Results
Interpopulation differentiation
After filtering, 9,031 high quality SNPs remained from the larval samples collected at ambient temperature across 68 individually genotyped larvae from the five families (n = 7 – 20 for each). The five families separated in multidimensional space, in which genetic variation explained 43% of the variation between multilocus genotypes from the five families (PC1-PC3: 25.8; 13.4; 4.1%) (Figure 1A). Families that shared the same sire (Tijou 2A) roughly clustered together in Principle Components Analysis (PCoA) space (WW1 and CW with WW2), whereas the other two families were more divergent.
Analysis of population structure using Discriminant Analysis of Principle Components (DAPC) further confirmed this pattern of clustering between WW1 and CW along the first two discriminant functions. The 95% confidence ellipses demonstrate that the dispersion of multilocus genotypes also varied across the families, where WW2 shows the greatest and WW1 shows the least variance. Cluster membership probabilities calculated from DAPC analysis also showed flat distributions in between-group and within-group allele variances, demonstrating that admixture has occurred, especially between WW3, WC, WW1, and to a lesser extent with WW2, and particularly CW.
Allele frequency profiles
The majority of loci were fixed at either 0 or 1. The average distribution of loci in population purebred crosses was significantly different compared to interpopulation crosses, with a greater abundance of loci at intermediate frequencies in the interpopulational families (compared to null distributions; Kolmogorov-Smirnov Two-Sample test,p = 2.1e-26 and 1.3e-21, Figure 2). This suggests that these interpopulation hybrids have significantly greater genetic diversity than within population purebreds. When individual purebred families were analysed, the majority of the on average reduced diversity (e.g. lower abundance of loci at intermediate frequencies) was attributed to crosses WW1 and WW2, whereas WW3 demonstrated a similarly greater abundance of loci at intermediate frequencies as the interpopulational crosses (Supplementary Figure 1). WW1 and WW2 exhibited significantly different distributions compared to the distribution of WC (p = 5.5e-15 and 2.4e-4, respectively) and CW (p = 1.2e-11 and 6.7e-4, respectively) (Figure 2). In contrast, the distribution of loci in the WW3 cross was not significantly different compared to WC (p =8.7e-2) or CW (p = 0.77). All families differed significantly from expected, modelled HW frequencies when modelled again populations of the same size (p = WW and WC: 1.03e-45; CW: 2.2e-46). Observed heterozygosity (Ho), and alpha Diversity (0Dα and2Dα at q = 0 and 2) was highest in the interpopulation crosses (WC, CW) compared to the purebreds, with the exception of WW2, which was also high (Table 1). Alpha Entropy at q = 1 was the highest in WC, WW2 and lowest in WW3.
QTL analysis
SNPs were correlated with the five families reared under ambient temperature. The number of SNPs associated with these five families varied depending on threshold cut-offs ranging from 2,224 SNPs (75% quartile) to 87 SNPs at the stricter threshold value (99% quartile). Twenty-six of the 87 SNPs (30%) could not be assigned annotations within the A. digitifera genome, a common issue in coral and symbiont ‘omics (Figure 3 inset, sensu (Barshis et al., 2013; Peng et al., 2010; Shinzato et al., 2011; Sogin, Putnam, Anderson, & Gates, 2016)).
In the 87 SNPs that contributed to the separation of crosses at ambient temperatures, we identified proteins involved in a range of biological processes, mainly immunity and stress, cell functioning and metabolism, and calcification (Figure 3 inset). Stress related proteins included lysosomal-trafficking regulator-like proteins, CEPU-1-like protein, E3 ubiquitin-protein ligase RNF213-like, spondin, NFX1-type zinc finger-containing protein, NACHT, MAP/microtubule affinity and other processes, NLRC3-like proteins. Proteins involved with collagen production varied across families, as well as 2 dTDP-glucose 4,6-dehydratase-like and two sodium bicarbonate transporter-like proteins. A host of kinase related proteins, including those involved in cell functioning and transcription were detected.
PCoA loadings defined cut-offs for significant SNPs and hence assigned proteins. Homeodomain-interacting proteins, NFX1-type zinc finger-containing proteins and src proteins (Figure 3A), which are broadly associated with transcription, immunity and stress, and cell functioning were broadly positively related to the separation of families WW2/WW1/CW and WW3 and WC along PCoA axis 1 (Figure 1A). Alternatively, differences in cell functioning proteins were more negatively associated with their separation. Proteins contributing to the separation of WW2/WW3, WW1 and CW, and WC along PCoA2 were associated with responses to UV radiation, and MAP proteins (associated with cell functioning and immunity/stress, respectively) whereas multiple proteins associated with immunity and transcription were negatively associated with their separation.
Narrow-sense heritability
Population of origin (WW, WC, CW), family cross identity (WW1, WW2, WW3, WC, CW) and symbiont type significantly influenced juvenile responses to growth, bleaching and survivorship at 27.5°C and 31°C (full summary in Quigley et al., 2020). Specifically, juveniles with at least one parental coral sourced from the northern GBR survived significantly better, grew more, and bleached less at 31°C compared to other crosses, especially if infected with the Symbiodiniaceae symbiont, D. trenchii .
Physiological data for juvenile survival, growth, and bleaching were previously presented in Quigley et al., 2020. These raw data were used to calculate Bayesian narrow-sense heritability (h2) estimates for survival, bleaching, and growth demonstrate that only survival of corals infected withC. goreaui was strongly affected by temperature (Figure 4, Supplementary Table 1). Bleaching responses were generally associated with low heritability estimates at both temperatures, although juveniles associated with C. goreaui exhibited more moderate heritability estimates. Heritable genotypic variation across families contributed little to differences in growth rates at 27.5 and 31°C (Table 2). With the other three Symbiodiniaceae taxa, h2 estimates across the three traits varied in juveniles from the five families infected at 27.5 and 31°C (Supplementary Figure 1, Supplementary Table 1).