Maternal effects in coral fitness are reflected in gene expression patterns
Maternal effects in recruit survival and size previously reported forA. loripes x A. tenuis hybrid corals were consistent with their gene expression patterns. At the time when the corals were sampled for gene expression analyses, hybrid LT and its maternal purebred LL had higher survival compared to hybrid TL and its maternal purebred TT (i.e., 36-49 % versus 7-23 %) under both ambient and elevated conditions (Chan et al., 2018). Although the corals did not differ in size at seven months of age, maternal effects on size were evident by one year of age (Chan et al., 2018). In addition to the above study, maternal effects have been reported in other Indo-PacificAcropora hybrid corals obtained via laboratory crossing. These include effects on: 1) morphology of interspecific hybrids from anA. pulchra x A. millepora cross (Willis et al., 2006), 2) survival of interspecific hybrid larvae from an A. florida xA. intermedia cross (Isomura et al., 2013), and 3) thermal tolerance of intraspecific A. millepora hybrid larvae from a higher and lower latitude population. In contrast, paternal effects were found on morphology of natural interspecific hybrids of A. palmata and A. cervicornis from the Caribbean (Vollmer & Palumbi, 2002), and additive effects on survival (i.e., hybrid survival was intermediate between the parental offspring) were observed in experimentally produced intraspecific hybrids of A. milleporafrom a higher and lower latitude cross (van Oppen et al., 2014).
While a few studies have reported maternal effects on coral fitness and morphology, little is known about maternal effects on gene expression. In addition to the coral host, the host-associated microbiome can also have an impact on host gene expression (Barfield et al., 2018; Helmkampf et al., 2019). In our study, however, the bacterial and microalgal endosymbiont communities of the corals were similar at the time of sampling (Chan et al., 2019). The consistency between host gene expression and phenotypic results thus suggests that maternal host-related factors were likely the drivers behind the observed fitness differences. Almost 2000 differentially expressed genes (DEGs) were found when comparing hybrid LT and its maternal purebred LL to hybrid TL, and maternal effects were evident in these corals based on PCA, heatmap and volcano plots. While a statistical comparison cannot be made back to the parental purebred TT due to small sample size, gene expression of hybrid TL was similar to the only TT sample tested based on PCA and the heatmap and was indicative of maternal effects.
Only a few studies have reported maternal effects in gene expression. Videvall et al. (Videvall et al., 2016) showed that gene expression patterns were distinct between parental populations of 12-week-old seedling of the perennial herb Arabidopsis lyrata , and expression in intraspecific hybrids was frequently more similar to that of the maternal than paternal population. Only 15 DEGs were found between the hybrid produced in one direction and its maternal population, yet > 8800 DEGs were found when compared to its paternal population (Videvall et al., 2016). Interestingly, maternal effects were weaker in the hybrid cross of the other direction, with 334 and 661 DEGs observed when compared to its maternal and paternal population respectively (Videvall et al., 2016). Only one previous study has examined maternal effects on coral hybrid gene expression and only coral larvae were studied. Consistent with our findings, Dixon et al. (2015) showed that gene expression of intraspecific A. millepora hybrid larvae was similar to that of their maternal population. Up to 2,000 genes in hybrids were found to follow the expression patterns of the maternal population (Dixon et al., 2015). In both studies however, maternal effects were examined in early life stages only (i.e., 12-week-old seedling and 6-day-old larvae). Our results show that maternal effects can continue to influence gene expression of hybrid corals up to the age of at least seven months, indicating the potential long-term nature of maternal effects.
While differences in gene expression patterns were obvious between reciprocal hybrids as well as between hybrid TL and its paternal purebred, it was unclear what pathways and mechanisms were linked to these differences. Gene ontology (GO) analyses revealed underrepresentation of a very broad GO category, “cytosol”, in both pairs of comparison, suggesting that a wide range of genes and pathways were potentially involved in the observed holobiont fitness differences. In contrast, clear pathways involved in maternal effects were observed in the intraspecific A. millepora hybrid larvae (Dixon et al., 2015). Analyses of cellular component categories of tolerance-associated genes (i.e., genes for which expression levels prior to stress predicted the probability of larval survival under stress) showed enrichment of nuclear-encoded mitochondrial membrane components in hybrid coral larvae whose parents come from a warmer latitude (Dixon et al., 2015). The most upregulated GO categories were energy production and conversion, and encompassed mitochondrial proteins, suggesting mitochondrial protein variation in larvae may have contributed to maternal effects on thermal tolerance (Dixon et al., 2015). In our study, however, no mitochondrial-related pathways or genes were differentially expressed. The difference in GO associated patterns between these two studies may due to 1) the parental populations chosen for hybridization, and 2) the symbiotic/aposymbiotic nature of the corals. Parental populations of the same species from different latitudes were selected in Dixon et al. (2015), whereas parental populations of two different species from the same reef were chosen for this study. The differences in parental thermal regimes in Dixon et al. (2015) may lead to clearer maternal effects on thermal stress-related GO categories. Moreover, gene expression responses of aposymbiotic larvae (in Dixon et al., 2015) were likely different from coral recruits (in this study) that are associated with a high density of microalgal endosymbionts. Hence, the contrasting results of the two studies are unsurprising.