition of pleiotropy restricted to
the number of associated expression traits which could have a multitude
of different functions. While gene expression is widely treated as a
quantitative trait, its association with other genes within a
co-expression network may not always reflect pleiotropy and could be
confounded by the effect size of unaccounted for cis-variants (Josephset al ., 2017. However, considering the large genome size of
conifers, and that most variants in our study are located in non-genic
regions, this would be a challenging task to accomplish without further
improvement in genomic resources.