Many species endemic to deep-sea methane seeps have broad geographical distributions, suggesting that they produce larvae with at least episodic long-distance dispersal. Cold-seep communities on both sides of the Atlantic share species or species complexes, yet larval dispersal across the Atlantic is expected to take prohibitively long at adult depths. Here, we provide direct evidence that the long-lived larvae of two cold-seep molluscs migrate hundreds of metres above the ocean floor, allowing them to take advantage of faster surface currents that may facilitate long-distance dispersal. We collected larvae of the ubiquitous seep mussel “Bathymodiolus” childressi and an associated gastropod, Bathynerita naticoidea, using remote-control plankton nets towed in the euphotic zone of the Gulf of Mexico. The timing of collections suggested that the larvae might disperse in the water column for more than a year, where they feed and grow to more than triple their original sizes. Ontogenetic vertical migration during a long larval life suggests teleplanic dispersal, a plausible explanation for the amphi-Atlantic distribution of “B.” mauritanicus and the broad western Atlantic distribution of B. naticoidea. These are the first empirical data to demonstrate a biological mechanism that might explain the genetic similarities between eastern and western Atlantic seep fauna.
Species endemic to deep-sea methane seeps can be broadly distributed, despite their reliance on chemosynthetic primary productivity. Cold-seep sites on the east side of the Atlantic share nearly 12% of their megafauna with seeps on the west sides of the Atlantic . These include mussels in the subfamily Bathymodiolinae, which always bear chemoautotrophic symbionts and, therefore, are found exclusively at cold seeps, hydrothermal vents, sunken wood or whale falls. Members of the “Bathymodiolus” childressi species complex (genus uncertain) form extensive beds at cold seeps that range in depth from 500 to more than 2000 m throughout the Gulf of Mexico . Recent morphological and genetic analyses reveal that one member of this complex, “B.” mauritanicus occurs at cold seeps distributed along the Atlantic Equatorial Belt from the Barbados accretionary prism across to the Nigerian margin seeps [1,3–5].
This broad amphi-Atlantic distribution suggests at least episodic connectivity via dispersive larvae, yet dispersal in slow deep-sea currents is expected to take too long for larvae to cross this ocean . Connections among the Atlantic Equatorial Belt seeps would be facilitated by ontogenetic vertical migration because dispersal distances are expected to increase in the faster currents of surface waters compared with the deep-sea [6,7]. However, initial hypotheses suggested that the larvae of bathymodiolin mussels probably do not migrate to surface waters, because migration would increase advection of larvae away from the required chemosynthetic habitats [8–10]. Moreover, in situ collections of hydrothermal vent larvae have suggested that transport takes place mostly at depth [11,12]. While indirect evidence from morphology, isotopic analysis and lipid profiles suggests that some vent and seep larvae might undergo vertical migrations (reviewed in [13,14]), direct evidence for ontogenetic vertical migration by organisms that rely on chemoautotrophy is scarce .
Within the Gulf of Mexico, there is minimal genetic differentiation among populations of “B.” childressi . Similarly, the amphi-Atlantic species, “B.” mauritanicus, which is distinct from “B.” childressi (approx. 5% sequence divergence for mitochondrial COI), exhibits minimal (less than or equal to 0.42%) genetic distances between eastern and western Atlantic populations . These genetic findings support a hypothesis of infrequent or historical connectivity of “B.” mauritanicus via widespread larval dispersal [1,4]. The neritid gastropod Bathynerita naticoidea is usually found in association with bathymodiolin mussels in the Gulf of Mexico and is also known from the southern Barbados Prism at depths from 400 to 1700 m, where it is the most abundant snail in mussel beds . Bathynerita-like neritids are found in Miocene cold-seep deposits from Italy  and a Middle Eocene deposit in western Washington, USA , suggesting that widespread dispersal has been a life-history feature of this clade for millions to tens-of-millions of years.
To determine whether cold-seep larvae migrate from the deep to the sea surface, we towed a remote-control plankton net sampling system through discrete depth horizons above cold seeps in the Gulf of Mexico in multiple years. We collected larvae of two molluscs, the mussel “B.” childressi and the snail B. naticoidea, which are both endemic to cold seeps and widespread throughout the Gulf of Mexico.
2. Material and methods
(a) Plankton sampling
We sampled plankton throughout the water column above the Brine Pool NR1 cold seep (approx. 650 m depth), located approximately 180 km south of New Orleans, LA, in the Gulf of Mexico (27°43′24″ N, 91°16′30″ W). We sampled nine times, in March 2002, December 2002, February 2003 and November 2003, using a Multiple Opening and Closing Net Environmental Sampling System (MOCNESS) towed through 50–100 m intervals. Each 150-µm net was towed at the maximum depth of its sampling interval for 10 min then pulled obliquely through 50- or 100-m at about 20–25 m min−1 before it was closed and the next net was opened. Larval samples were either preserved in 95% ethanol or fixed overnight in 10% seawater-buffered formalin then stored in 70% ethanol. Each larva was photographed under 10–20× magnification and tentatively identified based on morphology (shell length, shape and colour) . Larvae were then processed for identification to species based on either gene sequencing or scanning electron micrograph examination of shell morphology.
(b) Molecular identifications
Individual formalin-fixed larvae were extracted , amplified and sequenced for approximately 300 bp of the cytochrome-c-oxidase subunit I (COI) locus with the methods of Hoos et al. . These methods included two rounds of PCR; the primers COIG/H were used in the first round . The second round of PCR included 1 µl of product from the first round of PCR and the primers InternalF: 5′-AGA GTT CAT CCA GTC CCA-3′ and InternalR: 5′-TGC TAT GCC AGT TTT AGC TGC-3′ designed by P. Hoos. Individual ethanol-preserved larvae were extracted, amplified and sequenced for COI as in Johnson et al. . Adults of “B.” childressi collected from the Brine Pool cold seep were sequenced for COI for comparison. Sequences were then compared against the blast database (http://blast.ncbi.nlm.nih.gov/) to confirm the identities. For “B.” childressi, Kimura-2-parameter distances were also calculated with Mega (v. 6.01)  among other known bathymodiolin mussel species from the Gulf of Mexico and the Atlantic Ocean. For one bivalve larval morphotype, we were unable to successfully sequence the COI amplicon; instead, we successfully sequenced approximately 550 bp of the 18S ribosomal RNA amplicon with the DNA extraction methods of Johnson & Geller  and the primers from Giribet et al.  and compared it to the blast database for identification at the family level. Sequences were deposited in GenBank under accession numbers: KF739294–KF739297, KJ576847, KJ576848 and KJ585667.
(c) Scanning electron micrograph identifications
Scanning electron micrographs (SEMs) of larval shells were taken on a JEOL 6400F field emission scanning electron microscope. Shells were cleaned in 5% sodium hypochlorite solution, rinsed with distilled water, air-dried and mounted on adhesive carbon discs for SEM [27,28]. Procedures to accurately document the shapes and dimensions of the larval bivalve shells were modified from those in Fuller et al. . We also used larval tube traps placed at the Brine Pool cold seep to collect late-stage larvae of both species for comparison with plankton samples (figure 1). Briefly, larval tube traps were 30-cm tall PVC pipes (5-cm diameter opening, aspect ratio = 6 : 1) that were mounted on 2 kg iron discs, filled with 10% buffered formalin  and placed at the Brine Pool approximately 250 days . For larval bivalves, we measured height and length of the prodissoconch II, shell length, straight hinge length of the prodissoconch I (if possible), provinculum length and number of teeth and compared them to those of “B.” childressi in Arellano & Young . Length is the greatest dimension approximately parallel to the provinculum and height is the greatest dimension starting from and perpendicular to the hinge line. For neritid larval shells, we compared shell shape, shell length and aperture shape of MOCNESS-collected larvae with those collected at the Brine Pool.
(d) Laboratory cultures
Both species were cultured in the laboratory. Adults of both species and egg capsules of B. naticoidea were collected using the Johnson-Sea-Link I and II submersibles (Harbor Branch Oceanographic Institution) from the Brine Pool cold seep. Adults of both species were maintained at approximately 7°C in the laboratory at the Oregon Institute of Marine Biology.
Complete culturing procedures and results for “B.” childressi are detailed in [13,32]. B. naticoidea began to lay egg capsules on “B.” childressi mussel shells in the laboratory from winter 2003 to 2005. Egg capsules were separated from each other immediately after deposition and placed in 2-ml wells filled with cold (7°C) 0.45 μm-filtered seawater (FSW) until hatching. The water was changed once a week until hatching from May to July. Once hatched, veligers from each capsule were placed into either 175-ml glass dishes with FSW or combined with many capsules that hatched on the same day into 2-l glass jars, with 10 μg l−1 chloramphenicol. Veligers were fed a mixture of Thalassiosira pseudonana and Isochrysis galbana at concentrations of 5000–10 000 cells ml−1, and water was changed every other day .
(e) Temperature tolerances of larvae
Thermal tolerances of B. naticoidea larvae were tested by exposing them to a range of temperatures found throughout the water column using an aluminium thermal gradient block . B. naticoidea veligers (20 days post-hatching) in three replicate 20-ml scintillation vials (1 larva ml−1) of cold (7–8°C) FSW were place into five temperature treatments (15, 25, 29, 32 and 35°C). Per cent survival was scored after 72 h. Because all treatments except one resulted in either 100% survival or 0% survival, data were not analysed statistically. Thermal tolerances for trochophore larvae of “B.” childressi have been previously published in Arellano & Young . Those data were checked for normality and heteroscedasticity, arcsine transformed, then analysed with a one-way analysis of variance followed by a two-sided Dunnett's test against the control (7°C) .
3. Results and discussion
Eleven B. naticoidea and three “B.” childressi veligers were collected in the top 100 m of the Gulf of Mexico above the Brine Pool cold seep in February 2003 (table 1 and figure 1). Additional veligers of each species were collected at greater depths in November and February. Larval shells of mytilid mussels and neritid snails were easily identified due to their distinctive shell shapes (figure 1). We further narrowed our search for “B.” childressi and B. naticoidea based on colour and size: “B.” childressi and other bathymodiolin larvae are a distinctive pink colour [13,31], and B. naticoidea veligers are smaller than the coastal and estuarine neritid larvae that may be present in the Gulf of Mexico. Identifications based on morphology viewed under light microscopy were further corroborated with either analysis of the larval shell using SEM (table 2 for “B.” childressi) or sequencing (table 3 for “B.” childressi). COI and 16S mtRNA sequences of our B. naticoidea larva (table 1) were 99% similar to B. naticoidea isolate GM.1 (EU732361, EU732198).
A similar multistep approach to larval identification has been used to identify species of mytilid mussel post-settlers  and hydrothermal vent larvae . Table 1 includes individuals identified based on morphology as viewed under light microscopy only if we could confirm the identity of individuals with similar morphotypes via SEM or sequencing. Thus, we consider our estimates of the numbers of larvae of these two species that we captured in the plankton to be conservative; many of the individuals we tentatively identified as “B.” childressi or B. naticoidea based on morphology were not included in table 1 because methodological limitations prevented some identities from being confirmed. For one bivalve morphotype (figure 1 and table 1), we were unable to successfully sequence the COI amplicon; however, sequence comparison of the 18S rRNA amplicon showed 100% similarity to other members of the Bathymodiolinae in the GenBank database. In addition to “B.” childressi, at least four species of bathymodiolin mussels are known from the Gulf of Mexico seeps. While Idas macdonaldi and Tamu fisheri can be found on the upper Louisiana slope, Bathymodiolus heckerae and B. brooksi are generally found at the deeper Gulf of Mexico seep sites (e.g. Alaminos Canyon, Atwater Valley, Florida Escarpment) from approximately 2000–3000 m . Besides some estimated larval size ranges , there are virtually no data available on the larval development or reproductive timing of these other bathymodiolin species. Nevertheless, while we cannot confirm that those bathymodiolin larvae that we identified with 18S rRNA sequencing are “B.” childressi, the timing of their collection from November to February (table 1) is consistent with the reproductive season of “B.” childressi .
Our larval cultures of both species provided indirect evidence for vertical migration and long larval durations: neither species could be reared long enough to reach metamorphosis and the larvae of both species were tolerant of temperatures found above the permanent thermocline. Egg capsules of B. naticoidea held in the laboratory at 8°C released swimming veligers (length: x̄ ± s.d. = 170.6 μm ± 4.9; n = 28) after approximately four months of encapsulated development. The larvae were reared in the laboratory for up to 90 days after release and consumed microalgae but did not undergo metamorphosis. “B.” childressi embryos develop more than twice as slow as those of their shallow-water mytilid relatives, reaching D-shell veligers after 8 days . As with B. naticoidea, we were unable to rear the larvae of “B.” childressi to settlement, but by comparing settlement times to known spawning seasons, Arellano & Young  predicted that “B.” childressi larvae remain in the plankton anywhere from 2–3 months to more than 1 year before settling.
We tested the ability of our cultured larvae to survive at a range of temperatures (7–35°C) they might encounter if migrating vertically through the water column. The ambient temperature at the Brine Pool cold seep is 7–8°C year round, while the sea surface above the Pool ranges from 20–25°C in the winter (figure 2) to upwards of 30°C in the summer . High percentages of larvae survived at all temperatures, including at those temperatures representative of the sea surface (figure 3) . Tolerances of early mussel larvae to temperature are not as broad as those of B. naticoidea larvae (figure 3) , suggesting that “B.” childressi larvae must either widen their tolerances as they grow or acclimate to the gradual increase in water column temperature (figure 2) during migration to the shallow depths where we found them.
The size range and timing of occurrence of larvae collected in MOCNESS plankton samples are consistent with a prediction of long larval durations for both species. Previous histological work has shown that gametogenic cycles are strongly periodic for both species, with extended spawning periods from October to March annually [13,36,37]. In this study, adult B. naticoidea deposited egg capsules in the laboratory between October 2002 and March 2003, with peak oviposition between late December and February. Similar cycles of oviposition were observed in the field, where capsules were common in November and February 2003 but none were found in July 2004 . Assuming all populations of B. naticoidea in the Gulf of Mexico have a reproductive season from October to March, and considering the lengths of larvae obtained from MOCNESS tows in February (389–676 μm; table 1), the larvae of B. naticoidea would be planktonic for at least 7 to 12 months, tripling in size before settling at cold seeps. Settlement sizes were confirmed by examining the protoconch lengths of settlers collected in tubes traps ( ± s.d. = 667.6 ± 44.02 μm; n = 12) and of post-settlement juveniles collected at the Brine Pool ( ± s.d. = 622.4 ± 10.7 μm; n = 2; figure 1).
We collected “B.” childressi veligers that were nearly settlement-size ( ± s.d. = 427.1 ± 10.0 μm; n = 3) near the surface in February 2003 (table 1). If these collected “B.” childressi veligers developed from eggs spawned at the beginning of the spawning season (October 2003), they would be 4.5 months old with calculated growth rates of 3.2 μm d−1. This growth rate is comparable to those of veligers of the related intertidal mussel Mytilus edulis developing at 6°C with high food rations (approx. 3.4 μm d−1 with 10–40 cells μl−1) . On one hand, this may suggest that “B.” childressi veligers migrate upwards slowly, finding enough food below the photic zone to grow very quickly. However, our laboratory cultures show that early larvae of “B.” childressi grow two to four times slower than does the intertidal mussel M. trossulus at the same temperature and salinity . Alternatively, the settlement-sized “B.” childressi we collected between the surface and 100 m depth in February might have been spawned during the previous season and, thus, could have been up to 16.5 months old.
Although feeding larvae of some deep-sea taxa have been found in shallow plankton tows , direct evidence of ontogenetic vertical migration has never been shown for animals endemic to highly specialized and isolated chemosynthesis-based habitats like cold seeps, hydrothermal vents or wood- or whale-falls. Transport in the upper water column would explain the lack of genetic differentiation among populations of “B.” childressi throughout the Gulf of Mexico  and the equally widespread distribution of B. naticoidea throughout the Gulf of Mexico and at the Barbados Accretionary Prism. Using physical oceanographic data, we have modelled dispersal of “B.” childressi and B. naticoidea in the upper water column and near the bottom; these models show that maximum dispersal distance is increased by ontogenetic migration . Molluscan larvae with a 1-year larval period originating in the Gulf of Mexico have the potential of dispersing up the entire eastern seaboard of the US in a single generation , a result that explains the recent discovery of extensive beds of B. childressi mussels (identified by Katharine Coykendall, United States Geological Survey) in Northwest Atlantic canyons . Similarly, our results with “B.” childressi suggest a mechanism by which the larvae of its sister-species, “B.” mauritanicus, might drift in equatorial surface currents across the tropical Atlantic, connecting disjunct metapopulations off Barbados and West Africa and in the Gulf of Cadiz [1,5]. Teleplanic (far-wandering) molluscan larvae are well known among shallow-water Atlantic species . Our data now extend the concept of teleplanic larval dispersal to deep-sea species in isolated and distant chemosynthetic environments.
This work was supported by National Science Foundation grant nos. OCE-118733, OCE-0527139 and OCE-1030453 to C.M.Y. S.M.A. was supported by an NSF graduate research fellowship, a Ford Foundation pre-doctoral fellowship and the MBARI summer internship (David and Lucile Packard Foundation).
We thank the captains and crews of the R. V. Seward Johnsons I & II. Technical assistance for SEM work was given by the staff of the Center for Advanced Materials Characterization in Oregon (CAMCOR) at the University of Oregon. Two anonymous reviewers greatly improved this manuscript. S.M.A. collected data on “B.” childressi and drafted the manuscript. A.V.G. collected data on B. naticoidea. S.B.J. and S.M.A. completed molecular identifications. C.M.Y. conceived the broader project on dispersal of deep-sea larvae, supervised the project and led cruises. R.C.V. provided oversight for the molecular identifications. All authors approved the final manuscript.
- Received December 16, 2013.
- Accepted April 9, 2014.
- © 2014 The Author(s) Published by the Royal Society. All rights reserved.