Introduction

The role of oceanic rafting in structuring littoral and near-shore ecosystems is increasingly being demonstrated through molecular analyses, e.g., [1–7], and by ecological or observational studies, e.g., [8–10]; see also reviews [11–13]:. In particular, detached buoyant macroalgae can form rafts, facilitating trans-oceanic dispersal of themselves and algal-associated fauna [5]. The dispersal ability of buoyant macroalgal species such as southern bull-kelp (Durvillaea antarctica) and giant bladder kelp (Macrocystis pyrifera) is highlighted by their broad distributions [14,15]. If strong buoyancy enables some macroalgal taxa to traverse vast oceanic distances, non-buoyant taxa with poorly-dispersive gametes and spores might be expected to show both relatively restricted ranges and poor genetic connectivity; yet many such species have surprisingly broad distributions [16,17] and may therefore be able to disperse long distances [18,19].

There is thus a paradox between the broad distributions of some species and their inherently poor dispersal capacity [20]. Although marine biogeographic studies are increasingly revealing examples of this paradox, we still do not fully understand the factors that favour long-distance dispersal in some species over others, nor how common dispersal capacity is within and among closely related groups [17]. Using multi-gene (mtDNA: COI, cpDNA: rbcL, and rRNA: LSU) phylogeographic analyses of two widespread, nonbuoyant subantarctic macroalgal taxa, we here tested the hypothesis that coastal species with low inherent dispersal capacity can nonetheless readily achieve long-distance, trans-oceanic dispersal. We further aimed to assess whether either of the two species, or lineages within them, appear more or less capable of such dispersal. The first of these taxa, Bostrychia intricata (Bory de Saint-Vincent) Montagne, is a turfing red alga (Rhodophyta) which reproduces sexually, grows in the upper limits of the intertidal zone on rock platforms, and forms dense moss-like mats of intertwined individuals that are particularly common in and around shaded rock cracks and crevices ( [21]: as Stictosiphonia hookeri) on numerous subantarctic islands, as well as the coasts of southern New Zealand, Chile and Argentina and the Antarctica Peninsula. Bostrychia tetraspores, although slightly motile [22], are short-lived, surviving only about one day in culture (J. A. West, pers. comm.), and are thus unlikely to disperse long distances. The second taxon, Adenocystis utricularis (Bory de Saint-Vincent) Skottsberg, is a brown macroalga (Phaeophyceae), with a similar geographical range. It also has short-lived gametes, with female gametes settling on the substratum within 30 minutes of release, and the more motile male gametes settling within 24 hours [23]. This species has a biphasic life history, existing as diploid macroscopic sporophytes (tear-drop shaped sacs a few centimetres long) in summer, or solely as thin haploid filaments in winter, and can reproduce either sexually or asexually [23]. Thus, neither species has a life history that seems likely to promote unassisted trans-oceanic dispersal of propagules.

This study does not attempt to provide a comprehensive biogeography of the two target species throughout their entire distributional ranges, but rather focuses on samples collected from high-latitude Southern Hemisphere locations including southern Chile, New Zealand and several sub-Antarctic islands. These disjunct regions are connected oceanographically by the Antarctic Circumpolar Current, and as some genetic affinities of populations of buoyant macroalgae [14,15] and associated invertebrates [4] have previously been shown across vast oceanic distances in these regions, they constitute an ideal system in which to test our hypothesis that non-buoyant algal species can also achieve long-distance dispersal in this setting.

Materials and Methods

Sample collection, DNA extraction and sequencing

Samples were collected by hand from intertidal rock platforms around New Zealand, several subantarctic islands, southern Chile, and the Falkland Islands. Sampling sites are indicated on the maps in Figures 1 and 2; site co-ordinates and the number of samples sequenced for each marker from each site are provided in Table S1. Samples were identified in the field based on morphological features, and each was separated from the others by at least 30 cm to ensure that separate individuals were sampled. Although samples on the same rock platform could potentially be somewhat related (e.g., siblings or parents), sampling was carried out using the same criteria within each population so that, if diversity were underestimated due to relatedness of samples, this artefact should have been consistent across all sites. For Bostrychia, sampling involved taking small (< 0.5 cm²) pinches of macroalgal tuft from separate clumps on the rock platform. Each sample was wrapped in a fragment of CHUX® Superwipes®, and multiple samples from each site were then placed in 50 ml vials containing 96% ethanol. Examination of individuals from each clade (by G.C. Zuccarello) confirmed their identities as B. intricata according to morphological criteria. For Adenocystis, one thallus was taken from each individual group of thalli, where 'groups' were separated by at least 30 cm; each thallus was squeezed to remove as much internal moisture as possible, and multiple thalli were then placed in 50 ml vials containing 96% ethanol (up to 10 samples per vial). In the laboratory, samples were dried at 60°C for between two and five hours. Desiccated voucher tissue from each sample has been kept in the laboratory, preserved with silica gel.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Phylogeographic relationships within Adenocystis utricularis based on COI and rbcL data.

The phylogenetic trees indicate relationships among, and diversity within, major clades or lineage groups (distinguished by colour); support for these clades is shown by Bayesian Posterior Probability values above branches, and ML bootstraps below (those in parentheses show ML bootstraps of >50% for the concatenated analysis of all markers). Outgroup taxa have been removed for clarity. The map indicates clade distributions and proportions at each locality, with multiple haplotypes at a site indicated by pie divisions.

https://doi.org/10.1371/journal.pone.0069138.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Phylogeographic relationships within Bostrychia intricata based on COI and rbcL data.

The phylogenetic trees indicate relationships among, and diversity within, major clades or lineage groups (distinguished by colour); support for these clades is shown by Bayesian Posterior Probability values above branches, and ML bootstraps below (those in parentheses show ML bootstraps of >50% for the concatenated analysis of all markers). Outgroup taxa have been removed for clarity. The map indicates clade distributions and proportions at each locality, with multiple haplotypes at a site indicated by pie divisions.

https://doi.org/10.1371/journal.pone.0069138.g002

A small (< 1 mm²) fragment of tissue from each sample was isolated using flame-sterilised forceps. DNA extraction followed standard Chelex® procedure (Walsh et al., 1991). Extracted Adenocystis DNA was diluted one hundredfold before PCR; this step was not necessary for Bostrychia. For each species, a region of mtDNA (partial COI), cpDNA (partial rbcL; although in the case of Adenocystis the fragment amplified included only the last 198 bases of rbcL, followed by 224 bases of intergenic region), and rRNA (partial LSU). PCR amplification was carried out in 20µL volumes containing 1.0µL DNA, 0.5µM of each primer, 1 x buffer, 0.8 mM dNTP, 1.5 mM MgCl₂, and 1 U Taq polymerase (Bioline ‘BIOTAQ,’ London, UK). Early trials with some relatively ‘universal’ algal primers generally failed or yielded sequences contaminated by non-target DNA, perhaps from epiphytic material; specific primers were therefore developed for each marker for each genus (Table S2), with two exceptions for Adenocystis: the primers T01N(F) [24] and T13(R) [25] which successfully amplified a fragment of LSU, and cox 1-789F and cox 1-1378R [26] which sometimes successfully amplified a fragment of COI. Amplification was performed in an Eppendorf Mastercycler® ep gradient using the following profile: 94°C for 2 min; forty cycles of 15s at 94°C, 30s at 50°C, 1 min at 72°C, followed by a final 4 min extension at 72°C. PCR products were purified using an Omega BIO-TEK Ultra-Sep ® Gel Extraction Kit (D2510-02: 500 preps) purification kit – the initial incubation steps were omitted as samples did not require gel purification. DNA fragments were sequenced by the Genetic Analysis Service in the Department of Anatomy and Structural Biology at the University of Otago.

Results

Each data set comprised sequences of between 400 and 900 bases, with >150 samples amplified for COI, and a subset incorporating representatives from each major COI clade amplified for rbcL and LSU (Table 1).

Numerous highly divergent lineages and clades were resolved in each of the two target species (Figures 1 and 2, and Tables 2 and 3), particularly for the most informative marker, mitochondrial COI. In each case, the chloroplast marker (rbcL) was less variable, but also showed several divergent lineages. For both Adenocystis and Bostrychia, the nuclear marker (LSU) was considerably less informative than the mitochondrial and chloroplast markers. The Bostrychia LSU phylogeny comprised several poorly-supported clades, largely concurrent with COI clades with the exception of samples from COI clade 4, which did not form a clade (Figure S4 of the supplementary material). Most Adenocystis samples shared a single LSU sequence (sequence 1A in Figure S2). Adenocystis utricularis COI clades were monophyletic with respect to all outgroups. For the A. utricularis rbcL phylogeny, the position of Utriculidium durvillaei could not be resolved. Bostrychia intricata was monophyletic with respect to outgroups for COI, consistent with the well-supported branches of the rbcL and LSU phylogenies (Figures S3 and S4).

Given the high level of genetic divergence observed among clades within each of the study taxa (up to 9% for A. utricularis COI, and up to 12.1% for B. intricata COI: Tables 2 and 3), potentially representing cryptic species, interpretations of phylogenetic analyses primarily focused on within-clade patterns; thus, even if any of the major clades represent distinct species, interpretations of patterns of diversity and differentiation should not be confounded by taxonomic issues. Nevertheless, F_ST and AMOVA analyses to assess levels of population differentiation were carried out using complete COI data sets, as sample numbers were too low to allow meaningful within-clade population analyses.

For both A. utricularis and B. intricata, close phylogenetic relationships – in some cases even shared haplotypes – were observed among populations separated by large oceanic distances. For example, some COI A. utricularis haplotypes from Macquarie Island (haplotypes 4B and 4C) differed only by one to two base pairs (0.3-0.5%) from haplotypes detected in southern Chile (haplotypes 4A and 4D), and haplotype 1A in A. utricularis COI clade 1 was shared by populations in southern Chile and the Falkland Islands (Figures 1 and 3). Close relationships were also detected between South Georgian and Chile / Falkland Island haplotypes (1A and 1B vs1C-1F: 0.3-0.6% distant). Similarly close relationships among New Zealand, Chilean and Falkland Island A. utricularis were observed for the chloroplast marker (e.g., as low as 0.3% between haplotypes 3A or 3C in New Zealand and haplotypes 3I from the Falkland Islands or 3J from Chile and the Falkland Islands, respectively: see clade 3 in Figure S1 of the supplementary material), and almost no variation was observed across vast distances in the ribosomal marker (maximum 0.2%: Table 2; and see Figure S2 of supplementary material). For B. intricata, one haplotype (COI: haplotype 1A, Figures 2 and 4; rbcL: haplotype 1A, Figure S3) was shared among extremely distant sites on Marion Island, the Falkland Islands, and throughout southern Chile. Within COI clade 4 some New Zealand haplotypes were more closely related to some Chilean haplotypes than they were to other New Zealand haplotypes (e.g., for COI, 4F is more closely related to 4L (0.5%) than to 4N (2.1%): see Figure 2; and for rbcL, haplotype 4A was shared among New Zealand and Chilean sites: see Figure S3 of the supplementary material).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Haplotype networks for Adenocystis utricularis mtDNA (COI), illustrating the relationships among haplotypes within clades.

Colours correspond to clade colours in Figure 1.

https://doi.org/10.1371/journal.pone.0069138.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Haplotype networks for Bostrychia intricata mtDNA (COI), illustrating the relationships among haplotypes within clades. Colours correspond to clade colours in Figure 1.

https://doi.org/10.1371/journal.pone.0069138.g004

The population pairwise F_ST analyses (Supplementary tables S6 and S7) indicate that some disjunct populations were genetically similar (note that, as sample sizes at several populations were low, these results should be interpreted cautiously). For example, for B. intricata, Falkland Island and several Chilean populations were not significantly differentiated. Nevertheless, in some populations several divergent clades were present, which, combined with low sample sizes, appears to have led to the failure of many of the trans-oceanic genetic similarities noted in the preceding paragraph to be reflected in the populations pairwise F_ST comparisons. All AMOVAs were highly significant, and showed the vast proportion of genetic variation to be among rather than within populations (Table 4).

Discussion

Despite clear geographic partitioning of genetic lineages (i.e. some clades being confined to particular geographic regions), both A. utricularis and B. intricata also showed unequivocal evidence of long-distance, trans-oceanic dispersal. Although our genetic data indicate the possibility of cryptic species within both taxa (see below) the patterns of genetic diversity observed nonetheless allow assessment of the trans-oceanic dispersal hypothesis. For example, A. utricularis showed low genetic distances among the Falkland Islands, South Georgia and southern Chile, as well as across ~ 8000 km of the Pacific Ocean (Figure 1), and throughout the NZ subantarctic. Similarly, B. intricata COI exhibited evidence of dispersal throughout the NZ subantarctic, and between NZ and Chile (Figure 2), and a single B. intricata COI haplotype (1A) was shared by populations in Chile, the Falkland Islands, and Marion Island, implying dispersal of reproductively viable adult thalli across approximately 6,500 km of open ocean. Future research on whether all lineages – including those that are appear most ancient and have restricted distributions – are equally capable of dispersal (e.g., whether all can grow on floating substrata such as wood or kelp), would help to shed light on the evolution of dispersal mechanisms.

Despite the evidence that long-distance dispersal is possible for both B. intricata and A. utricularis, the maintenance of genetic diversity on small spatial scales suggests that it might only rarely result in gene flow or establishment of emigrants’ offspring. Research is increasingly indicating that density-dependent processes play a critical role in driving spatial genetic and ecological structure, with founding colonizers of disturbed or new habitat rapidly reaching high densities, preventing establishment of latecomers (‘founder takes all’ [39]). Thus, for many species, dispersal events may occur frequently but primarily permit opportunistic colonisation of newly-available habitat, rather than facilitating ongoing long-distance genetic connectivity (but see 7). Future research should aim to directly test whether such a process is indeed responsible for spatial genetic structure in these intertidal algal species.

Many macroalgal species may have some level of (usually short-lived) buoyancy, if only due to small air bubbles trapped around their fronds. Mature A. utricularis thalli consist of small sacs filled with liquid [23] and are only poorly buoyant (C. Fraser pers. obs.). In addition, they are considerably less robust than the large, tough kelps (e.g., Durvillaea and Macrocystis), and it is unlikely that they could survive long voyages floating at the sea surface. Nonetheless, the evidence from this study suggests that A. utricularis can indeed travel vast, trans-oceanic distances. Adenocystis utricularis grows in the low intertidal, and shallow subtidal, and it is possible (although as-yet unobserved) that adult thalli, or gametophytic filaments, can grow on - or entwine with - holdfasts of the large kelps, and are able to raft across oceans on these more-buoyant hosts.

High genetic diversity was detected within the New Zealand A. utricularis COI group 4I-4R (up to 7.3% for COI: Table 2), with samples from populations north of the Canterbury Bight being distinct from all more southern New Zealand populations. That the Canterbury Bight – a 200 km-long unbroken stretch of cobbled beach south of Banks Peninsula – forms a significant biogeographic barrier for rocky-shore marine macroalgal taxa has previously been demonstrated in two bull-kelp (Durvillaea) species [40,41], the chiton Sypharochiton pelliserpentis [42] and the gastropod Diloma arida [43].

Recently, patterns of genetic connectivity between New Zealand and South America have been observed in another non-buoyant, intertidal macroalga that was previously hypothesised to be a poor disperser. Boedeker et al. [44] noted close genetic (nuclear ITS) similarities among populations of the turfing green alga Wittrockiella lyalli on either side of the South Pacific Ocean. The authors suggested this alga may be able to traverse oceans by rafting on buoyant wood, as the species has been observed growing on tree logs in the intertidal zone. Similarly, during this study we observed B. intricata growing on detached driftwood in the intertidal zone, and it seems probable that this species is able to disperse using that buoyant substratum as a raft. Anthropogenic means of dispersal such as hull fouling or ballast water of ships [45] could also potentially allow both species to traverse oceans; although we could find no record of either species in a ship fouling or ballast environment, such dispersal cannot be categorically ruled out. Future work could aim to assess broad-scale phylogeographic patterns in these species in relation to major shipping routes. Nevertheless, it seems unlikely that the very short-lived gametes of either species (see Introduction) could survive an ocean traverse in ship ballast, and these macroalgae would be unlikely to be tolerated on ship hulls to the age and size at which they could release gametes.

Molecular studies are increasingly indicating that long-distance dispersal of both buoyant and non-buoyant species of macroalgae is possible (e.g. see 14,15,18,46), although some species and geographic regions show greater evidence of trans-oceanic dispersal than others [46]. Rafting, in particular, appears to have played a critical role in facilitating long-distance dispersal and structuring intertidal ecosystems [13]. Many species that lack obvious trans-oceanic dispersal ability (e.g., brooding invertebrates [4,5,7]:; small, non-buoyant macroalgae [44]: and this study; and even terrestrial vertebrates [47]:) are nonetheless evidently capable of capitalising on the buoyancy of floating wood or detached kelp to disperse long distances. The long-distance, trans-oceanic dispersal events inferred for both A. utricularis and B. intricata were most probably facilitated by the strong Antarctic Circumpolar Current, which has previously been implicated in dispersal of numerous taxa around the high-latitudes of the Southern Hemisphere (see references in [13]). Long-distance dispersal of macroalgae naturally relies on effective dispersal means and vectors (such as rafting on buoyant material in strong ocean currents), and its likelihood for any species or region must therefore be considered on an entirely case-by-case basis.

We observed considerable genetic diversity within and among populations of both B. intricata and A. utricularis in regions that are thought to have been affected by ice at the Last Glacial Maximum (high-latitude subantarctic islands and Chilean fiordland; see 14,41,48). Under a postglacial recolonisation hypothesis, populations of taxa that were driven locally extinct in glaciated regions, but have since recolonised these areas, are expected to show considerably lower genetic diversity in such regions than elsewhere [48,49]. In contrast to ice-scour susceptible kelps such as Durvillaea antarctica and Macrocystis pyrifera, which have been found to show negligible genetic diversity at high latitudes [14,15,41], the diversity observed in such regions for both B. intricata and A. utricularis suggests that these species may have persisted in some ice-affected regions throughout the LGM. Such a hypothesis is consistent with the biology of these macroalgae; D. antarctica and M. pyrifera are apparently unable to survive in ice-affected locations today [50], whereas B. intricata and A. utricularis can. Bostrychia intricata is recorded from contemporarily sea-ice affected regions, including Antarctica [51], and can survive several days of freezing [52]. Adenocystis utricularis has a circum-subantarctic distribution and occurs in ice-affected regions such as the South Shetland Islands [53], presumably due to its persistence as a filamentous gametophyte during winter when ice is most extensive.

Supporting Information

Acknowledgments

We are extremely grateful to the many who assisted with this research. For field support / sample collection, we thank: Carlos Olavarría (Centro de Estudios del Cuaternario, Punta Arenas, Chile); Emma Newcombe; Erasmo Macaya Horta; Mauricio Palacios; Cesar Cardenas; Jude Brown; Karen Neely (Shallow Marine Surveys Group, Falkland Islands); Bill Wagman; James Doube (Australian Antarctic Division); Susana Clusella-Trullas (Centre for Invasion Biology, University of Stellenbosch); Kalinka Rexer-Huber; Graham Parker; Elaine Leung; Sue Scott; Greg Landreth (Northanger expeditions); Keri Pashuk; Hayley Landreth; Raisa Nikula; Henk Haazen (Tiama expeditions); Tony whiting; Ross Watters; Maria Noel Serra; Rebecca Cumming. For collection / export permits we thank: Trevor Glass (Tristan da Cunha Conservation); Nick Rendell (Falkland Islands Government); Richard McKee (Government of South Georgia); Rob Green (Department of Primary Industries, Australia); Macquarie Island Research Advisory Group; Department of Environmental Affairs, South Africa; Pete McClelland and Doug Veint (Department of Conservation, Southland, NZ). For field accommodation support we thank: Sea Lion Lodge (Falkland Islands); Marco Quiroz, Luis Pardo and the UACH Marine Station at Calfuco, Chile. Comments by Olivier de Clerck, Martin Thiel and anonymous referees helped us improve the MS.