Introduction

Many coral reef ecosystems worldwide are threatened by both impacts of climate change and unprecedented rates of urbanization1,2. Coastal development has resulted in extensive transformation of natural shorelines through the installation of artificial structures, such as seawalls and other defences, to prevent coastal inundation3. As the intertidal surfaces of seawalls are typically species-poor, there is a growing interest in enhancing the ecological value of these structures via transplantation of reef-building corals4,5,6. However, the seawall environment in the intertidal zone is generally suboptimal for hard coral establishment7,8,9. In particular, the conditions during periods of emersion, with associated exposure to desiccation risk and high light levels and temperatures, can be especially detrimental for corals10,11.

Spatial structuring of hard coral communities is generally attributed to the surrounding environment in which presence or absence of species reflects their tolerance limits12,13. This is commonly exhibited as depth ranges, but also applies to salinity, wave energy, and sediment tolerances9,14,15,16. It is assumed that organisms living in extreme environments are more stress-resistant17,18. While previous reports have documented rich coral communities thriving in intertidal nearshore habitats8,9,14, few studies have attempted to determine how these intertidal corals survive such stressful conditions. In particular, Zhang et al.19 have shown that selective sweeps occur among intertidal coral populations in which gene ontology categories associated with regulation of apoptosis were significantly enriched.

Generally, intertidal species exhibit various massive morphological forms (e.g. massive, encrusting, foliose) but branching corals, including acroporids and pocilloporids, also exist in these areas9,15,20. In particular, the pocilloporid species Pocillopora acuta can inhabit a relatively wide range of habitats, including the upper subtidal and lower intertidal zones20,21,22,23. In Singapore, this species has been observed growing well in atypical coral habitats such as the base of seawalls and the sides of marina pontoons24. However, they exhibit among-genotype variation in physiological and molecular responses to changing conditions4,25,26,27. For instance, a previous experiment showed this species can withstand 2 h of emersion, but only certain genotypes were able to survive longer emersion periods4. This indicates substantial intraspecific variability in stress tolerance in P. acuta that is likely driven by genetic variation among genotypes.

To date, most studies on genome-wide DNA polymorphisms in corals aimed to determine population structure and genetic connectivity28,29,30,31 while the structural and functional effects of these variants are rarely examined. Nevertheless, some research has highlighted genomic signatures of adaptation in corals to challenging conditions [e.g. 32,33,34. In particular, Bay and Palumbi33 showed temperature adaptation in Acropora hyacinthus colonies living in a backreef pool that experiences highly-variable temperature. They found signals of selection on coding regions that interact with heat-shock proteins, as well as those that involved in antioxidant defense and apoptosis. Another study on Acropora corals by Cooke et al.32 showed selective sweeps for colonies found in inshore habitats of the Great Barrier Reef. The sweep loci are associated with osmotic regulation and skeletal development, which were linked to challenging water quality conditions (e.g. hyposaline) surrounding inshore coral populations.

Expression levels of genes are known to be influenced by many factors including the structural location and nature of DNA polymorphisms within the genome, which in turn can affect the associated functions that control various biological processes35. Generally, such inferences are obtained by identifying single nucleotide polymorphisms (SNPs), however, more recently, copy number variation (CNV) of putative genes has also emerged as an important genetic mechanism that can ultimately influence an organism's adaptive capacity to environmental change36. For example, some hypoxia-associated genes were identified to exhibit large copy variation among colonies of the massive Porites lutea, which is known to tolerate highstress environments such as the intertidal zone37. Although contributions of CNVs and SNPs to phenotypic variation are largely independent, they may act in concert with each other38,39. While knowledge about the effect of both types of variation co-existing in the same region is limited, it has been shown that co-existent CNVs can magnify gene-trait association power of SNPs40.

Studies aiming to identify genomic variations and their corresponding biological functions are fundamental to uncovering variants underlying trait differences among and within-coral species. Findings can also uncover genes that are potentially under selection which can suggest adaptation to certain environments41. Here, we hypothesize that (1) there is significant genomic variation, in the form of single nucleotide polymorphisms (SNPs), that differentiates intertidal and subtidal P. acuta colonies and, (2) identified regions harbouring SNPs also exhibit CNVs. Lastly, we aim to characterize the functional relevance of these genomic variations (i.e. SNPs and CNVs) in the context of the intertidal/subtidal environments they were sampled from.

Materials and methods

Sample collection

Coral nubbins (~ 2 cm size) from different colonies of Pocillopora acuta found on the intertidal seawalls at Pulau Hantu (n = 8), Kusu (n = 5) and Sentosa (n = 5) (Fig. 1) were collected in duplicate during low tide periods. Colonies of P. acuta were identified following Poquita-Du et al.42. Specifically, these corals were collected at the base of the seawalls and are submerged during high tide up to a depth of 2 to 3 m, depending on the location43. For P. acuta colonies in subtidal areas (~ 6 m depth), duplicate fragments were collected from different colonies at Kusu (n = 12) and another site: Raffles Lighthouse (n = 9) underwater using SCUBA as they are fully-submerged even during low tide. Collection sites and number of samples depended on the presence of P. acuta and colonies having approximately the same maximum diameter size (~ 20 cm).

Figure 1
figure 1

Map showing all sites within Southern Islands of Singapore where P. acuta colonies were collected from intertidal (Pulau Hantu, Kusu Island, Sentosa Island) and subtidal (Kusu, Raffles Lighthouse) areas.

Full size image

All 78 coral nubbins (samples hereafter) of P. acuta (39 colonies × 2 replicates) were immediately fixed in RNAlater and kept in an icebox while being transported to the laboratory. All samples were kept in a − 80 °C freezer until DNA extraction (conducted within a week of collection).

DNA isolation, whole genome sequencing and data quality check

Genomic DNA (gDNA) was extracted using Qiagen DNeasy Blood and Tissue DNA extraction kit following the manufacturer’s protocol with an extra step of RNA removal using RNase solution (Promega). The procedures for library preparation and sequencing followed Azenta Life Sciences (China). Briefly, the library preparation was carried out using ND607 VAHTS Universal Library Kit V3 following the manufacturer’s protocol. For each sample, 100 ng genomic DNA was randomly fragmented to < 500 bp by sonication (Covaris S220). The fragments were treated with End Prep Enzyme Mix for end repairing, 5’ phosphorylation and dA-tailing in one reaction, followed by a T-A ligation to add adaptor to both ends. Size selection of adaptor-ligated DNA was then performed using beads and fragments of ~ 470 bp (with approximate insert size of 350 bp) were recovered. Each sample was then amplified by polymerase chain reaction (PCR) using P5 and P7 primers, with both primers carrying sequences which can anneal with flow cell to perform bridge PCR and, P5/P7 primer carrying index allowing for multiplexing. The PCR products were subsequently purified using magnetic beads, validated using an Qsep 100 (Bioptic, Taiwan, China), and quantified by Qubit 3.0 Fluorometer (Invitrogen, Carlsbad, California, USA).

The libraries with different indices were multiplexed and paired-end (pe) sequencing was performed on the Illumina NovaSeq 6000 platform according to the manufacturer’s instructions (Illumina, San Diego, California, USA) to generate 150 bp pe reads (30 × coverage). Image analysis and base calling were conducted using the NovaSeq Control Software (NCS) + OLB _ GAPipeline-1.6 on the NovaSeq instrument. Adapter sequences, PCR primers, content of N bases more than 10% and bases of quality lower than 20 were removed with cutadapt (V1.9.1)44.

Sequence alignment and SNP detection

An annotated draft genome of P. acuta45 was used as the reference genome assembly in this study. While the P. acuta genome assembled by Vidal-Dupiol et al.45 was based on Illumina short-read sequencing, the quality is acceptable (BUSCO completeness of 89.4% and 91.4% based on Metazoa and Eukaryota datasets, respectively) and it was the only available P. acuta genome during the analysis. In support, the colony used to generate the P. acuta genome by Vidal-Dupiol et al.45was collected from the same region (i.e. Indonesia) not far from where our samples were collected (Singapore), therefore, it is an ideal reference genome for our study.

Sequence alignment was performed by individually mapping pe reads for all samples against the reference assembly using Burrowers-Wheeler Aligner (BWA, v 0.7.13) maximal exact match (mem) algorithm46,47. The reference assembly was indexed prior to mapping with sample sequences using the function ‘bwa index’. Unmapped and duplicate mapped reads were subsequently removed using SAMtools (v 1.10)48. Initial identification of genotype likelihoods at each genomic position, variant calling and subsequent normalization of variants were performed using bcftools v1.1149. The outputs were filtered to retain SNPs using Plink50. Further, SNPs that showed extended linkage disequilibrium (LD) in which a correlated (r2 = 0.5) pair of SNPs within a window of 50 kb were removed from the data set. In addition, markers that showed extensive deviation from Hardy–Weinberg equilibrium (HWE; p < 0.001), a missing rate of 1%, and low minor allele frequency (MAF) of 5% were excluded.

Examining population structure and admixture proportions

To examine the genetic divergence between P. acuta colonies from two tidal environments, the fixation index (FST) was estimated in Plink using the final output of SNPs generated above. Principal component analysis (PCA) was also performed in Plink and plotted in R51 to visualize the genetic structure of P. acuta colonies from different sites and tidal environment. Additionally, we assessed the proportions of ancestry for up to five potential K values, with computations of cross validation (CV) errors, using Admixture v1.3.052 and visualized using custom python scripts.

Association analysis and annotation of SNPs

Identification of specific regions harbouring SNPs that were significantly associated with intertidal vs. subtidal P. acuta colonies was performed using Plink53. Allele frequencies of quality-filtered SNPs were compared between intertidal and subtidal colonies using Fisher’s exact test to generate p values. SNPs with association p values of ≤ 0.01 were considered significantly associated with intertidal P. acuta colonies50,54,55. Results were visualized with a Manhattan plot using ggplot2 and qqman packages in R51,56,57.

Analysis for the distribution of all SNPs across genomic regions and structural annotation were performed using snpEff58. The output of the annotation run included translated protein sequences which were filtered to regions where SNPs were significantly associated with intertidal colonies using the sequence processing tool, ‘mothur’59. The output list of protein sequences obtained from the previous step were then aligned against the most updated Swissprot/Uniprot database using NCBI Blast (‘blastp’ command). Gene ontology (GO) enrichment analysis was subsequently performed using BiNGO (V 3.0.5; https://www.psb.ugent.be/cbd/papers/BiNGO/Home.html) available in Cytoscape (V 3.9.1, http://www.cytoscape.org/)60,61 by performing Hypergeometric test with Benjamin and Hochberg False Discovery Rate (FDR) multiple testing correction (significance level = 0.05) on the set of genes from the blastp output.

Orthology and copy number variation

Orthogroups were predicted to help determine whether genomic variations that differentiate intertidal and subtidal P. acuta colonies were also driven by CNVs. OrthoFinder v2.5.4, a tool that implements the phylogenetic orthology inference method62,63, was used to predict orthogroups. Sequences from protein-coding regions harbouring all variants, including SNPs, insertions, and deletions, for both tidal environments (i.e. intertidal and subtidal) were used for the orthology analysis (default settings). Orthogroups that exhibited CNVs, particularly those that were present in both environments, were identified and the corresponding protein sequences belonging to these orthogroups were retrieved for alignment process. Using the BLAST built-in tool (blastp) in CLC Genomics Workbench (V 10.0.1), protein sequences from the different orthogroups were aligned against the most recent version of the UniProt SwissProt protein database (accessed 14 April 2022). BLAST hits that showed the highest number of HSP (high-scoring segment pair) were used to search for corresponding gene IDs and putative functions on UniProtKB64.

Results

Sequencing and alignment

Illumina sequencing resulted in an average of 56.1 million pe raw reads per sample (Supplementary data). After removal of low quality and adapter reads, 55.5 million pe reads per sample were retained, with 38.76% GC content and Q phred scores of 97% Q20. For each sample, an average of ~ 96% of quality-filtered sequences mapped to the P. acuta draft genome45.

SNP detection and annotation

Overall, there were 29,619,409 variants obtained from the variant detection analysis, with average genotyping rate of approximately 99%. These variants consisted of 25,773,468 SNPs, 1,345,237 insertions, and 2,500,704 deletions. Following a series of stringent filtering to obtain independent and high-quality SNPs, only 74,391 SNPs were retained. These SNPs were detected in varying locations within 386 genomic regions, of which 87.31% were categorized as modifiers (Fig. 2a), which usually are variants affecting non-coding genes. The SNPs were found in different regions within the P. acuta genome (Fig. 2b), including ~ 29% within introns and only ~ 12% in functional regions (i.e. exons).

Figure 2
figure 2

Structural annotation showing percentage of SNP effects by impact categories (a) and distribution of SNPs in different genomic regions (b).

Full size image

Population structure and admixture proportions

Results from both PCA and Admixture (Figs. 3a and 4, respectively) showed no structuring based on site and tidal environments. FST values for all identified SNPs across various genomic positions (mean Global FST = 0.02) were generally low (Fig. 3b), indicating low genetic differentiation between P. acuta colonies found in the intertidal and subtidal environment.

Figure 3
figure 3

Genetic structure of Pocillopora acuta based on (a) Principal component analysis at the colony level and, (b) genetic divergence among colonies between the two tidal environments, showing estimates of FST for each identified SNP across various genomic positions.

Full size image
Figure 4
figure 4

Proportion of ancestry calculated using Admixture, showing up to five potential K values and the corresponding CV errors in brackets.

Full size image

Association analysis

Out of 386 regions that harbour SNPs, 134 of them were found to be significantly associated with intertidal colonies (p values < 0.01; Fig. 5). A complete inventory of all the regions with SNPs and their specific locations are provided in the Supplementary Data.

Figure 5
figure 5

Manhattan plot showing allele frequencies of quality-filtered SNPs from different chromosomes. SNPs that were highly associated with intertidal colonies (p values of < 0.01) are indicated by the red horizontal line.

Full size image

Gene ontology enrichment analysis for intertidal-associated SNPs

While the majority of detected SNPs highly associated with intertidal P. acuta colonies were observed in introns, some were found in exons and hence predicted to be functional regions. However, only a small portion (~ 28%) of these intertidal-associated SNPs could be annotated based on homologs from well-studied taxa which were used to search for GO terms under biological processes categories. Significant overrepresentation of GO was observed for genes that are associated with cellular homeostasis, metabolism, and signalling processes (Fig. 6; also refer to Supplementary data). Due to the interdependency of GO categories, whole branches for each GO hierarchy are shown in Fig. 6, but only coloured nodes that are farthest down the network were considered relevant (refer to60). In particular, the overrepresentation of categories—cellular homeostasis, metabolism, and signalling process—reflect the presence of genes associated with responses to glucose stimulus, fatty-acid beta-oxidation, and regulation of mitogen-activated protein kinase (MAPK) and nuclear factor-kappa B (NF-kappaB) cascades.

Figure 6
figure 6

Gene ontology enrichment analysis of genes significantly associated with intertidal P. acuta colonies, showing a gene network with significantly overrepresented genes (shaded nodes; cut-off p value 0.05) under the biological processes category.

Full size image

Orthology and copy number variation

OrthoFinder assigned 93.4% of the total number of genes (11,911 out of 12,755) into 2665 orthogroups, 570 of which were single copy orthogroups. Only genes that showed copy number differences > 2 between the two tidal environments are summarized in Table 1, with corresponding protein matches and putative functional profiles. The complete set of orthogroups that exhibit CNVs can be found in the Supplementary data.

Table 1 Summary of genes that exhibit copy number variations across two tidal environments, intertidal and subtidal, ordered by descending variance.
Full size table

Discussion

The environmental conditions surrounding intertidal areas are generally suboptimal for coral survival8,11, but several species have been observed to establish at the lower intertidal section of seawalls in Singapore including the common branching coral, Pocillopora acuta9,15. Here, we identified genomic regions harbouring SNPs that distinguish intertidal from subtidal P. acuta colonies. We also identified the presence of CNVs in all regions harbouring variants. The functional relevance of these regions, particularly coding regions, was further examined to gain insights into the possible involvement of these SNPs and CNVs in biological processes. Our results revealed that the genomic regions differentiating intertidal and subtidal colonies are mainly associated with cellular homeostasis, metabolism, and signalling processes, suggesting these are likely signatures of adaptation to extreme environments.

To account for population stratification and genetic ancestry, we examined the genetic structure and proportions of ancestry among P. acuta colonies studied here and found that there is no structuring based on site and tidal environment. While the five population divisions inferred when K = 5 had the lowest CV error, they do not coincide with the site or environmental groupings, implying that the limited structuring observed was not associated with these a priori subdivisions. Our observations align with findings in Afiq-Rosli et al.28, which also showed no clear among-site separation for P. acuta corals collected from reef sites within the Southern Islands of Singapore, including the collection sites in the present study [i.e., Hantu, Kusu, Satumu (Raffles Lighthouse)].

We found enrichment of genomic regions with SNPs that are linked to cellular homeostasis for intertidal P. acuta colonies. In particular, overrepresentation of responses to glucose stimulus was detected, likely due to corals relying on their endosymbionts for most of their daily energy budget through translocation of photosynthetic products (i.e. glucose)65. While photosynthesis is critical for corals to survive, it can contribute to the production of reactive oxygen species (ROS) which are known to be toxic for coral hosts and therefore need to be regulated to achieve homeostasis in the coral holobiont for efficient cellular functioning66,67. This is especially challenging for intertidal corals as they are exposed to high solar radiation and elevated temperatures for longer periods of time, which can destabilize the photosynthetic electron-transport chain of the endosymbiotic dinoflagellates, resulting in increased production of ROS67,68,69. Relatedly, there was a remarkably high number of gene copies coding for Protein O-mannosyl-transferase in the intertidal corals (Table 1). While the role of this protein is not known for corals, the function of O-mannosylation has been described for Saccharomyces cerevisae where it plays a role in solubilization of misfolded proteins during stressful conditions70. When there is an accumulation of misfolded proteins, the unfolded protein response (UPR) is triggered, however, prolonged UPR activation can cause production of ROS71. It is widely known that high levels of ROS can cause DNA damage which is detrimental to many organisms including corals67,72,73. Thus, high numbers of copies of genes coding for Protein O-mannosyl-transferase, which are protective against cell damage, may be beneficial for intertidal corals which are frequently exposed to extreme conditions.

Further, significant overrepresentation of genes associated with signalling pathways, MAPK and NF-kappaB were also detected in regions with SNPs associated with intertidal colonies. These pathways have been linked previously to oxidative stress in corals73,74. For instance, a significant activation of c-Jun N-terminal kinase (JNK, one of the major subgroups of MAPK gene family) has been shown to repress oxidative stress in Stylophora pistillata corals exposed to ultraviolet radiation (UVR) and thermal stress75. Similarly, genes associated with signal-transduction also displayed a relatively high number of copies for intertidal corals, particularly genes linked to Palmitoyltransferase specific to casein kinase 1 (AKR1) as well as calcium-dependent cell–cell adhesion via plasma (stan) and calcium ion export across the plasma membrane (Atp2b3). This suggests a highly coordinated signalling response to enable intertidal corals to adapt to extreme environments.

As with many other animals, metabolic activities in corals are known to be dynamic, responding to diel variations in environmental conditions such as light availability. For example, during daytime the coral’s diffusive boundary layer (DBL) can experience hyperoxia due to high photosynthetic activity76. Further, intense solar radiation can cause photoinhibition to endosymbionts, reducing their capacity to provide photosynthetically-fixed carbon to the coral host65. Our results show significant enrichment of metabolic processes associated with fatty acid beta-oxidation for intertidal corals. Fatty acids are important components of lipids in a wide range of living organisms, including corals77. Storage lipids play a crucial role in a coral’s metabolism, functioning as reserves to generate energy for survival during periods of adverse conditions78,79. A coral’s lipid stores can comprise up to ~ 40% of its total biomass, representing a substantial quantity of energy reserves80,81. The high occurrence of SNPs in metabolism-associated regions observed here implies a relatively high utilization of lipids for intertidal P. acuta corals, corroborating recent findings on metabolic plasticity of this species to withstand extreme environment82.

Overall, there is clear genomic variation within the species P. acuta living in two distinct tidal zones, suggesting population differentiation associated with these distinct environments. The genome-wide SNPs and CNVs reported here provide important insights into the biological functions (cellular homeostasis, metabolism, and signalling) that contribute to the capacity of P. acuta to withstand extreme conditions in the intertidal zone, but additional research on the expression of quantitative trait loci which involve direct association tests is required to confirm this. For instance, assessing lipid levels in colonies inhabiting two tidal environments can provide insights into the role of lipid metabolism in local acclimatization to different conditions. It is important to mention that this study is limited by the absence of P. acuta corals in the subtidal or intertidal area for some sites, thus, there is an unbalanced representation of the two tidal environments. Nevertheless, the genomic resources generated in this study can help facilitate discovery of functional variants associated with traits (e.g. stress tolerance) which can contribute towards active conservation efforts such as coral transplantation and selective breeding.