Foxed intra-and interspecific differentiation in Leptogorgia ( Octocorallia : Gorgoniidae ) . A description of a new species based on multiple sources of evidence

1 Dpto. Biodiversidad y Biología Evolutiva, Museo Nacional de Ciencias Naturales, CSIC, 28006 Madrid, Spain. (MMS-H) (Corresponding author) E-mail: mar.s.hurtado@mncn.csic.es. ORCID-iD: http://orcid.org/0000-0002-0284-9900 (AM) E-mail: annie@mncn.csic.es. ORCID-iD: http://orcid.org/0000-0003-0341-0809 2 Biodiversidad y Ecología de Invertebrados Marinos, Facultad de Biología, Universidad de Sevilla, 41012, Sevilla, Spain. (CM) E-mail: cmegina@us.es. ORCID-iD: http://orcid.org/0000-0002-9462-9330 (PJL-G) E-mail: pjlopez@us.es. ORCID-iD: http://orcid.org/0000-0002-7348-6270 3 Instituto Nacional de Biodiversidad, Museo Ecuatoriano de Ciencias Naturales. Rumipamba 341 y Av. Shyris, Quito, Ecuador.


INTRODUCTION
In the last decade, several studies have highlighted the difficulties of identifying marine species, especially invertebrates, using only morphological characters (Calvo et al. 2009, López-González et al. 2015, Alfaya et al. 2015), which can lead to an underestimation of biodiversity, and possibly adversely impact the conservation of endangered species or those with restricted distributions (see e.g.Rocha-Olivares et al. 2001, Dincă et al. 2011, Eberle et al. 2016).
Understanding species boundaries in corals, particularly in octocorals, is a challenge for several fields of study, including taxonomy, evolutionary biology, life history and ecology (Vermeij et al. 2007, Stefani et al. 2008, Gori et al. 2012).The identification of octocorals has traditionally relied on morphological features (from the seminal studies of Bayer et al. 1983, Breedy and Guzman 2003, Vargas et al. 2010); however, this is often problematic as some features are homoplastic (Sánchez and Wirshing 2005, Sánchez et al. 2007, Gori et al. 2012).Moreover, morphometric analyses have shown that the observed overlapping range for some continuous characters in octocorals is widely attributed to an apparent phenotypic plasticity, making consistent taxonomic units difficult to establish (Weinbauer andBranko 1995, Sánchez 2009).Evolutionary processes, such as interspecific hybridization, explosive radiation and incomplete lineage sorting (Hatta et al. 1999, Vollmer and Palumbi 2004, Forsman et al. 2010), may also contribute to the difficulty of distinguishing closely related species.
Although molecular studies of octocorals have been performed, resolution at the species level is not always observed, such as in the genera Pacifigorgia, Leptogorgia and Eugorgia (Vargas et al. 2014, Ament-Velásquez et al. 2016), and the Alcyonium species complex (Mc-Fadden and Hutchinson 2004), or even at the genus level, as in Plexaura, Pseudoplexaura, Eunicea and Eunicella, for instance (Sánchez et al. 2003, McFadden et al. 2006, Gori et al. 2012, Costantini et al. 2016).Based on the idea that "more is better" (Winkler et al. 2015), many authors have supported the use and concatenation of nuclear and mitochondrial markers, such as ITSs (internal transcribed spacers ITS1 and ITS2), 28S rRNA, mtMutS and Cox (partial Cox-II and I including the Igr1 region) (McFadden et al. 2006, 2011, Aguilar and Sánchez 2007, Sánchez et al. 2007).In addition, anthozoan mitochondrial genomes, especially those of octocorals, evolve 10 to 100 times more slowly than those of other metazoans (France and Hoover 2002, McFadden et al. 2004, Hellberg 2006), and possible incongruences between mitochondrial and nuclear phylogenetic reconstructions may be attributed to hybridization.
Methodological advances in morphological analysis, such as the use of morphometric statistics, can provide additional characteristics regarding the shape and structure of organisms (Gori et al. 2012).These techniques may better delimit difficult cases of species identification, reducing possible taxonomist subjectivity (Mutanen and Pretorius 2007) and molecular conflicts.Furthermore, these analyses have been shown to be at least as accurate as classical morphological and phylogenetic analyses in the delimitation of closely related species such as between Eunicellla singularis morphotypes (Gori et al. 2012), and within the Choristoneura fumiferana species complex (Lumley and Sperling 2010) and Ophion scutellaris (Thomson species group) (Schwarzfeld and Sperling 2014).
In other ecosystems, combining analyses has provided a stronger evaluation of species identification (Schwarzfeld and Sperling 2014).Thus, an integrative taxonomic approach is likely best for understanding octocoral benthic community diversity (Dayrat 2005, Padial et al. 2010, Pérez et al. 2016).
Using such an approach, we investigated two Ecuadorian morphotypes of the genus Leptogorgia Milne-Edwards and Haime, 1857, which is considered one of the most cosmopolitan genera of the Gorgoniidae family (Grasshoff 1988, Breedy and Guzman 2007, Soler-Hurtado and López-González 2012).Specifically, these two morphs may represent eco-typical variants of Leptogorgia alba Duchassaing and Michelotti, 1864, or they may constitute two distinct species coexisting in the same habitat.With more than 60 valid species (Williams and Chen 2012), Leptogorgia is widely distributed in the Mediterranean and Caribbean seas, Atlantic and Pacific oceans, and South African and sub-Antarctic coasts (Bayer 1961, Grasshoff 1988, Williams and Chen 2012).In the eastern Pacific, Leptogorgia is one of the most frequently encountered genera (in terms of species richness and abundance), with 27 described species known for this area (Breedy and Guzman 2007, Horvath 2011, Soler-Hurtado and López-González 2012).Leptogorgia species are generally restricted to shallow-water habitats (although Leptogorgia styx Bayer, 2000, has been found at a depth of 1900 m at the East Pacific Rise) and are distinguished from each other based on a few characters including the colour of the colony, branching patterns (filiform, dichotomous, or pinnate), the absence of anastomosis (in most cases), the polyp mound, and coenenchymal sclerites (spindles and captans) (Bayer et al. 1983, Grasshoff 1988, Breedy and Guzman 2007).
Here, we report our analyses of the morphological and genetic variation of two Leptogorgia alba Duchassaing and Michelotti, 1864 morphs, using molecular analyses and classical morphology, with the support of morphometric techniques, to determine whether they represent a single species or distinct species.This study contributes towards understanding the relationship between genetic and morphological variation in this case study.

Sampling
Leptogorgia colonies were collected by SCUBA diving from rocky bottoms in Machalilla National Park (Manabí, Ecuador), which is considered one of Ecuador′s most important marine-terrestrial reserves.Sampling was performed between February 2010 and June 2014.We collected 40 colonies from each of the two morphs of Leptogorgia.The attachment at the base of the holdfast with coenenchyme was left in place so that the colony could have the possibility to regenerate following collection.Growth plasticity and growth rates in closely related shallow-water species and genera are faster than initially suspected (e.g.Rossi et al. 2011, Viladrich et al. 2016).These morphs are among the most common gorgonians at 3-30 m depth in the littoral zone of Ecuador.Given the uncertainty in their species status, they are of special interest for conservation management.
To account for the high range of size variability, colonies were collected according to four size classes, determined by the distance between the holdfast and the most distant branch tip (class 1, ≤70 mm; class 2, between 71 and 140 mm; class 3, between 141 and 210 mm; and class 4, >211 mm).
Photographs of the sampled specimens were first taken underwater (Fig. 1) and then again on deck.Subsamples were either stored in absolute ethanol for molecular analysis or in 4% buffered formalin (after relaxation with menthol crystals) for morphological analysis.Formalin-fixed subsamples were subsequently transferred to 70% ethanol until analysis.The remaining colonies were air dried.
Collected specimens were deposited in the Museo Ecuatoriano de Ciencias Naturales (MECN), the octocoral reference collection of the research group "Biodiversidad y Ecología de Invertebrados Marinos" at the University of Seville (BEIM), the Museo Nacional de Ciencias Naturales in Madrid (MNCN-CSIC) or the Museu de Ciénces Naturals in Barcelona (MZB).

External morphology and SEM study
Colony fragments were prepared for scanning electron microscopy (SEM) according to standard methods (Bayer andStefani 1989, Alderslade 1998).Additionally, permanent mounts were prepared for light microscopy observation.The colonies were described and illustrated according to standard terminologies (Bayer et al. 1983, Breedy andGuzman 2007).

Morphometry of the colonies
For each colony, the total area and total area without gaps were measured (Fig. 2) using the ImageJ software (Java version of NIH image) (Abràmoff et al. 2004), calibrated using a ruler as a reference.
The following measurements were taken from each colony: maximum height, maximum width, and average and maximum lengths of the primary branches.The number of ramifications and the number of branches of each order were also counted.
The following colony parameters were measured and calculated: mean width (mean of three measurements, taken at equidistant positions at a right angle to the height) (Gori et al. 2012); height to width ra- tio; height to mean width ratio; ramification density (number of ramifications per surface area using both total area and area without gaps) (Weinbauer and Branko 1995); order of the colony [using a methodology developed by Horton (1945) and later modified by Strahler (1954)]; mean length of the primary branches; maximum length of the primary branches to maximum height ratio; tributary to source ratio of primary and secondary order branches (T/S) ("tributaries" (T) are branches that join branches of higher order (i.e.no change in order), and "sources" (S) are branches that join branches of equal order) (Brazeau and Lasker 1988); bifurcation ratio (the ratio of the number of branches of a given order to the number of branches of the next higher order) (Horton 1945, Strahler 1954, Brazeau and Lasker 1988); main thickness of the primary, secondary and tertiary order branches; angle formed between primary and secondary order branches; angle formed between secondary and tertiary order branches; and angle formed by basal branches with respect to the central axis.

Morphometric statistical analyses
As two highly collinear variables contain the same information and would be redundant for the purpose of analysis (Anderson et al. 2008), we checked for correlations among the morphometric variables measured, prior to modelling.Only variables having absolute inter-correlation values of less than 0.75 were subsequently used.The data were organized into a morphometric variable/sample matrix, and an Euclidean distance similarity matrix was calculated based on the normalized data (Anderson 2001).Differences in the multivariate structure were analysed in a distance-based permutational multivariate analysis of variance (PER-MANOVA) (Anderson 2001, McArdle andAnderson 2001).The experimental design included two crossed fixed factors: morph (with two levels) and size (with four levels).The sum of squares (SS) used was Type III SS, where every term in the model is fitted only after taking into account all other terms in the full model (Anderson et al. 2008).We used 9999 permutations of residuals under a reduced model (Anderson 2001).Homogeneity of dispersions was also tested with the PERMDISP routine, which performs a Levene-type test using the group means but obtains the p-values by permutations (Anderson 2006).Multivariate analyses were performed using the software PRIMER v6.1.11& PERMANOVA v1.0.1 statistical package (Clarke and Gorley 2006).
ML and MP supports were determined through 1000 bootstrap replicates (bootstrap values = bv).Bayesian inference of phylogenetic relationships was performed in MrBayes 3.1 (Huelsenbeck and Ronquist 2001), employing two parallel runs of 5 million generations, verifying their convergence in Tracer v1.4 (Rambaut and Drummond 2007) and looking that standard deviations of split frequencies were smaller than 0.01, with one cold and three heated Markov Chains Monte Carlo (MCMC) for each run, sampling one every 1000 replicates.The first 10% of sampled trees were discarded as burn-in, and support was evaluated based on posterior probabilities (pp).

RESULTS
Based on the evidence provided by the study of the differentiation of two Leptogorgia morphs, supported mainly by morphological and morphometric analyses, and framed by their evolutionary history, here we describe a new species within the family Gorgoniidae.
(Figs 3-4) Description of the holotype.The colony is 155 mm long and 120 mm wide, irregularly pinnate; branches slender, mostly in a plane (Fig. 3A).Living colony dark pink in colour, and pink or light pink in a dried state.Unbranched distal twigs up to 50 mm in length and 19 mm in diameter, compressed proximally, more cylindrical and slightly tapered distally (Fig. 3A, B).Slightly marked longitudinal grooves along the thick basal branches and near the base.Polyps retract within slightly raised polyp mounds, sparsely distributed all around the branches with oblong apertures (Fig. 3B).Coenenchymal sclerites colourless (Fig. 3C).Dominant sclerite type spindles, straight or bent, some with a marked waist, measuring up to 0.14 mm in length and 0.04 mm width with 4-6 whorls of tubercles (Figs 3C, 4A).Capstans measure up to 0.08 mm in length and 0.03 mm wide (Fig. 3B).Crosses not present.Anthocodial sclerites hyaline or colourless, rods measure up  to 0.10 mm in length and 0.02 mm wide, with some marginal projections (Figs 3C, 4C).
Variability.Colonies vary up to a maximum of 316 mm in length and 236 mm in width.Unbranched distal twigs measure up to 13.9 cm in length and 19 mm in diameter.Spindles measure 0.06-0.17mm in length and 0.03-0.06mm in width.Capstans measure 0.03-0.10mm in length and 0.02-0.07mm in width.Anthocodial sclerite rods measure 0.05-0.14mm in length and 0.01-0.06mm width.
Geographic and bathymetric distribution.Leptogorgia manabiensis n. sp. is known from the type locality in Los Ahorcados, Cope, Los Frailes, Punta Gruesa, Punta Mala, Isla de Salango and Isla de la Plata (continental coast of Ecuador), living on rocky bottoms in shallow waters at a depth of 5-23 m.
Etymology.The specific epithet refers to the Province of Manabí (Ecuador), the type locality where the new taxon was discovered.
Remarks.According to classical morphological characters, L. alba and L. manabiensis n. sp.show two main differences (Figs 3,4,5,6).The first is colony colour, which is clearly distinct between the two spe-cies (Fig. 1): L. alba is white (alive and dry) and L. manabiensis n. sp. is deep pink (alive) and pink/light pink (dry).The second difference is the length and diameter of the unbranched distal branches, which are slightly longer and thicker in L. alba, up to 153 mm in length and 2.7 mm in diameter.In L. manabiensis n. sp., unbranched distal twigs reach up to 139 mm in length and 1.9 mm in diameter.Other differences found between the two studied species are discussed in the morphometric analysis section below.
For a complete description and list of the materials of Leptogorgia alba examined for this study, see Soler-Hurtado et al. (2016).

Morphometric analyses
Nine morphological characteristics were selected: maximum height, maximum length of the primary branches to maximum height ratio, order of colony, bifurcation ratio, tributary to source ratio of secondary order branches, main thickness of the primary order branches, angle formed between primary and secondary order branches, angle formed between secondary and tertiary order branches, and angle formed by basal branches.There were significant differences in the multivariate structure between the species, L. alba and L. manabiensis n. sp., and among colony size class (Table 3).Interactions between species and size class were not detected.The multivariate dispersion did not show any significant difference, either between the levels of the factor "morph" (PERMDISP F1,75: 2.25; P(perm): 0.14), or among the levels of the factor "size" (PER-MDISP F1,73: 1.22; P(perm): 0.36).Figure 7 shows a graphical comparison of the nine morphological characteristics analysed in the two gorgonian species.Colonies of L. alba tended to be characterized by higher maximum length of the primary branches to maximum height ratio, the angle formed between primary and secondary order branches, the formed between secondary and tertiary order branches, and the angle formed by basal branches (Fig. 7).In contrast, colonies of L. manabiensis n. sp.tended to be characterized by a higher order of colony, bifurcation ratio, and tributary to source ratio of secondary order branches (Fig. 7).In summary, the two forms cannot be considered as ontogenetic stages of a single species.

Molecular analyses
One matrix was prepared with the concatenated data from the four marker alignments: mtMutS+Cox+ITSs+28S (16 specimens, 3353 characters).New sequences were deposited in GenBank (Table 2).According to jModelTest, the nucleotide substitution model that best fit the data was HKY+I.Trees obtained by Bayesian, MP and ML analyses all showed the same topology.In the concatenated alignment including the outgroups, a total of 152 sites were parsimony-informative.However, only 7 sites were parsimony-informative between L. manabiensis n. sp. and L. alba: six from the ITSs region and one from 28S.The mitochondrial fragments were uninformative.Thus, the topology of the concatenated tree largely matched the tree obtained using nuclear data alone (ITSs+28S) (not shown).
The reconstruction based on the data obtained from the four concatenated genes (Fig. 8) showed a well-supported Leptogorgia alba-L.manabiensis n. sp.clade (Clade I) (pp=1; bv=100).Specimens of L. alba and L. manabiensis n. sp. were divided into different subclades, but without taxonomic congruence.A well-supported polyphyletic assemblage consisting of Eugorgia daniana as sister group to the L. alba + L. manabiensis n. sp.clade and E. ahorcadensis (Clade II) was also observed (pp=1; bv≥85).Leptogorgia diffusa and L. obscura, which are closely related to L. mariarosae (Clade III), formed the sister group to Clade II (pp0=1; bv≥97).Pacifigorgia stenobrochis and Antilogorgia bipinnata (designated as an outgroup) were basal to clades I, II and III.

DISCUSSION
The morphological and morphometric results presented here, which showed significant differences in colony shape and branching, support the existence of two distinct species, L. alba and L. manabiensis n. sp., coexisting in Machalilla National Park (Ecuador).In addition, the molecular data has contributed to our understanding of the evolutionary history of this group of species.Overall, we have shown that the delimitation and identification of challenging species is possible through an integrative taxonomic approach based on morphology and on molecular and morphometric techniques.Indeed, in this study, evidence based on morphological, molecular and morphometric analyses have provided three semi-independent datasets.Taken together, these evidences help to better understand the true diversity and evolutionary history within this species group.
Previous morphological studies based on a limited set of characters, including L. alba (Breedy and Guzman 2007), suggested that variability in colony colour and shape represented intraspecific morphological variation, correlated with differences in environmental factors principally due to local currents and depth (Lewis and Wallis 1991, Carlon and Budd 2002, Gori et al. 2012).However, the two species analysed here were living in close proximity in the same study area; thus, differences in colony morphology between L. alba and L. manabiensis n. sp.cannot be attributed to phenotypic plasticity based on environmental conditions or ecological niches.Moreover, Guzman and Breedy (2008) divided the genus Leptogorgia into three species groups based on morphological characteristics, including an "alba-group" consisting of some species "with flat or slightly raised polyp-mounds, pinnate or dichotomous branching and white colonies" (Guzman and Breedy 2008).In addition, Ament-Velásquez et al. (2016), using molecular phylogenetic analyses, demonstrated that the white phenotype was consistent and revealed a synapomorphy for the previously defined "albagroup".In light of our analyses, based on the morphological definition of the "alba-group", L. manabiensis n. sp.must not be considered part of it due to its pink colour, reinforcing the idea that it is a differentiated species.However, on the other hand, the molecular similarity and relationship between L. manabiensis n. sp. and L. alba revealed the artificialness of using colour as a character to define this "alba-group".
Most of the morphometric characters considered in this study were not noted in previous descriptions and revisions.In addition, our study shows that the differences between L. alba and L. manabiensis n. sp. were significant in all size classes, so not restricted to only larger individuals or specimens in the first stage of development.In a speciation framework, some morphological characteristics might remain stable for each species (e.g. the colour of the colony, some morphometric branching measures), while others are shared between the two species (Forsman et al. 2010, Vollmer andPalumbi 2004).Given the importance of understanding the morphological variation present in these organisms, the mechanisms and heritability involved in this variation should be considered (Vermeij et al. 2007).Indeed, these variations could be due to genetic polymorphisms, in which the morphology of the species is determined by genotypes, independent of the ecological niche occupied (Carlon andBudd 2002, Vermeij et al. 2007).The significant differences observed in the morphometric and morphological analyses were not completely consistent with our molecular results, in which the relationship between L. alba and L. manabiensis n. sp.resulted in a polyphyletic assemblage.As expected, the mitochondrial data were only useful for comparing the ingroup (L.alba and L. manabiensis n. sp.) and outgroup, confirming the need to complement the use of these markers with nuclear ones in the study of close relationships (Yasuda et al. 2015, Aguilar and Sánchez 2007, McFadden et al 2010).Hybridization or incomplete lineage sorting, in which L. alba and L. manabiensis n. sp.unequally share a genetic polymorphism, may also explain the lack of congruence of the molecular nuclear data.This phenomenon has been observed between closely related species within other coral genera such as Heliopora (Yasuda et al. 2015), Seriatopora (Flot et al. 2008) and Acropora (Vollmer and Palumbi 2004), and in other invertebrates such as Cypraeidae marine gastropods (Meyer and Paulay 2005), Melanoplus grasshopper species (Carstens and Knowles 2007) and the butterfly family Lycaenidae (Wiemers and Fiedler 2007).
Although for some molecular phylogenies of corals, hybridization processes have been used to explain a lack of resolution, polyphyly or paraphyly (Hatta et al. 1999, Diekmann et al. 2001, McFadden and Hutchinson 2004), possible ancestral polymorphisms linked to a delayed process of evolution, such as a source of shared haplotypes between species, should also be considered (van Oppen et al. 2001, Vollmer andPalumbi 2004).Incomplete lineage sorting can be a particularly important factor to consider for rapidly and recently diverged species (Hoelzer and Melnick 1994, Henning and Meyer 2014, Eberle et al. 2016), as it makes species misidentifications more likely, thus leading to conflicting or misleading findings (van Velzen et al. 2012, Nater et al. 2015).Although the molecular analyses were unable to provide diagnostic characters for the complete delimitation of L. alba and L. manabiensis n. sp., they indicated that the phenotypic variation observed may be the result of a recent speciation and diversification event.Wirshing and Baker (2015) suggested that, due to the common limited genetic variation within gorgonian species, a re-evaluation of species-level morphological characters is needed.Followed this recommendation, we provide new diagnostic morphometric characters (e.g., bifurcation ratio, branch thickness and branch angle) to complement classical morphological characters used for species identification, within an evolutionary context.
Evolutionary status and the relatedness of species, and the traits used to identify them, may not have a direct positive relationship (Carlon and Budd 2002).However, this is not to say that diagnostic species criteria are not mutually exclusive: alternate criteria could indicate different aspects of biological information, providing more robust support about the identification of species boundaries (Rocha-Olivares et al. 2001).
Overall, the differentiation reported here between the new species L. manabiensis n. sp. and L. alba may help clarify previously cited discrepancies and reduce unexplained variability in these challenging gorgonian species.Moreover, this study confirms that the use of other tools and frameworks can be valuable for delimiting species for comparative studies.
Of the approximately 28 Leptogorgia species that have been described for the eastern Pacific (Breedy and Guzman 2007;Horvath 2011), six (including L. manabiensis n. sp.) are distributed along the relatively unknown coasts of Ecuador (Soler-Hurtado andLópez-González 2012, Soler-Hurtado et al. 2016, this paper).Many of the species descriptions of Leptogorgia in the eastern Pacific have come from specimens collected from northern localities such as California, Mexico, Costa Rica, and Panama (Breedy and Guzman 2005, 2007, Breedy et al. 2012).Therefore, this study increases our knowledge of the biodiversity of this group for other regions of this ocean.Furthermore, it reinforces the observation that Leptogorgia diversity is higher along the eastern Pacific coast of the Americas than along the western Atlantic coast, where only 12 valid recognized species are known to date (Bayer 1961, Devictor andMorton 2010).To better understand the evolution and biodiversity of gorgonian gardens, and to implement effective conservation strategies, future research focusing on the biological, ecological, and behavioural characters of these organisms are necessary, especially in the case of L. manabiensis n. sp., due to its restricted distribution area.

Fig. 1 .
Fig. 1. -Underwater images of the two Ecuadorian morphs collected in Los Frailes, Machalilla National Park (Ecuador), where L. alba (white, left) and L. manabiensis n. sp.(pink, right) were observed at a sympatric location (side by side).
Fig. 2. -Images showing the differences between total area of the colony (left) and area without gaps (right).

Table 1 .
-Primers and PCR conditions used to amplify fragments of nuclear and mitochondrial genes in the studied Leptogorgia species.In the PCR protocol column, we indicate annealing temperature, time in seconds for the denaturation (at 94ºC), annealing and elongation (at 72ºC), respectively, and the number of cycles for each gene, independent of the primer combination used.*PCRs were performed with the following primer combinations: Cox: 1+3, or 1+4 and 2+3; mtMutS: 5+6; ITSs: 7+8; 28S rRNA: 9+10 or 9+11.

Table 2 .
-Gorgoniidae species involved in the molecular comparisons carried out in this study.Materials in bold are species sequenced for this study.Sequences with duplicate complete names are also numbered (e.g.1-6) for the purpose of correctly identifying the sequence in Figure8.

Table 3 .
-PERMANOVA table of results.Two crossed fixed factors: morph and size.