Predominant east to west colonizations across major oceanic barriers: Insights into the phylogeographic history of the hydroid superfamily Plumularioidea, suggested by a mitochondrial DNA barcoding marker

Abstract We provide preliminary insights into the global phylogeographic and evolutionary patterns across species of the hydrozoan superfamily Plumularioidea (Cnidaria: Hydrozoa). We analyzed 1,114 16S sequences of 198 putative species of Plumularioidea collected worldwide. We investigated genetic connections and divergence in relation to present‐day and ancient biogeographic barriers, climate changes and oceanic circulation. Geographical distributions of most species are generally more constrained than previously assumed. Some species able to raft are dispersed widely. Human‐mediated dispersal explains some wide geographical ranges. Trans‐Atlantic genetic connections are presently unlikely for most of the tropical‐temperate species, but were probably more frequent until the Miocene–Pliocene transition, before restriction of the Tethys Sea and the Central American Seaway. Trans‐Atlantic colonizations were predominantly directed westwards through (sub)tropical waters. The Azores were colonized multiple times and through different routes, mainly from the east Atlantic, at least since the Pliocene. Extant geminate clades separated by the Isthmus of Panama have predominantly Atlantic origin. Various ancient colonizations mainly directed from the Indian Ocean to the Atlantic occurred through the Tethys Sea and around South Africa in periods of lower intensity of the Benguela upwelling. Thermal tolerance, population sizes, dispersal strategies, oceanic currents, substrate preference, and land barriers are important factors for dispersal and speciation of marine hydroids.


| INTRODUC TI ON
Spatial distributions and evolutionary diversifications of marine species are constrained by biological/physiological adaptations to disperse and survive in different environmental conditions. In general, dispersal ability, temperature, productivity, oceanic circulation and habitat complexity and configuration are major determinants of population abundance and intra-and interspecific ecological interactions that promote diversification and control biogeographic ranges (reviewed by Bowen et al., 2016;Costello & Chaudhary, 2017).
Contemporary and historic patterns of oceanic circulation, physical barriers, and availability of suitable habitat, as well as ecological interactions and historic demography, will then affect the spatial genetic structure and speciation patterns of marine organisms (e.g., Ayre, Minchinton, & Perrin, 2009;Palumbi, 1994;Vermeij & Grosberg, 2010).
These medusoids are nonfeeding and short-lived rudimentary medusa-like organelles that liberate gametes into the water column.
Although a few species produce planktonic medusoids, spatial proximity is required for fertilization, so long-distance dispersal should be highly constrained for even these species of Plumularioidea.
These characteristics, along with the frequent presence of plumularioids in benthic environments worldwide, both in shallow and deep waters, make these hydroids interesting subjects for the investigation of phylogeographic patterns and historical processes in oceanic settings.

| ME THODS
The present study uses 1,114 16S sequences of different Plumularioidea, collected from various depths and locations worldwide (see sampling data and Genbank accession numbers in Table   S1; Figure 2).  Table S1.

| Phylogenetic analyses
DNA sequences were entered in Geneious R10.0.2. The different datasets of sequences analyzed included all sequences of Plumularioidea together and sequences of each family separately.
DNA sequences were aligned with MAFFT (Katoh, Misawa, Kuma, & Miyata, 2002). Subsequently, for the datasets with sequences separated by families, we used the Gblocks server (Castresana, 2000) with relatively strict settings (excluding gapped positions and minimizing the number of contiguous nonconserved positions) to obtain alignments (alignments were as in Moura et al., 2018; see Table S1 of the "Supplementary Information" of ref Moura et al., 2018).
Phylogenetic reconstructions included maximum likelihood (ML) and Bayesian inference (BI) tree searches for all the generated alignments. The general time-reversible model of nucleotide evolution with gamma and invariant parameters (GTR + G+I) was F I G U R E 2 Geographical location and depth range of DNA sequences used in this study for each Plumularioidea family used for all analyses, after assessment of the most suitable model using the "Akaike Information Criterion." ML analyses were conducted in PhyML (Guindon & Gascuel, 2003) (Ronquist, Teslenko, & Mark, 2012) and consisted of two runs of four chains each of 100 million generations with trees sampled every 1,000 generations after a burn-in fraction of 25% of the trees (as in Moura et al., 2018). A summary of the nodal support obtained with the different ML and Bayesian analyses with the different sequence datasets is shown in Figure S1 and was contrasted with the "Ancestral state reconstruction" results presented in Figure 3 and Figure S2.
To estimate divergence times, we applied the Bayesian relaxed dating implemented in BEAST v.1.8.3 (Drummond, Suchard, Xie, & Rambaut, 2012), for the dataset with all the 16S sequences of Plumularioidea with indels present. We used a Yule speciation process prior for rates of cladogenesis (Drummond & Rambaut, 2007) in combination with an uncorrelated lognormal clock. As calibration points, we used clades presumed to have diverged due to the Central American Seaway closure ca. 3 Mya (e.g., Lessios, 2008); see subsection "Connectivity between the Atlantic and Eastern Pacific"). Thirty-seven runs ranging from 63.5-100 million generations each (average = 92.4 millions), with every 10,000 generations sampled, were performed on the online computer cluster CIPRES Science Gateway (Miller, Pfeiffer, & Schwartz, 2010 (Drummond et al., 2012) was used to combine tree files from the multiple runs, resampling states every one million generations and using a burn-in of 2,500 for each run. TreeAnnotator v1.8.3 (Drummond et al., 2012) was used to produce a consensus tree.

| Ancestral state reconstructions (ASR)
Likelihood-based ASR were conducted in Mesquite v.3.2 (Maddison & Maddison, 2017). Characters (East Pacific = 0, East Atlantic = 1, West Atlantic = 2, Indo-West Pacific = 3) were traced over the tree resulting from analyses performed with BEAST (Drummond et al., 2012). To account for topological uncertainty, we used the "trace character over trees" option, which summarizes the ASR over a series of trees. All reconstructions were integrated over 5,699 trees, obtained with LogCombiner v1.8.3 (Drummond et al., 2012) after combining the thirty-seven runs obtained from BEAST, and resampling states at a lower frequency of 600,000, using a burn-in of 2,200 for each run. As a model of evolution for the ML reconstructions, we employed the Markov k-state 1 (Mk1) parameter model, with equal probability for any particular character change.

| RE SULTS AND D ISCUSS I ON
The large number of 16S sequences of Plumularioidea used in this study provided particularly good representation of shallow and deep-water taxa of the N Atlantic ( Figure 2). Similarly, relatively strong sampling in the Indo-Pacific, especially for the families Aglaopheniidae and Plumulariidae, permitted an initial understanding of phylogeographic relationships between oceans. We use the "molecular clock" of Plumularioidea to assess hypotheses regarding approximate epochs of splitting of lineages, in relation to past climatic, geological, and oceanographic changes. We use the term "colonization" to denote that the extant species is found in a location different than the one inferred to be occupied by its ancestor. which induced extinctions in other groups (e.g., Sepkoski, 1992,).

| Evolutionary relationships over time
The great diversification of extant lineages started mainly thereafter, coinciding with the subsequent general cooling of oceans.
The high levels of cladogenesis observed since the last third of the Miocene, notably observed in Plumularioidea from NE Atlantic, match with the great constriction of the Tethys/Paratethys Seaway (Hüsing et al., 2009;Popov et al., 2006), followed by the rise of the Isthmus of Panama (Lessios, 2008), which led to major changes in oceanic circulation and climate (Butzin, Lohmann, & Bickert, 2011;Hamon, Sepulchre, Lefebvre, & Ramstein, 2013;Mudelsee & Raymo, 2005). Such events likely induced divergence of populations, but also extensive species radiations (similar to the history of cetaceans as reported by Steeman et al., 2009), as suggested for example by the radiation of Aglaophenia species of the NE Atlantic ( Figure S1).
Numerous lineage divergences occurred through the (Plio-) Pleistocene ( Figure 3), as they did in other marine life (Ludt & Rocha, 2015;Pinheiro et al., 2017), when high origination rates of biodiversity likely counterbalanced high extinction rates (Allmon, Rosenberg, Portell, & Schindler, 1993). Interestingly, Antarctic clades, of Schizotricha and Oswaldella, probably radiated into a number of species in a short amount of time (Pleistocene-Pliocene; Figure S1), possibly due to colonizations after deglaciation(s), at a time of increased Southern Ocean surface water productivity and elevated circum-Antarctic temperatures (Cook et al., 2013).
F I G U R E 3 Time-calibrated phylogeny of the hydroid superfamily Plumularioidea. Nodes collapsed presented posterior probabilities below 70%. Lineages were subsumed by putative species as determined by Moura et al. (2018). Circles at nodes indicate relative proportion of ancestral states (oceanic affinity) given by the ASR analyses: white-East Pacific; green-West Atlantic; blue-East Atlantic; black-Indo-West Pacific; gray-equivocal; red-node absent). Arrows indicate probable direction of colonization between oceanic regions: turned left means eastward colonization, turned right westward colonization. See Figure S2 for complete results of the ASR analyses
Nevertheless, more haplotype sampling, both in studied and little studied regions (e.g., cold NW Atlantic and S Atlantic), may reveal the presence of more haplotypes shared across the Atlantic, as well as other phylogeographic patterns.
Kirchenpaueria halecioides exhibits the same haplotype in Argentina, Madeira, Azores, and mainland Portugal, mostly in ports and marinas (including on a ship hull). This taxon is thus likely to disperse via boats to remote locations. Remarkably, the eight other haplotypes of K. halecioides, also mainly collected from ports or marinas, are not so widely dispersed (each one collected once). This suggests that the haplotype found simultaneously at eight sampling stations may be the best adapted for adverse/diverse abiotic conditions associated with boat travel and to different destinations.
Macrorhynchia philippina has a widespread haplotype occurring in W Africa, Brazil, the Caribbean, and the E Pacific, suggesting also dispersal mediated by boats. Two other haplotypes of that taxon are widely distributed across the Indo-Pacific and Indian Ocean, respectively. Besides boat traffic, the wide distribution of M. philippina, may also be explained by the liberation of medusoids as part of its life cycle, which could aid dispersal via oceanic currents, and its large population's sizes in warm waters.
Halopteris cf. alternata ("lineage 5") has small sessile colonies and is frequently an epibiont on algae, sponges, and rocks. Although it seems to not release medusoids, the rafting abilities of this taxon (e.g., perhaps on algae) may explain the presence of two haplotypes with amphi-Atlantic distribution, with one of these haplotypes also present in the central E Pacific. Although this taxon was not found in marinas or ports, we do not discard the hypothesis of its transport via boats as well.
Antennella similis ("lineage 3") forms small colonies collected from a wide variety of substrates, including rock, dead coral, sponges, and fishing lines. Therefore, rafting on natural of artificial substrates probably also explains its recent trans-oceanic dispersal.
That no other Plumularioidea were represented by shared haplotypes between the E and W Atlantic suggests that presently trans- A few other (putative) species contain genetically close haplotypes on both sides of the Atlantic, suggesting relatively recent trans-Atlantic genetic connections, namely Aglaophenia postdentata, Nemertesia cf. antennina ("cryptic lineage 1"), Plumularia cf. setacea ("lineage 11"), and Monotheca cf. margaretta ("lineage 2"). The latter two taxa are frequent epibionts (e.g., Calder, 1997) and may have crossed the Atlantic by rafting facilitated by the strong and warm South Equatorial Current (Lumpkin & Garzoli, 2005). The observation of Nemertesia "cf. antennina" in deep waters of the west and east North Atlantic is surprising; large population sizes and/or an unidentified deep-sea current may explain the amphi-Atlantic distribution of this species. The geographical range of Aglaophenia postdentata on both sides of the Atlantic, central E Pacific, and Indian Ocean discovered by Moura et al. (2018) is intriguing because this species had been recorded only from the Indo-Pacific and the Caribbean (Galea, 2010). Despite its wide geographical range, A. postdentata is rare, but perhaps its observed occasional growth on algae could have contributed to dispersal through rafting.  (Harzhauser, Piller, & Steininger, 2002;Vermeij, 2012;Vermeij & Rosenberg, 1993). It is not in agreement, however, with the predominantly eastward dispersal discovered in some fishes (Beldade et al., 2009;Floeter et al., 2008;Muss, Robertson, Stepien, Wirtz, & Bowen, 2001) and mollusks (Vermeij & Rosenberg, 1993). Our molecular dating analyses ( Figure 3) suggest that these trans-Atlantic genetic interconnections ceased around the end of the Miocene or in the Pliocene (e.g., in agreement with Baarli et al., 2017;Vermeij, 1997).

| Ancient trans-Atlantic dispersal
This timing coincides with the constriction of the Central American Seaway (CAS) and Indonesian Gateway, which followed that of the Tethys Sea, leading to the end of the global equatorial surface current, global cooling and change of the internal circulation of Atlantic waters von der Heydt & Dijkstra, 2005;Karas et al., 2017;Keigwin, 1982;Landau, Silva, & Vermeij, 2015;Silva, Landau, & La Perna, 2011). These oceanographic changes might have thus promoted vicariance between the two sides of the Atlantic, ceasing the trans-Atlantic distribution of many warm-water species.
Regarding trans-Atlantic dispersal through deep waters, there were two instances of past genetic connections in Nemertesia lineages from the east to western Atlantic that likely took place around the end of the Pliocene (Figure 3). Conversely, two past colonization events with reverse direction, that is, from the west to the east Atlantic, are suggested in the deep-water clade of Cladocarpus and Aglaophenopsis (Figure 3). Although the direction of dispersal is inconclusive in representatives of Cladocarpus sigma, the divergence between Cladocarpus carinatus and Aglaophenopsis cartieri suggests a colonization from the NW to the NE Atlantic through deep waters ( Figure 3), perhaps only up to the mid-Atlantic ridge, where A. cartieri seems to be endemic (Ramil & Vervoort, 1992). Analogously, Eilertsen and Malaquias, (2015)  Atlantic deep water (e.g., Billups, 2002;Karas et al., 2017;Keigwin, 1982;Wright, Miller, & Fairbanks, 1991).

| Colonization of the Azores
The importance of seawater temperature for dispersal is evidenced by the Azorean fauna (Cornelius, 1992b), located close to the mid- In terms of dispersal strategy, rafting on algae or other floating animals or detritus should explain the presence of many of the shallow-water hydroids in the Azores (as already hypothesized by Cornelius, 1992b); this is a dispersal method consistent with the multiple colonization events verified across various taxa (see paragraph above; Figure S2). Some hydroids have clearly arrived to the archipelago attached to boats, as is the case for Kirchenpaueria halecioides. The arrival of hydroids through deep waters should have been mostly through large or spatially close reproducing populations dispersed by gamete and larva liberation, carried by oceanic currents.

| Migrations across Central America
Only two of the species represented in this study-Macrorhynchia philippina and Halopteris cf. alternata (lineage 5), exhibit the same 16S haplotypes in the Atlantic and E. Pacific, which suggests recent or ongoing genetic interconnections between these oceanic basins.
These taxa reveal preference for warm/tropical waters and present amphi-Atlantic distributions as well (cf. last section). We therefore suspect these two species crossed the Isthmus of Panama with the aid of boats and/or rafting on other floating or swimming substrates (see Discussion on their dispersal strategies in the last subsection).
Eleven geminate clades of Plumularioidea were probably separated with the rise of the Isthmus of Panama, since the middle/end of the Miocene until the completion of that barrier by the Pliocene (see in Figure 3 lineages with white and green circles; e.g., Lessios, 2008). Excluding the clades mentioned in the previous paragraph for which we suspect human-mediated transport, we used this geological event for the molecular clock calibrations the following clades: Aglaophenia postdentata; A. trifida + A. prominens; Halopteris diaphana "lineage 1"; Halopteria alternata "lineage 3"; Monostaechas spp.; Plumularia setacea "lineage 12"; Plumularia floridana complex.
These sister taxa were collected from shallow waters of the Pacific and Caribbean coasts of Panama and/or Costa Rica, and share close morphological and genetic similarities, suggesting that they diverged at a time close to the completion of the Isthmus ca. 3 Mya.
The following two lineages probably migrated across the Central American Seaway but were excluded from "molecular clock" calibrations because their respective sister groups on the Caribbean side were not sampled or readily apparent in the phylogenetic results by phylogenetic analyses: Antennella "secundaria lineages 6 and 7" and Antennella "secundaria lineage 11" + Monostaechas sp. colonization patterns verified in a great variety of other taxa (e.g., Groves, 1997;Kronenberg & Lee, 2007;Roopnarine, 2001;Waller, 2007); reviewed by Leigh et al. (Leigh, O'Dea, & Vermeij, 2013). We do not discard also the suggestions of eastward flow across the CAS, before the seaway closure (Miocene to Early Pliocene), followed by the western Atlantic "dramatic extinctions during and after formation of the isthmus" (as reviewed by Leigh et al., 2013), considering the lack of Plumularioidea fossils and extant genetic lineages that could support that complementary hypothesis. Nevertheless, the predominant westward colonization across the CAS we note for the Plumularioidea hydroids fits with the general westward pattern of trans-Atlantic colonizations we verified until the Miocene restriction of the circum-tropical surface current, as discussed in the subsection "Trans-Atlantic genetic connectivity".

| Migrations around the southern tip of South America
Our data only display one evident example of colonization across the seaway between South America and Antarctica, represented by a cold-water clade of Plumularia species ("lineage 7", plus the nominal species P. duseni, P. virginiae, and P. lagenifera) sampled from W.
USA, Chile, Argentina, and New Zealand. That transition occurred probably from the south Pacific to the south Atlantic possibly before the early-middle Pleistocene (Figure 3), when glacial episodes likely induced lineage divergence. Interestingly, prior to that event, this clade of Plumularia likely reached the E Pacific through the NW Atlantic via a trans-Artic route (see below). Furthermore, that clade is apparently absent from the warm waters of the E. Pacific, presenting an antitropical distribution along the cold shallow waters of the E. Pacific, which reflects the sensitivity of these invertebrates to seawater temperature.  (Figure 3), in accordance with dates provided by fossil taxa, between ca. 5.3 and 3.5 Ma (Gladenkov, Oleinik, Marinkovitch, & Baranov, 2002;Vermeij, 1991Vermeij, , 2004. These trans-Artic Plumularioidea colonizations were probably from the Atlantic to the Pacific (Figure 3), which contradicts the prevalent eastward gene flow direction assumed for most marine taxa (e.g., Briggs & Bowen, 2013;Durham & MacNeil, 1967;Marincovich & Gladenkov, 1999;Vermeij, 1991;Vermeij, 2004) but agrees with inferred mollusk colonizations and predominant southward currents through the Bering Strait when it opened ca. 5.5-4.8 Mya (Marincovich, 2000;Marincovich, Barinov, & Oleinik, 2002).

| Migrations through the trans-Arctic route
However, due to an underrepresentation of Plumularioidea samples from the temperate/cold waters of the NW Atlantic, we cannot exclude the hypothesis of postglacial recolonization of the NW Atlantic from the NW Pacific.
The split of Nemertesia ramosa collected from the NE Atlantic and Mediterranean, in relation to Nemertesia sp. from Japan, seems to have occurred by the end of the Miocene -Pliocene (Figure 3; Figure   S2). By then, the Tethys Sea was already closed; thus, the divergence of these two clades is explained by the intensification of the Benguela upwelling system. This hypothesis agrees with the geographical distribution of N. ramosa in the eastern Atlantic through South Africa, and in the Indian Ocean in Mozambique (Ansín Agís, Ramil, & Vervoort, 2001). Our analyses suggest that in this case, colonization possibly occurred from the Atlantic to the Indian Ocean.
Aglaophenia postdentata seems to have an Indo-Pacific origin, with apparent colonization at (at least) two independent times to the Atlantic, because lineages from the SW Indian Ocean cluster independently from lineages from Central West Africa and from the Caribbean, respectively (Figure 3; Figure S2). Such transitions likely occurred around Cape Agulhas, before the Pleistocene, when the intensity of the Benguela upwelling and Agulhas Current experienced cyclic fluctuations (e.g., Cornelius, 1992a;Rocha et al., 2005;Shannon, Lutjeharms, & Nelson, 1990;Vermeij & Rosenberg, 1993).

Plumularia filicaulis is represented by two lineages, one from
Australia the other from the Atlantic side of South Africa, with the latter probably derived from the Indo-Pacific. Such vicariance seems to have occurred by the early Pleistocene (Figure 3). This species did not reach further north in the Atlantic (e.g., Watson & Wells, 1997), thus not surpassing the biogeographic barrier caused by the upwelling of the cold Benguela current. An hypothesis is that this taxon might have dispersed through surface currents of the thermohaline circulation connecting Australia with South Africa.

| Ancient migrations through the Tethys Sea
We found four probable examples of genetic colonization through the Tethys Sea, prior to its closure around the Middle Miocene (see in Figure 3 clades with circles and arrows with black and blue colors; e.g., Harzhauser et al., 2002;Vrielynck, Odin, & Dercourt, 1997;Harzhauser et al., 2007). Furthermore, our results suggest these colonization events occurred from the Indian Ocean toward the Atlantic (Figure 3), a similar pattern as in fishes (Briggs & Bowen, 2013), consistent with modeling studies that suggest predominant water flow from the Indian to the Atlantic Oceans through the proto-Mediterranean (e.g., Herold, Huber, Müller, & Seton, 2012;von der Heydt & Dijkstra, 2005). In contrast, Vermeij (2001) mentions invading clades of mollusks from tropical America and the European part of Tethys, toward the present Indian Ocean, before the Miocene.
The most interesting colonization between the Atlantic and Indian Ocean is represented by the clade containing Atlantic species of Aglaophenia (e.g., A. tubulifera, A. picardi, A. lophocarpa, A. pluma) that derived clearly from the Indo-Pacific and radiated extensively in the Atlantic into several species after the closure of the Tethys Sea ( Figure 3). Again, this may be evidence of the effect of the colonization of a new oceanic basin (i.e., the Atlantic), possibly with low biodiversity levels prior to that time, that likely experienced dramatic change in its oceanographic conditions after the closure of the Tethys Sea (e.g., Butzin et al., 2011;Hamon et al., 2013). Atlantic Gymnangium also clearly derived from the Indo-Pacific, but due to low support for some phylogenetic relationships between some clades of that genus, we cannot infer the route that the ancestors of Gymnangium speciosum (lineage 1) and G. sinuosum likely took to colonize the Caribbean Sea. However, an ancestor of Gymnangium montagui probably colonized the NE Atlantic from the Indian Ocean through the Tethys Sea; and subsequent colonizations occurred to the Caribbean side resulting in Gymnangium allmani (lineages 3 and 4) and Gymnangium speciosum ("lineage 2") ( Figure 3).

| Dispersal strategies
The majority of the Plumularioidea (cf. Bouillon et al., 2006) disperse only through the release to the water column of gametes and shortlived planula larvae without much swimming capability (e.g., Hughes, 1977;Hughes, 1986;Postaire, Gelin, et al., 2017a;Postaire, Gélin, et al., 2017b;Yund, 1990), and occasionally through rafting (e.g. , Calder 1995;Cornelius, 1992aCornelius, , 1992b or sporadic detachment of (pieces of) colonies from their substrate. Consequently, and as noted ( Figure   S1), these hydroids exhibit great population subdivisions often related to physical distance, which tends to increase with the decrease of organismal abundances locally (because close proximity of colonies is often required for successful sexual reproduction).
Macrorhynchia philippina presents a phylogeographic structure similar to other Plumularioidea ( Figure S1), conforming to generalized seawater circulation patterns. However, this species has been able to overcome hard and permeable barriers to dispersal easier than all other plumularioids (Figure 3). Macrorhynchia philippina, possibly with Indo-Pacific origin (Figure 3), likely takes advantage of boat traffic for dispersal, as it was sampled near ports (Moura et al., 2018) and from artificial substrates like shipwrecks and is also associated with ropes (Riera, Espino, & Moro, 2016) and found in very shallow depths (Ansín Agís et al., 2001). Additionally, this species releases medusoids , has a high concentration of nematocysts that provide protection, and develops prominent colonies (pers. obs.), allowing for large population sizes and high sexual fecundity.
Kirchenpaueria halecioides and Halopteris cf. alternata (lineage 5) are the other species represented with widely distributed haplotypes (Figure 3), probably as a result of dispersal by boat traffic.
Aglaophenia postdentata, despite our failure to detect widely dispersed haplotypes, seems to present a circum-tropical distribution, with relatively small genetic distances between populations of various oceanic areas. Only this species and M. philippina were found simultaneously in the Indo-Pacific and W and E Atlantic. Because A. postdentata was not detected near ports or marinas, we do not suspect human-mediated dispersal. Although it appears to have small population sizes at present, its ability to overgrow algae suggests that rafting assisted by oceanic currents is the most plausible explanation for its large-scale dispersal. However, we know little about its life cycle, and it is possible that medusoid release or longlived planula larvae could be discovered for this species.
Finally, reproduction and population sizes, and ultimately dispersal, are also likely to be affected by the growth mode of colonies, which impact the number of reproductive structures produced (Moura et al., 2012). For example, Nemertesia ramosa has large population sizes both in shallow and deep waters, and develops robust ramified colonies from which several reproductive structures can originate, favouring genetic connections across wide bathymetrical ranges. Inversely, demonstrating the correlation between hydroid growth and food availability (Di Camillo et al., 2017), "true" Nemertesia antennina develops prominent colonies in shallow waters only, leading to wide dispersal of haplotypes throughout the NE Atlantic and Mediterranean, whereas cryptic Nemertesia "antennina" has delicate colonies with few reproductive structures in deep waters, resulting in reduced geographic ranges of haplotypes and ultimately sympatric and parapatric speciation. The vulnerability or adaptability of species to thrive in different biotic and abiotic conditions, therefore, seems to affect population sizes and dispersal.

| CON CLUS IONS
With a large (growing) dataset of DNA barcodes and "molecular clock" calibrations, we may discover many phylogeographic and evolutionary patterns. Nevertheless, the present study is exploratory and a generator of phylogeographic hypotheses, which apply to marine benthic invertebrates. Much of marine genetic diversity (including that of the Plumularioidea) remains little sampled/investigated; but with an increased number of DNA Barcoding and genomic studies, we will gradually increase our understanding of marine evolutionary history.
To complement this particular study, it is desirable to sample more DNA barcodes from other oceanic settings, such as the southern and northwest Atlantic, and much of the cold waters of the Pacific and the deep sea. Furthermore, as other (combinations of) genes may suggest discordant phylogeographic patterns and thus more fine-scale complex movements of genetic lineages, the usage of further genetic markers will also increase robustness of evolutionary inferences.
The major constraints to dispersal and speciation we hypothesized were dispersal strategy (e.g., especially rafting or not, but perhaps also with or without medusoids), population sizes, water temperature, and consequent tolerance of the marine fauna, coupled with the effects of oceanic circulation and land barriers. For example, vicariance induced by the closures of the Tethys Sea and Central American Seaway, as well as by glacial periods and oceanic current modifications, seems to have promoted considerable diversification of hydroids. Additionally, the Indo-West Pacific seems to have played a major role as a source of genetic diversity to the Atlantic. Indeed, the majority of Atlantic species of Plumularioidea seem to have derived evolutionarily from the Indian Ocean, and from species radiations inside the Atlantic. We found some lineages that colonized the Atlantic via the ancient Tethys Sea. The maritime passage rounding South Africa served predominantly for westward genetic connections between the Atlantic and Indian Ocean, in periods of less activity of the Benguela current.
More recently, trans-Atlantic migrations seem unlikely for most Plumularioidea hydroids of tropical to temperate waters. Haplotypes with amphi-Atlantic distribution, suggestive of recent genetic interconnections across the Atlantic, were only detected in four hydroid species that likely disperse through rafting on natural and/or human-made floating substrates. However, various tropical taxa migrated across the Atlantic in the past, mainly from east to west, taking advantage of equatorial currents. The main genetic break of trans-Atlantic colonizations probably occurred around the Miocene-Pliocene transition, coinciding with the constriction of the Tethys Sea and the Central American Seaway, and consequent change of oceanic current velocities and other properties. Deep-water clades of the N Atlantic seem to have mostly crossed the Atlantic from east to west, but with some exceptions.
Presently, the Gulf Stream does not seem to contribute much to trans-Atlantic dispersal of Plumularioidea hydroids. The Azores, in the middle of the North Atlantic and greatly influenced by the Gulf Stream, is instead more phylogeographically related with the Mediterranean, Lusitanian province and Macaronesian islands and seamounts. The archipelago has been colonized multiple times and through different routes from the eastern Atlantic side, since the Pliocene to (close to) the present, sometimes more than once for a single species. Steppingstone dispersal across Macaronesian seamounts and islands, as well as eddies and deep-sea currents departing from the vicinity of the Mediterranean, may explain the occurrence of Plumularioidea hydroids in the Azores. Few species arrived at the archipelago with the aid of boats. We displayed evidence of one lineage that reached the mid-Atlantic ridge from the West Atlantic, but via deep waters.

ACK N OWLED G M ENTS
The first author was funded by the FCT postdoctoral grant -SFRH/ BPD/84582/2012. We acknowledge reviews of prof. Vermeij on a previous version of this manuscript, and of two anonymous reviewers.

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial and nonfinancial interests.

AUTH O R CO NTR I B UTI O N S
The present work makes part of the postdoctoral research of Allen G. Collins https://orcid.org/0000-0002-3664-9691