This is the first study to use full mitogenomes to reconstruct relationships within the family Keratoisididae. It is also the first to compare phylogenies of Keratoisididae generated from complete mitochondrial genomes and conserved elements of the same individuals, or closely related individuals, which allowed us to test for congruence between the mitochondrial and nuclear genomes. Our study comprises nine of 12 of the keratoisidid groups defined by Watling et al. (2022) and generates the first complete mitogenomes for six of those groups. Furthermore, this study is the second to report unknown ORFs within the Order Scleralcyonacea, and the first to find them across multiple species and genera within the same family.
Phylogeny and mitonuclear discordance
Both phylogenies recovered a deep divergence in keratoisidids, referred to as Clade I and Clade II. This divergence has been reported previously in other conserved element based phylogenomic reconstruction (Morrissey et al. 2023a), but was not found in previous phylogenies that used a handful of exclusively mitochondrial markers (mtMuts-5', cox1 + igr1, 16s-nad2, and igr4, (Morrissey et al. 2022) or a combination of nuclear and mitochondrial gene regions (mtMuts-5' and 3' and partial 18s, Watling et al. 2022) probably due to a lack of phylogenetic signal. The topology of the conserved element phylogeny presented in this study is identical to Morrissey et al.’s (2023a) conserved element phylogenomic tree, despite the reduced taxon sampling herein (to match the mitochondrial genome sampling).
Group D2 was retuned as monophyletic in the mitochondrial phylogeny, with maximum support for the three specimens forming this clade. However, in the conserved elements reconstruction, one specimen within group D2, CE-21-039, was found as sister to the wider D1-D2 group. Only a fragment of this coral exists, but in-situ images show this specimen is a large sparsely branched colony similar to the appearance of Eknomisis dalioi. In addition, morphological examination of CE-21-039 identified characteristics of Eknomisis; sclerites arranged diagonally along the polyp body and internodal branching. The polyphyly of D2 was also seen in a larger conserved element phylogeny of the family (Morrissey et al. 2023a) with species of the genus Eknomisis, previously one of two clades that typify the diversity within group D2 (Watling et al., 2022), forming a distinct clade outside of a larger D1-D2-H1 clade.
By building a new conserved element phylogeny with an almost identical taxon set to the complete mitochondrial genome phylogeny, we provide evidence that differences are due to mitonuclear discordance and not insufficient or uneven taxon sampling. Mitonuclear discordance has been previously reported across deep to shallow divergences within Anthozoa (Quattrini et al. 2023b). This is the first time mitonuclear discordance has been reported within the family Keratoisididae.
Considerations for phylogenetic and barcoding studies using single or multiple genes
The sliding window analysis demonstrated that the three most variable regions within the keratoisidid mitogenome are the 5’ end of cob and the 3’ ends of both nad6 and mtMutS. The 5’ of mtMutS is the most commonly sequenced gene region for octocorals due to the gene itself being the most variable when comparing across Octocorallia (Muthye et al. 2022). However, studies within Keratoisididae have previously shown that other markers hold similar genetic variation for delimitation e.g., igr4 (van der Ham et al. 2009; Morrissey et al. 2023a) and 16S rRNA – nad2 (Morrissey et al. 2022). Our results suggest that the true diversity to be gained from single gene barcodes for Keratoisididae is not being represented by these commonly used gene regions. More research is warranted into family specific variability within the mitogenome as other families may have increased variability across less commonly sequenced regions of the mitogenome which may aid in species discover within a particular group after an initial broadscale assessment using mtMutS. Furthermore, when generating reference barcode libraries for use with eDNA, these more variable regions may better represent the diversity of the community present.
While next generation sequencing studies have become common place for resolving specific phylogenomic issues, it is still not financially feasible for exploratory studies. Single- and multi-gene phylogenies will continue to be used into the future and our results demonstrate that there may be more informative regions of the mitogenome that can be used to construct phylogenies, regions with higher variability than the commonly sequenced 5-mtMutS. However, the presence of discordance between mitochondrial and nuclear inferences provides evidence that concatenating sequences of nuclear and mitochondrial origin is inappropriate within Keratoisididae. Interfamilial studies using both nuclear and mitochondrial genomes have also found mitonuclear discordance at the familial rank within Anthozoa (Quattrini et al. 2023b). Therefore, until more studies investigate within familial relationships, caution should be applied when making phylogenetic inferences inside any anthozoan family. Where the purpose of a study is species discovery, then using multiple gene regions that maximise genetic diversity, regardless of origin, can effectively be used to discover and perhaps assign molecular operational taxonomic units to individuals.
Presence of Unknown ORFs
Three unknown ORFS within the mitogenomes of Keratoisididae. ORF1 was found in individuals belonging to Clade 1 (B1, C1, D1, D2, F1, I4, and J3), and ORF2 and ORF 3 were found in individuals assigned to I1 (Clade II). ORF2 and ORF 3 were much shorter than ORF 1, 84–289 bp (28–96 amino acids) and 243–510 bp (78–170 amino acids) respectively. There were no unknown ORFS in specimens assigned to A1 which only had a 10 bp intergenic region between cob – nad6. Unknown ORFs have only been reported twice within Octocorallia, in the mitogenome of Calicogorgia granulosa (Malacalcyonacea; Paramuriceidae) (Park et al. 2011) and in Paraminabea aldersladei (Scleralcyonacea; Coralliidae) (Brockman and McFadden 2012), although they have been found in species belonging to most hexacoral orders including Actiniaria (Emblem et al. 2014; Foox et al. 2016; Zhang et al. 2017), Antipatharia (Barrett et al. 2020), Scleractinia (Flot and Tillier 2007), and Zoantharia (Chi and Johansen 2017), suggesting they may be more common than currently assumed.
The arrangement of the genes in the keratoisidid mitochondrial genome conforms to Octocoral Gene Order B (Brugler and France 2008, except for S1 specimens), an arrangement also found in Anthoptilum sea pens (Hogan et al. 2019). This gene order is believed to have evolved independently in keratoisidids and Anthoptilum, and the large region between cob – nad6 is exclusive to bamboo corals, emphasising the difference between these genomes. As an exclusively deep-sea family, it could be reasonable to believe that the ORF in this region, if expressed, provides some functional role which has allowed keratoisidids to adapt to deep-sea conditions. To date, there have only been two studies that have focused on the characteristics of keratoisidid mitogenomes and a few studies that have used igr4 as a genetic marker for barcoding (van der Ham et al. 2009; Morrissey et al. 2022) and phylogenetics (Dueñas et al. 2014). Our study is the first to report on ORFs within this region. Previous studies were interested in the variability of the entire intergenic region, or in describing the gene order and thus could have overlooked them.
Utility of off-target reads and genome skimming for future phylogenomic studies.
Twelve keratoisidid mitogenomes for this study were assembled from off-target reads from target-capture data of Morrissey et al. (2023a). Target-capture data for anthozoans has been steadily increasing (Quattrini et al. 2019, 2020, 2022; Cowman et al. 2020; Untiedt et al. 2021; Erickson et al. 2021; McFadden et al. 2022; Quek et al. 2023; Morrissey et al. 2023a) since the initial conserved element bait set was designed (Quattrini et al. 2018). At the time of writing, off-target reads have been used to assemble over four hundred complete or partial mitogenomes from anthozoans (Muthye et al. 2022; Quattrini et al. 2023b, a; this study). The use of conserved elements is expected to continue as conserved elements have applications at every evolutionary scale, thus mining SRAs for additional genes and sequences of interest in the future will be a valuable tool for researchers.
In this study, 600–938 conserved elements were extracted from three specimens which only had genome skim libraries, 626–1640 conserved elements from 14 keratoisidid with just target-capture libraries, and 1535–1737 loci from seven individuals which had combined genome skim and target-capture libraries. As the cost of sequencing decreases, genome skimming is becoming a common method to extract full mitogenomes which have been used to increase our understanding of anthozoan phylogenomics (e.g., Hogan et al. 2019; Barrett et al. 2020; Ramos et al. 2023) and it has been found that similar or slightly greater numbers of conserved elements can be bioinformatically extracted from octocoral genome skim libraries as compared with target-capture libraries (Quattrini et al. 2023a). The utility of genome skim libraries can go beyond that of just phylogenomic studies, and may also be used to extract species barcodes for rapid biodiversity assessments as the cost is equivalent to approximately 6 or 7 loci sequenced by Sanger sequencing (Quattrini et al. 2023a), and also can be used to extract reference sequences of a wide range of loci that can be applied to eDNA studies (McCartin et al. 2023). Currently, many studies that use genome skimming to extract complete mitogenomes do not deposit the raw reads in an appropriate online repository which limits the potential of these data for future use. A shift towards depositing these reads, as is standard practice with target-capture data, will accelerate biodiversity and evolutionary studies across Anthozoa.