Aim: Genetic connectivity is a key component of species resilience to climate change in terms of recovery capacity following disturbance and capacity to disperse to novel locations as the climate warms and isotherms shift poleward. We aimed to strengthen our understanding of resilience in this context by characterizing patterns of connectivity and genetic diversity in a broadcast spawning coral across a tropical-temperate transition zone. We hypothesize genetic differentiation between tropical and temperate populations and decreasing genetic diversity with higher latitudes.

Location: Western Australia (WA).

Taxon: Turbinaria species complex. Turbinaria reniformis Oken, 1815 (Dendrophylliidae).

Methods: Samples from 930 target corals were collected from ten locations between 13 - 32 ^o latitude spanning a 9° C mean temperature range. In-situ species identification of T. reniformis is hindered by morphological plasticity and homoplasy with sister species. We combined Sanger sequencing of two mitochondrial DNA markers and high-throughput genotyping by sequencing (GBS) to isolate a single genetic Turbinaria lineage from our dataset through which patterns of genetic flow and diversity along the WA coastline could be explored using population- and individual-based analyses.

Results: Mitochondrial DNA sequence variation was low among Turbinaria samples and could not resolve individual species. Using GBS, we identified three genetically distinct lineages. Subsequent analyses within one of these lineages revealed strong spatial subdivision with 2-3 genetic clusters. While temperate populations were genetically diverged from more tropical sites, we did not observe declines in genetic diversity with latitude.

Main Conclusions: Temperate coral populations in Western Australia are genetically isolated from their tropical counterparts. Tropical populations of T. ‘reniformis’ exhibit adequate connectivity. Shark Bay represents the current southern limit of the tropical population of T. ‘reniformis’. Interestingly, temperate T. ‘reniformis’ in this study exhibit some genetic resilience due to their relatively high genetic diversity, yet the strong patterns of genetic subdivision for this widely dispersing coral species potentially limit their resilience to future climate scenarios.

In-situ species identification of corals is difficult. At a species level, existing studies based on COI and rDNA show most species are polyphyletic (Shimpi, Patel, & Haldar, 2019) and there is a high level of genetic overlap amongst morphospecies and particularly between T. mesenterina and T. reniformis (Arrigoni et al., 2014). With this in mind, we included three general morphotypes in our collections - Turbinaria reniformis, T. mesenterina, and T. frondens; however, samples with morphotypes matching the descriptions for T. conspicua, T. bifrons and T. stellulata listed in Veron & Pichon (1980) and Veron (2000) were also included for comparison. Tissue samples from a total of 930 colonies were collected on SCUBA at ten locations along the Western Australian coastline, spanning more than 18 degrees of latitude and a mean temperature range of 9°C (Figure 1). Samples of 2 cm² were collected using side cutters, at least 5 metres apart to avoid clone mates, but within 100m², and were preserved in 100% AR grade ethanol. In addition, eleven previously submitted Museum voucher specimens for Turbinaria were included in the study. Genomic DNA was extracted and purified using a Qiagen DNeasy Blood & Tissue kit (plate format) following the manufacturers protocol. Samples included 11 specimens previously accessioned at the WA Museum.

Genotyping-by-Sequencing

A subset of 536 samples were selected for reduced representation library sequencing. This subset of samples included representatives from each site and from varied morphotypes and COI haplotypes. In addition, ten Museum voucher specimens for Turbinaria that had suitable sequences for the COI analysis were included. Genome-wide single nucleotide polymorphism (SNP) data were generated at Diversity Arrays Technology (DArT; http://www.diversityarrays.com) using a GBS approach (DArTseq) with the PstI and HpaII restriction enzymes. Genomic DNA was processed in digestion/ligation reactions principally as per Kilian et al. (2012), but replacing a single PstI-compatible adaptor with the two different adaptors corresponding to different restriction enzyme overhangs. The PstI-compatible adapter included the Illumina flow cell attachment sequence, sequencing primer and barcode. The reverse adapter contained the flow cell attachment region and HpaII-compatible overhang sequence. Sequencing was carried out on two lanes of an Illumina Hiseq2500 and processed using proprietary DArT analytical pipelines (DArTsoft14).

SNP QC and Filtering

All filtering steps and statistical analyses were implemented using R software, version 3.0.1 unless noted (R Core Team, 2015). The initial genotype matrix of 546 individuals across 88,743 binary SNPs was filtered using ‘DArTR’ (Gruber, Unmack, Berry, & Georges, 2018). We removed loci that deviated from Hardy-Weinberg equilibrium (after adjusting for multiple comparisons at FDR 0.05), with minor allele frequencies less than 0.05, with call rates less than 0.90, and with read depth less than 10x. We retained only one SNP locus where there was more than one locus per sequence tag to avoid issues associated with linkage disequilibrium. We also removed any individuals with poor data (<90% call rate) and utilised a reproducibility statistic to filter out all loci with < 0.90 correct calls across individuals. Finally, to remove any possible symbiont data, we aligned all sequences to available symbiont genomes (Aranda et al., 2016; Lin et al., 2015; Liu et al., 2018; Shoguchi et al., 2013) and removed any loci with a blastn e-value below 10-3.

Initial screening

The initial SNP dataset of 546 individuals representing a range of different Turbinaria species, demonstrated by a number of genetically distinct lineages within the genus Turbinaria. To refine the broad clustering of our samples into genetically independent lineages, and to identify a single genetic lineage through which we can explore patterns of gene flow and population genetic structure, we used individual-based principle components analysis with “ade4” (Dray & Dufour, 2007) using Euclidean Distance matrices, and constructed a neighbour-Joining (NJ) tree based on Nei’s genetic distance in ‘poppr’ (Kamvar, Tabima, & Grünwald, 2014).

See the ReadMe file attached.