Marine Biology

, Volume 153, Issue 4, pp 545–563

Geographic clines and stepping-stone patterns detected along the East Pacific Rise in the vetigastropod Lepetodrilus elevatus reflect species crypticism

Authors

    • Muséum National d’Histoire Naturelle, Département Milieux et Peuplements AquatiquesUMR BOME 5178 (MNHN, UPMC, CNRS)
  • E. Thiébaut
    • Station Biologique de RoscoffCNRS, UMR 7144
    • Université Pierre et Marie Curie-Paris 6UMR 7144, Station Biologique de Roscoff
  • D. Le Guen
    • Station Biologique de RoscoffCNRS, UMR 7144
    • Université Pierre et Marie Curie-Paris 6UMR 7144, Station Biologique de Roscoff
  • F. Sadosky
    • Muséum National d’Histoire Naturelle, Département Milieux et Peuplements AquatiquesUMR BOME 5178 (MNHN, UPMC, CNRS)
  • D. Jollivet
    • Station Biologique de RoscoffCNRS, UMR 7144
    • Université Pierre et Marie Curie-Paris 6UMR 7144, Station Biologique de Roscoff
  • F. Bonhomme
    • Biologie IntégrativeISEM, CNRS-Université Montpellier II, UMR 5554
Research Article

DOI: 10.1007/s00227-007-0829-3

Cite this article as:
Matabos, M., Thiébaut, E., Le Guen, D. et al. Mar Biol (2008) 153: 545. doi:10.1007/s00227-007-0829-3

Abstract

Three different molecular markers (i.e. seven allozyme loci, two nuclear gene loci and, mtCOI DNA sequences) were used to assess the genetic structure of the vent gastropod Lepetodrilus elevatus collected from three vent fields along the East Pacific Rise (13°N, 9°50′N and 17°S). While allozymes and nuclear loci suggested a strong stepping-stone pattern, a multivariate analysis performed on allozymic frequencies showed the presence of two distinct evolutionary lineages: the first situated in the north from 13°N to 9°50′N and the second in the south from 9°50′N to 17°S. The analysis of mitochondrial DNA sequences confirmed the separation of L. elevatus into two distinct clades with a divergence of 6.5%, which is consistent with the interspecific level of sequence variation in other vent species. A divergence time of 6–14 Mya was estimated between the two clades from previous clock calibrations. Our results suggest that these taxa followed an allopatric speciation between the northern and southern parts of the EPR with a recent demographic expansion of the southern clade to the north and a subsequent secondary contact (clade hybridisation). This speciation was probably reinforced by a habitat specialisation of the two cryptic species because the southern clade was mainly found associated with mussel-dominated communities and the northern clade with tubeworm-dominated communities. However, the analysis of shell morphology failed to separate the two cryptic species based on this sole criterion although they differed from Lepetodrilus elevatus galriftensis (Galapagos population) by a higher shell elevation. Within each clade, genetic differentiation was not related to the distance across populations and could be within vent field as important as between fields. While both clades appear to be in expansion since their speciation, significant excesses in heterozygotes suggest a very recent and local bottleneck at 17°S, probably due to massive site extinction in this region.

Introduction

Hydrothermal vents are scattered along the mid-ocean ridge systems, back-arc basins and off-axis submarine volcanoes where they host dense benthic communities characterised by a high level of endemism, a patchily distribution around vent sites separated by few hundred metres to thousand kilometres, and an ephemeral life time related to tectonic events and volcanic activities (Van Dover 2000). Such spatio-temporal fluctuations in vent activity can result in local extinction creating large gaps in species distribution, or in the establishment of new habitats for colonisation. Dispersal abilities of larvae are then essential to maintain regional populations. Although several authors have shown that larvae could be transported by either neutrally buoyant hydrothermal plumes (Mullineaux et al. 1995) or near-bottom flows (Metaxas 2004; Mullineaux et al. 2005), spatial scales of dispersal remain poorly known due to the lack of knowledge of larval lifespan for most species (but see Marsh et al. 2001 for the sole well-documented example concerning the vestimentiferan tubeworm Riftia pachyptila and Pradillon et al. 2001 for the worm Alvinella pompejana).

Recently, molecular genetics tools appeared to be a valuable indirect method to infer larval dispersal of hydrothermal species (Vrijenhoek 1997). With the exception of some off-axis seamounts, hydrothermal vents are confined to the axial graben, which is 100–500 m wide along the ridge. Thus, genetic structures of vent organisms should be in agreement with the level of gene flow estimated under assumptions of one-dimensional population models (Chevaldonné et al. 1997). In the classical island model, which is characteristic of organisms with high dispersal abilities, dispersing individuals are drawn from a well-mixed pool of migrants, and estimated rates of gene flow are independent of geographical distances (Slatkin 1993). In the stepping-stone model, genetic differentiation should increase with distance between individuals of disjoint populations (Wright 1943; Kimura and Weiss 1964). Gene flow and geographical distance are then correlated and species are expected to exhibit patterns of isolation-by-distance (Slatkin 1993). From these theoretical considerations, a stepping-stone pattern is expected to be widespread for the majority of hydrothermal organisms given their potentially limited dispersal abilities (Jollivet 1996; Vrijenhoek et al. 1998). Reported along NEPR for vestimentiferan tubeworms (Black et al. 1994, 1998), no pattern of isolation-by-distance was observed for different species including bivalves (Karl et al. 1996), alvinellid polychaetes (Jollivet et al. 1995) or gastropods (Craddock et al. 1997).

Deviations of hydrothermal species from theoretical assumptions of population models could be explained by the migration modes of vent organisms that do not exactly fit the stepping-stone or island models but also to extinction/recolonisation events (Jollivet et al. 1999). Thus, migration and drift may not have the time to equilibrate if dispersal is not homogeneous in space or if populations are highly fluctuating in size over time. The shifting of hydrothermal activity along the ridge axis generates numerous patchy and ephemeral vents subjected to rapid and frequent extinction and recolonisation events (Chevaldonné et al. 1997), which can either, increase or decrease genetic differentiation between populations (Vrijenhoek 1997; Jollivet et al. 1999). Vrijenhoek et al. (1998) indeed showed that the level of genetic diversity was highly related to the proportion of empty habitats in a given vent field and not proportional to the estimated migration rate. Topographic features such as transform faults, ridge offsets, bathymetric inflation and intersecting microplates, can also alter along-axis dispersal efficiency over large distances and form geographic barriers to gene flow (Vrijenhoek et al. 1998; Van Dover et al. 2002).

On the other hand, apparent isolation-by-distance patterns may also be the result of a clinal admixture of well-differentiated genetic populations and/or sibling species. The genetic survey of individuals from both the sides of a hybrid zone, more or less marked depending on selective pressure, also induces a cline in allelic frequencies that may be interpreted as a decline in gene flow with geographical distance (O’Mullan et al. 2001). Alternatively, the mixing of partially sympatric sibling species displaying a spatial gradient of abundance can also produce an apparent stepping-stone pattern of dispersal. Occurrence of sibling species is widespread in marine environments (Knowlton 1993) and has been reported at hydrothermal vents for many marine taxa including vesycomid clams (Goffredi et al. 2003), mussels (Won et al. 2003), crustaceans (Guinot and Hurtado 2003) and polychaetes (Hurtado et al. 2004).

The vetigastropod Lepetodrilus elevatus is one of the main gastropod species present on the East Pacific Rise (EPR), from 21°N to 21°S, and on the Galapagos Rift (GAR) (McLean 1988; Warén and Bouchet 2001; Jollivet et al. 2005; Desbruyères et al. 2006). While McLean (1988) identified two discrete subspecies, L. e. elevatus on the EPR and L. e. galriftensis on the GAR, Craddock et al. (1997) suggested from genetic data that they should represent two discrete species. They reported the presence of L. galriftensis from 21°N to 9°N in sympatry with L. elevatus. However, while L. e. galriftensis sensu McLean was described from the Galapagos Rift, Craddock et al. (1997) did not include specimens from this locality in their analyses. On the other hand, these authors pointed out an apparent decline of gene flow with increasing distances on a portion of the NEPR, for both species, although the number of sampled populations was insufficient to reject the null hypothesis that genetic structure is independent of geographical distance. To clarify the status of Lepetodrilus elevatus along the EPR, we investigated its genetic structure from 13°N to 17°S using three kind of molecular markers. First, ten allozyme and two nuclear DNA loci were analysed in nine populations sampled at 13°N, 9°50′N and 17°S to check for genetic heterogeneities within the species at a global scale. Second, a partial fragment of the mitochondrial cytochrome oxidase subunit I gene (mtCOI) was sequenced from 77 individuals sampled over 5 representative populations to investigate the occurrence of phylogeographic breaks along the EPR. Finally, an analysis of shell morphology was performed to discriminate between putative sibling species and L. e. galriftensissensu McLean.

Materials and methods

Collection of samples

Lepetodrilus elevatus specimens were collected in April 1999 and April–May 2004 from three hydrothermal vent fields along the EPR (i.e. 13°N, 9°50′N and 17°S) using the submersibles Nautile (IFREMER) and Alvin (WHOI) (Table 1). At each vent field, two to four discrete vent sites were sampled within vestimentiferan clumps, mussel beds or an admixture of the two kinds of taxa. At 13°N, three samples were located inside the axial valley (i.e. Julie, Genesis and Elsa sites) while the Caldera sample came from an off-axis site. After sampling, individuals were stored in liquid nitrogen or ethanol until DNA or protein extractions. All shells were kept for morphological analyses. Additional individuals of L. pustulosus and L. cristatus were also sampled from the Caldera site at 13°N in order to be used as outgroups. As the whole limpet body was necessary for protein extraction, individuals screened for allozymes were not available for further DNA analysis.
Table 1

Locations of the Lepetodrilus elevatus populations sampled along the East Pacific Rise with the types of habitat and the number of individuals (N) analysed for each genetic marker

Vent field

Vent site

Latitude

Longitude

Dive no.

Date

Habitat

NAllozymes

NLep1

NLep3

NCOI

13°N

Caldera

12°42.70′N

103°54.40′W

1,364

20.04.99

R. pachyptila dominated and some B. thermophilus

64

63

8

Elsa

12°48.11′N

103°56.33′W

1,359

15.04.99

R. pachyptila

34

91

50

11

Genesis

12°48.67′N

103°56.43′W

1,357–1,361

12–17.04.99

R. pachyptila

52

116

61

4

Julie

12°49.05′N

103°56.56′W

1,360

16.04.99

R. pachyptila dominated and some B. thermophilus

45

80

80

9°50′N

Biotransect

9°50.44′N

104°17.52′W

4,002

14.04.04

R. pachyptila and B. thermophilus

19

 

East-wall

9°50.54′N

104°17.51′W

1,377

08.05.99

R. pachyptila and B. thermophilus

44

50

27

27

BioVent

9°50.79′N

104°17.60′W

1,372

03.05.99

R. pachyptila and B. thermophilus

53

 

17°S

Oasis-BS6

17°25.38′S

113°12.29′W

1,579

19.05.04

C. magnifica and B. thermophilus

34

30

18

27

Oasis-BS13

17°25.42′S

113°12.28′W

1,590

01.05.04

B. thermophilus dominated and R. pachyptila

49

 

Allozyme genotyping

Enzyme electrophoreses were conducted on 12% starch gel and electrophoretic procedures were performed according to Pasteur et al. (1987). Each frozen sample of the whole limpet body were homogenized in a grinding buffer (0.01 M Tris, 0.002 M EDTA, 0.05% β-mercaptoethanol, 0.0001 M phenylmethylsulphonyl fluoride, 0.25 M sucrose, pH 6.8) before centrifugation at 13,500g for 25 min. The supernatant was then absorbed onto 4 × 12 mm filter-paper wicks (Whatman No. 1) and electrophoresed using the following buffers: (1) Tris–citrate pH 8 (TC8), for phosphoglucomutase (Pgm, E.C. 2.7.5.1), mannose phosphate isomerase (Mpi, E.C. 5.3.1.8), Leu-Tyr peptidase (Pep, E.C. 3.4.11.×) and aspartate amino-transferase (Aat, E.C. 2.6.1.1); (2) Tris–citrate, electrodes pH 6.3, gel pH 6.7 (TC6.7) for malate deshydrogenase (Mdh, E.C. 1.1.1.37), glucose-6-phosphate isomerase (Gpi, E.C. 5.3.1.9), and isocitrate deshydrogenase (Idh, E.C. 1.1.1.42); (3) Tris−citric–boric–lithium hydroxide, electrodes pH 8.1, gel pH 8.3, (LiOH8.3), for malic enzyme (Me, E.C. 1.1.1.40). Buffer system (1) was run at 80 mA for 4 h, buffer system (2) at 60 mA for 4 h and buffer system (3) at 250 mA for 4 h. All eight enzyme systems encoded by ten gene loci were visualised using enzyme-specific stains according to Pasteur et al. (1987). Only seven loci (six enzymes) were consistently interpretable and were scored for all populations. Three additional loci (not scorable for one clade) were considered for five out of the eight populations, involving ten enzyme loci overall. Loci were numbered according to the decreasing anodal electromorph mobility in multi-loci systems, and alleles were assigned given to their relative distance to the most frequent allele (100).

Nuclear DNA loci genotyping

Polymorphic anonymous nuclear DNA markers were developed for Lepetodrilus elevatus using direct amplification of length polymorphism (DALP, Desmarais et al. 1998). First, a multiple-loci scan (DNA fingerprint) was performed from lepetodrilid genomic DNA using “universal” M13 derived-primers and standard PCR amplification (see Desmarais et al. 1998 for PCR conditions). DNA fragments were then run on a polyacrylamide gel and, bands associated with putative polymorphic loci were removed from the gel. PCR-products were purified using a QIAquick™ PCR purification kit, T/A-end ligated into a BlueScript™ T-vector plasmid at 16°C overnight and subsequently cloned into DH5α competent cells. Positive clones were then sequenced using a VISTRA™ 725 automatic sequencer in order to identify alleles within each locus. Two loci (i.e. Lep1 and Lep3) displaying an indel polymorphism were then selected and, specific primers were designed from the allele alignments. Primer sequences are: (Lep1-R: 5′-AAAGATCCTCCCTTTGTAATGG-3′; Lep1-F: 5′-CTAAACCTTAAAGTTCGA-3′; Lep3-R: GAAAGATCCTCCCTTTGTAATGG; Lep3-F: TAAACCTTAAAGTTCGAGAC).

For all individuals, total genomic DNA was extracted from a piece of muscle by phenol–chloroform protocol (Sambrook et al. 1989). Amplification reactions were performed in a 25 μl mixture containing 5 μl of template DNA, 1× PCR buffer, 1.5 mM MgCl2, 0.4 μM of each primer, 3.5 U Taq DNA polymerase, 0.14 mM of each dNTP, and sterile H2O to final volume. Reverse primers were labelled with the 6-FAM fluorochrome. Amplifications corresponded to a 2 min initial denaturation step at 94°C followed by 35 cycles of 30 s of denaturation at 91°C, 30 s of annealing at 55°C and 1 min of extension at 72°C with a 5 min of final elongation at 72°C. PCR products were electrophoresed at 50 W for 1 h 45 min on a denaturing polyacrylamide gel (8%, 0.5×) in 0.5× Tris–Borate–EDTA buffer, and then read with the FMBIO II HITACHI scanner.

Statistical analyses

For each population, allele frequencies, mean number of allele per locus (Nall), observed (Ho) and expected (HNB) heterozygosities were estimated using Genetix 4.05 (Belkhir et al. 2004); allelic richness (RS) was estimated using Fstat 2.9 (Goudet 2001). The null hypothesis of independence between loci was tested using Genetix 4.05. Deviations from Hardy–Weinberg equilibrium were analysed for each population, at each locus, by the calculation of Wright’s fixation index Fis as estimated by Weir and Cockerham’s (1984) f. The level of genetic differentiation between pairwise combinations of populations was estimated by calculating the Weir and Cockerham’s (1984) estimator \( \hat{\theta } \) of Wright’s Fst index for each locus and across loci. Deviations from 0 of Fis and Fst indexes were then tested by permuting genotypes in populations using Genetix 4.05 (Belkhir et al. 2004). Isolation-by-distance was assessed by testing the correlation between pairwise genetic distances estimated by \( \hat{\theta }/(1 - \hat{\theta }) \) and the shortest geographical distance between the samples using a Mantel test conducted with Genetix 4.05 software. Distances between sites were calculated using the “Great Circle Distance Calculator” (available at http://www.gb3pi.org.uk/great.html). To illustrate the relationships among individuals, a correspondence analysis on allozyme data based on genotypic frequencies was performed using Genetix 4.05 (Belkhir et al. 2004).

The Bottleneck software was used in order to test demographic disequilibria (Cornuet and Luikart 1996). In a recently bottlenecked or founded population, the observed Hardy–Weinberg gene diversity (He) is higher than the expected gene diversity (Heq), computed from the number of alleles (k), at the mutation-drift equilibrium under the assumption of a constant-size population (Cornuet and Luikart 1996; Luikart et al. 1998). Conversely, an expanding population is characterised by a decrease in He compared to Heq. To test such a deviation, a Wilcoxon test was performed under the infinite allele model (IAM) following the recommendations of Cornuet and Luikart (1996).

Mitochondrial DNA analyses

DNA was extracted and purified using a phenol–chloroform procedure (Sambrook et al. 1989). For most individuals, partial sequences of mitochondrial cytochrome c oxidase subunit I gene (mtCOI) sequence were amplified with the universal primers LCO1490 and HCO2198 described by Folmer et al. (1994). The 50 μl amplification mixture contained 3 μl of template DNA, 1× PCR buffer, 2 mM MgCl2, 0.4 μM of each primer, 40 μM of each dNTP, 2 U Taq DNA polymerase, and sterile H2O to final volume. Polymerase chain reactions (PCR) were performed as follows: (a) a 3 min initial denaturation step at 94°C, (b) 40 cycles of 45 s of denaturation at 94°C, 45 s of annealing at 50°C and 90 s of elongation at 72°C and, (c) a 7 min final elongation at 72°C. For Elsa, Genesis and several 9°50′N samples, a “nested PCR” was conducted due to DNA amplification difficulties. First, DNA was pre-amplified with universal primers (Folmer et al. 1994) and used as a target (1 μl of PCR-product) for slightly degenerated specific primers obtained from a COI alignment of lepetodrilid sequences (COI-R: 5′-TAACTTCAGGGTGACCAAAAAATCA-3′; COI-F: 5′-GTTCAAATCATAAAGATATTGG-3′). Amplifications were carried out using (a) a 3 min initial elongation step at 94°C, (b) 30 cycles with a 35 s initial denaturation at 94°C, a 35 s annealing at 56°C and 80 s of elongation at 72°C, and (c) a final 10 min elongation step at 72°C. The PCR products were purified and sequenced on ABI 3100 using BigDye® terminator chemistry (Applied Biosystems) following the manufacturer’s protocol. Sequences were proofread in Chromas 2.3 and aligned manually using BioEdit Sequence Alignment 7.0.1.

To analyse phylogenetic relationships between the individuals, a maximum-likelihood (ML) tree was constructed using PAUP 4.0b10 (Swofford 2002). The optimal substitution model was selected using the hierarchical likelihood ratio test (hLRTs) implemented by Modeltest 3.07 (Posada and Crandall 1998). The ML analysis was performed with heuristic searches using 50 stepwise random sequences addition and TBR branch swapping. Bootstrap support values were calculated from a 50% majority-rule consensus tree, based on 1,000 resampling of the data set. Sequences of Lepetodrilus cristatus and L. pustulosus were used as outgroups while published mtCOI sequences from L. elevatus (Genbank Accession nos AY923923, U56846) were included.

A measure of genetic diversity of each population was calculated using the haplotype (He-HAP), and nucleotide (π) diversities with DNAsp 4.0 (Rozas et al. 2003). Fu and Li’s F statistic was used to test the null hypothesis of the mutation-drift equilibrium (Fu and Li 1993). Mismatch curves were also drawn for each population and each lineage and observed values were compared to the expected curves fitted for constant population size model and for the population growth/decline model. The Ramos-Onsins and Rozas R2 test was used to detect population expansion, which is most powerful for small samples size than the Fu’s Fs test (Ramos-Onsins and Rozas 2002). The observed R2 value was compared to a theoretical value simulated assuming a constant population size using DNAsp 4.0 (Rozas et al. 2003). Median-joining networks were constructed using Network v. 4.1.0.9 (Bandelt et al. 1999) to infer the most parsimonious branch connections between the sampled haplotypes. Pairwise Fst values were computed from haplotype frequencies and exact tests of genetic differentiation between populations were conducted using Arlequin 2.0 (Schneider et al. 2000).

Morphometry

As variations in shell proportions were reported among Lepetodrilus species and subspecies (McLean 1988), shell morphometrics were obtained on 415 L. elevatus individuals from the nine sample sites for which allozyme and/or mtDNA analyses were performed. An additional sample of 85 L. elevatus galriftensis from the Galapagos Rift (collection from the Los Angeles County Museum of Natural History, LACM 2528) was included. Individuals with undamaged shells were only considered in the analysis. Eight shell characters were measured using the “Image tool” image analysis software (University of Texas; http://ddsdx.uthscsa.edu/dig/itdesc.html) (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0829-3/MediaObjects/227_2007_829_Fig1_HTML.gif
Fig. 1

Shell morphometrics of Lepetodrilus elevatus. Lcurv curvilinear shell length (total length from the anterior edge of the shell to the lip of the protoconch); H shell height (greatest vertical distance from the apex of the shell to the plane of the aperture); Lbas basal length (length of the aperture); L1 anterior length (distance from the anterior edge of the shell to the apex); L2 posterior length (distance from the posterior edge of the shell to the apex); Lob oblique length (maximal distance from the anterior edge of the shell to the posterior part of the shell); Ltot total shell length (greatest distance between posterior and anterior end); W shell width (greatest distance perpendicular to the anterior–posterior axis)

A principal component analysis (PCA) was performed on log-shape ratios after elimination of allometric changes following Mosimann (1970) and, Mosimann and James (1979). All variables were first log-transformed and size for each individual was defined as the arithmetic mean of all variables. The log-shape ratio was calculated by subtracting the log-size value from each variable for each individual.

Results

Deviation to Hardy–Weinberg equilibrium and allelic clines

Heterozygote deficiencies

Allele frequencies per population at the ten polymorphic enzyme loci are given in Table 2. Populations from 17°S were monomorphic at two loci (Gpi and Idh2) for Oasis-BS13 and at three loci (Gpi, Idh1 and Idh2) for Oasis-BS6. “Private alleles” were found at three loci (Idh2, Pgm2 and Leu-Tyr) for three northern populations (Julie, Biovent and Biotransect) and at two loci (Pgm1 and Me1) for one sample from 17°S (Oasis-BS13). All of them except one at the Idh2 locus corresponded to rare alleles (q < 0.05). Five loci (Idh1, Idh2, Mdh, Mpi and Pgm1) displayed at least one rare allele that occurred in two populations or more. Estimated allelic richness (Rs) and multilocus heterozygosity (HNB) ranged from 1.773 to 3.691 and from 0.304 to 0.550, respectively, with minimal values at populations from 17°S and maximal values at populations from 9°50′N (Table 3). Except for populations from 13°N, genotypic frequencies did not conform to Hardy–Weinberg expectations for almost all loci (Table 3). At 13°N, significant single-locus Fis values were only reported at one locus (i.e. Gpi and Mdh) for Julie and Genesis populations, respectively. Multi-locus Fis values showed significant heterozygote deficiencies for populations from 9°50′N and a slight excess at 17°S (Table 3). When mixing all samples from a single vent field, Fis values were all significantly different from zero for the three vent fields. Numerous linkage disequilibria were also observed among samples and appear concentrated at 9°50′N and to a lesser extent at 17°S when considering each population individually (Table 4). Anonymous nuclear (DALP) loci were considered separately. Nine and 12 alleles were detected for the lep1 and lep3 loci respectively, including 5 rare alleles each (q < 0.05, Table 2). Each population displayed a high number of alleles which ranged from 7 to 8 alleles for lep 1 and 6 to 11 alleles for lep 3 (Table 3). For both loci, allelic richness (RS) ranged from 5.64 to 7.12 while multilocus heterozygosity (HNB) varied from 0.57 to 0.75. Population from Oasis-BS6 presented a high gene diversity due to the high frequency of allele 550 which was often found in homozygous condition. At 13°N, both allelic richness and heterozygosity were higher for populations situated in the axial valley (i.e. Julie, Genesis and Elsa) than for the off-axis population (i.e. Caldera). Genotypic frequencies generally conformed to random mating expectations within each population except for Oasis-BS6 at lep 1 locus which displayed a heterozygote deficiency (Table 3).
Table 2

Allele frequencies at enzyme and anonymous nuclear loci

Locus allele

13°N

9°50′N

17°S

N

Caldera

Elsa

Genesis

Julie

Biotransect

BioVent

East Wall

Oasis-BS6

Oasis-BS13

0

34

52

45

19

53

44

34

49

Gpi

 10

0.029

0.956

0.368

0.623

0.227

1.000

1.000

 100

0.485

0.567

0.033

0.395

0.151

0.432

 140

0.088

0.038

0.026

0.028

0.034

 180

0.397

0.394

0.011

0.211

0.198

0.307

Idh1

 70

0.077

0.053

0.010

0.011

 80

0.030

0.038

0.033

0.010

0.010

 100

0.970

0.885

0.967

0.947

0.980

0.989

1.000

0.990

Idh2

 10

0.029

0.022

0.684

0.075

1.000

1.000

 80

0.105

 100

0.559

0.558

0.522

0.211

0.745

0.611

 115

0.010

0.047

 130

0.412

0.433

0.433

0.132

0.389

 145

0.022

Mdh

 70

0.227

0.327

0.433

0.079

0.059

0.200

 85

0.106

0.038

0.342

0.373

0.089

0.500

0.429

 100

0.667

0.635

0.533

0.553

0.529

0.678

0.500

0.571

 120

0.033

0.026

0.039

0.033

Aat2

 50

0.071

0.094

0.044

0.382

0.347

 75

0.029

0.034

0.357

0.811

0.267

0.603

0.633

 100

0.971

1.000

0.966

0.571

0.094

0.689

0.015

0.020

Mpi

 80

0.029

0.009

 90

0.050

0.019

0.025

0.026

0.019

0.022

 100

0.700

0.692

0.788

0.526

0.057

0.467

 105

0.342

0.170

0.100

0.515

0.500

 110

0.250

0.260

0.188

0.026

0.009

0.200

 115

0.079

0.736

0.211

0.485

0.500

Pgm2

 80

0.010

 90

0.125

0.130

0.410

0.158

0.029

0.224

 95

0.404

0.066

0.485

0.296

 100

0.813

0.870

0.564

0.132

0.058

0.474

 105

0.316

0.423

0.158

0.353

0.510

 110

0.063

0.026

0.211

0.039

 115

0.105

0.077

0.039

0.162

0.194

Pgm1

 90

0.033

0.039

 95

0.010

 100

0.800

0.856

0.868

1.000

0.970

 105

0.010

 110

0.167

0.077

0.053

0.010

 120

0.029

0.079

Leu-Tyr

 40

0.250

0.726

0.188

0.300

0.355

 60

0.625

0.179

0.146

0.400

0.307

 90

0.125

0.024

0.042

0.300

0.339

 100

0.048

0.542

 110

0.012

0.083

 120

0.012

Me1

 40

0.016

 60

0.016

 95

0.286

0.028

0.183

0.594

 100

0.714

0.953

0.967

0.550

0.219

 105

0.019

0.033

0.267

0.156

lep1

  N

 

64

91

116

80

0

0

50

30

0

 50

0.016

0.011

0.009

0.013

0.010

0.017

 100

0.617

0.560

0.509

0.556

0.440

0.333

 200

0.006

0.013

0.013

0.017

 300

0.047

0.066

0.095

0.069

0.130

0.083

 400

0.188

0.181

0.181

0.194

0.150

0.183

 450

0.008

0.017

0.022

0.006

 500

0.117

0.148

0.168

0.144

0.220

0.067

 550

0.010

0.300

 600

0.008

0.011

0.006

0.040

 700

0.004

lep3

 N

 

63

50

61

80

27

18

 190

0.006

 192

0.119

0.130

0.123

0.081

0.185

0.111

 193

0.008

0.030

0.066

0.069

 194

0.016

0.020

0.008

0.006

 195

0.175

0.190

0.139

0.131

0.056

0.250

 196

0.008

0.200

0.066

0.063

 197

0.048

0.030

0.082

0.044

0.074

0.111

 198

0.010

0.008

0.019

0.019

0.028

 200

0.587

0.560

0.492

0.569

0.630

0.444

 201

0.040

0.010

0.008

0.006

0.037

0.056

 202

0.006

 206

0.008

0.010

Individuals from eight samples were genotyped at seven enzyme loci. In addition, samples from 9°50′N and 17°S were genotyped at three additional loci (only scorable for the southern clade). Additional specimens were also genotyped at both nuclear loci for six populations because of the small sample size of Biovent, Biotransect and Oasis-BS13 collections. N sample size at each population

Table 3

Fis values at enzyme and anonymous nuclear loci (*P < 0.05; **P < 0.01; ***P < 0.001), and allozyme and nuclear genetic variability for each population

Locus

13°N

9°50 N

17°S

Caldera

Elsa

Genesis

Julie

Biotransect

BioVent

East Wall

Oasis-BS6

Oasis-BS13

All

Allozymes

 Fis

  Gpi

−0.215

−0.171

0.746***

0.543***

0.797***

0.430***

0.300***

  Idh1

−0.016

0.094

−0.023

−0.029

−0.005

0.000

0.000

0.040

  Idh2

0.218

0.051

0.227

0.789***

0.601***

0.215

0.298***

  Mdh

−0.030

−0.329**

−0.001

0.378

−0.353***

−0.119

−1.000***

−0.745***

−0.311***

  Aat2

−0.015

−0.018

1.000***

0.541***

0.807***

0.234

0.242

0.483***

  Mpi

0.192

0.159

0.068

0.578***

0.168

0.459***

−0.941****

−1.000***

−0.051

  Pgm2

−0.155

−0.140

0.063

−0.163

−0.387***

0.102

−0.425***

−0.186

−0.174***

  All

0.002

−0.068

0.112

0.461***

0.164***

0.311***

−0.527***

−0.408***

0.033

  NALL

2.8

2.7

2.8

3.7

4.3

3.7

1.8

2

 

  Ho

0.360

0.370

0.275

0.301

0.36

0.349

0.462

0.426

 

  HNB

0.361

0.346

0.309

0.550

0.430

0.504

0.305

0.304

 
 

(±0.241)

(±0.236)

(±0.218)

(±0.314)

(±0.346)

(±0.224)

(±0.478)

(±0.436)

 

  Rs

2.621

2.453

2.33

3.691

3.411

3.332

1.773

1.825

 

Nuclear DNA

 Lep1

  NALL

7

8

8

8

7

7

 

  Ho

0.562

0.648

0.664

0.600

0.660

0.567

 

  HNB

0.572

0.630

0.674

0.631

0.724

0.753

 

  Rs

5.637

6.128

6.083

5.967

6.177

7.000

 

  Fis

0.017

−0.030

0.014

0.050

0.089

0.264**

0.042

 Lep3

  NALL

8

9

10

11

6

6

 

  Ho

0.571

0.640

0.705

0.588

0.519

0.778

 

  HNB

0.611

0.637

0.714

0.646

0.570

0.732

 

  Rs

5.725

6.388

7.039

7.115

5.517

6.000

 

  Fis

0.065

−0.005

0.013

0.091

0.091

-0.065

0.042

Nall mean number of alleles per population; Ho and HNB observed and expected multilocus heterozygosity (±SD); Rs allelic richness

Table 4

Number of linkage disequilibrium observed between allozyme loci for each population

 

13°N

9°50 N

17°S

ALL

Elsa

Genesis

Julie

Biotransect

Biovent

East Wall

Oasis BS13

Oasis BS6

Number of linkage disequilibrium (N)

2

1

2

7

8

15

5

4

7

(21)

(15)

(20)

(21)

(21)

(21)

(10)

(6)

(21)

N Number of possible combinations between all loci

Overall genetic differentiation between populations

For allozymes, the global test for genetic differentiation among populations exhibited an unexpectedly high and significant level of heterogeneity in allele frequencies (Fst = 0.356, P = 0.000). A significant isolation-by-distance across fields (Mantel test; R2 = 0.5661; P < 0.01) was evidenced (Fig. 2) and corresponds to a clinal variation of allele frequencies from north to south at the Aat2, Mpi, Idh2, Pgm2 and Mdh loci (Table 2; Fig. 3). Populations from 9°50′N showed the largest variations in their relative allele frequencies between the nearby sites of the field. A similar clinal variation was also observed at the Gpi locus (Gpi-100, -180 vs. Gpi-10) except for Julie (13°N), which exhibited allele frequencies similar to 17°S populations. Within-field Fst values greatly vary between the fields with significant values at 13°N (Fst = 0.030, P < 0.01) and 9°50′N vent fields (Fst = 0.228, P < 0.01) but with no genetic differentiation at 17°S vent field (Fst = 0.011, P > 0.05). The high level of differentiation observed within fields correlates with the heterozygote deficiencies which were maximum at the 9°50′N vent field.
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0829-3/MediaObjects/227_2007_829_Fig2_HTML.gif
Fig. 2

Isolation-by-distance for enzyme and anonymous nuclear (DALP) loci. Axes on the left and right stand for the allozymes and the anonymous nuclear loci, respectively

https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0829-3/MediaObjects/227_2007_829_Fig3_HTML.gif
Fig. 3

Clinal distribution of allele frequencies for enzymes and anonymous nuclear loci between 13°N and 17°S

For the DALP loci, the overall genetic differentiation was not significantly different from zero for both the loci within (Fst = 0.000; P = 1.000) and between (Fst = 0.011, P > 0.05) the vent fields. A Mantel test computed on both the loci, however, provided evidence for a significant isolation-by-distance between 13°N and 17°S (R2 = 0.9333; P = 0.033) (Fig. 2). This was mainly associated with a clinal variation of the allele frequencies at lep 1 locus (Table 2; Fig. 3). Absent from 13°N samples, the lep1-550 allele was present at a low frequency at East Wall and balanced the more frequent lep1-100 allele at Oasis-BS6 suggesting that selection may affect this locus or that a proportion of individuals came from another differentiated population.

Evidence of crypticism in Lepetodrilus elevatus

Genetic evidence from molecular data

A correspondence analysis performed on allozyme data distinguished two groups on the first axis which accounted for 26.41% of the total variance: one grouping all individuals from 13°N and some individuals from 9°50′N, principally from East Wall (hereafter called the northern clade), and one clustering all individuals from 17°S and most individuals from 9°50′N (hereafter called the southern clade) (Fig. 4). Within the northern clade, Julie was slightly separated from Elsa–Genesis, while in the southern clade, the second axis separated 17°S populations from the 9°50′N ones (Fig. 4). On the first factorial plane, few individuals displayed an intermediate situation between the two principal groups and may be attributed to the presence of hybrids. They correspond to four specimens from 9°50′N and one specimen from Julie site (13°N).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0829-3/MediaObjects/227_2007_829_Fig4_HTML.gif
Fig. 4

Factorial correspondence analysis of enzyme loci. Ellipses symbolize vent fields (13°N, 9°50′N and 17°S)

Because two genetically-distinct groups of individuals have been distinguished from the correspondence analysis, a 361-bp fragment of the mitochondrial COI gene was examined from 79 specimens of Lepetodrilus elevatus coming from 17°S, 9°50′N and 13°N, 7 L. cristatus and 3 L. pustulosus. These COI sequences correspond to the accession numbers from EF486360 to EF486445 in Genbank. From L. elevatus individuals, a total of 31 haplotypes were identified that revealed 108 polymorphic sites including 9 singletons and 99 parsimoniously informative sites. The maximum likelihood tree revealed two distinct lineages highly supported by bootstraps values (Fig. 5). The two lineages were separated by 15 fixed nucleotide substitutions that did not lead to any fixed amino acid substitutions according to the invertebrate translation code. This leads to an average divergence (K2Pavg) of 6.5% between lineages. The two lineages are geographically well-discriminated: a northern lineage with Lepetodrilus elevatus populations from 13°N and a southern lineage with populations from 17°S. Only one individual belonging to the off-axis Caldera site at 13°N and several specimens from 9°50′N (i.e. East Wall) clustered with the southern lineage.
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0829-3/MediaObjects/227_2007_829_Fig5_HTML.gif
Fig. 5

Maximum likelihood tree based on the 31 Lepetodrilus elevatus haplotypes and Lepetodrilus spp. outgroups. Number next to nodes corresponds to the bootstrap values. Number in brackets corresponds to the number of individuals from the different locations

Genetic variation for each population is given in Table 5. Total haplotype diversity (He-HAP) and nucleotide diversity (π) were 0.882 ± 0.027 and 0.0306 ± 0.002, respectively. The individuals from Caldera and East Wall exhibited the highest nucleotidic diversities, but this was due to the admixture of the two lineages at these sites.
Table 5

Mitochondrial COI genetic variability for each population

 

13°N

9°50 N

17°S

Elsa

Genesis

Caldera

East Wall

Oasis

MtCOI

 NHAP

5

4

4

10

15

 HeHAP

0.764

1.000

0.750

0.800

0.863

(±0.107)

(±0.177)

(±0.139)

(±0.069)

(±0.062)

π

0.0027

0.0074

0.0165

0.0212

0.0060

π intra clade

0.0027a

0.0074a

0.0021a

0.0036b

0.0060b

NHAP Number of haplotypes, He-HAP haplotypic diversity, π nucleotide diversity

Northern clade

Southern clade

Lack of evidence from morphological data

To discriminate individuals from the two different clades as identified by allozyme and mtCOI data using shell morphology, two principal component analyses (PCA) were carried out on the log-shape ratio for each type of markers (Fig. 6). Individuals of Lepetodrilus galriftensis were used as a control. Potential hybrids identified from the correspondence analysis performed on allozyme data were removed to improve the discrimination between the two clades.
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0829-3/MediaObjects/227_2007_829_Fig6_HTML.gif
Fig. 6

Principal component analysis on the log-shape ratio (LSR) of shell morphology. a Eigenvalues and b individual scores in the plane F1 × F2 for the two clades identified by allozymes. c Eigenvalues and d individual scores in the plane F1 × F2 for the two clades identified by mtCOI. On both analyses, Lepetodrilus elevatus galriftensis individuals were added

In the PCA on individuals used in the allozyme analysis, axis 1 explained 39.03% and axis 2, 28.77% of the total variance. In the second PCA, axes 1 and 2 explained 38.46 and 32.63 % of the variation, respectively, for the individuals used in the mtCOI analysis. For both the analyses, total length, basal length and height (Ltot, Lbas and H) provided the heaviest loading on axis 1 while shell width (W) has the heaviest loading on axis 2. Axis 1 separated partially L. elevatus galriftensis from L. elevatus according to the flatter shell morphology of the former species. PCAs however failed to clearly discriminate between the northern and southern clades of L. elevatus when using shell morphology.

Demographic analyses within each clade

To test genetic homogeneity within clades, each clade was then analysed separately. Although the Gpi locus appeared to be of great interest to discriminate the two evolutionary lineages, large gaps in the data for this locus lead us to remove it for the intra clade analysis. The high number of alleles observed at 9°50′N in the global analysis (i.e. Nall = 3.7−4.3, see Table 3) was no longer observed when considering each clade separately. Likewise, when each lineage was considered separately, the nucleotide diversity of mitochondrial sequences ranged between 0.0027 and 0.0074 and was quite homogeneous among populations. For allozymes, heterozygote deficiencies greatly decreased in the northern clade with no more Fis values significantly different from zero (Table 6) and genotypic disequilibria were no longer detected for both clades. Significant heterozygote excesses as compared to Hardy–Weinberg equilibrium only remained for the southern clade. Both populations deviated from the null hypothesis of demographic equilibrium with a significant excess of heterozygotes according to Cornuet and Luikart (1996) (Wilcoxon test, P < 0.05) supporting a recent bottleneck in those populations.
Table 6

Fis values at enzyme loci for each clade (*P < 0.05; **P < 0.01; ***P < 0.001), and allozyme and nuclear genetic variability

Locus

13°N

9°50 N

17°25 S

Elsa

Genesis

Julie

Biotransect

BioVent

East Wall

Oasis-BS6

Oasis-BS13

All

Northern clade

Fis

Gpi

−0.215

−0.171

0.746***

0.048

0.543

0.46*

0.058

Idh1

−0.016

−0.094

−0.012

−0.059

−0.000

−0.000

0.045

Idh2

0.218

0.051

0.252*

0.714**

0.272

−0.085

0.166**

Mdh

−0.030

−0.329**

−0.019

−0.125

−0.200

−0.223

−0.156**

Aat2

−0.015

−0.018

−0.014

Mpi

0.192

0.159

0.068

0.000

0.077

0.167

0.143

Pgm2

−0.155

−0.140

0.068

−0.500*

−0.391

−0.290

−0.156*

All

0.002

−0.068

0.116

0.048

0.104

0.036

0.017

NALL

2.9

2.7

2.9

2.3

2.7

2.6

 

Ho

0.360

0.370

0.272

0.329

0.429

0.359

 

HNB

0.361

0.346

0.307

0.344

0.473

0.372

 
 

(±0.223)

(±0.202)

(±0.234)

(±0.270)

(±0.280)

(±0.252)

   

Rs

2.146

1.975

1.865

2.072

2.7

2.04

 

Southern clade

Fis

Gpi

1.000

0.863***

0.402

0.750***

Idh1

0.000

0.000

0.063***

Idh2

0.579***

1.000*

0.626***

Mdh

−0.018

−0.426***

−0.072

−1.000***

−0.745***

−0.600***

Aat2

1.000

0.113

−0.143

0.234

0.243

0.221**

Mpi

−0.167

−0.071

0.189

−0.941***

−1.000***

−0.634***

Pgm2

−0.577*

−0.580***

−0.029

−0.425*

−0.186

−0.373***

All

0.106

0.017

0.155

−0.527***

−0.408***

−0.237***

NALL

-

2

2.9

2.6

1.9

2

 

Ho

0.268

0.353

0.332

0.462

0.426

 

HNB

0.295

0.359

0.390

0.305

0.304

 
    

(±0.243)

(±0.205)

(±0.255)

(±0.288)

(±0.281)

 

Rs

1,917

2.172

2.262

1.712

1.736

 

Nall mean number of alleles per population; Ho and HNB observed and expected multilocus heterozygosity (±SD); Rs allelic richness

The northern clade

In the northern clade, pairwise Fst values from allozymes data indicated that the Julie (13°N) and Biotransect (9°50′N) samples differed from all other samples (Table 7). However, pairwise Fst values estimated between the vent fields 9°50′N and 13°N as a whole revealed no significant differentiation (Fst = 0.010, P > 0.05). Similarly, pairwise Fst estimated from haplotype frequencies were also significant between the Elsa population and all the other populations from 13°N and 9°50′N vent fields (Table 8). The haplotype network based on nine haplotypes showed the occurrence of two equally-frequent haplotypes that are not geographically specific (Fig. 7). Mismatch distribution exhibited a unimodal shape with an excess of rare variants when compared to the expected fitting curve for stable populations but fitted well with the expected curve obtained under the population growth/decline model (Fig. 8). This result was further confirmed by the Ramos-Onsins and Rozas R2 test that rejected the null hypothesis of constant size (i.e. observed R2 = 0.0597; P = 0.006). If the global Fu and Li’s F statistics was negative and deviated significantly from neutral expectations (i.e. F = −2.899; P < 0.05), this statistics was not significant for the Caldera and Elsa–Genesis populations (P > 0.1 and 0.05 < P < 0.1, respectively).
Table 7

Pairwise Fst values based on enzyme loci for the northern clade (above diagonal) and the southern clade (below diagonal) (*P < 0.05; **P < 0.01; ***P < 0.001)

 

13°N

9°50 N

17°S

Elsa

Genesis

Julie

Biotransect

BioVent

East Wall

Oasis-BS6

13°N

Elsa

 

−0.002

0.231***

0.093***

0.007

0.008

Genesis

 

0.251**

0.118***

0.040

0.021*

Julie

 

0.298**

0.138*

0.172***

9°50 N

Biotransect

 

0.074

0.102***

BioVent

0.321***

 

−0.006

East Wall

0.349**

0.038*

 

17°S

Oasis-BS6

0.070*

0.296***

0.370***

 

Oasis-BS13

0.072*

0.296***

0.357***

0.011

Table 8

Pairwise Fst values based on haplotype frequencies for the northern clade (above diagonal) and the southern clade (below diagonal) (* P < 0.05; ** P < 0.01; *** P < 0.001)

 

13°N

9°50 N

Genesis

Caldera

East Wall

13°N

Elsa

0.145*

0.249***

0.267*

Genesis

 

0.149

0.148

Caldera

 

0.291

9°50N

East Wall

 

17°S

Oasis-BS6

0.212***

https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0829-3/MediaObjects/227_2007_829_Fig7_HTML.gif
Fig. 7

Haplotypic network showing the evolutionary relationships between mitochondrial haplotypes of Lepetodrilus elevatus for the northern clade. Circles represent haplotypes with the surface of each circle representative of the frequency with which it occurred in the total sample

https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0829-3/MediaObjects/227_2007_829_Fig8_HTML.gif
Fig. 8

Observed and simulated match-mismatch curves for both northern and southern Lepetodrilus elevatus clades under constant population size (left) and population growth/decline model (right)

The southern clade

Based on allozyme data, pairwise Fst values in the southern clade showed a significant level of differentiation between all samples except the two samples within the 17°S field (Table 7). These values were maximal between the southern and northern vent fields (average Fst estimate = 0.243, P < 0.01). Haplotypic frequencies indicated that populations from East Wall and Oasis-BS6 were significantly differentiated (Table 8). As only one sequence represented the Caldera population, this latter was not included in the analysis. The haplotype network established from 22 haplotypes indicated that the southern lineage displayed a unique ancestral haplotype from which several variants derived by two to three substitutions (Fig. 9). Mismatch distribution showed a unimodal shape with an excess of rare variants when compared to the expected fitting curve for stable populations and rather followed the expected curve drawn under the population growth/decline model (Fig. 8). The Ramos-Onsins and Rozas R2 test rejected the null hypothesis of constant population size (R2 = 0.0361; P = 0.000). Fu and Li’s F statistics was negative and deviated significantly from neutral expectations (i.e. F = −3.926; P < 0.05).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0829-3/MediaObjects/227_2007_829_Fig9_HTML.gif
Fig. 9

Haplotypic network showing the evolutionary relationships between mitochondrial haplotypes of Lepetodrilus elevatus for the southern clade. Circles represent haplotypes with the surface of each circle representative of the frequency with which it occurred in the total sample

Discussion

Identification of cryptic species

The analysis of allele frequencies at allozyme and anonymous nuclear loci along the East Pacific Rise, from 13°N to 17°S, for the vetigastropod Lepetodrilus elevatus showed that significant patterns of isolation-by-distance together with high significant Fis values and linkage disequilibria for allozymes are the result of a clinal admixture of two distinct lineages. Craddock et al. (1997) reported the existence of two species genetically differentiated along the EPR between 21°N and 9°N which were tentatively identified as L. elevatus and L. galriftensis. These authors indeed stated that no allozymes (i.e. electrophomorphs) were shared between these two morphological forms of Lepetodrilus. In the present study, all enzyme loci display shared alleles [loci are similar to those analysed by Craddock et al. (1997), see Table 2] and thus do not correspond to the species referred as L. galriftensis by Craddock et al. (1997) nor to L. e. galriftensis sensu McLean on the basis of the shell morphology. These results demonstrate therefore the occurrence of two cryptic species within L. elevatussensu stricto that were not detected previously.

The first lineage is restricted to the northern part of the EPR, from 13°N to 9°50′N, while the second lineage is mostly found in the south, from 9°50′N to 17°S. Only one individual from the off-axis Caldera site at 13°N was assigned to the southern lineage. This assertion was highly supported by both mitochondrial haplotypes and allozyme frequencies. The two clades of mtCOI haplotypes, separated by 15 substitutions on the ML tree were concordant with groups already discriminated by the correspondence analysis performed on the allozyme dataset. If no diagnostic allele has been detected in the present study, latitudinal clines in allele frequencies are observed for five out of seven allozyme loci. Linkage disequilibria at enzyme loci as well as highly significant heterozygote deficiencies for the whole 9°50′N sample, confirmed the admixture of two differentiated genetic entities in this region. Conversely, the two nuclear loci failed to be diagnostic for the discrimination of these two clades but displayed a significant isolation-by-distance relationship related to an allelic cline at the lep1 locus. Two hypotheses can explain this result. On one hand, indels that cause allelic differentiation may have the same size but different positions in the sequence and, thus could represent homoplasious states between the clades. On the other hand, this may indicate gene flow across a semi-permeable barrier for neutral alleles, while other genetic markers would be counter-selected and not traverse the barrier. Lastly, indel polymorphisms found at these two loci may have been present in the ancestor predating the speciation and (re)colonisation events. Several evolutionary mechanisms have been proposed to explain the retention of ancestral polymorphism. First, selection can maintain specific allelic forms despite different speciation events. Second, alleles behaving neutrally can also persist if effective population size is large enough to limit their elimination by genetic drift (Goodacre and Wade 2001). Only one allele (i.e. lep1-550) restricted to the population from 17°S seemed to have appeared after the lineage separation. At any rate, the different status of lep and allozymes alleles as regards ancestral polymorphism, secondary exchanges and allelic frequency shifts, remains a pending question.

Divergence between mtCOI sequences ranged from 0 to 1.1% within the northern clade, and from 0 to 2.8% within the southern clade, but reached 6.5% between the two clades. Although Knowlton (2000) pointed out that the degree of genetic divergence between the distinct species can be highly variable, these distances correspond to intra- and inter-specific levels of divergence previously reported in other vent species. In the Lepetodrilus genus, a divergence of 7.3% was observed between mtCOI sequences of L. fuscensis and L. gordensis in the northeastern Pacific (Johnson et al. 2006). In bivalves, such a divergence ranged from 3.9 to 10.3% between distinct lineages of two species-complex of vesycomid clams (Goffredi et al. 2003) whereas a 4.4% divergence separated cryptic species of Bathymodiolus sp. across the Eastern microplate (Won et al. 2003). Multiple cryptic species of Oasisia tubeworms along the EPR exhibited a divergence ranging 2.6–9.7% (Hurtado et al. 2002). The two lineages of L. elevatus observed along the EPR could therefore be recognized as two sibling species. Even if criteria to erect new species are diverse, the fact that these lineages can be distinguished from multiple and independent genes, increase our confidence that they correspond to two discrete evolutionary units (Avise and Wollemberg 1997). Nevertheless, the presence of a few intermediate individuals on nuclear genes suggested that these lineages may hybridise at 9°50′N and probably, locally, at 13°N (Julie and Caldera sites). This was confirmed by the presence of one individual from 9°50′N assigned to the southern clade from mitochondrial DNA and to the northern clade from allozymes.

Even if there is an overlap between the three species, the morphological analysis indicated that the shell of the two sibling species of L. elevatus differed from that of L. e. galriftensis. As reported by McLean (1988), L. e. galriftensis is characterised by a lower shell elevation. By contrast, no difference has been detected between the two sibling species in their shell morphology. This absence of difference supported previous genetic studies on other vent species that led to the unexpected discovery of highly divergent sympatric taxa with the same morphology (Peek et al. 1997; Goffredi et al. 2003; Won et al. 2003). For mollusks, differences in shell shape and morphology can be phylogenetically constrained but can also reflect an environmental morphological plasticity which appears to be a common trait for vent endemic organisms (Black et al. 1994; Southward et al. 1996). Although differences in shell morphology like shape, decoration and colour, are diagnostic for the identification of Lepetodrilidae species, observations of the soft part of the body could be essential to discriminate between sibling species (Johnson et al. 2006).

Speciation processes

Vicariance plays an important role in the differentiation of hydrothermal vent populations through topographical barriers and global deep-sea oceanic circulation that alter along axis larval dispersal (Van Dover et al. 2002). A shift between NEPR and SEPR with various distributional limits has been already described along the East Pacific Rise in relation to different dispersal filters for a number of vent species like mussels (Won et al. 2003) and polychaetes (Hurtado et al. 2004). These filters included (1) a strong westward current that crosses the ridge axis at 15°S, (2) strong eastward deep-sea equatorial currents that create abrupt northern and southern gyres on both sides of the Equator, (3) the triple junction between the Galapagos Rift (GAR) and the EPR and (4) the succession of large transform faults reported between 17°S and the triple junction between GAR and EPR (e.g. Garrett, Wilkes and Discovery; see Chevaldonné et al. 1997). In the present study, the presence of the southern lineage of L. elevatus at 9°50′N suggests that dispersal filters mentioned above are probably inefficient to explain the speciation processes among the L. elevatus complex.

Tectonic history of mid-ocean ridges could also be responsible for vicariant events on the hydrothermal fauna that often led to genetic breaks in numerous species (Tunnicliffe and Fowler 1996). A tectonic model proposed for the evolution of the southern East Pacific Rise from the end of the Oligocene till recent days highlighted (1) a major reorganisation and reorientation of the spreading centre between 20 and 18.5 Mya, which remains active until the present south to 13°S, and (2) the formation of a temporary plate (i.e. Bauer plate) between 8.2 and 6.5 Mya north of 13°S (Mammerickx et al. 1980). Similar to the present-day Easter Microplate which is an important geographical barrier to gene flow (Won et al. 2003, Hurtado et al. 2004), this temporary plate could have led to a vicariant event that might account for allopatric speciation of the two lineages. Different molecular rates have been used to estimate the time of divergence between the two lepetodrilid lineages. With a substitution rate of 0.23% defined for vent annelids on the basis of the age separation of the EPR from the northern ridge complex (i.e. Juan de Fuca/Gorda/Explorer), (Chevaldonné et al. 2002), the separation would have occurred 14 Mya. Using the same period, Johnson et al. (2006) refined a substitution rate of 0.56% for mtCOI in Lepetodrilidae species, which in our case, decreases the time of divergence to 6 Mya. Although evolutionary rates can vary greatly through time and among taxa, and should be cautiously interpreted, a divergence time ranging between 6 and 14 Mya since the separation of the two evolutionary lineages of L. elevatus would corroborate the role of the Bauer plate in the lineages differentiation. Following that the southern population has, in a recent past, undergone a demographic expansion, colonising progressively new venting areas, this may explain why a secondary contact actually occurs at the 9°50′N with potential hybridisation.

Moreover, this vicariant event has been probably reinforced by a partial habitat specialisation of each lineage as suggested by the differences in the fauna they were associated with. While both lineages were sampled at 9°50′N, their mixing was found highly dependant on the proportion of mussels and vestimentiferan tubeworms in the assemblages. The southern clade was mainly reported from mussel beds and colonies of clams at 17°S and communities dominated by mussel beds at 9°50′N. Conversely, the northern clade was highly dominant at 13°N where communities were mainly composed of Riftia pachyptila clumps with some scarce mussels. Then, the geographical expansion of the southern lineage to the north could be limited to mussel-dominated habitats for which the individuals may be pre-adapted.

Given the high spatial and temporal heterogeneity of hydrothermal environment, organisms are likely to adapt to a particular resource and/or physico-chemical environment. The hypothesis of sympatric speciation between the two Lepetodrilus elevatus lineages can therefore not be ruled out, and different processes could be proposed to explain such a speciation: (1) habitat preferences may split a species by strong diversifying selective processes if gene flow is reduced and/or populations greatly fluctuate in size, (2) disruptive selection combined with habitat-based assortative mating can rapidly lead to reproductive isolation (Palumbi 1994). More geographic samples are therefore needed to discriminate between these alternative speciation processes.

Past and recent demographic histories in the southern and northern clades

According to a faster evolution rate in mitochondrial genes as compared to enzyme genes, contrasting patterns in allozyme and mitochondrial diversities occurred when populations deviate from the mutation-drift equilibrium in response to continuous population reduction and expansion as reported for different vent species (Hurtado et al. 2004). Both the mitochondrial nucleotidic diversity and gene heterozygosity were almost similar between populations of the two clades, suggesting that these clades are still expending since speciation. Nevertheless, significant deviation from neutral expectation could also be the result of nearly simultaneous selective sweep since mtDNA can be subjected to recurrent adaptative selection (Bazin et al. 2005). However, the 17°S samples displayed a low allelic richness, a significant excess of heterozygotes at almost all loci and the presence of fixed alleles that may indicate the occurrence of a very recent and local bottleneck in populations of the southern EPR. This assumption is greatly strengthened by significant departure to the mutation-drift equilibrium under the Infinite Allele model of mutations. This fits well with in situ observations of the southern venting fields during the Biospeedo cruise that indicated the occurrence of numerous vent extinctions in the vicinity of 17°S as compared to previous observations made on deep-sea assemblages in 1993 (Naudur cruise, chief scientist: J.M. Auzende) and 1999 (SEPR Alvin cruise, chief scientists: R.C Vrijenhoek and J. Lupton). Such demographic changes could be then explained by frequent volcanic eruptions which favour the local extinction of existing vent communities and the recolonisation of new vents along ultra fast-spreading ridge systems like SEPR (Haymon et al. 1991). However, these metapopulation processes are likely to alter the relationship between genetic differentiation and gene flow by influencing F Statistics (Jollivet et al. 1999).

For each clade, genetic differentiation between nearby vents distant from 0.1 to 1 km was as important as between vent fields distant from 330 to 3,000 km, suggesting that dispersal processes may be greatly balanced by extinction/recolonisation processes and thus may play a minor role in structuring populations or can be locally biased by hybridisation between lineages. Such result is coherent with previous observations showing a relative genetic homogeneity in Lepetodrilus elevatus populations from 21°N to 9°50′N (Craddock et al. 1997). Although these authors suggested a potential “stepping-stone” dispersal pattern, their analysis could be partly biased by either recolonisation processes or the mixing of the two sibling species in samples from 9°50′N.

Little is known about larval life span and dispersal of lepetodrilids although previous records supported the hypothesis of reduced effective dispersal capabilities. Morphological analyses of the larval shell suggested that these species possess a nonplanktotrophic mode of development and dispersal limitation even if numerous factors may affect larval life span (Lutz et al. 1986; Craddock et al. 1997). Furthermore, most Lepetodrilus spp. larvae being confined within a few meters above the seafloor could be mainly transported by relatively slow near-bottom currents (Mullineaux et al. 2005). By contrast, megaplumes associated with frequent seafloor volcanic eruptions could favour pulses of communication between vent invertebrate populations along ridge segments of fast-spreading ridges like the SEPR (Chevaldonné et al. 1997; Jollivet et al. 1999) and could then promote genetic homogenisation of populations at a large scale.

Acknowledgements

We thank the crew and pilots of the RVAtalante and the DSV Nautile for their assistance and technical support during the cruises HOPE’99 and BioSpeedo’04. We also thank François Lallier, chief scientist of the HOPE’99 cruise who allowed us to collect material. We are indebted to T. Shank and J.H. McLean who provided samples from the 9°50′N vent field and Galapagos Rift, respectively. We also thank the platform GENOMER for sequences acquisition and P. Labbe and J. de Barry for their help in DALP markers production. This work was partly funded by the programme Dorsales and the GDR Ecchis (Ifremer, CNRS). It is a part of M.M. PhD thesis supported by a grant from the French Ministry of National Education and Research.

Copyright information

© Springer-Verlag 2007