An integrated morpho-molecular approach to delineate boundaries of Millepora from the Red Sea

Item Type Article

Authors Arrigoni, Roberto; Maggioni, Davide; Montano, Simone; Hoeksema, Bert W.; Seveso, Davide; Shlesinger, Tom; Terraneo, Tullia Isotta; Tietbohl, Matthew; Berumen, Michael L.

Citation Arrigoni R, Maggioni D, Montano S, Hoeksema BW, Seveso D, et al. (2018) An integrated morpho-molecular approach to delineate species boundaries of Millepora from the Red Sea. Coral Reefs. Available: http://dx.doi.org/10.1007/s00338-018-01739-8.

Eprint version Post-print

DOI 10.1007/s00338-018-01739-8

Publisher Springer Nature

Journal Coral Reefs

Rights Archived with thanks to Coral Reefs

Download date 05/10/2021 19:48:14

Link to Item http://hdl.handle.net/10754/629424 Manuscript Click here to access/download;Manuscript;CORE Arrigoni et al Millepora.docx Click here to view linked References 1 2 3 4 1 An integrated morpho-molecular approach to delineate species boundaries of Millepora from the Red Sea 5 6 2 7 8 3 Roberto Arrigoni1, Davide Maggioni2,3, Simone Montano2,3, Bert W. Hoeksema4, Davide Seveso2,3, Tom Shlesinger5, 9 10 4 Tullia Isotta Terraneo1,6, Matthew D. Tietbohl1, Michael L. Berumen1 11 12 5 13 14 6 Corresponding author: Roberto Arrigoni, [email protected] 15 16 7 1Red Sea Research Center, Division of Biological and Environmental Science and Engineering, King Abdullah 17 18 8 University of Science and Technology, Thuwal 23955-6900, Saudi Arabia 19 20 9 2Dipartimento di Scienze dell’Ambiente e del Territorio (DISAT), Università degli Studi di Milano-Bicocca, Piazza 21 22 10 della Scienza 1, Milano 20126, Italy 23 24 11 3Marine Research and High Education (MaRHE) Center, Faafu Magoodhoo 12030, Republic of the Maldives 25 26 12 4Taxonomy and Systematics Group, Naturalis Biodiversity Center, P.O. Box 9517, 2300, RA Leiden, the Netherlands 27 28 5School of Zoology, The George S. Wise Faculty of Life Sciences, Tel-Aviv University, Tel-Aviv 69978, Israel 29 13 30 6College of Marine and Environmental Science, James Cook University, Townsville, QLD 4811, Australia 31 14 32 33 15 34 35 16 Keywords Fire corals ∙ Phylogeny ∙ Pore ∙ Nematocyst ∙ Eumedusoid 36 37 17 38 39 18 Abstract Fire corals of the hydrocoral Millepora provide an important ecological role as framework 40 41 19 builders of coral reefs in the Indo-Pacific and the Atlantic. Recent works have demonstrated the incongruence between 42 43 20 molecular data and the traditional of Millepora spp. based on overall skeleton growth form and pores. In an 44 45 21 attempt to establish a reliable and standardized approach for defining species boundaries in Millepora, we focused on 46 47 22 those from the Red Sea. In this region, three species are currently recognized the fan-shaped branching M. dichotoma, 48 49 23 the blade-like M. platyphylla, and the massive/encrusting M. exaesa. A total of 412 colonies were collected from six 50 51 24 localities. Two mitochondrial marker genes (COI and 16S rDNA) were sequenced to obtain phylogeny reconstructions 52 53 25 and haplotype networks. Eight morphological traits of pores and the nematocysts of both and eumedusoid stages 54 55 26 were measured to determine if significant morphological differences occur among the three species. Both markers 56 57 27 clearly resolved M. dichotoma, M. platyphylla, and M. exaesa as distinct, monophyletic lineages in the Red Sea. 58 59 28 Nevertheless, they also revealed deep genetic breaks with Southwestern Indian Ocean populations of the three species. 60 61 62 1 63 64 65 1 2 3 4 29 In the Red Sea, the three species were further distinguished based on their pore and nematocyst features. A 5 6 30 discriminant analyses revealed dactylopore density, number of dactylopores per gastropore, dactylopore distance, and 7 8 31 gastropore diameter as the most informative discriminative characters. The heteronemes, the large and small stenoteles 9 10 32 of polyps, and the distribution of mastigophores of eumedusoids also showed significantly interspecific differences. 11 12 33 An integrated morpho-molecular approach proved to be decisive in defining species boundaries of Millepora 13 14 34 supported by a combination of pore and nematocyst characters which may be phylogenetically informative. 15 16 35 17 18 36 Introduction 19 20 37 Species of the hydrocoral genus Millepora are popularly known as fire corals (Pourtales 1877). They are 21 22 38 colonial organisms building persistent calcareous skeletons and as such playing an important ecological role as 23 24 39 framework builders of coral reefs (Lewis 2006). These hydrocorals are among the most relevant reef builders in 25 26 40 shallow-water tropical seas, second only to scleractinians (Lewis 1989; Edmunds 1999; Smith et al. 2014). Millepora 27 28 occurs in tropical and sub-tropical coral reefs of both the Atlantic and the Indo-Pacific and its depth distribution is 29 41 30 restricted from less than 1 m to about 50 m deep because of the obligate symbiosis with zooxanthellae of the genus 31 42 32 33 43 Symbiodinium (Boschma 1948, 1956; Cairns et al. 1999). 34 35 44 Despite the wide geographical distribution and the relevant ecological role, Millepora has scarcely been 36 37 45 investigated in coral reef studies and, in particular, its taxonomy and systematics are controversial and much debated 38 39 46 (Hickson 1898, 1899; Crossland 1948; Boschma 1948a, 1966; Moshchenko 1992, 1997; Razak and Hoeksema 2003; 40 41 47 de Souza et al. 2017). Fire corals have been historically identified based on morphological features such as colony 42 43 48 growth form, skeletal pores, and ampullae (Klunzinger 1879; Quelch 1884; Hickson 1898; Boschma 1948b, 1949, 44 45 49 1956). Nevertheless, all these morphological characters are intraspecifically variable and even in a single specimen as 46 47 50 a result of ecophenotypic variation (de Weerdt et al. 1981, 1984; Boschma 1948a; Moshchenko 1994, 1995a, 1996a; 48 49 51 Amaral et al. 1997; Dubé et al. 2017a, b). Although growth form has been traditionally used as the main diagnostic 50 51 52 character for species identification (Forskål 1775; Ehrenberg 1834; Duchassaing and Michelotti 1860; Klunzinger 52 53 53 1879; Hickson 1898; Crossland 1941; Boschma 1948a, b), this feature has been shown to be highly variable. For 54 55 54 example, Millepora platyphylla Hemprich and Ehrenberg, 1834 in Vietnam displayed distinct morphotypes associated 56 57 55 to different reef zones and the occurrence of different colony growth forms was hypothesized to be related to factors 58 59 56 such as radiation availability, water movement, and settling suspended matter (Moshchenko 1995a). A complicating 60 61 62 2 63 64 65 1 2 3 4 57 factor herein is that juvenile Millepora corals usually start with an encrusting growth form (see e.g. Fig. 2g, Hoeksema 5 6 58 et al. 201: Fig. 1), which makes it difficult to distinguish species at early stage. Similarly, pore characters are known 7 8 59 to noticeably vary in response to the rapidity of growth and the amount of light on the skeleton surface and some 9 10 60 taxonomists have suggested to discard their use in Millepora taxonomy (Hickson 1898, 1899; Boschma 1948a, b). 11 12 61 However, Moshchenko (1994, 1995b, 1996b) proposed a quantitative approach for the analysis of pore structures and 13 14 62 was able to tell M. platyphylla apart from the branching Millepora species of the Indian Ocean. As a consequence of 15 16 63 this morphological variation, traditional morphology-based species identification since the first valid description of a 17 18 64 by Linnaeus (1758) generated more than 50 nominal species of Millepora, nine of which are now considered 19 20 65 valid in the Indo-Pacific and six in the Atlantic (Razak and Hoeksema 2003; Amaral et al. 2008; de Souza et al. 2017). 21 22 66 Recent outcomes of molecular phylogenetic analyses and DNA taxonomy have provided advantages for the 23 24 67 understanding of Millepora systematics and connectivity (López et al. 2015; Hoeksema et al. 2014; de Souza et al. 25 26 68 2017; Dubé et al. 2017c). The use of genetics has elucidated the identification and the geographic origin of fire corals 27 28 in some isolated and remote islands of the central and eastern Atlantic Ocean, including for example the Canaries, 29 69 30 Cape Verde Islands, and Ascension Island (López et al. 2015; Hoeksema et al. 2017). Moreover, the three endemic 31 70 32 33 71 species of the Brazilian province, i.e. Millepora braziliensis Veriill, 1868, Millepora nitida Veriill, 1868, and 34 35 72 Millepora laboreli Amaral, 2008, were successfully resolved based on molecular data although the traditional 36 37 73 morphological characters of pores were not sufficient to discriminate these taxa (de Souza et al. 2017). Nevertheless, 38 39 74 other studies have shown discordances between morphology-based taxonomy and genetics suggesting the need to 40 41 75 combine new molecular and morphological data for the definition of species boundaries (Meroz-Fine et al. 2003; 42 43 76 Ruiz-Ramos et al. 2014; Takama et al. 2018). For example, Ruiz-Ramos et al. (2014) genetically characterized four 44 45 77 Caribbean Millepora representatives, i.e. Linnaeus, 1758, Millepora complanata Lamarck, 1816, 46 47 78 Millepora squarrosa Lamarck, 1816, and Millepora striata Duchassaing and Michelotti, 1864, and recovered the latter 48 49 79 two species as distinct molecular lineages while the otherss form an unresolved species complex. Furthermore, a recent 50 51 80 study from Japan demonstrated that the three branching species Millepora dichotoma Forskål, 1775, Millepora 52 53 81 intricata Milne Edwards, 1860, and Boschma, 1949 were receovered in a single molecular clade 54 55 82 while a previously unknown lineage of an encrusting Millepora form was discovered, which overgrows living 56 57 83 scleractinian corals (Takama et al. 2018). 58 59 60 61 62 3 63 64 65 1 2 3 4 84 Fire corals are abundant and common throughout the entire Red Sea, where they represent an ecologically 5 6 85 important group in shallow-water reef habitats. Three species are supposed to occur in the Red Sea (Boschma 1948), 7 8 86 i.e. Millepora dichotoma, Millepora platyphylla, and Millepora exaesa Forskål, 1775. The reef crest of the Red Sea 9 10 87 reef systems is usually dominated by conspicuous colonies of M. dichotoma and M. platyphylla in addition to 11 12 88 scleractinians (Loya 1972, 1976; Benayahu and Loya 1977; Perkol-Finkel and Benayahu 2004). Consequently, Loya 13 14 89 and Slobodkin (1971) introduced the concept of “Millepora zone” to identify the reef portion at depths of 0.2 to 3 m. 15 16 90 Because of this abundance, early taxonomists of the Red Sea coral fauna focused their attention on fire corals and 17 18 91 provided detailed descriptions of their morphological characters. Notably, the three aforementioned species were 19 20 92 described from the Red Sea (Forskål 1775; Ehrenberg 1834; Klunzinger 1879; Crossland 1941). The morphological 21 22 93 variation and the colony development of both M. dichotoma and M. exaesa from the northern tip of the Red Sea (the 23 24 94 Gulf of Eilat and Aqaba) were examined and described by Vago et al. (1997) and Meroz-Fine et al. (2003). Although 25 26 95 these authors considered M. exaesa as an ecomorph of M. dichotoma, they found significant differences in molecular 27 28 data (rDNA internal transcribed spacers), nematocyst capsule size, and colony growth form. Hence, theysuggested 29 96 30 that the two morphs represent two distinct entities (Meroz-Fine et al. 2003). Nevertheless, these works did not include 31 97 32 33 98 M. platyphylla and evaluated fire corals only from the northern Gulf of Eilat, an isolated part of the Red Sea 34 35 99 (Klausewitz 1989; Wolf-Vecht et al. 1992; DiBattista et al. 2016a, b). 36 37 100 With the aim to establish a reliable and standardized approach for species boundaries definition in Millepora, 38 39 101 this study uses a combination of molecular and morphological datasets in to address the degree of concordance 40 41 102 between genetic and morphological variation of fire corals from the Red Sea. Two mitochondrial genes and two 42 43 103 distinct sets of morphological characters were investigated and their efficacy to detect species boundaries was 44 45 104 discussed in light of our results. 46 47 105 48 49 106 Materials and methods 50 51 107 Sample collection 52 53 108 A total of 412 specimens of Millepora were collected while SCUBA diving between 1 and 20 m depth at six 54 55 109 localities along the Red Sea (Fig. 1), consisting of 152 colonies of M. dichotoma, 128 of M. platyphylla, and 132 of 56 57 110 M. exaesa (Table S1). Specimens were identified to species level following original descriptions (Forskål 1775, 58 59 111 Ehrenberg 1834) and the taxonomic revision of Indonesian Millepora by Razak and Hoeksema (2003). The two main 60 61 62 4 63 64 65 1 2 3 4 112 diagnostic morphological characters used for identification were the colony growth form and the cyclosystem 5 6 113 arrangement. During the sampling process, we included all possible morphotypes per species (Fig. 2). To minimize 7 8 114 the likelihood of sampling clones, colonies ascribed to the same species were collected at a minimum distance of 5 m 9 10 115 (following Ruiz-Ramos et al. 2014). Each coral specimen was photographed underwater (Fig. 2), collected with 11 12 116 hammer and chisel, tagged, and approximately 2 cm3 of the collected colony was sampled and fixed in 95% ethanol. 13 14 117 The remaining material was immersed in sodium hypochlorite for 48 hours to remove all tissues, rinsed in freshwater, 15 16 118 and air-dried for identification and microscope observations. Specimens were deposited at King Abdullah University 17 18 119 of Science and Technology (KAUST, Thuwal, Saudi Arabia) and at Steinhardt Museum of Natural History, Tel Aviv 19 20 120 University (Tel Aviv, Israel). 21 22 121 23 24 122 DNA extraction, amplification and sequencing 25 26 123 DNA extractions were carried out with the DNeasy Blood and Tissue kit (Qiagen Inc., Hilden, Germany). A 27 28 portion of two mitochondrial genes, the cytochrome c oxidase subunit I (COI) and the large rRNA subunit (16S 29 124 30 rDNA), were obtained from all the collected samples. PCR amplifications of COI were performed using the previously 31 125 32 33 126 published primers COIF and COIR (Schweinsberg et al. 2016) and the protocol from López et al. (2015), while 16S 34 35 127 rDNA was amplified using the primers SHA and SHB (Cunningham and Buss 1993) and the protocol from de Souza 36 37 128 et al. (2017). All PCR products were purified with Illustra ExoStar (GE Healthcare, Buckinghamshire, UK) and 38 39 129 directly sequenced in forward and reverse directions with the described primers using an ABI 3130xl Genetic Analyzer 40 41 130 (Applied Biosystems, Carlsbad, CA, USA). Sequences obtained in this study were deposited at GenBank, and 42 43 131 accession numbers are listed in Table S1 (MH824796-MH825619). 44 45 132 46 47 133 Genetic data analyses 48 49 134 Chromatograms of the forward and reverse DNA strands were assembled and edited using Sequencher 5.3 50 51 135 (Gene Codes Corp., Ann Arbor, MI, USA). Sequences were aligned with MAFFT 7.397 (Katoh and Standley 2013) 52 53 136 and the iterative refinement method E-INS-i. General statistics concerning the obtained sequences and the variability 54 55 137 of the two markers were calculated with DnaSP 6.11.1 (Librado and Rozas 2009). Intra and interspecific genetic 56 57 138 distances were calculated as p-distance based on the separated COI and 16S rDNA datasets using DnaSP 6.11.1, and 58 59 139 variance was estimated with 1000 bootstrap replicates (Tables 1 and 2). 60 61 62 5 63 64 65 1 2 3 4 140 In order to define species boundaries of the Red Sea Millepora, a median-joining network (Bandelt et al. 5 6 141 1999) for each mitochondrial gene was constructed including exclusively the newly-obtained sequences from the Red 7 8 142 Sea using Network 5.0 (http://www.fluxus-technology.com). The median-joining method uses a maximum parsimony 9 10 143 approach to search for all the shortest phylogenetic trees of given dataset (Bandelt et al. 1999). 11 12 144 All available COI and 16S rDNA sequences of Millepora were downloaded from GenBank and included in 13 14 145 our phylogenetic analyses. Because these sequences derive from different sources and include several species from 15 16 146 different localities, we performed separate phylogeny reconstructions for each mitochondrial locus. 17 18 147 secunda was selected as outgroup for both phylogenetic analyses (Nawrocki et al. 2010; Maggioni et al. 2017a, 2018). 19 20 148 Two phylogenetic tree optimality criteria were employed, namely Bayesian inference (BI) using MrBayes (Ronquist 21 22 149 et al. 2012) and maximum likelihood (ML) using RAxML (Stamatakis 2014). Both BI and ML analyses were run on 23 24 150 the CIPRES server (Miller et al. 2010). Prior to the phylogenetic analyses, we determined the best partition scheme 25 26 151 and nucleotide substitution models using PartitionFinder 1.1.1 (Lanfear et al. 2012). We used unlinked branch lengths, 27 28 the greedy search algorithm for nucleotide sequence, and considered four partitions: 16S rDNA and the three codon 29 152 30 positions of COI. Partitioning scheme comparison was performed using the corrected Akaike Information Criterion 31 153 32 33 154 (AIC) and the Bayesian Information Criterion (BIC), testing the influence on the tree topology. PartitionFinder 1.1.1 34 35 155 selected for RAxML the evolutionary model GTR+G for all the four partitions and for MrBayes the models of 36 37 156 evolution HKY+G for 16S rDNA, K80+G for COI_pos1, F81 for COI_pos2, and HKY+G for COI_pos3. BI runs 38 39 157 were performed using four Markov Chain Monte Carlo (MCMC) chains for 10 million generations, saving a tree every 40 th 41 158 1,000 generation. The tree searches were stopped when all parameters reached the stationarity for effective sampling 42 43 159 size and unimodal posterior distribution using Tracer 1.6 (Rambaut et al. 2014). The first 25% trees sampled were 44 45 160 discarded as burn-in following indications by Tracer. ML topology was obtained under the default parameters shown 46 47 161 on the CIPRES server with a multiparametric bootstrap analyses composed of 1,000 bootstrap replicates. 48 49 162 50 51 163 Morphological examination 52 53 164 Two sets of morphological features were analyzed in this study, i.e. skeletal pores and polyp and eumedusoid 54 55 165 nematocysts. For the examination of skeletal pores, a total of eight characters previously used for taxonomic studies 56 57 166 of Millepora were selected (Boschma 1948a; de Weerdt 1984; de Weerdt and Glynn 1991; Moshchenko 1995b, 1996b; 58 59 167 Razak and Hoeksema 2003) (Fig 3): (1) gastropore diameter, (2) dactylopore diameter, (3) distance between 60 61 62 6 63 64 65 1 2 3 4 168 gastropores, (4) distance between dactylopores, (5) distance from gastropore to nearest dactylopore, (6) number of 5 6 169 dactylopores per gastropore, (7) gastropore density, and (8) dactylopore density. Because M. exaesa is characterized 7 8 170 by absent or limited cyclosystem arrangement, for the analyses we selected only colonies of this species showing a 9 10 171 limited but recognizable cyclosytem arrangement. A total of ten colonies selected from the ones that were genetically 11 12 172 characterized were analysed for each of the three Red Sea Millepora species and a total of ten measurements per 13 14 173 colony were taken for each single trait (Table S2). Photographs of the skeleton surface were obtained at x50 15 16 174 magnification with a Leica IC80 HD microscope and its integrated standalone digital camera. On the basis of the 17 18 175 digital images with a reference scale, the eight skeletal variables were measure using ImageJ 1.49 (Rasband 1997). 19 20 176 For the examination of polyp nematocysts, five colonies of each Millepora species chosen from the ones that 21 22 177 were molecularly investigated were decalcified using 10% formic acid for approximately 24 hours. Soft tissues were 23 24 178 then washed in seawater, observed with a Nikon Eclipse E600 compound microscope, photographed with its integrated 25 26 179 camera, and measured using the NIS-Elements Viewer 4.30. For the examination of eumedusoid nematocysts, newly 27 28 released eumedusoid were collected and anesthetized with menthol crystals. Thereafter, nematocysts were observed 29 180 30 with the compound microscope in order to assess their distribution and they were measured. Nematocysts were 31 181 32 33 182 identified according to Bouillon et al. (2006) and 30-90 undischarged capsules were measured for each type, life stage, 34 35 183 and species. A total of six nematocyst features were analyzed, namely (1) polyp large stenotele, (2) polyp small 36 37 184 stenotele, (3) polyp macrobasic apotrichous eurytele, (4) polyp macrobasic apotrichous mastigophore, (5) eumedusoid 38 39 185 microbasic mastigophore, and (6) eumedusoid stenotele. For each feature, four measurements were obtained, namely 40 푙푒푛𝑔푡ℎ 푤𝑖푑푡ℎ 푙푒푛𝑔푡ℎ 41 186 length, width, area ( 푥 푥 휋), and ratio ( ), for a total of 24 analyzed characters. Additionally, polyp 2 2 푤𝑖푑푡ℎ 42 43 187 heteronemes (euryteles and mastigophores) were discharged adding a drop of sodium hypochlorite to the tissues, and 44 45 188 the length of the exploded shaft was measured (Table S3). 46 47 189 48 49 190 Statistical analyses 50 51 191 Morphological variables were log+1 transformed when the normality distribution was violated and, after 52 53 192 addressing assumptions, a sub-set of the eight pore characters (all but gastropore density) was used in a multivariate 54 55 193 analysis of variance (MANOVA) to test for differences in pore characters among M. dichotoma, M. exaesa, and M. 56 57 194 platyphylla from the Red Sea. Hereafter, a multiple univariate analysis of variance (ANOVA) was run to determine 58 59 195 which characters showed differences among the three investigated species. Wilks’ lambda criterion was used to test 60 61 62 7 63 64 65 1 2 3 4 196 for group differences in the MANOVA. A discriminant function analysis (DFA) was used to examine the pore 5 6 197 characters in discriminating individuals among species and to investigate whether the measured pore characters could 7 8 198 be used to classify samples into their original group. A classification with cross-validation was carried out to assign 9 10 199 individuals to their original group, and the scatterplots of the first two discriminant scores were drawn to depict the 11 12 200 species distribution on the graph. 13 14 201 Furthermore, the nematocyst measurements were investigated in order to detect statistically significant 15 16 202 differences among the three investigated species of fire corals. In particular, for the large and small stenoteles, which 17 18 203 are found in both the polyp and the eumedusoid stages, and for the two types of heteronemes, i.e. euryteles and 19 20 204 mastigophores, which are found only in the polyp stage, the following four measurements were taken into account: 21 22 205 length, width, area, and ratio. The measurements were taken by moving in the same direction across the slide in order 23 24 206 to measure each nematocyst only once within each specimen in order to avoid pseudoreplication. Statistical differences 25 26 207 between the nematocysts of the three fire coral species were tested using the Kruskal-Wallis test when we violated the 27 28 assumption of normality, and ANOVA when the normality distribution and the homoscedasticity were satisfied. 29 208 30 All the statistical analyses were carried out in SPSS 25 (SPSS, Chicago, IL, USA). All data report mean ± 31 209 32 33 210 dev. St. otherwise stated. 34 35 211 36 37 212 Results 38 39 213 Species identification 40 41 214 Based on the colony growth form and the cyclosystem arrangement, the sampled specimens were grouped in 42 43 215 three species: M. dichotoma showing fan-shaped branching growth form with distinct cyclosystems (Figs. 2a-c, 3b), 44 45 216 M. platyphylla showing a blade-like growth form with distinct cyclosystems on a flat surface (Figs. 2d-f, 3c), and M. 46 47 217 exaesa showing massive or sub-massive growth form with absent or limited cyclosystem arrangement (Figs. 2g-i, 3d). 48 49 218 50 51 219 Molecular results 52 53 220 COI and 16S rDNA sequences were obtained from all 412 collected fire coral colonies (Table S1). The COI 54 55 221 dataset including exclusively our Red Sea Millepora samples resulted in 426 bp with 104 variables nucleotides and 56 57 222 identified 33 haplotypes. The 16S rDNA alignment included 597 positions with 113 variable nucleotides for a total of 58 59 223 37 haplotypes. Both mitochondrial networks clearly showed the presence of three main clusters of haplotypes (Figs. 60 61 62 8 63 64 65 1 2 3 4 224 4a-b, Fig. S1) corresponding to the three groups of specimens assigned to M. dichotoma, M. platyphylla, or M. exaesa 5 6 225 based on growth colony form and cyclosystem arrangement. Individuals ascribed to the same species were very closely 7 8 226 related, being separated by a total number of mutations ranging from one to seven. The three species groups were 9 10 227 genetically isolated with no shared haplotypes and were distinguished by a high number of mutations, from 28 (M. 11 12 228 platyphylla – M. dichotoma based on 16S rDNA) to 76 (M. platyphylla – M. exaesa based on COI and M. dichotoma 13 14 229 – M. exaesa based on 16S rDNA). 15 16 230 The two single-gene trees are shown in Figs. 4c-d. In both phylogeny reconstructions, BI and ML analyses 17 18 231 were largely congruent with no topology conflicts and node support values were high across the ingroup. COI and 19 20 232 16S rDNA yielded similar topologies with well-supported clades although some sister relationships among the 21 22 233 recovered molecular lineages varied between the two phylogenetic trees. Both phylogenetic trees recovered Millepora 23 24 234 as composed of two main lineages, the Atlantic cluster and the Indo-Pacific one. Within the latter, all the Red Sea 25 26 235 samples clustered in three distinct and highly supported molecular clades matching morphological differences based 27 28 on traditional taxonomy. In fact, clade I includes M. dichotoma (COI Bayesian posterior probabilities PP = 1, ML 29 236 BI 30 bootstrap B = 99; 16S rDNA PP = 1 and B = 100), clade II is composed of M. platyphylla (COI PP = 1 and 31 237 ML BI ML BI 32 33 238 BML = 100; 16S rDNA PPBI = 1 and BML = 82), and clade III grouped M. exaesa (COI PPBI = 1 and BML = 100; 16S 34 35 239 rDNA PPBI = 1 and BML = 93). The three species from the Red Sea exhibited very high interspecific genetic distances 36 37 240 and extremely low intraspecific values (< 1) under both mitochondrial loci (Tables 1 and 2). Interspecific genetic 38 39 241 distances ranged from 9.08 ± 1.32% (mean ± standard deviation) (M. dichotoma – M. platyphylla) to 18.77 ± 1.32% 40 41 242 (M. platyphylla – M. exaesa) based on COI and between 5.76 ± 0.97% (M. dichotoma – M. platyphylla) and 13.67 ± 42 43 243 1.44% (M. dichotoma – M. exaesa) based on 16S rDNA. Notably, these three molecular lineages include exclusively 44 45 244 our newly-obtained sequences from the Red Sea. The sequences downloaded from GenBank of conspecific deriving 46 47 245 from other localities of the Southwestern Indo-Pacific did not cluster with any Red Sea sequence and constituted 48 49 246 additional and distinct molecular clades. As a result of this, Millepora dichotoma and M. platyphylla were recovered 50 51 247 as polyphyletic taxa while M. exaesa was recovered as monophyletic but composed of two divergent and distinct 52 53 248 clades. The COI phylogeny reconstruction recovered clade IVa including M. platyphylla from French Polynesia (Leray 54 55 249 et al. 2013; Schweinsberg et al. 2016) and clade Va composed of M. dichotoma from Australia (Schweinsberg et al. 56 57 250 2016). The phylogenetic tree based on 16S rDNA resolved clade IVb with M. platyphylla from Réunion Island and 58 59 251 Northern Mariana Islands (Souza et al. 2017), clade Vb consisted of M. dichotoma from Northern Mariana Islands (de 60 61 62 9 63 64 65 1 2 3 4 252 Souza et al. 2017), and clade VII grouping M. exaesa from Réunion Island. These molecular divergences between the 5 6 253 Red Sea populations and the Southwestern Indo-Pacific ones of M. dichotoma, M. platyphylla, and M. exaesa were 7 8 254 also clearly highlighted by the high genetic distances among these clades (Tables 1, 2). Indeed, the COI genetic 9 10 255 distances between Red Sea M. dichotoma (clade I) and the one from Australia (clade Va) (9.17 ± 1.22%) and between 11 12 256 Red Sea M. platyphylla (clade II) and the one from French Polynesia (clade IVa) (10.78 ± 1.34%) were comparable 13 14 257 to the one between Red Sea M. dichotoma (clade I) and Red Sea M. platyphylla (clade II) (9.08 ± 1.32%). Similarly, 15 16 258 the genetic distances based on 16S rDNA between Red Sea M. dichotoma (clade I) and the one from Northern Mariana 17 18 259 Islands (clade Vb) (4.3 ± 0.81%), Red Sea M. platyphylla (clade II) and the one from Réunion Island and Northern 19 20 260 Mariana Islands (clade IVb) (5.4 ± 0.92%), and Red Sea M. exaesa (clade III) and the one from Réunion Island (clade 21 22 261 VII) (5.14 ± 1.01%) were comparable to the one between Red Sea M. dichotoma (clade I) and Red Sea M. platyphylla 23 24 262 (clade II) (5.76 ± 0.97%). 25 26 263 27 28 Morphological results 29 264 30 The measurements of the eight pore characters are summarized in Table 3. Overall, M. dichotoma is 31 265 32 33 266 characterized by small gastropore diameter (0.12 ± 0.02%), large distance between dactylopores (0.3 ± 0.06%), small 34 35 267 number of dactylopores per gastropore (3.9 ± 0.89%), and small dactylopore density (24.39 ± 5.39%). Millepora 36 37 268 platyphylla has small dactylopore diameter (0.04 ± 0.01%), large distance between gastropores (0.97 ± 0.17%), small 38 39 269 distance from gastropore to nearest dactylopore (0.11 ± 0.02%), and an intermediate dactylopore density (43.71 ± 40 41 270 7.97%). Millepora exaesa shows a high number of dactylopores per gastropore (6.93 ± 0.84%), small gastropore 42 43 271 density (5.27 ± 1.54%), and high dactylopore density (54.86 ± 10.7%). 44 45 272 The analysis of the three Millepora species revealed the presence of a common set of nematocysts shared by 46 47 273 all species, with some species-specific variations. Regarding the polyp stage, all species showed the presence of small 48 49 274 and large stenoteles and macrobasic apotrichous heteronemes (Figs. 5a-o). The heteronemes could be further classified 50 51 275 as mastigophores and euryteles. Specifically, mastigophores were found in M. dichotoma (Figs. 5c-e) and M. 52 53 276 platyphylla (Figs. 5h-j), whereas M. exaesa was characterized by euryteles (Figs. 5m-o). Regarding the eumedusoid 54 55 277 stage, all fire coral species displayed the presence of large stenoteles in the bulbs but differences were found in the 56 57 278 presence and the distribution of microbasic mastigophores on the exumbrella (Figs. 5p-w). Millepora dichotoma 58 59 279 showed mastigophores grouped in a band in the middle portion of the exumbrella (Fig. 5p), M. platyphylla had 60 61 62 10 63 64 65 1 2 3 4 280 mastigophores grouped in the lower part of the exumbrella, in between bulbs (Fig. 5s), and M. exaesa had an 5 6 281 exumbrella deprived of nematocysts (Fig. 5v). 7 8 282 The MANOVA was run to test for differences in the multivariate pore characters measured for the three 9 10 283 species of Millepora from the Red Sea. All but one pore characters were assessed. The differences among the three 11 12 284 species based on the combined dependent variables were statistically significant, (F 2,294 = 178,395, p < 0,0005; Wilks’ 13 14 285 lambda 0.036, partial ƞ2 0.811). The follow-up univariate ANOVAs showed that all pore characters were statistically 15 16 286 significantly different among the three analyzed species (Table 4). The Wilks’ lambda test of discriminant function 17 18 287 analysis showed significant differences in morphometric measurements for all the three species (P < 0.001) (Table 19 20 288 S4). In the discriminant function analysis, the first DF (discriminant function) accounted for 66.6 %, the second DF 21 22 289 accounted for 33.4 %, showing clear interspecific differentiation. The dactylopore density, number of dactylopores 23 24 290 per gastropore, dactylopore distance, and gastropore diameter contributed to first DF while dactylopore diameter, 25 26 291 distance between gastropores, and distance from gastropore to nearest dactylopore contributed to second DF (Table 27 28 5) indicating that the former four pore characters were the most important measurements in discriminating the three 29 292 30 fire coral species. The DFA using cross-validation classification showed 99 % correct classification of the analyzed 31 293 32 33 294 individuals into their original species. Indeed, the proportion of correctly classified M. dichotoma samples was the 34 35 295 highest (100%), followed by M. exaesa samples (99 %), and M. platyphylla samples (98%). In accordance with the 36 37 296 species identification, the DF I vs. DF II graphs depicted the presence of three units although the presence of limited 38 39 297 overlap between M. exaesa and M. platyphylla(Figure 6). The analyses of the nematocysts measurements revealed 40 41 298 statistically significant differences among the three fire coral species of for the large stenoteles in both the polyp and 42 43 299 the eumedusoid stages (Table 6), as well as for the small stenoteles and heteronemes in the polyp stage (Table 7). 44 45 300 46 47 301 Discussion 48 49 302 Our results confirmed the presence of three distinct species of Millepora in the Red Sea, i.e. M. dichotoma, 50 51 303 M. exaesa, and M. platyphylla, based on new molecular and morphological data. The two sequenced mitochondrial 52 53 304 genes (COI and 16S rDNA) clearly resolved the three species as distinct lineages and provided a robust phylogenetic 54 55 305 hypothesis of the genus. Based on pore characters the analyzed samples were grouped in three clusters in full 56 57 306 agreement with the in-situ species identification. Finally, significant differences in the nematocysts of both polyps and 58 59 60 61 62 11 63 64 65 1 2 3 4 307 eumedusoids examined for the first time in material from this region were found among the three analyzed species of 5 6 308 fire corals from the Red Sea. 7 8 309 Although early taxonomic works suggested the presence of a single widespread species of Millepora 9 10 310 throughout the Indo-Pacific and the Caribbean-Atlantic (Hickson 1898, 1899), several taxonomists recognized three 11 12 311 well-defined species of fire corals in the Red Sea, i.e. M. dichotoma, M. exaesa, and M. platyphylla (Klunzinger 1879; 13 14 312 Crossland 1941; Boschma 1948a, b, 1966). These authors stated that these species were clearly distinct based on the 15 16 313 colony growth form and provided a detailed description of their morphological characters. Our study confirmed these 17 18 314 findings and included innovative information about the genetics, the skeletal pores, and the nematocysts of the three 19 20 315 species, discussed hereafter. 21 22 316 The resolution of the species boundaries in the Red Sea fire corals using a combination of genetic data and 23 24 317 morphology at two distinct levels, i.e. skeletal pore and nematocysts, showed that an integrated morpho-molecular 25 26 318 approach is useful for a better understanding of the systematics of Millepora. Our phylogeny reconstructions based 27 28 on COI and 16S rDNA resolved well-supported clades within this genus of hydrozoans and, for the first time, these 29 319 30 two molecular markers were used to build a phylogenetic hypothesis of the Indo-Pacific fire corals. In fact, the former 31 320 32 33 321 locus successfully resolved the species boundaries among M. alcicornis, M. complanata, M. squarrosa, and M. striata 34 35 322 from the Caribbean (Ruiz-Ramos et al. 2014). The latter has been already used by de Souza et al. (2017) to define the 36 37 323 phylogenetic relationships and the connectivity of the Caribbean-Atlantic fire corals. The resulting phylogenetic 38 39 324 hypothesis successfully resolved all the analyzed species, i.e. M. alcicornis, M. braziliensis, M. laboreli, and M. nitida 40 41 325 (de Souza et al. 2017). Therefore, COI and 16S rDNA may be selected for future works aimed to evaluate the species 42 43 326 boundaries of Millepora throughout its entire geographic distribution. A potential additional marker could be the ITS 44 45 327 region that has been successfully used for the identification of Millepora from Japan (Takama et al. 2018) The recently 46 47 328 developed set of 15 microsatellites Dubé et al. (2017c) seemed, however, to be only valid for the Indo-Pacific species. 48 49 329 Pores have been usually considered extremely variable and their use as informative taxonomic characters has 50 51 330 been criticized by several taxonomists (Hickson 1898, 1899; Crossland 1948; Boschma 1948a, 1966). Nevertheless, 52 53 331 Moshchenko (1994, 1995b, 1996b) examined in a rigorous way the utility of several pore characters, including the 54 55 332 ones we analyzed in the present work, and concluded that pores were able to define Millepora species. Indeed, a 56 57 333 quantitative evaluation of pore characters in fire corals from Vietnam successfully discriminated M. platyphylla from 58 59 334 the five branching species occurring in that area, i.e. M. cruzi Nemenzo, 1975 (probably M. intricata; see Razak and 60 61 62 12 63 64 65 1 2 3 4 335 Hoeksema, 2003), M. dichotoma, M. intricata, M. murrayi Quelch, 1884, and M. tenera, while a continuous gradient 5 6 336 was found among the branching species (Moshchenko 1994, 1995b, 1996b). These results were further confirmed by 7 8 337 investigations of growth forms, enzymes, and anatomy of the skeleton (Moshchenko 1994, 1995b, 1996b). 9 10 338 Interestingly, a recent molecular study demonstrated that the three branching species of fire corals found in Japan, i.e. 11 12 339 M. dichotoma, M. intricata, and M. tenera, were nested together in a single molecular clade (Takama et al. 2018). 13 14 340 More in general, Takama et al. (2018) recovered four molecular lineages of Millepora that were not in agreement with 15 16 341 the traditional morphology-based species identification (Razak and Hoeksema 2003). Nevertheless, they also showed 17 18 342 significant differences in gastropore diameter, dactylopore diameter, distances between gastropores among the genetic 19 20 343 clades, with the only exception of the absence of significant differences between two clades. Overall, based on our 21 22 344 results and the published studies, we conclude that pores are informative to tell apart at least some of the molecular 23 24 345 clades of Millepora. 25 26 346 The cnidome of Millepora, i.e. type, size, and distribution of nematocysts, was scarcely investigated in the 27 28 past, with a few exceptions. For instance, a first detailed assessment of the polyp nematocysts was presented by Calder 29 347 30 (1988), whose description of Millepora alcicornis from Bermuda included information and images of stenotele and 31 348 32 33 349 mastigophore capsules. More recently, other authors partially investigated the cnidome of a few Millepora species 34 35 350 from both Atlantic and Indo-Pacific localities (Meroz-Fine et al. 2003; Garcia-Arredondo et al. 2012; Bourmaud et al. 36 37 351 2013), finding in some cases differences in the size and the distribution of the nematocysts among fire coral 38 39 352 morphotypes (Meroz-Fine et al. 2003; Bourmaud et al. 2013). In this work, we analyzed and compared for the first 40 41 353 time the complete set of the nematocysts found in both the polyp and the eumedusoids stages of three Millepora 42 43 354 species from the Red Sea, finding significant differences among them and demonstrating the utility of the cnidome in 44 45 355 species discrimination within Millepora hydrozoans. Not only the size, but also the distribution of the nematocysts of 46 47 356 Red Sea Millepora species differed among taxa. Indeed, similarly to the observations of Bourmaud et al. (2013) on 48 49 357 eumedusoids from Réunion Island, we found that the eumedusoids of the three Red Sea species showed a different 50 51 358 organization of the mastigophores on the exumbrella. Millepora exaesa eumedusoids from both the Red Sea and 52 53 359 Réunion Island lacked mastigophores and this morphological similarity is also supported by the fact that these two 54 55 360 taxa clustered together in a fully supported clade in the 16S rRNA phylogeny reconstruction. Moreover, M. dichotoma 56 57 361 from the Red Sea and M. tenera from Réunion Island showed a similar pattern of exumbrellar nematocyst organization, 58 59 60 61 62 13 63 64 65 1 2 3 4 362 and indeed are closely related in the 16S rRNA phyogenetic tree. Contrarily, M. platyphylla from the Red Sea and 5 6 363 Réunion Island had mastigophores organized in different ways and were phylogenetically divergent. 7 8 364 Our single-gene phylogeny reconstructions revealed unexpected phylogenetic patterns for fire corals in the 9 10 365 Indo-Pacific. Although the genetic analyses of our samples confirmed the presence of three lineages at the species 11 12 366 level within the Red Sea, the inclusion of available Millepora sequences from the Indo-Pacific in our analyses resulted 13 14 367 in the polyphyly of both M. dichotoma and M. platyphylla and in the presence of two divergent and allopatric lineages 15 16 368 of M. exaesa. Unfortunately, no specimen photos were published for these sequences downloaded from GenBank with 17 18 369 the only exception of the samples of M. dichotoma from Australia included in the COI phylogeny reconstruction 19 20 370 (Schweinsberg et al. 2016). Nevertheless, M. platyphylla and M. exaesa are both supposedly morphologically 21 22 371 distinctive species in the Indo-Pacific as they are told apart based on their unique colony growth form (Boschma 23 24 372 1949a, b; Razak and Hoeksema 2003), although a certain degree of ecophenotypic variation occurs especially in M. 25 26 373 platyphylla (Dubé et al. 2017a, b). Therefore, we are prone to reject the hypothesis of a misidentification by the authors 27 28 based on colony morphology of both M. platyphylla from Réunion Island and French Polynesia (Leray et al. 2013; 29 374 30 Schweinsberg et al. 2016) and M. exaesa from Réunion Island. This being said, the deep intraspecific genetic 31 375 32 33 376 divergences between populations of M. dichotoma, M. exaesa, and M. platyphylla from the Red Sea and various 34 35 377 localities of the Southwestern Indo-Pacific suggested that these populations represented distinct evolutionary units 36 37 378 that have not been detected in previous taxonomic works. These three species have been historically considered 38 39 379 widespread throughout the Indo-Pacific, from the Red Sea to French Polynesia (Klunzinger 1879; Crossland 1941; 40 41 380 1948; Boschma 1949a; Razak and Hoeksema 2003), but our genetic analyses clearly demonstrated that they all 42 43 381 constitute species complexes. Each of these three species complexes is composed of at least two species that share a 44 45 382 similar colony growth form. Therefore, while the growth of the colony is a diagnostic character to distinguish the Red 46 47 383 Sea species between each other (Forskål 1775; Ehrenberg 1834; Klunzinger 1879), it is clearly not informative to 48 49 384 define boundaries within the three species complexes. The genetic distances between the two distinct lineages for each 50 51 385 of the three nominal species are comparable to those occurring between valid species in the genus for the examined 52 53 386 markers (Tables 1 and 2). Given that the type locality of M. dichotoma, M. exaesa, and M. platyphylla is the Red Sea, 54 55 387 a future taxonomic revision of Millepora would maintain these names for the three Red Sea molecular lineages while 56 57 388 new names or synonymised names would be assigned to the other Indo-Pacific lineages of these three species. In this 58 59 389 respect, it will be relevant to re-examine Millepora tuberosa Boschma, 1966, since its type specimens from Mauritius 60 61 62 14 63 64 65 1 2 3 4 390 resemble corals of M. exaesa. Boschma (1966) considered M. exaesa a doubtful species and he suspected that 5 6 391 specimens from outside the Red Sea are actually M. tuberosa. Both taxa were synonymized by Razak and Hoeksema 7 8 392 (2003), who gave priority to M. exaesa, being the senior synonym. Since Réunion Island is only 225 km away from 9 10 393 Mauritius, this may also concern specimens from this locality that were studied by Dubé et al. (2017c), who did not 11 12 394 refer to Boschma’s (1966) work. Our results remove the doubts expressed by Boschma (1966) concerning the status 13 14 395 of M. exaesa as a valid species. 15 16 396 From a morphological point of view, either morphological convergence of the colony growth form or 17 18 397 character state conservation from the ancestor may have occurred. On the base of our phylogenetic trees, the former 19 20 398 hypothesis may be applied to M. dichotoma and M. platyphylla, given the absence of a sister relationships between 21 22 399 the Red Sea clades and the Southwestern Indo-Pacific ones. The second explanation is more likely for M. exaesa 23 24 400 considering that the two clades from the Red Sea and Réunion Island are sister taxa, possibly M. exaesa and M. 25 26 401 tuberosa. Although the absence of variation in the colony growth form, our integrated morphological analyses on Red 27 28 Sea Millepora may suggest to further evaluate both pore structures and nematocysts of the Southwestern Indo-Pacific 29 402 30 clades to evaluate the occurrence of possible differences of these morphological traits between the Red Sea and the 31 403 32 33 404 Southwestern Indo-Pacific molecular lineages. Indeed, nematocysts of M. dichotoma, M. exaesa, and M. platyphylla 34 35 405 from the Southwestern Indo-Pacific have never been studied. Moreover, no comparisons of pores between the Red 36 37 406 Sea and the Southwestern Indo-Pacific colonies of these three species have been undertaken so far. From a biological 38 39 407 point of view, the distinction between the Red Sea and the Southwestern Indo-Pacific clades may suggest that either 40 41 408 the Red Sea lineages have a small-distance dispersal capability or the Southwestern Indo-Pacific lineages have failed 42 43 409 to disperse within the Red Sea. At least for the case of clade IVb (M. platyphylla from Réunion Island and Northern 44 45 410 Mariana Islands), this evolutionary unit is capable of long-distance dispersal. This contrasting pattern of short- and 46 47 411 long-distance dispersal capabilities among species is similar to the one occurring in the Caribbean-Atlantic where M. 48 49 412 alcicornis showed a high connectivity within the Caribbean while the three endemic species of Brazil, i.e. M. 50 51 413 braziliensis, M. nitida, and M. laboreli, are restricted to few locations and isolated by strong biogeographic barriers 52 53 414 (de Souza et al. 2017). 54 55 415 A growing number of studies has revealed that several hydrozoans traditionally considered to display a large 56 57 416 geographic distribution are indeed species complexes composed of geographically circumscribed lineages (e.g., 58 59 417 Schuchert 2014; Postaire et al. 2016, 2017; Maggioni et al. 2017b; Montano et al. 2017). For example, the near- 60 61 62 15 63 64 65 1 2 3 4 418 cosmopolitan species Plumularia setacea (Linnaeus, 1758) is subdivided in a multitude of distinct molecular clades 5 6 419 showing distinct geographic distributions, such as South Africa, the Western Indian Ocean, the Caribbean, and the 7 8 420 Mediterranean Sea. Similarly, Postaire et al. (2017) revealed a deep and strong genetic differentiation between the 9 10 421 Indian and the Pacific populations of Macrorhynchia phoenicea (Busk, 1852). Several analogous cases have been 11 12 422 recently described in many hard corals throughout the Indo-Pacific (Flot et al. 2008; Ladner and Palumbi 2012; Pinzón 13 14 423 et al., 2013; Richards et al. 2016; Gélin et al., 2017; Arrigoni et al. 2016a). Moreover, the existence of unexpected 15 16 424 intraspecific evolutionary breaks between the Red Sea and the Southwestern Indo-Pacific populations/species have 17 18 425 been discovered based on genetic studies in several marine phyla, including scleractinian corals (Arrigoni et al. 2012, 19 20 426 2016b; Huang et al. 2014), octocorals (Reijnen et al. 2014), sea urchins (Bronstein et al. 2016), and fishes (DiBattista 21 22 427 et al. 2013; Fernandez et al. 2015; Ahti et al. 2016; Waldrop et al. 2016). In most cases, these genetic divergences 23 24 428 between Oceans reflect the importance of geological features, such as tectonic plates and mantle plume tracks. Indeed, 25 26 429 it is now largely accepted that the actual distribution of most hard corals depends more on geological events than 27 28 contemporary environmental conditions and habitats (Keith et al. 2013; Leprieur et al. 2016). For example, important 29 430 30 tectonic changes in what is now the Indian Ocean occurred during both the Eocene and Oligocene, resulting in the 31 431 32 33 432 fragmentation of the Tethys Sea (Schettino and Turco, 2011), and may have promoted isolation and speciation evens. 34 35 433 In conclusion, the obtained results clearly resolved the species boundaries of Millepora in the Red Sea. Given 36 37 434 the long-term challenges of morphology-based species identification of these organisms (Hickson 1898; Boschma 38 39 435 1948b, 1949) and the recent unexpected molecular findings in discordance with traditional systematics (Ruiz-Ramos 40 41 436 et al. 2014; Takama et al. 2018), we believed that the presented approach may be useful to define the relationships 42 43 437 among the fire corals. Moreover, the recovery of multiple molecular lineages of M. dichotoma, M. exaesa, and M. 44 45 438 platyphylla between the Red Sea and the Indo-Pacific suggested the need of investigate more material from additional 46 47 439 localities of the Indo-Pacific. The integration of genetics and morphology (pore structure and nematocysts of both 48 49 440 polyp and eumedusoid stages) may be the key to elucidate what Boschma (1948a) defined “the species problem in 50 51 441 Millepora” more than 50 years ago. 52 53 442 54 55 443 Acknowledgements 56 57 444 This research was undertaken in accordance with the policies and procedures of the King Abdullah University 58 59 445 of Science and Technology (KAUST). Permissions relevant for KAUST to undertake the research have been obtained 60 61 62 16 63 64 65 1 2 3 4 446 from the applicable governmental agencies in the Kingdom of Saudi Arabia. We wish to thank Amr Gusti (KAUST), 5 6 447 the captain and crew of the MV Dream-Master and the KAUST Coastal and Marine Resources Core Lab for fieldwork 7 8 448 logistics in the Red Sea. This project was supported by funding from KAUST (award #FCC/1/1973-21 and baseline 9 10 449 research funds to MLB). TS would like to acknowledge Yossi Loya and the Israeli Taxonomy Initiative for funding 11 12 450 his work. We are deeply grateful to the editor and three anonymous referees for their comments which greatly 13 14 451 improved the manuscript. 15 16 452 17 18 453 Compliance with ethical standards 19 20 454 Conflict of interest 21 22 455 On behalf of all authors, the corresponding author states that there is no conflict of interest. 23 24 456 25 26 457 References 27 Ahti PA, Coleman RR, DiBattista JD, Berumen ML, Rocha LA, Bowen BW (2016) Phylogeography of Indo‐Pacific 28 458 reef fishes: sister wrasses Coris gaimard and C. cuvieri in the Red Sea, Indian Ocean and Pacific Ocean. J 29 459 Biogeogr 43:1103–1115 30 460 31 461 Amaral FD, Silva RS, Mauricio-da-Silva L, Sole-Cava AM (1997) Molecular systematics of Millepora alcicornis 32 462 Linnaeus, 1758 and M. braziliensis Verrill, 1868 (: Milleporidae) from Brazil. Proc 8th Int Coral 33 463 Reef Symp 2:1577–1580 34 464 Amaral FD, Steiner AQ, Broadhurst MK, Cairns SD (2008) An overview of the shallow-water calcified hydroids from 35 465 Brazil (Hydrozoa: ), including the description of a new species. Zootaxa 1930:56–68 36 466 Arrigoni R, Stefani F, Pichon M, Galli P, Benzoni F (2012) Molecular phylogeny of the robust clade (Faviidae, 37 467 Mussidae, Merulinidae, and Pectiniidae): an Indian Ocean perspective. Mol Phylogenet Evol 65:183–193 38 468 Arrigoni R, Berumen ML, Chen CA, Terraneo TI, Baird AH, Payri C, Benzoni F (2016a) Species delimitation in the 39 469 reef coral genera Echinophyllia and Oxypora (Scleractinia, Lobophylliidae) with a description of two new 40 470 species. Mol Phylogenet Evol 105:146–159 41 471 Arrigoni R, Benzoni F, Huang D, Fukami H, Chen CA, Berumen ML, Hoogenboom M, Thomson DP, Hoeksema 42 472 BW, Budd AF, Zayasu Y, Terraneo TI, Kitano YF, Baird AH (2016b) When forms meet genes: revision of 43 473 the scleractinian genera Micromussa and Homophyllia (Lobophylliidae) with a description of two new 44 474 species and one new genus. Contrib Zool 85:387–422 45 475 Bandelt HJ, Forster P, Röhl A (1999) Median-joining networks for inferring intraspecific phylogenies. Mol Biol Evol 46 476 16:37–48 47 477 Benayahu Y Loya Y (1977) Space partitioning by stony corals soft corals and benthic algae on the coral reefs of the 48 478 northern Gulf of Eilat (Red Sea). Helgol Wiss Meeresunters 30:362 49 479 Boschma H (1948a) The species problem in Millepora. Zool Verh Leiden 1:1–115 50 480 Boschma H (1948b) Specific characters in Millepora. Proc Kon Ned Akad Wet 51:818–823 51 481 Boschma H (1949) The ampullae of Millepora. Proc Kon Ned Akad Wet 52:3–14 52 482 Boschma H (1956) Milleporina and Stylasterina. In: Moore RC (ed) Treatise on paleontology. Part F, 53 483 Coelenterata. Geological Society of America & University of Kansas Press. 54 484 Boschma H (1966) On a new species of Millepora from Mauritius, with notes on the specific characters of Millepora 55 485 exaesa. Proc Kon Ned Akad Wet 69:409–419 56 486 Bouillon J, Gravili C, Gili J-M, Boero F. 2006. An introduction to Hydrozoa. Ann Mus Hist Nat Paris 194: 1–591. 57 487 Bourmaud CAF, Leung JKL, Bollard S, Gravier‐Bonnet N (2013) Mass spawning events, seasonality and reproductive 58 488 features in Milleporids (Cnidaria, Hydrozoa) from Reunion Island. Mar Ecol 34:1424 59 489 Bronstein O, Kroh A, Haring E (2016) Do genes lie? Mitochondrial capture masks the Red Sea collector urchin’s true 60 490 identity (Echinodermata: Echinoidea: Tripneustes). Mol Phylogenet Evol 104:1–13 61 62 17 63 64 65 1 2 3 4 491 Cairns SD, Hoeksema BW, Van der land J (1999) List of extant stony corals. Atoll Res Bull 459:1–46 5 492 Calder DR (1988) Shallow-water hydroids of Bermuda: the Athecatae. R Ontario Mus Life Sci Contrib 148:1–107 6 493 Crossland C (1941) On Forskål's collection of corals in the Zoological Museum of Copenhagen. Spolia ZooL Haun. 7 494 Skr Vdgivet Univ Zool Mus Kbh 1:5–63, pls 1–12 8 495 Crossland C (1948) Reef corals of the South African Coast. Ann Nat Mus 9:169–205, pls 1–14 9 496 Cunningham CW, Buss LW (1993) Molecular evidence for multiple episodes of paedomorphosis in the Family 10 497 Hydractiniidae. Biochem Syst Ecol 21:57–69 11 498 de Souza JN, Nunes FL, Zilberberg C, Sanchez JA, Migotto AE, Hoeksema BW, Serrano XM, Baker AC, Lindner A 12 499 (2017) Contrasting patterns of connectivity among endemic and widespread fire coral species (Millepora 13 500 spp.) in the tropical Southwestern Atlantic. Coral Reefs 36:701–716. 14 501 de Weerdt WH (1981) Transplantation experiments with Caribbean Millepora species (Hydrozoa, Coelenterata), 15 502 including some ecological observations on growth forms. Bijdr Dierkd 51:1–19 16 503 de Weerdt WH (1984) Taxonomic characters in Caribbean Millepora species (Hydrozoa, Coelenterata) including 17 504 some ecological observations on growth forms. Bijdr Dierkd 54:243–362 18 505 de Weerdt, WH, Glynn PW (1991) A new and presumably now extinct species of Millepora (Hydrozoa) in the eastern 19 506 Pacific. Zool Med Leiden 65:267–276 20 507 DiBattista JD, Berumen ML, Gaither MR, Rocha LA, Eble JA, Choat JH, Craig MT, Skillings DJ, Bowen BW (2013) 21 508 After continents divide: comparative phylogeography of reef fishes from the Red Sea and Indian Ocean. J 22 509 Biogeogr 40:1170–1181 23 510 DiBattista JD, Choat JH, Gaither MR, Hobbs JP, Lozano-Cortés DF, Myers R, Paulay G, Rocha LA, Toonen RJ, 24 511 Westneat M, Berumen ML (2016a) On the origin of endemic species in the Red Sea. J Biogeogr 43:13–30 25 512 DiBattista JD, Roberts MB, Bouwmeester J, Bowen BW, Coker DJ, Lozano‐Cortés DF, Choat JH, Gaither MR, Hobbs 26 513 JP, Khalil MT, Kochzius M, Myers RF, Paulay G, Robitzch V, Saenz-Agudelo P, Salas E, Sinclair-Taylor 27 TH, Toonen RJ, Westneat MW, Williams ST, Berumen ML (2016b) A review of contemporary patterns of 28 514 endemism for shallow water reef fauna in the Red Sea. J Biogeogr 43:423–439 29 515 Dubé CE, Mercière A, Vermeij MJ, Planes S (2017a) Population structure of the hydrocoral Millepora platyphylla in 30 516 31 517 habitats experiencing different flow regimes in Moorea, French Polynesia. PloS ONE 12:e0173513 32 518 Dubé CE, Boissin E, Maynard JA, Planes S (2017b) Fire coral clones demonstrate phenotypic plasticity among reef 33 519 habitats. Mol Ecol 26:3860–3869 34 520 Dubé CE, Planes S, Zhou Y, Berteaux-Lecellier V, Boissin E (2017c) Genetic diversity and differentiation in reef- 35 521 building Millepora species, as revealed by cross-species amplification of fifteen novel microsatellite loci. 36 522 PeerJ 5:e2936 37 523 Duchassaing P, Michelotti J (1860) Mémoire sur les Coralliaires des Antilles. Mém Ac Sc Turin Sér 2, 19:279–365, 38 524 pls 1–10 39 525 Edmunds PJ (1999) The role of colony morphology and substratum inclination in the success of Millepora alcicornis 40 526 on shallow coral reefs. Coral Reefs 18:133–140 41 527 Ehrenberg CG (1834) Beitrage zur physiologischen Kenntniss der Corallenthiere im Allgemeinen und besunders des 42 528 Rothen Meeres, nebst einem Versuche zur physiologischen Systematik derselben. Abh Kon Akad Wiss 43 529 Berlin 1832:225–380 44 530 Fernandez‐Silva I, Randall JE, Coleman RR, DiBattista JD, Rocha LA, Reimer JD, Meyer CG, Bowen BW (2015) 45 531 Yellow tails in the Red Sea: phylogeography of the Indo‐Pacific goatfish Mulloidichthys flavolineatus reveals 46 532 isolation in peripheral provinces and cryptic evolutionary lineages. J Biogeogr 42:2402–2413 47 533 Flot JF, Licuanan WY, Nakano Y, Payri C, Cruaud C, Tillier S (2008) Mitochondrial sequences of Seriatopora corals 48 534 show little agreement with morphology and reveal the duplication of a tRNA gene near the control region. 49 535 Coral Reefs 27:789–794 50 536 Forskål P (1775) Descriptiones animalium avium, amphibiorum, piscium, insectorum, vermium; quae in itinere 51 537 orientali observavit. IV Corallia. Heineck et Faber, Hauniae, pp 131–139 52 538 García-Arredondo A, Rojas A, Iglesias-Prieto R, Zepeda-Rodriguez A, Palma-Tirado L (2012) Structure of 53 539 nematocysts isolated from the fire corals Millepora alcicornis and Millepora complanata (Cnidaria: 54 540 Hydrozoa). J Venom Anim Toxins Incl Trop Dis 18:109–115 55 541 Gélin P, Postaire B, Fauvelot C, Magalon H (2017) Reevaluating species number, distribution and endemism of the 56 542 coral genus Pocillopora Lamarck, 1816 using species delimitation methods and microsatellites. Mol 57 543 Phylogenet Evol 109:430–446 58 544 Hickson SJ (1898) On the species of the genus Millepora: a preliminary communication. J Zool 66:246–257 59 545 Hickson SJ (1809) Report on the specimens of the genus Millepora collected by Dr Willey. J Zool 119:661–672 60 61 62 18 63 64 65 1 2 3 4 546 Hoeksema BW, Nunes FLD, Lindner A, de Souza JN (2017) Millepora alcicornis (Hydrozoa: ) at Ascension 5 547 Island: confirmed identity based on morphological and molecular analyses. J Mar Biol Assoc UK 97:709– 6 548 712 7 549 Huang D, Benzoni F, Arrigoni R, Baird AH, Berumen ML, Bouwmeester J, Chou LM, Fukami H, Licuanan WY, 8 550 Lovell ER, Meier R (2014) Towards a phylogenetic classification of reef corals: the Indo‐Pacific genera 9 551 Merulina, Goniastrea and Scapophyllia (Scleractinia, Merulinidae). Zool Scripta 43:531–548 10 552 Katoh K, Standley DM (2013) MAFFT multiple sequence alignment software version 7: improvements in 11 553 performance and usability. Mol Biol Evol 30:772–780 12 554 Keith SA, Baird AH, Hughes TP, Madin JS, Connolly SR (2013) Faunal breaks and species composition of Indo- 13 555 Pacific corals: the role of plate tectonics, environment and habitat distribution. Proc R Soc B 280:20130818 14 556 Klausewitz W (1989) Evolutionary history and zoogeography of the Red Sea ichthyofauna. Fauna Saudi Arabia 15 557 10:310–337 16 558 Klunzinger CB (1879) Die Korallthiere des Rothen Meeres. Gutmann, Berlin. 3:1–100 17 559 Ladner JT, Palumbi SR (2012) Extensive sympatry, cryptic diversity and introgression throughout the geographic 18 560 distribution of two coral species complexes. Mol Ecol 21:2224–2238 19 561 Lanfear R, Calcott B, Ho SYW, Guindon S (2012) PartitionFinder: combined selection of partitioning schemes and 20 562 substitution models for phylogenetic analyses. Mol Biol Evol 29:1695–1701 21 563 Leprieur, F., Descombes, P., Gaboriau, T., Cowman, P.F., Parravicini, V., Kulbicki, M., Melian, C.J., de Santana CN, 22 564 Heine C, Mouillot D, Bellwood DR, Pellissier L (2016) Plate tectonics drive tropical reef biodiversity 23 565 dynamics. Nat Commun 7:11461 24 566 Leray M, Yang JY, Meyer CP, Mills SC, Agudelo N, Ranwez V, Boehm JT, Machida RJ (2013) A new versatile 25 567 primer set targeting a short fragment of the mitochondrial COI region for metabarcoding metazoan diversity: 26 568 application for characterizing coral reef fish gut contents. Front Zool 10:34 27 Lewis JB (1989) The ecology of Millepora. Coral Reefs 8:99–107 28 569 Lewis JB (2006) Biology and ecology of the Millepora on coral reefs. Adv Mar Biol 50:1–55 29 570 Librado P, Rozas J (2009) DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. 30 571 31 572 Bioinformatics 25:1451–1452 32 573 Linnaeus C (1758) Systema Naturae, ed. X, vol. I. Laurentii Salvii, Holmiae 33 574 López C, Clemente S, Almeida C, Brito A, Hernandez M (2015) A genetic approach to the origin of Millepora sp. in 34 575 the eastern Atlantic. Coral Reefs 34:631–638 35 576 Loya Y (1972) Community structure and species diversity of hermatypic corals at Eilat, Red Sea. Mar Biol 13:100– 36 577 123 37 578 Loya Y (1976) Recolonization of Red Sea corals affected by natural catastrophes and man‐made perturbations. 38 579 Ecology 57:278–289 39 580 Loya Y, Slobodkin LB (1971) The coral reefs of Eilat. Symp Zool Soc Lond 28:117–139 40 581 Maggioni D, Galli P, Berumen ML, Arrigoni R, Seveso D, Montano S (2017a) Astrocoryne cabela, gen. nov. et sp. 41 582 nov.(Hydrozoa: Sphaerocorynidae), a new sponge-associated hydrozoan. Invert Syst 31:734–746 42 583 Maggioni D, Montano S, Arrigoni R, Galli P, Puce S, Pica D, Berumen ML (2017b) Genetic diversity of the Acropora- 43 584 associated hydrozoans: new insight from the Red Sea. Mar Biodivers 47:1045–1055 44 585 Maggioni D, Arrigoni R, Galli P, Berumen ML, Seveso D, Montano S (2018) Polyphyly of the genus Zanclea and 45 586 family (Hydrozoa, Capitata) revealed by the integrative analysis of two bryozoan-associated 46 587 species. Contrib Zool 87:87–104 47 588 Meroz-Fine E, Brickner I, Loya Y, Ilan M (2003) The hydrozoan coral Millepora dichotoma: speciation or phenotypic 48 589 plasticity? Mar Biol 143:1175–1183 49 590 Miller MA, Pfeiffer W, Schwartz T (2010) Creating the CIPRES Science Gateway for inference of large phylogenetic 50 591 trees. Proceedings of the Gateway Computing Environments Workshop (GCE), New Orleans, IEEE 51 592 Montano S, Maggioni D, Galli P, Hoeksema BW (2017) A cryptic species in the Pteroclava krempfi species complex 52 593 (Hydrozoa, Cladocorynidae) revealed in the Caribbean. Mar Biodivers 47:83–89 53 594 Moshchenko AV (1992) Morphology and variability of colonies of branched milleporids (Hydrozoa, Athecata, 54 595 Milleporidae) from Vietnam. Zool Zhur 71:5–12 55 596 Moshchenko AV (1994) Method of quantitative evaluation of the structure of the pore apparatus of millepore hydroids. 56 597 Rus J Mar Biol 20:358–365 57 598 Moshchenko AV (1995a) Environmental variations of the colony form of hydroid Millepora platyphylla in 58 599 Vietnamese Reefs. Rus J Mar Biol 21:174–183 59 600 Moshchenko AV (1995b) A quantitative evaluation of the structure of the pore apparatus of Millepora hydroids of 60 601 Vietnam. Rus J Mar Biol 21: 265–274 61 62 19 63 64 65 1 2 3 4 602 Moshchenko AV (1996a) Growth and development of Millepora colonies (Athecata, Milleporidae). Zool Zhur 5 603 75:485–493 6 604 Moshchenko AV (1996b) Variability of the pore apparatus of milleporine hydroids of Vietnam. Rus J Mar Biol 22:32– 7 605 40 8 606 Moshchenko AV (1997) On the species composition of millepores in the Indo-Pacific. Rus J Mar Biol 23:238–247 9 607 Nawrocki AM, Schuchert P, Cartwright P (2010) Phylogenetics and evolution of Capitata (Cnidaria: Hydrozoa), and 10 608 the systematics of . Zool Scr 39:290–304 11 609 Perkol-Finkel S, Benayahu Y (2004) Community structure of stony and soft corals on vertical unplanned artificial 12 610 reefs in Eilat (Red Sea): comparison to natural reefs. Coral Reefs 23:195–205 13 611 Pinzón JH, Sampayo E, Cox E, Chauka LJ, Chen CA, Voolstra CR, LaJeunesse TC (2013) Blind to morphology: 14 612 genetics identifies several widespread ecologically common species and few endemics among Indo‐Pacific 15 613 cauliflower corals (Pocillopora, Scleractinia). J Biogeogr 40:1595–1608 16 614 Postaire B, Magalon H, Bourmaud CA, Bruggemann JH (2016) Molecular species delimitation methods and 17 615 population genetics data reveal extensive lineage diversity and cryptic species in Aglaopheniidae (Hydrozoa). 18 616 Mol Phylogenet Evol 105:36–49 19 617 Postaire B, Gélin P, Bruggemann JH, Pratlong M, Magalon H (2017) Population differentiation or species formation 20 618 across the Indian and the Pacific Oceans? An example from the brooding marine hydrozoan Macrorhynchia 21 619 phoenicea. Ecol Evol 7:8170–8186 22 620 Pourtales LF (1877) Effects of urticating organs of Millepora on the tongue. Nature 17:27 23 621 Ruiz-Ramos DV, Weil E, Schizas NV (2014) Morphological and genetic evaluation of the hydrocoral Millepora 24 622 species complex in the Caribbean. Zool Stud 53:4 25 623 Quelch JJ (1884) The Milleporidae. Nature 30:539 26 624 Rambaut A, Suchard MA, Xie D, Drummond AJ (2014) Tracer v1.6. http://beast.bio.ed.ac.uk/Tracer/ 27 Rasband WS (1997) ImageJ. http://rsb.info.nih.gov/ij/ 28 625 Razak TB, Hoeksema BW (2003) The hydrocoral genus Millepora (Hydrozoa: Capitata: Milleporidae) in Indonesia. 29 626 Zool Verh Leiden 345:313–336 30 627 31 628 Richards ZT, Berry O, van Oppen MJ (2016) Cryptic genetic divergence within threatened species of Acropora coral 32 629 from the Indian and Pacific Oceans. Conserv Genet 17:577–591 33 630 Reijnen BT, McFadden CS, Hermanlimianto YT, van Ofwegen LP (2014) A molecular and morphological exploration 34 631 of the generic boundaries in the family Melithaeidae (Coelenterata: Octocorallia) and its taxonomic 35 632 consequences. Mol Phylogenet Evol 70:383–401 36 633 Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A, Höhna S, Barget B, Liu L, Suchard MA, Huelsenbeck 37 634 JP (2012) MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model 38 635 space. Syst Biol 61:539–542 39 636 Ruiz-Ramos DV, Weil E, Schizas NV (2014) Morphological and genetic evaluation of the hydrocoral Millepora 40 637 species complex in the Caribbean. Zool Stud 53:4 41 638 Schettino A, Turco E (2011) Tectonic history of the western Tethys since the Late Triassic. Geol Soc Am Bull 123:89– 42 639 105 43 640 Schuchert P (2014) High genetic diversity in the hydroid Plumularia setacea: a multitude of cryptic species or 44 641 extensive population subdivision? Mol Phylogenet Evol 76:1–9 45 642 Schweinsberg M, Tollrian R, Lampert KP (2016) Inter‐and intra‐colonial genotypic diversity in hermatypic 46 643 hydrozoans of the family Milleporidae. Mar Ecol 38:e12388 47 644 Smith TB, Glynn PW, Maté JL, Toth LT, Gyory J (2014) A depth refugium from catastrophic coral bleaching prevents 48 645 regional extinction. Ecology 95:1663–1673 49 646 Stamatakis A (2014) RAxML Version 8: A tool for phylogenetic analysis and post-analysis of large phylogenies". 50 647 Bioinformatics 30:1312–1313 51 648 Takama O, Fernandez-Silva I, López C, Reimer JD (2018) Molecular phylogeny demonstrates the need for taxonomic 52 649 reconsideration of species diversity of the hydrocoral genus Millepora (Cnidaria: Hydrozoa) in the Pacific. 53 650 Zool Sci 35:123–133 54 651 Vago R, Achituv Y, Vaky L, Dubinsky Z, Kizner Z (1998) Colony architecture of Millepora dichotoma Forskal. J 55 652 Exp Mar Biol Ecol 224:225–235 56 653 Waldrop E, Hobbs JP, Randall JE, DiBattista JD, Rocha LA, Kosaki RK, Berumen ML, Bowen BW (2016) 57 654 Phylogeography, population structure and evolution of coral‐eating butterflyfishes (Family Chaetodontidae, 58 655 genus Chaetodon, subgenus Corallochaetodon). J Biogeogr 43:1116–1129 59 656 Wolf-Vecht A, Paldor N, Brenner S (1992) Hydrographic indications of advection/convection effects in the Gulf of 60 657 Eilat. Deep Sea Research Part A. Oceanogr Res Pap 39:1393–1401 61 62 20 63 64 65 1 2 3 4 658 5 6 659 Figures legends 7 8 660 Fig. 1 Map of Millepora sampling locations along the Red Sea 9 10 661 Fig. 2 In-vivo photos of the three Millepora species occurring in the Red Sea. a-c M. dichotoma; d-f M. 11 12 662 platyphylla; g-i M. exaesa. Scale bars: 5 cm 13 14 663 Fig. 3 a Pore characters analyzed in this study, gastropore diameter (1), dactylopore diameter (2), distance 15 16 664 between gastropores (3), distance between dactylopores (4), distance from gastropore to nearest dactylopore (5), and 17 18 665 number of dactylopores per gastropore (6). b-c Skeleton surface of the three Millepora species occurring in the Red 19 20 666 Sea. b M. dichotoma; c M. platyphylla; d M. exaesa. Scale bars: 2 mm 21 22 667 Fig. 4 a COI and b 16S rDNA most parsimonious median-joining networks of the three Red Sea Millepora 23 24 668 species. Each circle represents a unique haplotype and its size is proportional to its total frequency. Thin branches and 25 26 669 black cross-bars represent a single nucleotide change, thick black bars represent greater than one nucleotide change 27 28 670 (as indicated), small black circles represent missing haplotypes. c COI and d 16S rDNA phylogeny reconstructions of 29 30 671 Millepora inferred by Bayesian inference. The clade support values are Bayesian posterior probabilities (≥ 0.7) and 31 32 672 maximum likelihood bootstrap replicates (≥ 70). Colours denote each distinct molecular lineage as reported in the 33 34 673 figure box. 35 36 674 Fig. 5 Cnidome of the polyp (a-o) and eumedusoid stages (p-w) of the three Millepora species analyzed in 37 38 675 this work. a Large and b small undischarged stenoteles, c macrobasic apotrichous mastigophores undischarged 39 40 676 capsule, d detail of the discharged distal portion of the shaft, and e discharged capsule in M. dichotoma. f Large and 41 42 677 g small stenoteles, h macrobasic apotrichous mastigophores undischarged capsule, i detail of the discharged distal 43 44 678 portion of the shaft, and j discharged capsule in M. platyphylla. k Large and l small stenoteles, m macrobasic 45 46 679 apotrichous euryteles undischarged capsule, n detail of the discharged distal portion of the shaft, and j discharged 47 48 680 capsule in M. exaesa. p Nematocyst distribution, q stenoteles, and r microbasic mastigophores in the eumedusoid of 49 50 681 M. dichotoma. s Nematocyst distribution, t stenoteles, and u microbasic mastigophores in the eumedusoid of M. 51 52 682 platyphylla. v Nematocyst distribution, and w stenoteles in the eumedusoid of M. exaesa. Black and white dots in 53 54 683 Figs. p, s, v represent stenoteles and microbasic mastigophores, respectively. Scale bars: 10 µm 55 56 684 Fig. 6 Discriminant analysis plot for the analyzed pore characters of Millepora species from Red Sea. 57 58 685 Millepora dichotoma (blue), M. platyphylla (red), and M. exaesa (green). 59 60 61 62 21 63 64 65 1 2 3 4 686 5 6 687 Supporting Information 7 8 688 Fig. S1 a COI and b 16S rDNA phylogeny reconstructions of Millepora inferred by Bayesian inference. The 9 10 689 two trees are identical to the ones presented in Fig. 4 but the graphics is different: each single tip represent a single 11 12 690 sequence. The clade support values are Bayesian posterior probabilities (≥ 0.7) and maximum likelihood bootstrap 13 14 691 replicates (≥ 70). Colours denote each distinct molecular lineage as reported in the figure box. 15 16 692 Table S1 List of Millepora samples collected from the Red Sea for this study, including voucher number, 17 18 693 depth, site, latitude, longitude, GenBank accession numbers of the two mitochondrial genes (COI and 16S rDNA) 19 20 694 Table S2 Measurements of pore characters analyzed in this study for the three Red Sea Millepora species 21 22 695 Table S3 Measurements of nematocyst characters of both polyp and medusa stages analyzed in this study 23 24 696 for the three Red Sea Millepora species 25 26 697 Table S4 Wilks’ lambda test for verifying differences among Millepora species with pore character 27 28 measurements using the DFA 29 698 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 22 63 64 65 Figure 1 Click here to access/download;Figure;Figure 1.pdf

34° 36° 38° 40° 42° 44°

30° Eilat

28° Duba

26° Al Wajh

24° Yanbu

22° Thuwal

20° Al Lith

18°

16°

14°

N 12°

0 Km 500 Figure 2 Click here to access/download;Figure;Figure 2_rev.pdf

a b c

d e f

g h i Figure 3 Click here to access/download;Figure;Figure 3_rev.pdf

a 4 b 5

6

3 1 2

c d Figure 4 Click here to access/download;Figure;Figure 4_rev.pdf

a b

140 samples 140 samples

2 2

missing haplotype missing haplotype 76 mutations

52 mutations

18 mutations 26 mutations

14 mutations

0.99/90 IX 1/94 VIIIb c VIIIa d ATLANTIC 1/100 0.99/90 XII ATLANTIC 1/97 X 0.75/72 0.78/100 XI 1/90 XIII 1/100 1/92 1/100 XIV

1/100 III III 1/93

INDO-PACIFIC 1/100 0.97/96 1/91 IVa 1/97 Va VIII

1/100 INDO-PACIFIC 0.97/90 II II 1/82 1/79 1/100

1/99 IVa 1/97 Va 1/98 VI -/79 0.79/73

0.83/76

I I 1/99 1/100

Solanderia secunda 0.9 Solanderia secunda 0.8

Millepora dichotoma (Red Sea) - I Millepora platyphylla (Red Sea) - II Millepora exaesa (Red Sea) - III Millepora platyphylla (Indo-Pacific) - IV Millepora dichotoma (Pacific) - V Millepora tenera (La Reunion) - VI Millepora exaesa (La Reunion) - VII Millepora alcicornis (Atlantic Ocean) - VIII Millepora striata (Atlantic Ocean) - IX Millepora complanata (Atlantic Ocean) - X Millepora squarrosa (Atlantic Ocean) - XI Millepora braziliensis (Atlantic Ocean) - XII Millepora laboreli (Atlantic Ocean) - XIII Millepora nitida (Atlantic Ocean) - XIV Figure 5 polyp Click here to access/download;Figure;Figureeumedusoid 5_rev.pdf a c e f

b d g h M. dichotoma

i k m n

j l o p M. platyphylla

q s u v

r t w M. exaesa Figure 6 Click here to access/download;Figure;Figure 6_rev.pdf

5.0

2.5

0.0 DF 2

-2.5

-5.0

-5.0 -2.5 0.0 2.5 5.0

DF 1 Table Click here to access/download;Table;Tables.docx

Table 1 Pairwise comparisons of genetic distance values (p-distance) within and between Millepora species based on COI. Interspecific pairwise comparisons of genetic distance are in bold, intraspecific genetic (± standard deviations) distances are indicated along the diagonal, and standard deviations are reported on the upper right hand portions for each set of comparisons

M. M. M. dichoto M. M. exaes dichoto ma platyphyl platyphyl a M. M. M. M. ma Red Pacific la Red la Pacific Red striat alcicorn complana squarro Sea Ocean Sea Ocean Sea a is ta sa M. dichotom 0.84 ± a Red Sea 0.27 1.22 1.32 1.35 1.86 1.79 1.75 1.80 1.82 M. dichotom a Pacific 1.13 ± Ocean 9.17 0.4 1.35 1.33 1.89 1.70 1.67 1.69 1.79 M. platyphyll 0.2 ± a Red Sea 9.08 9.76 0.04 1.34 1.87 1.80 1.78 1.75 1.94 M. platyphyll a Pacific 3.12 ± Ocean 10.61 10.39 10.78 0.68 1.93 1.70 1.69 1.71 1.83 0.38 M. exaesa ± Red Sea 18.11 19.48 18.77 0.19 0.16 1.75 1.76 1.76 1.96 0.26 ± M. striata 16.83 17.53 17.95 0.17 16.93 0.25 0.42 0.85 1.73 M. 0.63 ± alcicornis 16.81 16.85 0.18 0.18 17.10 1.15 0.27 0.82 1.73 M. complana 0.52 ± ta 17.01 16.75 16.57 17.43 17.58 4.03 3.90 0.21 1.80 M. 0.26 ± squarrosa 17.00 16.95 19.87 16.67 19.92 16.36 16.45 17.53 0.24

Table 2 Pairwise comparisons of genetic distance values (p-distance) within and between Millepora species based on 16S rDNA. Interspecific pairwise comparisons of genetic distance are in bold, intraspecific genetic (± standard deviations) distances are indicated along the diagonal, and standard deviations are reported on the upper right hand portions for each set of comparisons. n.c. = not calculable

M. exae M. M. M. M. sa M. dichoto platyph platyph exae Indi dichoto ma M. ylla ylla M. M. M. M. sa an ma Red Pacific tene Indo- Red alcicor brazilie labor niti Red Oce Sea Ocean ra Pacific Sea nis nsis eli da Sea an M. dichoto ma Red 0.33 ± Sea 0.08 0.81 0.83 0.86 0.97 1.52 1.59 1.66 1.70 1.44 1.52 M. dichoto ma Pacific 0.38 ± Ocean 4.30 0.20 0.38 0.82 0.88 1.47 1.50 1.49 1.61 1.34 1.44 2.09 M. ± tenera 4.90 1.74 0.73 0.82 0.97 1.46 1.53 1.55 1.65 1.46 1.50 M. platyphy lla Indo- 0.27 ± Pacific 4.42 4.27 4.88 0.13 0.92 1.45 1.39 1.47 1.54 1.48 1.47 M. platyphy lla Red 0.14 ± Sea 5.76 5.62 6.41 5.40 0.05 1.55 1.56 1.60 1.70 1.45 1.52 M. alcicorn 12.9 0.93 ± is 12.13 13.21 8 11.91 13.13 0.25 1.25 1.19 1.34 1.69 1.71 M. brazilie 15.4 1.12 ± nsis 15.69 15.52 9 14.43 15.54 12.86 0.31 0.63 0.83 1.64 1.50 M. 16.3 laboreli 16.69 16.03 5 15.54 16.17 12.98 3.28 0 ± 0 0.87 1.65 1.55 0.77 M. 17.2 ± nitida 16.99 17.06 9 16.45 16.81 14.51 5.16 4.80 0.27 1.68 1.56 M. 0.1 exaesa 15.5 18.2 6± Red Sea 13.67 14.72 1 14.65 14.27 15.16 16.40 17.52 1 0.03 1.01 M. exaesa Indian 14.1 17.7 Ocean 13.28 14.01 6 13.69 13.70 14.40 15.36 16.73 0 5.14 n.c.

Table 3 List of pore and nematocyst characters analyzed in this study. Mean values, standard deviations, and ranges of these features obtained from the three Red Sea Millepora species

Morphological feature M. dichotoma M. platyphylla M. exaesa Skeleton gastropore diameter 0.12 ± 0.02 (0.09 – 0.17) 0.15 ± 0.02 (0.11 – 0.21) 0.14 ± 0.02 (0.1 – 0.2) dactylopore diameter 0.06 ± 0.01 (0.04 – 0.08) 0.04 ± 0.01 (0.02 – 0.07) 0.06 ± 0.01 (0.04 – 0.08) distance between 0.69 ± 0.17 (0.35 – 1.3) 0.97 ± 0.17 (0.68 – 1.46) 0.68 ± 0.12 (0.38 – 1.02) gastropores distance between 0.3 ± 0.06 (0.19 – 0.48) 0.16 ± 0.04 (0.09 – 0.29) 0.15 ± 0.05 (0.05 – 0.32) dactylopores distance from gastropore 0.15 ± 0.04 (0.1 – 0.27) 0.11 ± 0.02 (0.06 – 0.16) 0.16 ± 0.03 (0.05 – 0.22) to nearest dactylopore number of dactylopore 3.9 ± 0.89 (1 – 6) 6.06 ± 0.96 (4 – 8) 6.93 ± 0.84 (5 – 9) per gastropore gastropore density (0.25 5.71 ± 1.6 (3 – 11) 5.63 ± 2.22 (2 – 14) 5.27 ± 1.54 (2 – 9) mm2) dactylopore density (0.25 24.39 ± 5.39 (13 – 39) 43.71 ± 7.97 (24 – 61) 54.86 ± 10.7 (35 – 78) mm2) Polyp large stenotele length 16.9 1.53 (14.4 – 20.33) 16.3 ± 1.33 (14.24 – 18.92 ± 2.84 (14.74 – 19.11) 25.51) large stenotele width 13.3 ± 0.98 (11.54 – 13.01 ± 1.04 (10.82 – 15.6 ± 2.04 (12.49 – 14.88) 14.6) 20.21) small stenotele length 7.72 ± 0.67 (6.67 – 9.85) 9.16 ± 0.84 (7.93 – 11.02) 10.03 ± 0.75 (8.28 – 12.85) small stenotele width 6.02 ± 0.63 (5.19 - 8) 7.04 ± 0.95 (5.35 – 9.4) 7.69 ± 0.68 (5.65 – 9.43) heteroneme type mastigophore mastigophore eurytele heteroneme length 27.91 ± 1.4 (26.06 – 28.88 ± 1.67 (26.78 – 31.55 ± 2.06 (27.46 – 32.95) 34.67) 35.29) heteroneme width 22.66 ± 1.27 (20.28 – 22.14 ± 1.32 (19.97 – 22.91 ± 1.67 (18.39 – 25.96) 25.7) 25.97) heteroneme shaft 162.33 ± 15.72 (140.38 – 410.8 ± 15.91 (394.17 – 331.34 ± 19.58 (307.21 – 190.25) 445.15) 367.15) Eumedusoid stenotele length 14.42 ± 0.71 (13.03 – 14.37 ± 0.9 (12.76 – 14.78 ± 0.92 (13.14 – 16.08) 16.47) 17.35) stenotele width 11.54 ± 0.73 (9.48 – 11.99 ± 0.66 (10.47 – 12.42 ± 0.81 (9.98 – 12.76) 13.33) 14.67) mastigophore length 11.82 ± 0.83 (10.16 – 10.83 ± 0.51 (9.3 – 11.9) absent 14.13) mastigophore width 9.56 ± 0.73 (7.6 – 11.21) 9.06 ± 0.33 (8.27 – 9.87) absent mastigophore area 88.89 ± 10.56 (64.53 – 77.16 ± 5.22 (60.4 – absent 107.2) 89.15) mastigophore ratio 1.24 ± 0.1 (1.04 – 1.49) 1.2 ± 0.06 (1.05 – 1.4) absent medusa size length 740.45 ± 68.32 (645.4 – 765.01 ± 40.15 (717.38 – 648.58 ± 47.44 (592.03 – 849.58) 822.85) 730.13) medusa size width 765.72 ± 91.18 (679.86 – 749.65 ± 32.99 (718.31 – 647.59 ± 40.72 (583 – 906.01) 791.22) 703.14) medusa size ratio 0.97 ± 0.05 (0.91 – 1.04) 1.02 ± 0.03 (0.99 – 1.06) 1.01 ± 0.11 (0.87 – 1.18)

Table 4 One-way ANOVA results for differences in the analyzed pore characters among the three species of Millepora from the Red Sea

Sum of Mean Partial df F Sign. Pore character squares square eta ƞ2 gastropore diameter 0,063 2 0,031 77,011 0,000 0,341 dactylopore diameter 0,029 2 0,014 206,020 0,000 0,581 distance between gastropores 5,463 2 2,731 112,733 0,000 0,432 distance between dactylopore 0,150 2 0,075 240,462 0,000 0,618 distance from gastropore to 0,018 2 0,009 71,856 nearest dactylopore 0,000 0,326 number of dactylopore per 486,780 2 243,390 299,759 gastropore 0,000 0,669 dactylopore density 47533,527 2 23766,763 343,049 0,000 0,698

Table 5 Contribution of each pore character to the discriminant functions for the three species of Millepora from the Red Sea. * Largest correlation between each variable and any DF

Pore character DF1 DF2 dactylopore density ,579* 0,317 number of dactylopore per gastropore ,558* 0,229 distance between dactylopore -,520* -0,014 distance between gastropore ,281* -0,122 dactylopore diameter -0,229 ,597* gastropore diameter 0,175 -,438* distance from gastropore to nearest dactylopore, -0,118 ,365*

Table 6 One-way ANOVA results for differences in the polyp and eumedusoid large stenoteles among the three species of Millepora from the Red Sea

Character Medusa large stenotele Polyp large stenotele Sum of Mean Sum of Mean df F Significance df F Significance squares squares squares squares

Lenght 4,851 2 2,426 3,238 0,042 119,627 2 59,813 14,444 0,000 Width 16,492 2 8,246 14,975 0,000 127,295 2 63,647 30,127 0,000

Area 4282,191 2 2141,095 9,258 0,000 86947,724 2 43473,862 22,059 0,000

Ratio 0,083 2 0,042 9,985 0,000 0,059 2 0,030 6,031 0,003

Table 7 Kruskal-Wallis test results for differences in the polyp small stenoteles and heteronemes among the three species of Millepora from the Red Sea

Character Small Stenotele Heteroneme H Kruskal-Wallis df Significance H Kruskal-Wallis df Significance Length 99,577 2 0,000 69,526 2 0,000 Width 71,654 2 0,000 12,026 2 0,002 Area 88,231 2 0,000 31,020 2 0,000 Ratio 2,319 2 0,314 87,219 2 0,000

Figure S1

Click here to access/download Supplementary Material Figure S1_rev.pdf Table S1

Click here to access/download Supplementary Material Table S1_rev.xlsx Table S2

Click here to access/download Supplementary Material Table S2.xlsx Table S3

Click here to access/download Supplementary Material Table S3_rev.xlsx Table S4

Click here to access/download Supplementary Material Table S4.docx Revised manuscript with track Changes

1 An integrated morpho-molecular approach to delineate species boundaries of Millepora from the Red Sea

2

3 Roberto Arrigoni1, Davide Maggioni2,3, Simone Montano2,3, Bert W. Hoeksema4, Davide Seveso2,3, Tom

4 Shlesinger5, Tullia Isotta Terraneo1,6, Matthew D. Tietbohl1, Michael L. Berumen1

5

6 Corresponding author: Roberto Arrigoni, [email protected]

7 1Red Sea Research Center, Division of Biological and Environmental Science and Engineering, King Abdullah

8 University of Science and Technology, Thuwal 23955-6900, Saudi Arabia

9 2Dipartimento di Scienze dell’Ambiente e del Territorio (DISAT), Università degli Studi di Milano-Bicocca, Piazza

10 della Scienza 1, Milano 20126, Italy

11 3Marine Research and High Education (MaRHE) Center, Faafu Magoodhoo 12030, Republic of the Maldives

12 4Department of Marine ZoologyTaxonomy and Systematics Group, Naturalis Biodiversity Center, P.O. Box 9517,

13 2300, RA Leiden, the Netherlands

14 5School of Zoology, The George S. Wise Faculty of Life Sciences, Tel-Aviv University, Tel-Aviv 69978, Israel

15 6College of Marine and Environmental Science, James Cook University, Townsville, QLD 4811, Australia

16

17 Keywords Fire corals ∙ Phylogeny ∙ Pore ∙ Nematocyst ∙ Eumedusoid

18

19 Abstract Fire corals of Tthe hydrocoral genus Millepora provides an important ecological role as

20 framework builders of coral reefs in the Indo-Pacific and the Atlantic. Recent works have demonstrated the

21 incongruence between molecular data and the traditional taxonomy of fire coralsMillepora spp. based on overall

22 skeleton growth form and pores. In an attempt to establish a reliable and standardized approach for defining species

23 boundaries in Millepora, we focused on fire coralsthose from the Red Sea. In this region, three species are currently

24 recognized have been traditionally identified: the fan-shaped branching M. dichotoma, the blade-like M. platyphylla,

25 and the massive/encrusting M. exaesa. A total of 412 colonies were collected from six localities in the Red Sea. Two

26 mitochondrial marker genes (COI and 16S rDNA) were sequenced to obtain phylogeny reconstructions and

27 haplotype networks. Eight morphological traits of pores and the nematocysts of both polyp and eumedusoid stages

28 were measured to determine if significant morphological differences occur among the three species. Both markers

1

29 clearly resolved M. dichotoma, M. platyphylla, and M. exaesa as distinct, monophyletic lineages in the Red Sea.

30 Nevertheless, they also revealed deep genetic breaks with Southwestern Indian Ocean populations of the three

31 species. In the Red Sea, the three species were further distinguished on the based onof their pore and the nematocyst

32 features. The A discriminant analyseis revealed dactylopore density, number of dactylopores per gastropore,

33 dactylopore distance, and gastropore diameter as the most informative discriminative characters. The heteronemes,

34 the large and small stenoteles of polyps, and the distribution of mastigophores of eumedusoids also showed also

35 significantly interspecific differences. An integrated morpho-molecular approach proved to be decisiveis helpful in

36 defining species boundaries of Millepora supported by a combination of pore and nematocyst characters which may

37 be phylogenetically informative.

38

39 Introduction

40 Species of the hydrocoral genus Millepora are popularly known as fire corals (Pourtales 1877). They are

41 colonial organisms that secrete calcium carbonate in order to building persistentmassive calcareous skeletons and as

42 such, providplaying an important ecological role as framework builders of coral reefs (Lewis 2006). These

43 hydrocorals are among the most relevant reef builders several in shallow-water tropical seas, second only to

44 scleractinians (Lewis 1989;, Edmunds 1999;, Smith et al. 2014). Millepora occurs in tropical and sub-tropical coral

45 reefs of both the Atlantic and the Indo-Pacific and its depth distribution is restricted from less than 1 m deep to about

46 50 m deep because of the obligate symbiosis with zooxanthellae of the genus Symbiodinium which require sunlight

47 for photosynthesis (Boschma 1948, 1956;, Cairns et al. 1999).

48 Despite the wide geographical distribution and the relevant ecological role, Millepora has been scarcely

49 been investigated in coral reef studies and, in particular, its taxonomy and systematics are controversial and long-

50 standing much debated (Hickson 1898, 1899;, Crossland 1948;, Boschma 1948a, 1966;, Moshchenko 1992, 1997;,

51 Razak and Hoeksema 2003; de Souza et al. 2017). Fire corals have been historically identified based on

52 morphological features such as colony growth form, skeletal pores, and ampullae (Klunzinger 1879;, Quelch 1884;,

53 Hickson 1898;, Boschma 1948b, 1949, 1956). Nevertheless, all these morphological characters are extremely

54 intraspecifically variable within a single species and even in a single specimen as a result of morphological plasticity

55 due to ecophenotypic variation variationbility and morphological plasticity (de Weerdt et al. 1981, 1984;, Boschma

56 1948a;, Moshchenko 1994, 1995a, 1996a;, Amaral et al. 1997;, Dubé et al. 2017a, b). Although growth form has

2

57 been traditionally used as the main diagnostic character for species identification (Forskål 1775;, Ehrenberg 1834;,

58 Duchassaing and Michelotti 1860;, Klunzinger 1879;, Hickson 1898;, Crossland 1941;, Boschma 1948a, b), this

59 feature has been showndemonstrated to be highly variableplastic. For example, Millepora platyphylla Hemprich and

60 Ehrenberg, 1834 in Vietnam displayed distinct morphotypes associated to different reef zones and the occurrence of

61 specific different colony growth forms was hypothesized to be related to factors such as radiation availability, water

62 movement, and settling suspended matter (Moshchenko 1995a). A complicating factor herein is that juvenile

63 Millepora corals usually start with an encrusting growth form (see e.g. Fig. 2g, Hoeksema et al. 20174: Fig. 1),

64 which makes it difficult to distinguish species at early stage. Similarly, pore characters are known to noticeablye

65 vary in response toupon the rapidity of growth and the amount of light on the skeleton surface and some taxonomists

66 have suggested to discardavoid their use in Millepora taxonomy (Hickson 1898, 1899;, Boschma 1948a, b).

67 However, Moshchenko (1994, 1995b, 1996b) proposed a quantitative approach for the evaluation analysis of pore

68 structures and was able to tell M. platyphylla apart from the branching Millepora species of the Indian Ocean. As a

69 consequence of this morphological variation, traditional morphology-based species identification since the first valid

70 description of a fire coral by Linnaeus (1758) generated more than 50 nominal species of Millepora, nine of which

71 nine are now considered valid in the Indo-Pacific and six in the Atlantic (de Weerdt 1989, Razak and Hoeksema

72 2003;, Amaral et al. 2008;, de Souza et al. 2017).

73 Recent outcomes of molecular phylogenetic analyses and DNA taxonomy have provided advantages for the

74 understanding of Millepora systematics and connectivity (López et al. 2015;, Hoeksema et al. 2014;, de Souza et al.

75 2017;, Dubé et al. 2017c). The use of genetics has elucidated the identification and the geographic origin of fire

76 corals in some isolated and remoted islands of the central and eastern Atlantic Ocean, including for example the

77 Canaries, Capeo Verde Islands, and Ascension Island (López et al. 2015;, Hoeksema et al. 20142017). Moreover,

78 the three endemic species of the Brazilian province, i.e. Millepora braziliensis Veriill, 1868, Millepora nitida

79 Veriill, 1868, and Millepora laboreli Amaral, 2008, have beenwere successfully resolved based on molecular data

80 although the traditional morphological characters of pores were not sufficient to discriminate these taxa (de Souza et

81 al. 2017). Nevertheless, other studies have showed shown the presence of several instances of discordances between

82 morphology-based taxonomy and genetics data suggesting the need to combine new molecular and morphological

83 data for appropriate the definitions of species boundaries (Meroz-Fine et al. 2003;, Luiz-RamosRuiz-Ramos et al.

84 2014;, Takama et al. 2018). For example, Luiz-RamosRuiz-Ramos et al. (2014) genetically characterized four

3

85 Caribbean Millepora representatives, i.e. Millepora alcicornis Linnaeus, 1758, Millepora complanata Lamarck,

86 1816, Millepora squarrosa Lamarck, 1816, and Millepora striata Duchassaing and Michelotti, 1864, and recovered

87 the latter two species as distinct molecular lineages while the others formers two taxa represented form an

88 unresolved species complex. Furthermore, a recent study from Japan demonstrated that the three branching species

89 Millepora dichotoma Forskål, 1775, Millepora intricata Milne Edwards, 1860, and Millepora tenera Boschma, 1949

90 were receovered inrepresented a single molecular clade while a previously unknown lineage of an encrusting

91 Millepora form was discovered, which overgrows living scleractinianhard corals (Takama et al. 2018).

92 Fire corals are abundant and common throughout the entire Red Sea, where they represent an ecologically

93 important taxa group in shallow-water reef habitats. Three species are supposed to occur in the Red Sea (Boschma

94 1948), i.e. Millepora dichotoma, Millepora platyphylla, and Millepora exaesa Forskål, 1775. The reef crest of the

95 Red Sea reef systems is usually dominated by conspicuous colonies of M. dichotoma and M. platyphylla in addition

96 to many scleractinians coral species (Loya 1972, 1976;, Benayahu and Loya 1977;, Perkol-Finkel and Benayahu

97 2004). Consequently, Loya and Slobodkin (1971) introduced the concept of “Millepora zone” to identify the reef

98 portion at a depths of 0.2 to 3 meters. Because of this abundance, early taxonomists ofn the Red Sea coral fauna

99 have focused their attention on fire corals and provided detailed descriptions of their morphological characters. and,

100 Nnotably, all the three aforementioned species have were originally been described from the Red Sea (Forskål

101 1775;, Ehrenberg 1834;, Klunzinger 1879;, Crossland 1941). The morphological plasticity variation and the colony

102 development of both M. dichotoma and M. exaesa from the northern tip of the Red Sea (the Gulf of Eilat and Aqaba)

103 have beenwere examined and described by Vago et al. (1997) and Meroz-Fine et al. (2003). Although these authors

104 considered M. exaesa as an ecomorph of M. dichotoma, they found significant differences in molecular data (rDNA

105 internal transcribed spacers), nematocyst capsule size, and the colony growth form. Hence, they and suggested that

106 the two morphs represented two distinct entities (Meroz-Fine et al. 2003). Nevertheless, these works did not include

107 M. platyphylla and evaluated fire corals only from the northern Gulf of Eilat and Aqaba, an isolated partregion of the

108 Red Sea (Klausewitz 1989;, Wolf-Vecht et al. 1992;, DiBattista et al. 2016a, b). A complicating factor herein is that

109 juvenile Millepora corals usually start with an encrusting growth form (see e.g. Fig. 2g, Hoeksema et al. 2014: Fig.

110 1), which makes it difficult to distinguish species at early stage.

111 With the aim to establish a reliable and standardized approach for defining species boundaries definition in

112 Millepora, this study uses a combination of molecular and morphological datasets in order to address the degree of

4

113 concordance between genetic and morphological variation of fire corals from the Red Sea. Two mitochondrial genes

114 and two distinct levels sets of morphological characters were investigated and their efficacy to detectisentangle

115 species boundaries was discussed in light of the obtainedour results.

116

117 Materials and methods

118 Sample collection

119 A total of 412 specimens of Millepora were collected while SCUBA diving between 1 and 20 m depth at

120 six sites localities along the Red Sea (Fig. 1), consisting of 152 colonies of M. dichotoma, 128 of M. platyphylla, and

121 132 of M. exaesa (Table S1). Specimens were identified to species level following original descriptions (Forskål

122 1775, Ehrenberg 1834) and the taxonomic revision of Indonesian Millepora by Razak and Hoeksema (2003). The

123 two main diagnostic morphological characters used for identification were the colony growth form and the

124 cyclosystem arrangement. During the sampling process, we maximized the morphological variability of Millepora

125 includeding all possible morphotypes per species (Fig. 2). To minimize the likelihood of sampling clones, colonies

126 ascribed to the same species were collected at a minimum distance of 5 m (following Ruiz-Ramos et al. 2014). Each

127 coral specimen was photographed underwater (Fig. 2), collected with hammer and chisel, tagged, and approximately

128 2 cm23 of the collected colony was sampled and fixed in 95% ethanol. The remaining material was immersed in

129 sodium hypochlorite for 48 hours to remove all tissues, rinsed in freshwater, and air-dried for identification and

130 microscope observations of the cleaned corallum. Specimens were deposited at King Abdullah University of Science

131 and Technology (KAUST, Thuwal, Saudi Arabia) and at Steinhardt Museum of Nnatural Hhistory, Tel Aviv

132 University (Tel Aviv, Israel).

133

134 DNA extraction, amplification and sequencing

135 DNA extractions were carried out with the DNeasy Blood and Tissue kit (Qiagen Inc., Hilden, Germany).

136 A portion of two mitochondrial genes, the cytochrome c oxidase subunit I (COI) and the large rRNA subunit (16S

137 rDNA), were obtained from all the collected samples. PCR amplifications of COI were performed using the

138 previously published primers COIF and COIR (Schweinsberg et al. 2016) and the protocol from López et al. (2015),

139 while 16S rDNA was amplified using the primers SHA and SHB (Cunningham and Buss 1993) and the protocol

140 from de Souza et al. (2017). All PCR products were purified with Illustra ExoStar (GE Healthcare,

5

141 Buckinghamshire, UK) and directly sequenced in forward and reverse directions with the described primers using an

142 ABI 3130xl Genetic Analyzer (Applied Biosystems, Carlsbad, CA, USA). Sequences obtained in this study were

143 deposited at GenBank, and accession numbers are listed in Table S1 (MH824796-MH825619).

144

145 Genetic data analyses

146 Chromatograms of the forward and reverse DNA strands were assembled and edited using Sequencher 5.3

147 (Gene Codes Corp., Ann Arbor, MI, USA). Sequences were aligned with MAFFT 7.397 (Katoh and Standley 2013)

148 and the iterative refinement method E-INS-i. General statistics concerning the obtained sequences and the variability

149 of the two employed markers were calculated with DnaSP 6.11.1 (Librado and Rozas 2009). Intra- and interspecific

150 genetic distances were calculated as p-distance based on the separated COI and 16S rDNA datasets using DnaSP

151 6.11.1, and variance was estimated with 1000 bootstrap replicates (Tables 1 and 2).

152 In order to define species boundaries of the Red Sea Millepora, a median-joining network (Bandelt et al.

153 1999) for each mitochondrial gene was constructed including exclusively the newly-obtained sequences from the

154 Red Sea using Network 5.0 (http://www.fluxus-technology.com). The median-joining method uses a maximum

155 parsimony approach to search for all the shortest phylogenetic trees of given dataset (Bandelt et al. 1999).

156 All available COI and 16S rDNA sequences of Millepora were downloaded from GenBank and included

157 inused for our phylogenetic analyses. Because these sequences derive from different sources and include several

158 species from different localities, we performed separate phylogeny reconstructions for each mitochondrial locus.

159 Solanderia secunda was selected as outgroup for both phylogenetic analyses , according to previous hypotheses

160 (Nawrocki et al. 2010;, Maggioni et al. 2017a, 2018). Two phylogenetic tree optimality criteria were employed,

161 namely Bayesian inference (BI) using MrBayes (Ronquist et al. 2012) and maximum likelihood (ML) using

162 RAxML (Stamatakis 2014). Both BI and ML analyses were run on the CIPRES server (Miller et al. 2010). Prior to

163 the phylogenetic analyses, we determined the best partition scheme and nucleotide substitution models using

164 PartitionFinder 1.1.1 (Lanfear et al. 2012). We used unlinked branch lengths, the greedy search algorithm for

165 nucleotide sequence, and considered four partitions: 16S rDNA and the three codon positions of COI. Partitioning

166 scheme comparison was performed using the corrected Akaike Information Criterion (AIC) and the Bayesian

167 Information Criterion (BIC), testing the influence on the tree topology. PartitionFinder 1.1.1 selected for RAxML

168 the evolutionary model GTR+G for all the four partitions and for MrBayes the models of evolution HKY+G for 16S

6

169 rDNA, K80+G for COI_pos1, F81 for COI_pos2, and HKY+G for COI_pos3. BI runs were performed using four

170 Markov Chain Monte Carlo (MCMC) chains for 10 million generations, saving a tree every 1,000th generation. The

171 tree searches were stopped when all parameters reached the stationarity for effective sampling size and unimodal

172 posterior distribution using Tracer 1.6 (Rambaut et al. 2014). The first 25% trees sampled were discarded as burn-in

173 following indications by Tracer. ML topology was obtained under the default parameters shown on the CIPRES

174 server with a multiparametric bootstrap analyses composed of 1,000 bootstrap replicates.

175

176 Morphological examination

177 Two levels sets of morphological characters features were analyzed in this study, i.e. skeletal pores and

178 polyp and eumedusoid nematocysts. For the examination of skeletal pores, a total of eight features characters that

179 have been previously used for taxonomic studies on of Millepora were selected (Boschma 1948a;, de Weerdt 1984;,

180 de Weerdt and Glynn 1991;, Moshchenko 1995b, 1996b;, Razak and Hoeksema 2003) (Fig 3): (1) gastropore

181 diameter, (2) dactylopore diameter, (3) distance between gastropores, (4) distance between dactylopores, (5)

182 distance from gastropore to nearest dactylopore, (6) number of dactylopores per gastropore, (7) gastropore density,

183 and (8) dactylopore density. Because M. exaesa is characterized by absent or limited cyclosystem arrangement, for

184 the analyses we selected only colonies of this species showing a limited but recognizable cyclosytem arrangement.

185 A total of ten colonies selected from the ones that were genetically characterized were analysed for each of the three

186 Red Sea Millepora species were analyzed and a total of ten measurements per colony were taken for each single trait

187 (Table S2). Photographs of the skeleton surface were obtained at x50 magnification with a Leica IC80 HD

188 microscope and its integrated standalone digital camera. On the basis of the digital images with a reference scale, the

189 eight skeletal variables were measure using ImageJ 1.49 (Rasband 1997).

190 For the examination of polyp nematocysts, five coloniesfragments of each Millepora species

191 choosenchosen from the ones that were molecularly investigated were decalcified using 10% formic acid for

192 approximately 24 hours. Soft tissues were then washed in seawater, observed with a Nikon Eclipse E600 compound

193 microscope, photographed with its integrated camera, and measured using the NIS-Elements Viewer 4.30. For the

194 examination of eumedusoid nematocysts, newly released eumedusoid were collected and anesthetized with menthol

195 crystals. Thereafter, nematocysts were observed with the compound microscope in order to assess their distribution

196 and they were measured. Nematocysts were identified according to Bouillon et al. (2006) and 30-90 undischarged

7

197 capsules were measured for each type, life stage, and species. A total of six nematocyst features were analyzed,

198 namely (1) polyp large stenotele, (2) polyp small stenotele, (3) polyp macrobasic apotrichous eurytele, (4) polyp

199 macrobasic apotrichous mastigophore, (5) eumedusoid microbasic mastigophore, and (6) eumedusoid stenotele. For

푙푒푛𝑔푡ℎ 푤𝑖푑푡ℎ 200 each featurecapsule, four measurements were obtained, namely: length, width, area ( 푥 푥 휋), and ratio 2 2

푙푒푛𝑔푡ℎ 201 ( ), for a total of 24 analyzed characters. Additionally, polyp heteronemes (euryteles and mastigophores) were 푤𝑖푑푡ℎ

202 discharged adding a drop of sodium hypochlorite to the tissues, and the length of the exploded shaft was measured

203 (Table S3).

204

205 Statistical analyses

206 Morphological variables were log+1 transformed when the normality distribution was violated and, after

207 addressing assumptions, a sub-set of the eight pore characters (all but gastropore density) was used in a multivariate

208 analysis of variance (MANOVA) to test for the presence of differences in pore characters composition among M.

209 dichotoma, M. exaesa, and M. platyphylla from the Red Sea. Hereafter, a multiple univariate analysis of variance

210 (ANOVA) was run to determine which characters showed differences among the three investigated species. Wilks’

211 lambda criterion was used to test for group differences in the MANOVA. A discriminant function analysis (DFA)

212 was used to examine the pore characters in discriminating individuals among species and to investigate whether the

213 measured pore characterfeatures could be used to classify samples into their original group. A classification with

214 cross-validation was carried out to assign individuals to their original group, and the scatterplots of the first two

215 discriminant scores were drawn to depict the separation among the species distribution on the graph.

216 Furthermore, the nematocyst measurements were investigated in order to detect statistically significant

217 differences among the three investigated species of fire corals. In particular, for the large and small stenoteles, which

218 are found in both the polyp and the eumedusoid stages, and for the two types of heteronemes, i.e. euryteles and

219 mastigophores, which are found only in the polyp stage, the following four measurements were taken into account:

220 length, width, area, and ratio. The measurements were taken by moving in the same direction across the slide in

221 order to measure each nematocyst only once within each specimen in order to avoid pseudoreplication. Statistical

222 differences between the nematocysts of the three fire coral species were tested using the Kruskal-Wallis test when

223 we violated the assumption of normality, and ANOVA when the normality distribution and the homoscedasticity

224 were satisfied.

8

225 All the statistical analyses were carried out in PERMANOVA+ for PRIMER 7 (PRIMER-E Ltd, Plymouth,

226 UK) and SPSS 25 (SPSS, Chicago, IL, USA). All data report mean ± dev. St. otherwise stated.

227

228 Results

229 Species identification

230 All collected samplesSpecimens were identified to species level following original descriptions (Forskål

231 1775, Ehrenberg 1834) and the taxonomic revision of Indonesian Millepora by Razak and Hoeksema (2003). The

232 two main diagnostic morphological characters used for species identification were the colony growth form and the

233 cyclosystem arrangement. Based on the colony growth form and the cyclosystem arrangementhese two

234 charactersfeatures, the sampled specimens were grouped in three species: M. dichotoma showing fan-shaped

235 branching growth form with distinct cyclosystems (Figs. 2a-c, 3b), M. platyphylla showing a blade-like growth form

236 with distinct cyclosystems on a flat surface (Figs. 2d-f, 3c), and M. exaesa showing massive or sub-massive growth

237 form with absent or limited cyclosystem arrangement (Figs. 2g-i, 3d).

238

239 Molecular results

240 Overall, COI and 16S rDNA sequences were obtained from all 412 collected fire coral colonies (Table S1).

241 The COI dataset including exclusively our Red Sea Millepora samples resulted in 426 bp with 104 variables

242 nucleotides and identified 33 haplotypes. The 16S rDNA alignment included 597 positions with 113 variable

243 nucleotides for a total of 37 haplotypes. Both mitochondrial networks clearly showed the presence of three main

244 clusters of haplotypes (Figs. 4a-b, Fig. S1) that corresponded tocorresponding to the three groups of specimens

245 species identified based on growth colony form and cyclosystem arrangementassigned to, i.e. M. dichotoma, M.

246 platyphylla, and or M. exaesa based on growth colony form and cyclosystem arrangement. Individuals ascribed to

247 the same species were very closely related, being separated between each other by a total number of mutations

248 ranging from one to seven. The three species groups were genetically isolated with no shared haplotypes and were

249 distinguished by a high number of mutations, from 28 (M. platyphylla – M. dichotoma based on 16S rDNA) to 76

250 (M. platyphylla – M. exaesa based on COI and M. dichotoma – M. exaesa based on 16S rDNA).

251 The two single-gene trees are shown in Figs. 4c-d. In both phylogeny reconstructions, BI and ML analyses

252 were largely congruent with no topology conflicts and node support values were high across the ingroup. COI and

9

253 16S rDNA yielded similar topologies with well-supported clades although some sister relationships among the

254 recovered molecular lineages varied between the two phylogenetic trees. Both phylogenetic trees recovered the

255 group leading to Millepora as composed of two main lineages, the Atlantic cluster and the Indo-Pacific one. Within

256 the latter one, all the collected Red Sea samples clustered in three distinct and highly supported molecular clades

257 according matching to our species identificationmorphological differences based on traditional taxonomy. IndeedIn

258 fact, clade I includes M. dichotoma (COI Bayesian posterior probabilities PPBI = 1, ML bootstrap BML = 99; 16S

259 rDNA PPBI = 1 and BML = 100), clade II is composed of M. platyphylla (COI PPBI = 1 and BML = 100; 16S rDNA

260 PPBI = 1 and BML = 82), and clade III grouped M. exaesa (COI PPBI = 1 and BML = 100; 16S rDNA PPBI = 1 and BML

261 = 93). The three species from the Red Sea exhibited very high interspecific genetic distances and extremely low

262 intraspecific values (< 1) under both mitochondrial loci (Tables 1 and 2). Interspecific genetic distances ranged from

263 9.08 ± 1.32% (mean ± standard deviation) (M. dichotoma – M. platyphylla) to 18.77 ± 1.32% (M. platyphylla – M.

264 exaesa) based on COI and between 5.76 ± 0.97% (M. dichotoma – M. platyphylla) and 13.67 ± 1.44% (M.

265 dichotoma – M. exaesa) based on 16S rDNA. Notably, these three molecular lineages include exclusively our

266 newly-obtained sequences from the Red Sea. The conspecific sequences downloaded from GenBank of conspecific

267 deriving from other localities of the Southwestern Indo-Pacific and downloaded from GenBank did not cluster with

268 any Red Sea sequence and constituted additional and distinct molecular clades. As a result of this, Millepora

269 dichotoma and M. platyphylla were recovered as polyphyletic taxa while M. exaesa was recovered as monophyletic

270 taxon but composed of two divergent and distinct clades. The COI phylogeny reconstruction recovered clade IVa

271 including M. platyphylla from French Polynesia (Leray et al. 2013;, Schweinsberg et al. 2016) and clade Va

272 composed of M. dichotoma from Australia (Schweinsberg et al. 2016). The phylogenetic tree based on 16S rDNA

273 resolved clade IVb with M. platyphylla from Réunion Island and Northern Mariana Islands (Souza et al. 2017),

274 clade Vb consisted of M. dichotoma from Northern Mariana Islands (de Souza et al. 2017), and clade VII grouping

275 M. exaesa from Réunion Island. These molecular divergences between the Red Sea populations and the

276 Southwestern Indo-Pacific ones of M. dichotoma, M. platyphylla, and M. exaesa were also clearly highlighted by the

277 extremely high genetic distances among these clades (Tables 1, and 2). Indeed, the COI genetic distances between

278 Red Sea M. dichotoma (clade I) and the one from Australia (clade Va) (9.17 ± 1.22%) and between Red Sea M.

279 platyphylla (clade II) and the one from French Polynesia (clade IVa) (10.78 ± 1.34%) were comparable to the one

280 between Red Sea M. dichotoma (clade I) and Red Sea M. platyphylla (clade II) (9.08 ± 1.32%). Similarly, the

10

281 genetic distances based on 16S rDNA between Red Sea M. dichotoma (clade I) and the one from Northern Mariana

282 Islands (clade Vb) (4.3 ± 0.81%), Red Sea M. platyphylla (clade II) and the one from Réunion Island and Northern

283 Mariana Islands (clade IVb) (5.4 ± 0.92%), and Red Sea M. exaesa (clade III) and the one from Réunion Island

284 (clade VII) (5.14 ± 1.01%) were comparable to the one between Red Sea M. dichotoma (clade I) and Red Sea M.

285 platyphylla (clade II) (5.76 ± 0.97%).

286

287 Morphological results

288 The measurements of the eight pore characters are summarized in Table 3. Overall, M. dichotoma was is

289 characterized by small gastropore diameter (0.12 ± 0.02%), large distance between dactylopores (0.3 ± 0.06%),

290 small number of dactylopores per gastropore (3.9 ± 0.89%), and small dactylopore density (24.39 ± 5.39%).

291 Millepora platyphylla hads small dactylopore diameter (0.04 ± 0.01%), large distance between gastropores (0.97 ±

292 0.17%), small distance from gastropore to nearest dactylopore (0.11 ± 0.02%), and an intermediate dactylopore

293 density (43.71 ± 7.97%). Millepora exaesa showsed a high number of dactylopores per gastropore (6.93 ± 0.84%),

294 small gastropore density (5.27 ± 1.54%), and high dactylopore density (54.86 ± 10.7%).

295 The analysis of the three Millepora species revealed the presence of a common set of nematocysts shared

296 by all species, with some species-specific variations. Regarding the polyp stage, all species showed the presence of

297 small and large stenoteles and macrobasic apotrichous heteronemes (Figs. 5a-o). However, tThe heteronemes could

298 be further classified more in detail as mastigophores and euryteles. Specifically, mastigophores were found in M.

299 dichotoma (Figs. 5c-e) and M. platyphylla (Figs. 5h-j), whereas M. exaesa was characterized by euryteles (Figs. 5m-

300 o). Regarding the eumedusoid stage, all fire coral species displayed the presence of large stenoteles in the bulbs but

301 differences were found regarding in the presence and the distribution of microbasic mastigophores on the

302 exumbrella (Figs. 5p-w). Millepora dichotoma showed mastigophores grouped in a band in the middle portion of the

303 exumbrella (Fig. 5p), M. platyphylla had mastigophores grouped in the lower part of the exumbrella, in between

304 bulbs (Fig. 5s), and M. exaesa had an exumbrella deprived of nematocysts (Fig. 5v).

305 The MANOVA was run to determine the possible presence of test for differences in the multivariate pore

306 characters measured for the three species of Millepora from the Red Sea. All but one pore characters were assessed.

307 The differences among the three species based on the combined dependent variables were statistically significant, (F

2 308 2,294 = 178,395, p < 0,0005; Wilks’ lambda 0.036, partial ƞ 0.811). The follow-up univariate ANOVAs showed that

11

309 all pore characters were statistically significantly different among the three analyzed species (Table 4). The Wilks’

310 lambda test of discriminant function analysis showed significant differences in morphometric measurements for all

311 the three species (P < 0.001) (Table S4). In the discriminant function analysis, the first DF (discriminant function)

312 accounted for 66.6 %, the second DF accounted for 33.4 %, showing clear interspecificamong-sample

313 differentiation. The pore characters dactylopore density, number of dactylopores per gastropore, dactylopore

314 distance, and gastropore diameter contributed to first DF while dactylopore diameter, distance between gastropores,

315 and distance from gastropore to nearest dactylopore contributed to second DF (Table 5) indicating that the former

316 four pore characters were the most important measurements in discriminating the three fire coral species. The DFA

317 was able to discriminate the three species of Millepora among the origin allocations with a high degree of accuracy.

318 The DFA using cross-validation classification showed 998.7 % correct classification of the analyzed individuals into

319 their original species. Indeed, the proportion of correctly classified M. dichotoma samples to their original group was

320 the highest (100%), followed by M. exaesa samples (99 %), and M. platyphylla samples (98%). In accordance with

321 the species identification, tThe DF I vs. the DF II graphs depicted the presence of three distinct units although the

322 presence of limited overlap between M. exaesa and M. platyphyllain accordance with the species identification

323 (Figure 6). The analyseis of the nematocysts measurements regarding the nematocysts revealed statistically

324 significant differences among the three fire coral analyzed species of fire corals for the large stenoteles in both the

325 polyp and the eumedusoid stages (Table 6), as well as for the small stenoteles and heteronemes in the polyp stage

326 (Table 7).

327

328 Discussion

329 The presentedOur results confirmed the presence of three distinct species of Millepora in the Red Sea, i.e.

330 M. dichotoma, M. exaesa, and M. platyphylla, based on new molecular and morphological data. The two sequenced

331 mitochondrial genes (COI and 16S rDNA) clearly resolved the three species as distinct lineages and provided a

332 robust phylogenetic hypothesis of the genus. The Based on pore characters grouped the analyzed samples were

333 grouped in three clusters according toin full agreement with the in-situgrowth form-based species identification.

334 Finally, significant differences in the nematocysts of both polyps and eumedusoids examined for the first time in

335 material from this region were found among the three analyzed species of fire corals from the Red Sea.

12

336 Although early taxonomic works suggested the presence of a single widespread species of Millepora

337 throughout the Indo-Pacific and the Caribbean-Atlantic (Hickson 1898, 1899), several taxonomists recognized three

338 well-defined species of fire corals in the Red Sea, i.e. M. dichotoma, M. exaesa, and M. platyphylla (Klunzinger

339 1879;, Crossland 1941;, Boschma 1948a, b, 1966). These authors stated that these three species were clearly distinct

340 based on the colony growth form and provided an excellent a detailed description of their morphological characters.

341 Our study confirmed these findings and included innovative information about the genetics, the skeletal pores, and

342 the nematocysts of the three species, which we discussed in the below paragraphshereafter.

343 The resolution of the species boundaries in the Red Sea fire corals using a combination of genetic data and

344 morphology at two distinct levels, i.e. skeletal pore and nematocysts, showed that an integrated morpho-molecular

345 approach may beis useful for a better understanding of the systematics of Millepora. Our phylogeny reconstructions

346 based on COI and 16S rDNA resolved well-supported clades within this genus of hydrozoans and, for the first time,

347 these two molecular markers were used to build a phylogenetic hypothesis of the Indo-Pacific fire corals. Indeed, tIn

348 fact, the former locus successfully resolved the species boundaries among M. alcicornis, M. complanata, M.

349 squarrosa, and M. striata from the Caribbean (Luiz-RamosRuiz-Ramos et al. 2014). Moreover, tThe latter gene has

350 been already used by de Souza et al. (2017) to define the phylogenetic relationships and the connectivity of the

351 Caribbean-Atlantic fire corals. The resulting phylogeneticy hypothesis successfully resolved all the analyzed

352 species, i.e. M. alcicornis, M. braziliensis, M. laboreli, and M. nitida (de Souza et al. 2017). Therefore, COI and 16S

353 rDNA may be selected for future works aimed to evaluate the species boundaries of Millepora throughout its entire

354 geographic distribution. A potential n eventual additional marker could be the ITS region that has been successfully

355 used for the identification of Millepora from Japan (Takama et al. 2018) T, while the recently developed set of 15

356 microsatellites Dubé et al. (2017c) seemed, however, to be only valid only for the Indo-Pacific species.

357 Pores have been usually considered extremely variable and their use as informative taxonomic characters

358 has been criticized by several taxonomists (Hickson 1898, 1899;, Crossland 1948;, Boschma 1948a, 1966).

359 Nevertheless, Moshchenko (1994, 1995b, 1996b) examined in a rigorous way the utility of several pore

360 featurecharacters, including the ones we analyzed in the present work, and concluded that pores were able to define

361 Millepora species. Indeed, a quantitative evaluation of pore featurecharacters in fire corals from Vietnam

362 successfully discriminated M. platyphylla from the five branching species occurring in that area, i.e. M. cruzi

363 Nemenzo, 1975 (probably M. intricata; see Razak and Hoeksema, 2003), M. dichotoma, M. intricata, M. murrayi

13

364 Quelch, 1884, and M. tenera, while a continuous gradient was found among the branching species (Moshchenko

365 1994, 1995b, 1996b). These results were further confirmed by investigations of growth forms, enzymes, and

366 anatomy of the skeleton (Moshchenko 1994, 1995b, 1996b). Interestingly, a recent molecular study demonstrated

367 that the three branching species of fire corals found in Japan, i.e. M. dichotoma, M. intricata, and M. tenera, were

368 nested together in a single molecular clade (Takama et al. 2018). More in general, Takama et al. (2018) recovered

369 four molecular lineages of Millepora that were not in agreement with the traditional morphology-based species

370 identification (Razak and Hoeksema 2003). Nevertheless, they also showed significant differences in gastropore

371 diameter, dactylopore diameter, distances between gastropores among the genetic clades, with the only exception of

372 the absence of significant differences between two clades. Considering these findings together with the results

373 obtained from the Red SeaOverall, based on our results and the published studies, we suggest conclude that pores

374 may beare informative to tell apartdefine at least some of the molecular clades of Millepora.

375 The cnidome of Millepora, i.e. type, size, and distribution of nematocysts, has beenwas scarcely

376 investigated in the past, with a few exceptions. For instance, a first detailed assessment of the polyp nematocysts

377 was presented by Calder (1988), whose description of Millepora alcicornis from Bermuda included information and

378 images of stenotele and mastigophore capsules. More recently, other authors partially investigated the cnidome of a

379 few Millepora species from both Atlantic and Indo-Pacific localities (Meroz-Fine et al. 2003; Garcia-Arredondo et

380 al. 2012;, Bourmaud et al. 2013), finding in some cases differences in the size and the distribution of the

381 nematocysts among fire coral morphotypes (Meroz-Fine et al. 2003;, Bourmaud et al. 2013). In this work, we

382 analyzed and compared for the first time the complete set of the nematocysts found in both the polyp and the

383 eumedusoids stages of three Millepora species from the Red Sea, finding significant differences among them and

384 demonstrating the utility of the cnidome in species discrimination within Millepora hydrozoans. Not only the size,

385 but also the distribution of the nematocysts of Red Sea Millepora species differed among taxa. Indeed, similarly to

386 the observations of Bourmaud et al. (2013) on eumedusoids from Réunion Island, we found that the eumedusoids of

387 the three Red Sea species showed a different organization of the mastigophores differentially organized on the

388 exumbrella. Millepora exaesa eumedusoids from both the Red Sea and Réunion Island lackedwere deprived of

389 mastigophores and this morphological similarity is also supported by the fact that these two taxa clustered together

390 in a fully supported clade in the 16S rRNA phylogeny reconstruction. Moreover, M. dichotoma from the Red Sea

391 and M. tenera from Réunion Island showed a similar pattern of exumbrellar nematocyst organization, and indeed are

14

392 closely related in the 16S rRNA phyogenetic tree. Contrarily, M. platyphylla from the Red Sea and Réunion Island

393 had mastigophores organized in different ways and were indeed phylogenetically divergent.

394 Our single-gene phylogeny reconstructions revealed unexpected phylogenetic patterns among species offor

395 fire corals in the Indo-Pacific. Although the genetic analyses of our samples confirmed the presence of three

396 lineages at the species level within the Red Sea, the inclusion of available Millepora sequences from the Indo-

397 Pacific within the our analyses resulted in the polyphyly of both M. dichotoma and M. platyphylla and in the

398 presence of two divergent and allopatric lineages of M. exaesa. First of all, it is worth to highlight thatUnfortunately,

399 no specimen photos have beenwere published for these sequences downloaded from GenBank with the only

400 exception of the samples of M. dichotoma from Australia which are included in the COI phylogeny reconstruction

401 (Schweinsberg et al. 2016). Nevertheless, M. platyphylla and M. exaesa are both supposedly morphologically

402 distinctive species in the Indo-Pacific as they are easily recognizabletold apart based on their unique colony growth

403 form (Boschma 1949a, b;, Razak and Hoeksema 2003), although a certain degree of ecophenotypic

404 variationphenotypic plasticity occurs especially in M. platyphylla (Dubé et al. 2017a, b) . Therefore, we confidently

405 are prone to reject the hypothesis of a misidentification by the authors based on colony morphology of both M.

406 platyphylla from Réunion Island and French Polynesia (Leray et al. 2013;, Schweinsberg et al. 2016) and M. exaesa

407 from Réunion Island. This being said, the deep intraspecific genetic divergences between populations of M.

408 dichotoma, M. exaesa, and M. platyphylla from the Red Sea and various localities of the Southwestern Indo-Pacific

409 suggested that these populations represented distinct evolutionary units that have not been detected in previous

410 taxonomic worksobscured by traditional taxonomy. These three species have been historically considered

411 widespread throughout the Indo-Pacific, from the Red Sea to French Polynesia (Klunzinger 1879;, Crossland 1941;,

412 1948;, Boschma 1949a;, Razak and Hoeksema 2003), but our genetic analyses clearly demonstrated that they all

413 constituted species complexes. Each of these three species complexes is composed of at least two species that share

414 a similar colony growth form. Therefore, while the growth of the colony is a diagnostic character to distinguish the

415 Red Sea species between each other (Forskål 1775;, Ehrenberg 1834;, Klunzinger 1879), this feature it is clearly not

416 informative to define boundaries within the three species complexes. The genetic distances between the two distinct

417 lineages for each of the three nominal species are comparable to those occurring between valid species in the genus

418 for the examined markers (Tables 1 and 2). Given that the type locality of M. dichotoma, M. exaesa, and M.

419 platyphylla is the Red Sea, a future taxonomic revision of Millepora would maintain these names for the three Red

15

420 Sea moleculare lineages while new names or synonymised names would be assigned to the other Indo-Pacific

421 lineages of these three species. In this respect, it will be relevant to re-examine Millepora tuberosa Boschma, 1966,

422 since its type specimens from Mauritius resemble corals of M. exaesa. Boschma (1966) considered M. exaesa a

423 doubtful species and he suspected that specimens from outside the Red Sea are actually M. tuberosa. Both taxa

424 were synonymized by Razak and Hoeksema (2003), who gave priority to M. exaesa, being the senior synonym.

425 Since Réunion Island is only 225 km away from Mauritius, this may also concern specimens from this locality that

426 were studied by Dubé et al. (2017c), who did not refer to Boschma’s (1966) work. Our results remove the doubts

427 expressed by Boschma (1966) concerning the status of M. exaesa as a valid species.

428 From a morphological point of view, either morphological convergence of the colony growth form or

429 character state conservation from the ancestor may have occurred. On the base of our phylogenetic trees, the former

430 hypothesis may be applied to M. dichotoma and M. platyphylla, given the absence of a sister relationships between

431 the Red Sea clades and the Southwestern Indo-Pacific ones. while tThe second explanation is more likely for M.

432 exaesa considering that the two clades from the Red Sea and Réunion Island are sister taxa, possibly M. exaesa and

433 M. tuberosa. Although the absence of variation in the colony growth form, our integrated morphological analyses on

434 Red Sea Millepora may suggest to further evaluate both pore structures and nematocysts of the Southwestern Indo-

435 Pacific clades to evaluate the occurrence of possible differences of these morphological traits between the Red Sea

436 and the Southwestern Indo-Pacific molecular lineages. Indeed, nematocysts of M. dichotoma, M. exaesa, and M.

437 platyphylla from the Southwestern Indo-Pacific have never been studied. Moreover, no comparisons of pores

438 between the Red Sea and the Southwestern Indo-Pacific colonies of these three species have been undertaken so far.

439 From a biological point of view, the distinction between the Red Sea and the Southwestern Indo-Pacific clades may

440 suggest that either the Red Sea lineages have a small-distance dispersal capability or the Southwestern Indo-Pacific

441 lineages have failed to disperse within the Red Sea. At least for the case of clade IVb (M. platyphylla from Réunion

442 Island and Northern Mariana Islands), this evolutionary unit is capable of long-distance dispersal. This contrasting

443 pattern of shortmall- and long-distance dispersal capabilities among species is similar to the one occurring in the

444 Caribbean-Atlantic where M. alcicornis showed a high connectivity within the Caribbean while the three endemic

445 species of Brazil, i.e. M. braziliensis, M. nitida, and M. laboreli, are restricted to few locations and isolated by

446 strong biogeographic barriers (de Souza et al. 2017).

16

447 A growing number of studies has revealed that several hydrozoans traditionally considered to display a

448 large geographic distribution are indeed species complexes composed of geographically circumscribed lineages

449 (e.g., Schuchert 2014;, Postaire et al. 2016, 2017;, Maggioni et al. 2017b;, Montano et al. 2017). For example, the

450 near-cosmopolitan species Plumularia setacea (Linnaeus, 1758) is subdivided in a multitude of distinct molecular

451 clades showing distinct geographic distributions, such as South Africa, the Western Indian Ocean, the Caribbean,

452 and the Mediterranean Sea. Similarly, Postaire et al. (2017) revealed a deep and strong genetic differentiation

453 between the Indian and the Pacific populations of Macrorhynchia phoenicea (Busk, 1852). Several analogous cases

454 have been recently described in many hard corals throughout the Indo-Pacific (Flot et al. 2008;, Ladner and Palumbi

455 2012;, Pinzón et al., 2013;, Richards et al. 2016;, Gélin et al., 2017;, Arrigoni et al. 2016a). Moreover, the existence

456 of unexpectedsurprising intraspecific evolutionary breaks between the Red Sea and the Southwestern Indo-Pacific

457 populations/species have been discovered based on genetic studies in several marine phyla, including scleractinian

458 corals (Arrigoni et al. 2012, 2016b;, Huang et al. 2014), octocorals (Reijnen et al. 2014), sea urchins (Bronstein et al.

459 2016), and fishes (DiBattista et al. 2013;, Fernandez et al. 2015;, Ahti et al. 2016;, Waldrop et al. 2016). In most

460 cases, these genetic divergences between Oceans reflect the importance of geological features, such as tectonic

461 plates and mantle plume tracks. Indeed, it is now largely accepted that the actual distribution of most hard corals

462 depends more on geological events than contemporary environmental conditions and habitats (Keith et al. 2013;,

463 Leprieur et al. 2016). For example, important tectonic changes in what is now the Indian Ocean occurred during

464 both the Eocene and Oligocene, resulting in the fragmentation of the Tethys Sea (Schettino and Turco, 2011), and

465 may have promoted isolation and speciation evens.

466 In conclusion, the obtained results clearly resolved the species boundaries of Millepora from in the Red

467 Sea. Given the long-term challenges of morphology-based species identification of these organisms (Hickson 1898;,

468 Boschma 1948b, 1949) and the recent unexpected molecular findings in discordance with traditional systematics

469 (Ruiz-Ramos et al. 2014;, Takama et al. 2018), we believed that the presented approach may be useful to define the

470 relationships among the fire corals. Moreover, the recovery of multiple molecular lineages of M. dichotoma, M.

471 exaesa, and M. platyphylla between the Red Sea and the Indo-Pacific suggested the need of investigate more

472 material from additional localities of the Indo-Pacific. The integration of genetics and morphology (pore structure

473 and nematocysts of both polyp and eumedusoid stages) may be the key to elucidate what Boschma (1948a) defined

474 “the species problem in Millepora” more than 50 years ago.

17

475

476 Acknowledgements

477 This research was undertaken in accordance with the policies and procedures of the King Abdullah

478 University of Science and Technology (KAUST). Permissions relevant for KAUST to undertake the research have

479 been obtained from the applicable governmental agencies in the Kingdom of Saudi Arabia. We wish to thank Amr

480 Gusti (KAUST), the captain and crew of the MV Dream-Master and the KAUST Coastal and Marine Resources

481 Core Lab for fieldwork logistics in the Red Sea. This project was supported by funding from KAUST (award

482 #FCC/1/1973-21 and baseline research funds to MLB Berumen). TS would like to acknowledge Yossi Loya and the

483 Israeli Taxonomy Initiative for funding his work. We are deeply grateful to the editor and threefour anonymous

484 referees for their comments which greatly improved the manuscript.

485 486 References 487 Ahti PA, Coleman RR, DiBattista JD, Berumen ML, Rocha LA, Bowen BW (2016) Phylogeography of Indo‐Pacific 488 reef fishes: sister wrasses Coris gaimard and C. cuvieri in the Red Sea, Indian Ocean and Pacific Ocean. J 489 Biogeogr 43:1103–1115 490 Amaral FD, Silva RS, Mauricio-da-Silva L, Sole-Cava AM (1997) Molecular systematics of Millepora alcicornis 491 Linnaeus, 1758 and M. braziliensis Verrill, 1868 (Hydrozoa: Milleporidae) from Brazil. Proc 8th Int Coral 492 Reef Symp 2:1577–1580 493 Amaral FD, Steiner AQ, Broadhurst MK, Cairns SD (2008) An overview of the shallow-water calcified hydroids 494 from Brazil (Hydrozoa: Cnidaria), including the description of a new species. Zootaxa 1930:56–68 495 Arrigoni R, Stefani F, Pichon M, Galli P, Benzoni F (2012) Molecular phylogeny of the robust clade (Faviidae, 496 Mussidae, Merulinidae, and Pectiniidae): an Indian Ocean perspective. Mol Phylogenet Evol 65:183–193 497 Arrigoni R, Berumen ML, Chen CA, Terraneo TI, Baird AH, Payri C, Benzoni F (2016a) Species delimitation in the 498 reef coral genera Echinophyllia and Oxypora (Scleractinia, Lobophylliidae) with a description of two new 499 species. Mol Phylogenet Evol 105:146–159 500 Arrigoni R, Benzoni F, Huang D, Fukami H, Chen CA, Berumen ML, Hoogenboom M, Thomson DP, Hoeksema 501 BW, Budd AF, Zayasu Y, Terraneo TI, Kitano YF, Baird AH (2016b) When forms meet genes: revision of 502 the scleractinian genera Micromussa and Homophyllia (Lobophylliidae) with a description of two new 503 species and one new genus. Contrib Zool 85:387–422 504 Bandelt HJ, Forster P, Röhl A (1999) Median-joining networks for inferring intraspecific phylogenies. Mol Biol 505 Evol 16:37–48 506 Benayahu Y Loya Y (1977) Space partitioning by stony corals soft corals and benthic algae on the coral reefs of the 507 northern Gulf of Eilat (Red Sea). Helgol Wiss Meeresunters 30:362 508 Boschma H (1948a) The species problem in Millepora. Zool Verh Leiden 1:1–115 509 Boschma H (1948b) Specific characters in Millepora. Proc Kon Ned Akad Wet 51:818–823 510 Boschma H (1949) The ampullae of Millepora. Proc Kon Ned Akad Wet 52:3–14 511 Boschma H (1956) Milleporina and Stylasterina. In: Moore RC (ed) Treatise on invertebrate paleontology. Part F, 512 Coelenterata. Geological Society of America & University of Kansas Press. 513 Boschma H (1966) On a new species of Millepora from Mauritius, with notes on the specific characters of 514 Millepora exaesa. Proc Kon Ned Akad Wet 69:409–419 515 Bouillon J, Gravili C, Gili J-M, Boero F. 2006. An introduction to Hydrozoa. Ann Mus Hist Nat Paris 194: 1–-591. 516 Bourmaud CAF, Leung JKL, Bollard S, Gravier‐Bonnet N (2013) Mass spawning events, seasonality and 517 reproductive features in Milleporids (Cnidaria, Hydrozoa) from Reunion Island. Mar Ecol 34:14–24 518 Bronstein O, Kroh A, Haring E (2016) Do genes lie? Mitochondrial capture masks the Red Sea collector urchin’s 519 true identity (Echinodermata: Echinoidea: Tripneustes). Mol Phylogenet Evol 104:1–13

18

520 Cairns SD, Hoeksema BW, Van der land J (1999) List of extant stony corals. Atoll Res Bull 459:1–46 521 Calder DR (1988) Shallow-water hydroids of Bermuda: the Athecatae. Roy Ontario Mus Life Sci Contrib 148:1– 522 107 523 Crossland C (1941) On Forskål's collection of corals in the Zoological Museum of Copenhagen. Spolia ZooL Haun. 524 Skr Vdgivet Univ Zool Mus Kbh J1:5–63, pls 1–12 525 Crossland C (1948) Reef corals of the South African Coast. Ann Nat Mus 9:169–205, pls 1–142 526 Cunningham CW, Buss LW (1993) Molecular evidence for multiple episodes of paedomorphosis in the Family 527 Hydractiniidae. Biochem Syst Ecol 21:57–69 528 de Souza JN, Nunes FL, Zilberberg C, Sanchez JA, Migotto AE, Hoeksema BW, Serrano XM, Baker AC, Lindner 529 A (2017) Contrasting patterns of connectivity among endemic and widespread fire coral species (Millepora 530 spp.) in the tropical Southwestern Atlantic. Coral Reefs 36:701–716. 531 de Weerdt WH (1981) Transplantation experiments with Caribbean Millepora species (Hydrozoa, Coelenterata), 532 including some ecological observations on growth forms. Bijdr Dierkd 51:1–19 533 de Weerdt WH (1984) Taxonomic characters in Caribbean Millepora species (Hydrozoa, Coelenterata) including 534 some ecological observations on growth forms. Bijdr Dierkd 54:243–362 535 de Weerdt, WH, Glynn PW (1991) A new and presumably now extinct species of Millepora (Hydrozoa) in the 536 eastern Pacific. Zool Med Leiden 65:267–276 537 DiBattista JD, Berumen ML, Gaither MR, Rocha LA, Eble JA, Choat JH, Craig MT, Skillings DJ, Bowen BW 538 (2013) After continents divide: comparative phylogeography of reef fishes from the Red Sea and Indian 539 Ocean. J Biogeogr 40:1170–1181 540 DiBattista JD, Choat JH, Gaither MR, Hobbs JP, Lozano-Cortés DF, Myers R, Paulay G, Rocha LA, Toonen RJ, 541 Westneat M, Berumen ML (2016a) On the origin of endemic species in the Red Sea. J Biogeogr 43:13–30 542 DiBattista JD, Roberts MB, Bouwmeester J, Bowen BW, Coker DJ, Lozano‐Cortés DF, Choat JH, Gaither MR, 543 Hobbs JP, Khalil MT, Kochzius M, Myers RF, Paulay G, Robitzch V, Saenz-Agudelo P, Salas E, Sinclair- 544 Taylor TH, Toonen RJ, Westneat MW, Williams ST, Berumen ML (2016b) A review of contemporary 545 patterns of endemism for shallow water reef fauna in the Red Sea. J Biogeogr 43:423–439 546 Dubé CE, Mercière A, Vermeij MJ, Planes S (2017a) Population structure of the hydrocoral Millepora platyphylla 547 in habitats experiencing different flow regimes in Moorea, French Polynesia. PloS ONE 12:e0173513 548 Dubé CE, Boissin E, Maynard JA, Planes S (2017b) Fire coral clones demonstrate phenotypic plasticity among reef 549 habitats. Mol Ecol 26:3860–3869 550 Dubé CE, Planes S, Zhou Y, Berteaux-Lecellier V, Boissin E (2017c) Genetic diversity and differentiation in reef- 551 building Millepora species, as revealed by cross-species amplification of fifteen novel microsatellite loci. 552 PeerJ 5:e2936 553 Duchassaing P, Michelotti J (1860) Mémoire sur les Coralliaires des Antilles. Mém Ac Sc Turin Sér 2, 19:279–365, 554 pls 1–10 555 Edmunds PJ (1999) The role of colony morphology and substratum inclination in the success of Millepora 556 alcicornis on shallow coral reefs. Coral Reefs 18:133–140 557 Ehrenberg CG (1834) Beitrage zur physiologischen Kenntniss der Corallenthiere im Allgemeinen und besunders des 558 Rothen Meeres, nebst einem Versuche zur physiologischen Systematik derselben. Abh Kon Akad Wiss 559 Berlin 1832:225–380 560 Ehrenberg CG (1834) Die corallenthiere des Rothen Meeres. Berlin. 561 Fernandez‐Silva I, Randall JE, Coleman RR, DiBattista JD, Rocha LA, Reimer JD, Meyer CG, Bowen BW (2015) 562 Yellow tails in the Red Sea: phylogeography of the Indo‐Pacific goatfish Mulloidichthys flavolineatus 563 reveals isolation in peripheral provinces and cryptic evolutionary lineages. J Biogeogr 42:2402–2413 564 Flot JF, Licuanan WY, Nakano Y, Payri C, Cruaud C, Tillier S (2008) Mitochondrial sequences of Seriatopora 565 corals show little agreement with morphology and reveal the duplication of a tRNA gene near the control 566 region. Coral Reefs 27:789–794 567 Forskål P (1775) Descriptiones animalium avium, amphibiorum, piscium, insectorum, vermium; quae in itinere 568 orientali observavit. IV Corallia. Heineck et Faber, Hauniae, pp 131–139 569 García-Arredondo A, Rojas A, Iglesias-Prieto R, Zepeda-Rodriguez A, Palma-Tirado L (2012) Structure of 570 nematocysts isolated from the fire corals Millepora alcicornis and Millepora complanata (Cnidaria: 571 Hydrozoa). J Venom Anim Toxins Incl Trop Dis 18:109–-115 572 Gélin P, Postaire B, Fauvelot C, Magalon H (2017) Reevaluating species number, distribution and endemism of the 573 coral genus Pocillopora Lamarck, 1816 using species delimitation methods and microsatellites. Mol 574 Phylogenet Evol 109:430–446 575 Hickson SJ (1898) On the species of the genus Millepora: a preliminary communication. J Zool 66:246–257

19

576 Hickson SJ (1809) Report on the specimens of the genus Millepora collected by Dr Willey. J Zool 119:661–672 577 Hoeksema BW, Nunes FLD, Lindner A, de Souza JN (20142017) Millepora alcicornis (Hydrozoa: Capitata) at 578 Ascension Island: confirmed identity based on morphological and molecular analyses. J Mar Biol Assoc 579 UK 97:709–712 580 Huang D, Benzoni F, Arrigoni R, Baird AH, Berumen ML, Bouwmeester J, Chou LM, Fukami H, Licuanan WY, 581 Lovell ER, Meier R (2014) Towards a phylogenetic classification of reef corals: the Indo‐Pacific genera 582 Merulina, Goniastrea and Scapophyllia (Scleractinia, Merulinidae). Zool Scripta 43:531–548 583 Katoh K, Standley DM (2013) MAFFT multiple sequence alignment software version 7: improvements in 584 performance and usability. Mol Biol Evol 30:772–780 585 Keith SA, Baird AH, Hughes TP, Madin JS, Connolly SR (2013) Faunal breaks and species composition of Indo- 586 Pacific corals: the role of plate tectonics, environment and habitat distribution. Proc R Soc B 280:20130818 587 Klausewitz W (1989) Evolutionary history and zoogeography of the Red Sea ichthyofauna. Fauna Saudi Arabia 588 10:310–337 589 Klunzinger CB (1879) Die Korallthiere des Rothen Meeres. Gutmann, Berlin. 3:1–100 590 Ladner JT, Palumbi SR (2012) Extensive sympatry, cryptic diversity and introgression throughout the geographic 591 distribution of two coral species complexes. Mol Ecol 21:2224–2238 592 Lanfear R, Calcott B, Ho SYW, Guindon S (2012) PartitionFinder: combined selection of partitioning schemes and 593 substitution models for phylogenetic analyses. Mol Biol Evol 29:1695–1701 594 Leprieur, F., Descombes, P., Gaboriau, T., Cowman, P.F., Parravicini, V., Kulbicki, M., Melian, C.J., de Santana 595 CN, Heine C, Mouillot D, Bellwood DR, Pellissier L (2016) Plate tectonics drive tropical reef biodiversity 596 dynamics. Nat Commun 7:11461 597 Leray M, Yang JY, Meyer CP, Mills SC, Agudelo N, Ranwez V, Boehm JT, Machida RJ (2013) A new versatile 598 primer set targeting a short fragment of the mitochondrial COI region for metabarcoding metazoan 599 diversity: application for characterizing coral reef fish gut contents. Front Zool 10:34 600 Lewis JB (1989) The ecology of Millepora. Coral Reefs 8:99–107 601 Lewis JB (2006) Biology and ecology of the Millepora on coral reefs. Adv Mar Biol 50:1–55 602 Librado P, Rozas J (2009) DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. 603 Bioinformatics 25:1451–1452 604 Linnaeus C (1758) Systema Naturae, ed. X, vol. I. Laurentii Salvii, Holmiae 605 López C, Clemente S, Almeida C, Brito A, Hernandez M (2015) A genetic approach to the origin of Millepora sp. in 606 the eastern Atlantic. Coral Reefs 34:631–638 607 Loya Y (1972) Community structure and species diversity of hermatypic corals at Eilat, Red Sea. Mar Biol 13:100– 608 123 609 Loya Y (1976) Recolonization of Red Sea corals affected by natural catastrophes and man‐made perturbations. 610 Ecology. Ecology 57:278–57289 611 Loya Y, Slobodkin LB (1971) The coral reefs of Eilat. Symp Zool Soc Lond 28:117–139 612 Maggioni D, Galli P, Berumen ML, Arrigoni R, Seveso D, Montano S (2017a) Astrocoryne cabela, gen. nov. et sp. 613 nov.(Hydrozoa: Sphaerocorynidae), a new sponge-associated hydrozoan. Invert Syst 31:734–746 614 Maggioni D, Montano S, Arrigoni R, Galli P, Puce S, Pica D, Berumen ML (2017b) Genetic diversity of the 615 Acropora-associated hydrozoans: new insight from the Red Sea. Mar Biodivers 47:1045–1055 616 Maggioni D, Arrigoni R, Galli P, Berumen ML, Seveso D, Montano S (2018) Polyphyly of the genus Zanclea and 617 family Zancleidae (Hydrozoa, Capitata) revealed by the integrative analysis of two bryozoan-associated 618 species. Contrib to ZoolContrib Zool 87:87–104 619 Meroz-Fine E, Brickner I, Loya Y, Ilan M (2003) The hydrozoan coral Millepora dichotoma: speciation or 620 phenotypic plasticity? Mar Biol 143:1175–1183 621 Miller MA, Pfeiffer W, Schwartz T (2010) Creating the CIPRES Science Gateway for inference of large 622 phylogenetic trees. Proceedings of the Gateway Computing Environments Workshop (GCE), New Orleans, 623 IEEE 624 Montano S, Maggioni D, Galli P, Hoeksema BW (2017) A cryptic species in the Pteroclava krempfi species 625 complex (Hydrozoa, Cladocorynidae) revealed in the Caribbean. Mar Biodivers 47:83–89 626 Moshchenko AV (1992) Morphology and variability of colonies of branched milleporids (Hydrozoa, Athecata, 627 Milleporidae) from Vietnam. Zool Zhur 71:5–12 628 Moshchenko AV (1994) Method of quantitative evaluation of the structure of the pore apparatus of millepore 629 hydroids. Rus J Mar Biol 20:358–365 630 Moshchenko AV (1995a) Environmental variations of the colony form of hydroid Millepora platyphylla in 631 Vietnamese Reefs. Russ J Mar Biol 21:174–183

20

632 Moshchenko AV (1995b) A quantitative evaluation of the structure of the pore apparatus of Millepora hydroids of 633 Vietnam. Rus J Mar Biol 21: 265–274 634 Moshchenko AV (1996a) Growth and development of Millepora colonies (Athecata, Milleporidae). Zool Zhur 635 75:485–493 636 Moshchenko AV (1996b) Variability of the pore apparatus of milleporine hydroids of Vietnam. Rus J Mar Biol 637 22:32–40 638 Moshchenko AV (1997) On the species composition of millepores in the Indo-Pacific. Rus J Mar Biol 23:238–247 639 Nawrocki AM, Schuchert P, Cartwright P (2010) Phylogenetics and evolution of Capitata (Cnidaria: Hydrozoa), and 640 the systematics of Corynidae. Zool Scr 39:290–304 641 Perkol-Finkel S, Benayahu Y (2004) Community structure of stony and soft corals on vertical unplanned artificial 642 reefs in Eilat (Red Sea): comparison to natural reefs. Coral Reefs 23:195–205 643 Pinzón JH, Sampayo E, Cox E, Chauka LJ, Chen CA, Voolstra CR, LaJeunesse TC (2013) Blind to morphology: 644 genetics identifies several widespread ecologically common species and few endemics among Indo‐Pacific 645 cauliflower corals (Pocillopora, Scleractinia). J Biogeogr 40:1595–1608 646 Postaire B, Magalon H, Bourmaud CA, Bruggemann JH (2016) Molecular species delimitation methods and 647 population genetics data reveal extensive lineage diversity and cryptic species in Aglaopheniidae 648 (Hydrozoa). Mol Phylogenet Evol 105:36–49 649 Postaire B, Gélin P, Bruggemann JH, Pratlong M, Magalon H (2017) Population differentiation or species formation 650 across the Indian and the Pacific Oceans? An example from the brooding marine hydrozoan 651 Macrorhynchia phoenicea. Ecol Evol 7:8170–8186 652 Pourtales LF (1877) Effects of urticating organs of Millepora on the tongue. Nature 17:27 653 Ruiz-Ramos DV, Weil E, Schizas NV (2014) Morphological and genetic evaluation of the hydrocoral Millepora 654 species complex in the Caribbean. Zool Stud 53:4 655 Quelch JJ (1884) The Milleporidae. Nature 30:539 656 Rambaut A, Suchard MA, Xie D, Drummond AJ (2014) Tracer v1.6. http://beast.bio.ed.ac.uk/Tracer/ 657 Rasband WS (1997) ImageJ. http://rsb.info.nih.gov/ij/ 658 Razak TB, Hoeksema BW (2003) The hydrocoral genus Millepora (Hydrozoa: Capitata: Milleporidae) in Indonesia. 659 Zool Verh Leiden 2003345:313–336 660 Richards ZT, Berry O, van Oppen MJ (2016) Cryptic genetic divergence within threatened species of Acropora 661 coral from the Indian and Pacific Oceans. Conserv Genet 17:577–591 662 Reijnen BT, McFadden CS, Hermanlimianto YT, van Ofwegen LP (2014) A molecular and morphological 663 exploration of the generic boundaries in the family Melithaeidae (Coelenterata: Octocorallia) and its 664 taxonomic consequences. Mol Phylogenet Evol 70:383–401 665 Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A, Höhna S, Barget B, Liu L, Suchard MA, 666 Huelsenbeck JP (2012) MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a 667 large model space. Syst Biol 61:539–542 668 Ruiz-Ramos DV, Weil E, Schizas NV (2014) Morphological and genetic evaluation of the hydrocoral Millepora 669 species complex in the Caribbean. Zool Stud 53:4 670 Schettino A, Turco E (2011) Tectonic history of the western Tethys since the Late Triassic. Geol Soc Am Bull 671 123:89–105 672 Schuchert P (2014) High genetic diversity in the hydroid Plumularia setacea: a multitude of cryptic species or 673 extensive population subdivision? Mol Phylogenet Evol 76:1–9 674 Schweinsberg M, Tollrian R, Lampert KP (2016) Inter‐and intra‐colonial genotypic diversity in hermatypic 675 hydrozoans of the family Milleporidae. Mar Ecol 38:e12388 676 Smith TB, Glynn PW, Maté JL, Toth LT, Gyory J (2014) A depth refugium from catastrophic coral bleaching 677 prevents regional extinction. Ecology 95:1663–1673 678 Stamatakis A (2014) RAxML Version 8: A tool for phylogenetic analysis and post-analysis of large phylogenies". 679 Bioinformatics 30:1312–1313 680 Takama O, Fernandez-Silva I, López C, Reimer JD (2018) Molecular phylogeny demonstrates the need for 681 taxonomic reconsideration of species diversity of the hydrocoral genus Millepora (Cnidaria: Hydrozoa) in 682 the Pacific. Zool Sci 35:123–133 683 Vago R, Achituv Y, Vaky L, Dubinsky Z, Kizner Z (1998) Colony architecture of Millepora dichotoma Forskal. J 684 Exp Mar Biol Ecol 224:225–235 685 Waldrop E, Hobbs JP, Randall JE, DiBattista JD, Rocha LA, Kosaki RK, Berumen ML, Bowen BW (2016) 686 Phylogeography, population structure and evolution of coral‐eating butterflyfishes (Family Chaetodontidae, 687 genus Chaetodon, subgenus Corallochaetodon). J Biogeogr 43:1116–1129

21

688 Wolf-Vecht A, Paldor N, Brenner S (1992) Hydrographic indications of advection/convection effects in the Gulf of 689 Eilat. Deep Sea Research Part A. Oceanogr Res Pap 39:1393–1401 690 691 Figures legends

692 Fig. 1 Map of Millepora sampling locations along the Red Sea

693 Fig. 2 In-vivo photos of the three Millepora species occurring in the Red Sea. a-c M. dichotoma; d-f M.

694 platyphylla; g-i M. exaesa. Scale bars: 5 cm

695 Fig. 3 a Pore characters analyzed in this study, gastropore diameter (1), dactylopore diameter (2), distance

696 between gastropores (3), distance between dactylopores (4), distance from gastropore to nearest dactylopore (5), and

697 number of dactylopores per gastropore (6). b-c Skeleton surface of the three Millepora species occurring in the Red

698 Sea. b M. dichotoma; c M. platyphylla; d M. exaesa. Scale bars: 2 mm

699 Fig. 4 a COI and b 16S rDNA most parsimonious median-joining networks of the three Red Sea Millepora

700 species. Each circle represents a unique haplotype and its size is proportional to its total frequency. Thin branches

701 and black cross-bars represent a single nucleotide change, thick black bars represent greater than one nucleotide

702 change (as indicated), small black circles represent missing haplotypes. c COI and d 16S rDNA phylogeny

703 reconstructions of Millepora inferred by Bayesian inference. The clade support values are Bayesian posterior

704 probabilities (≥ 0.7) and maximum likelihood bootstrap replicates (≥ 70). Colours denote each distinct molecular

705 lineage as reported in the figure box.

706 Fig. 5 Cnidome of the polyp (a-o) and eumedusoid stages (p-w) of the three Millepora species analyzed in

707 this work. a Large and b small undischarged stenoteles, c macrobasic apotrichous mastigophores undischarged

708 capsule, d detail of the discharged distal portion of the shaft, and e discharged capsule in M. dichotoma. f Large and

709 g small stenoteles, h macrobasic apotrichous mastigophores undischarged capsule, i detail of the discharged distal

710 portion of the shaft, and j discharged capsule in M. platyphylla. k Large and l small stenoteles, m macrobasic

711 apotrichous euryteles undischarged capsule, n detail of the discharged distal portion of the shaft, and j discharged

712 capsule in M. exaesa. p Nematocyst distribution, q stenoteles, and r microbasic mastigophores in the eumedusoid of

713 M. dichotoma. s Nematocyst distribution, t stenoteles, and u microbasic mastigophores in the eumedusoid of M.

714 platyphylla. v Nematocyst distribution, and w stenoteles in the eumedusoid of M. exaesa. Black and white dots in

715 Figs. p, s, v represent stenoteles and microbasic mastigophores, respectively. Scale bars: 10 µm

22

716 Fig. 6 Discriminant analysis plot for the analyzed pore characters of Millepora species from Red Sea.

717 Millepora dichotoma (blue), M. platyphylla (red), and M. exaesa (green).

718

719 Supporting Information

720 Fig. S1 a COI and b 16S rDNA phylogeny reconstructions of Millepora inferred by Bayesian inference.

721 The two trees are identical to the ones presented in Fig. 4 but the graphics is different: each single tip represent a

722 single sequence. The clade support values are Bayesian posterior probabilities (≥ 0.7) and maximum likelihood

723 bootstrap replicates (≥ 70). Colours denote each distinct molecular lineage as reported in the figure box.

724 Table S1 List of Millepora samples collected from the Red Sea for this study, including voucher number,

725 depth, site, latitude, longitude, GenBank accession numbers of the two mitochondrial genes (COI and 16S rDNA)

726 Table S2 Measurements of pore characters analyzed in this study for the three Red Sea Millepora species

727 Table S3 Measurements of nematocyst characters of both polyp and medusa stages analyzed in this study

728 for the three Red Sea Millepora species

729 Table S4 Wilks’ lambda test for verifying differences among Millepora species with pore character

730 measurements using the DFA

23