Solving the coral species delimitation conundrum

Distinguishing coral species is not only crucial for physiological, ecological and evolutionary studies, but also to enable effective management of threatened reef ecosystems. However, traditional hypotheses that delineate coral species based on morphological traits from the coral skeleton are frequently at odds with tree-based molecular approaches. Additionally, a dearth of species-level molecular markers has made species delimitation particularly challenging in species-rich coral genera, leading to the widespread assumption that inter-specific hybridization might be responsible for this apparent conundrum. Here, we used three lines of evidence – morphology, breeding trials and molecular approaches – to identify species boundaries in a group of ecologically important tabular Acropora corals. In contrast to previous studies, our morphological analyses yielded groups that were congruent with experimental crosses as well as with coalescent-based and allele sharing-based multilocus approaches to species delimitation. Our results suggest that species of the genus Acropora are reproductively isolated and independently evolving units that can be distinguished morphologically. These findings not only pave the way for a taxonomic revision of coral species but also outline an approach that can provide a solid basis to address species delimitation and provide conservation support to a wide variety of keystone organisms.

Methods
Dataset S1. Morphological Data. Color (No. 1) was assessed from photographs taken of each coral colony. Descriptive characters (2 – 16) were recorded from the overall observation of skeletal fragments. Morphometric characters (17 – 19) were measured directly from the branches using Vernier callipers. Corallite features (20 – 27) were obtained using a stereo microscope and an ocular graticule (except for 23 that was counted from above).

Dataset S2. Breeding Trials Data. In order to evaluate fertilization compatibility between the different morphospecies, approximately 100 washed eggs of each individual were added to each sperm dilution according to the breeding trial matrix (Fig. 2a). Briefly, crosses were performed with gametes from 6 colonies for a total of n=6 eggs only controls and n=36 crosses: 6 self-control, 6 within morphospecies, and 24 between morphospecies, with at least two replicates for each combination (see Dataset S2 - Breeding trials data). The numbers of regularly shaped embryos (prawn chip stage) and unfertilized eggs were counted under a stereomicroscope approximately ten hours after the breeding trials started. Mean fertilization success (%) was calculated as the average proportion of embryos divided by the number of embryos plus the remaining unfertilized eggs.

Dataset S3 (separate files). Chromatograms Data. Sequences were obtained by PCR-based amplification followed by Sanger sequencing. Chromatograms were edited and phasing was performed using two different complementary phasing approaches. Sequences of heterozygous individuals displaying alleles of the same length (without indel), were phased using SeqPHASE [step 1 and 2, https://eeg-ebe.github.io/SeqPHASE/]⁠ and PHASE v2.1.1. When length-variant heterozygotes were found in the dataset, Champuru v1.0 [https://eeg-ebe.github.io/Champuru/] was used to phase those sequences in a first step. Subsequently, they were inputted as “known haplotype pairs” during SeqPHASE’s step 1, thereby contributing to the phasing of the other individuals. Allele pairs with posterior probability ≥ 0.9 were chosen, except when more than one possible pair with similar posterior probabilities was found. In such cases, alleles shared with the highest number of individuals or that were connected with the most frequent haplotypes in the network were selected.