Marine Biology

, Volume 145, Issue 6, pp 1257–1264

Phylogeographic differentiation of storm petrels (Hydrobates pelagicus) based on cytochrome b mitochondrial DNA variation

Authors

    • Laboratoire d’Ecologie Moléculaire, EA3525Université de Pau et des Pays de l’Adour, UFR Sciences et Techniques
  • B. Lauga
    • Laboratoire d’Ecologie Moléculaire, EA3525Université de Pau et des Pays de l’Adour, UFR Sciences et Techniques
  • G. Hémery
    • Centre de Recherches sur la Biologie des Populations d’Oiseaux, Station Maritime de RechercheMuséum National d’Histoire Naturelle
  • C. Mouchès
    • Laboratoire d’Ecologie Moléculaire, EA3525UFR Sciences et Techniques Côte Basque
Research Article

DOI: 10.1007/s00227-004-1407-6

Cite this article as:
Cagnon, C., Lauga, B., Hémery, G. et al. Marine Biology (2004) 145: 1257. doi:10.1007/s00227-004-1407-6

Abstract

We evaluated mitochondrial DNA (mtDNA) sequence variation in a 910 bp region of the cytochrome b gene of the storm petrel, Hydrobates pelagicus. Samples from birds collected from five populations in the North Atlantic Ocean and the Mediterranean Sea were investigated. Genetic differentiation within the Mediterranean basin was low but high in the Atlantic. Strong differences were noted between the Atlantic and the Mediterranean populations, confirming the distinction of the subspecies H. p. pelagicus and H. p. melitensis for the Atlantic and Mediterranean seabirds, respectively. Divergence between the two subspecies probably resulted from paleogeographic changes in the Strait of Gibraltar, which was likely the route used by H. pelagicus to invade the Mediterranean Sea. Current and past demography and ecology of the storm petrel is regarded as an explanation for the level of differentiation observed within each oceanic basin. We compare the phylogeographic pattern of the storm petrel to other seabirds that breed in the same regions.

Introduction

During the Pleistocene, several major glaciations took place in the higher latitudes of the northern hemisphere. As a consequence, marine and vegetation zones were compressed toward the equator, as polar ice sheets spread considerably. Moreover, the reduction in sea level produced land bridges in several parts of the world (see Hewitt 2000 for review). These paleoclimatic events, marked by glacial advances, stabilization, and retreat, have been described as major events for the structuring of numerous extant populations under these latitudes (Hewitt 1999). The marine environment in western and southern Europe has been subjected to severe changes, marked in the Atlantic Ocean by a seasonal reduction of most of its surface due to ice sheets and in the Mediterranean, at the Strait of Gibraltar, by the emergence of land due to the decrease in sea level (Collina-Girard 2001; Hewitt 2000).

Geographic survey of nucleotide sequence variation via gene trees (i.e. phylogeography) can be a useful tool to assess the impact of global historical events on the structure of modern populations (Avise 1989). Partitioning of some conspecific populations between the Mediterranean Sea and the Atlantic Ocean have been attributed to these climate oscillations. However, most of the studies reporting partitioning are on aquatic organisms: for example, marine fishes (Kotoulas et al. 1995; Roldan et al. 1998), cetaceans (Garcia-Martinez et al. 1999), and molluscs (Perez-Losada et al. 1999). To develop a greater understanding of the effect of climate change and ice age scenarios on the marine ecosystems, we examined the phylogeographic pattern of the storm petrel (Hydrobates pelagicus; Hydrobatidae). We believe that marine birds are a useful model as they are strongly dependant on marine resources. They breed only on islands or the coastline and do not forage inland or migrate through continental areas (Zotier et al. 1999).

Hydrobates pelagicus is among the smallest marine bird species (average mass 30 g). Its breeding area is limited to the northeastern Atlantic Ocean and the western Mediterranean Sea (Fig. 1). Moreover, the birds of the two regions seem geographically and demographically isolated as several decades of observations and banding efforts have not demonstrated bird exchange or movement between the Atlantic Ocean and the Mediterranean Sea (Hémery and d’Elbée 1985). Finally, the distinction of two subspecies in H. pelagicus has been a topic of interest for some naturalists for a long time. Indeed, Schembri (1843) then Mathews (1934) first used plumage coloring to recognize two subspecies: H. pelagicus melitensis for the birds that breed around the Mediterranean Sea, and H. pelagicus pelagicus for the North Atlantic populations. However, this distinction was erroneous, since darkness of the plumage depended on the difference between fresh and worn feathers (Mayaud 1941, 1949). Later, based essentially on several measurements of the bill of a large sample (about 100 birds) from both museum and live specimens, Hémery and d’Elbée (1985) restored the distinction of the two subspecies. Using the same morphometric criteria, this result has been confirmed in a subsequent analysis (Lalanne et al. 2001). Although statistically significant, the distinction found by using this set of measures is not easy to achieve, since strict rules of use must be followed (Hémery and d’Elbée 1985).
Fig. 1

Geographical location of the Atlantic and Mediterranean populations. The shaded pie charts display haplotype frequencies; sample sizes are shown under each pie chart. H1–H8 haplotype names

Hence we use mitochondrial DNA (mtDNA) sequences from the cytochrome b gene (1) to test whether the distinction of two subspecies is supported with molecular data, (2) to investigate the phylogeography of the species, (3) to assess its genetic structure, and finally (4) to compare the observed pattern of diversity with data published on other bird species that breed in both the Mediterranean Sea and the Atlantic Ocean, to investigate the forces that promote population differentiation in these areas.

Materials and methods

Samples, DNA isolation, PCR, and sequencing

A total of 65 specimens of H. pelagicus were used for this study. They belong to five populations: Faeroe Islands, Banneg and Lorient (Brittany), La Roche Ronde and Bouccalot islets of Biarritz (Cantabrian population), Planne and Jarre islets of the Riou archipelago near Marseille, and Cerbicales islands in Corsica (Fig. 1). Samples from the Faeroe Islands were obtained from the tissue collection of the Zoological Museum of Copenhagen. Two samples, one from Lorient and one from Banneg, came from muscle tissue of specimens that had died. Blood samples were collected from birds belonging to each previously described colony. About 100 µl were preserved in APS buffer (Arctander 1988) and stored at –20°C. Total DNA was extracted by using the Easy-DNA kit (Invitrogen). A portion of mitochondrial DNA, covering nearly the entire sequence of the cytochrome b gene (970 bp of the 1,123 bp), was amplified by polymerase chain reaction (PCR). The primers used were HP1056 (5′-AACATCTCAGCATGATGAAA-3′), derived from one of the general primers; L14841, defined by Kocher et al. (1989); and a specific primer HP1070 (5′-GAGGCTAGTTGGCCGATGAT-3′), defined from the H. pelagicus cytochrome b gene (AF076059; Nunn et al. 1996). PCR products were purified using the Qiaquick gel extraction kit (Qiagen). Sequencing was done in both directions with the BigDye terminator cycle sequencing kit on an ABI PRISM-370 DNA sequencer (Perkin Elmer Applied Biosystem) providing unambiguous sequences of 910 bp for all 65 individuals. Sequences were confirmed as those of the mitochondrial cytochrome b gene by comparison with previously published sequences (Heidrich et al. 1998; Nunn et al. 1996).

Population genetic analysis

All sequences were aligned using CLUSTAL X version 1.8 (Thompson et al. 1997). The specimen from Lorient was grouped with the Banneg samples for all the analyses since it is single and was sampled from just around 100 km away from Banneg’s colony. The two colonies sampled from Biarritz, La Roche Ronde and Bouccalot, and the two colonies sampled from Marseille, Jarre and Planne, were also grouped into two populations (Biarritz and Marseille, respectively) since the birds of each locality are only a short distance apart (<1 km). Genetic diversity within populations was estimated with nucleotide (π) and haplotype (h) measures of diversity (Nei 1987). Differences in nucleotide diversity between colonies were tested using the formula t1−π2/[V1)2+V2)2]1/2 with V(π) variance of π (Nei 1987). Analysis of population genetic structure was carried out using F statistic calculated according to Hudson et al. (1992b). We also performed additional tests of population differentiation using the statistic K*ST (Hudson et al. 1992a) between all pairwise comparisons of the five populations, all populations together, and between the populations of the two basins. K*ST is a weighted measure of the ratio of the average pairwise differences within populations to the total average pairwise differences. This measure has been shown to have the highest power to detect population differentiation under simple models of population structure (Hudson 2000). Significance levels for these statistics were assessed using permutation tests, with 10,000 permutations (Hudson et al. 1992a). Diversity, population structure analyses, and permutation tests were estimated using DnaSP version 3.98 (Rozas and Rozas 1999). MEGA version 2.1 (Kumar et al. 2001) was used to calculate pairwise genetic distances using a Kimura two-parameter (K2P) model that corrects for multiple substitutions (Kimura 1980). Net sequence divergence between the populations of the two regions was corrected for within-region variation as described in Wilson et al. (1985).

Haplotype genealogy analyses

Phylogenetic trees using maximum-parsimony (MP; Fitch 1971) and neighbor-joining (NJ) methods (Saito and Nei 1987) were constructed using MEGA version 2.1 (Kumar et al. 2001). To test for statistical significance of the generated trees, data were resampled 1,000 times to obtain bootstrap P values (Felsenstein 1985) using the bootstrap option in the MEGA software. The Oceanodroma furcata (AF076063) sequence for the homologous cytochrome b fragment was used as an outgroup to root the tree since it is only about 9.7% divergent from the storm petrel sequences. A molecular-clock approach (Zuckerkandl and Pauling 1965) was applied to provide approximate evolutionary time frames for phylogenetic branching events between the populations of the two regions. Net divergence was used to correct for ancestral diversity. Based on fossil calibration, Nunn and Stanley (1998) estimated the rate of cytochrome b gene evolution of the Oceanitinae subfamily to be 1.29% sequence divergence between lineages per million years. However, since the method used to evaluate genetic divergence might underestimate this rate, we also applied the widely used divergence time for mtDNA in birds of 2% per million years (Shields and Wilson 1987). These molecular clocks must be regarded as provisional.

A statistical parsimony tree was constructed using the program TCS (Clement et al. 2000) to estimate genealogical relationships among sequences according to the method of Templeton et al. (1992) and to visualize the number of mutations between haplotypes.

Results

H. pelagicus mitochondrial cytochrome b sequence

The nearly complete sequence of the mitochondrial cytochrome b gene was analyzed for 43 H. pelagicus from the Atlantic Ocean and 22 from the Mediterranean Sea. We found eight haplotypes within all H. pelagicus samples. Twelve variable nucleotide positions were found over the 910 bp portion analyzed. No insertions or deletions were observed; all changes were point mutations with a transition:transversion ratio of 11:1. Furthermore all substitutions were in the third base position of the codon and none led to replacement in the coding frame. Average base composition was biased with a deficiency of guanine (13.8% G, 34.4% C, 27.8% A, and 24% T). Nine of the 12 variable sites were phylogenetically informative (Table 1).
Table 1

Variable nucleotide positions in mitochondrial cytochrome b DNA sequences between Hydrobates pelagicus haplotypes and haplotype frequencies. Dots indicate a match with the H1 sequence. Numbers above each site correspond to the first base position of the aligned products. Transitions (Ts) and transversions (Tv) are indicated. Accession numbers for the sequences are AF469067 to AF469071

Haplotype

Nucleotide position

Total

8

59

65

74

113

230

422

692

719

770

785

883

Ts

Ts

Ts

Ts

Ts

Ts

Ts

Ts

Tv

Ts

Ts

Ts

Atlantic

  H1

G

G

C

C

C

T

T

A

C

A

A

C

29

  H2

.

.

.

.

.

.

.

G

.

.

.

.

2

  H3

.

.

.

.

.

.

C

.

.

.

.

.

1

  H4

.

.

.

.

.

.

.

.

.

.

.

T

10

  H5

A

.

.

.

.

.

.

.

.

.

.

T

1

Mediterranean

  H6

.

.

T

T

T

C

.

.

A

.

G

.

2

  H7

.

.

T

T

T

C

.

.

A

G

G

.

19

  H8

.

A

T

T

T

C

.

.

A

G

G

.

1

Nucleotide diversity and genetic structure

Haplotype and nucleotide diversities were quite heterogeneous among sample sites (Table 2). In the Mediterranean, nucleotide diversity of the Marseille population was an order of magnitude higher than in the population of Corsica. This result was possibly due to the small sample size of the Marseille population. The Atlantic group showed higher haplotype and nucleotide diversity values than the Mediterranean group. These differences are, however, not significant. In contrast, the Faeroe specimens, the northernmost Atlantic samples, exhibited significantly lower nucleotide diversity than the Biarritz population (P<0.05).
Table 2

Levels of haplotype (h) and nucleotide (π) diversities in populations of H. pelagicus. Nhap Number of haplotypes

Sampling size

NHap

h±SD

π×10−2±SD

Populations

  Banneg/Lorient

12

3

0.318±0.164

0.037±0.02

  Biarritz

20

4

0.658±0.065

0.090±0.015

  Faeroe

11

1

0

0

  Corsica

18

2

0.110±0.096

0.012±0.011

  Marseille

4

3

0.833±0.222

0.110±0.037

Group of populations

  Atlantic

43

5

0.499±0.47

0.063±0.012

  Mediterranean

22

3

0.255±0.116

0.02±0.014

Total

65

8

0.7±0.037

0.396±0.028

The haplotype composition of the H. pelagicus sample showed obvious geographical structure as evidenced from Fig. 1. The FST estimate among all five populations was very high (90.1%). This is mainly due to strong differentiation between the Atlantic and Mediterranean regions, as evidenced by very high pairwise FST estimates between populations from the different regions (from 0.88 to 0.99, see Table 3) or between the two regions (0.943). No evidence of introgression was found in our sample. In contrast, the pairwise FST estimates between populations within the Atlantic region were much lower (from 0 to 0.37, see Table 3) and there is no evidence of differentiation between the two Mediterranean populations. The negative FST between Marseille and Corsica reflects a low level of differentiation combined with a small within-population sample, due mainly to the Marseille population. Permutation tests performed on the K*ST statistics confirmed the genetic differentiation among the Mediterranean and Atlantic populations. Comparisons of populations within each basin detected genetic structuring in the Atlantic basin, which appeared to be due to the Biarritz population (Table 3).
Table 3

Genetic difference between populations and variation within and between populations. Above the diagonal FST estimates for between-populations comparison and significance of K*ST statistics represented by asterisks; below the diagonal interpopulation nucleotide divergence (in percent) estimate with Kimura two parameters; in bold on the diagonal intrapopulation diversity

Banneg/Lorient

Biarritz

Faeroe

Marseille

Corsica

Banneg/Lorient

0.037

0.217*

0 ns

0.907***

0.969***

Biarritz

0.081

0.090

0.373**

0.881***

0.939***

Faeroe

0.018

0.072

0

0.929***

0.992***

Marseille

0.793

0.847

0.775

0.110

−0.053 ns

Corsica

0.787

0.841

0.768

0.058

0.012

Significance of the permutation test performed with 10,000 replicates

ns not significant; * 0.01<P<0.05; ** 0.001<P<0.01; *** P<0.001

The average pairwise sequence divergence of all populations was 0.396%. Within populations, diversity ranged from 0 (Faeroe) to 0.110%, and between populations divergence ranged from 0.018 to 0.847% (Table 3). The net average pairwise sequence divergence between the two groups of populations, Atlantic and Mediterranean, was 0.762% (0.802% for uncorrected p distance).

Genealogical relationships among H. pelagicus haplotypes

The mitochondrial cytochrome b sequences obtained in this study and two previously published sequences of H. pelagicus from Malta in the Mediterranean (sequences AJ004180 and AJ004182; Heidrich et al. 1998) were used to build a gene tree (Fig. 2A). The genotypes are clearly differentiated into two clades corresponding to the two regions, Mediterranean and Atlantic. Phenetic (NJ) and cladistic (MP) methods lead to the same tree topology, where each group of populations appears to be monophyletic. Bootstrap values were similar for the two methods and indicate that monophyly of each group is well supported. Divergence between Malta specimens and Mediterranean or Atlantic haplotypes ranged from 0.2 to 1.22% and from 1 to 2%, respectively. In addition to the NJ and MP bifurcating methods, we constructed, using the same set of data, a minimum spanning tree. The resulting tree shows that the Mediterranean and Atlantic clades are six steps apart (Fig. 2B). Consistency between the two methods strongly confirms the distinction of the Mediterranean clade and the Atlantic clade.
Fig. 2A, B

Gene trees of the mitochondrial cytochrome b gene of Hydrobates pelagicus including haplotypes from Malta (Mediterranean, Heidrich et al. 1998). A The tree constructed by the neighbor-joining method using the Kimura distance. Maximum parsimony analyses resulted in an identical topology. Numbers along the branches indicate the percentage of bootstrap replications (1,000) in which the node is supported; only values above 70% are given. Black and white icons refer to Atlantic and Mediterranean origins, respectively. All the haplotypes of each population are included in the tree. B Minimum spanning network for mitochondrial cytochrome b haplotypes of H. pelagicus. Sizes of circles are proportional to frequencies of haplotypes. White circles haplotypes found in the Mediterranean; black circles haplotypes found in the Atlantic; grey dots missing haplotypes. Haplotype names are indicated inside each circle. M1 and M2 correspond to Malta specimens AJ004180 and AJ004182, respectively

Discussion

H. p. pelagicus and H. p. melitensis: two distinct subspecies

We have demonstrated that mitochondrial data based on cytochrome b sequence variation is consistent with the distinction of two subspecies of H. pelagicus: H. p. melitensis for birds that breed around the Mediterranean Sea and H. p. pelagicus for birds that breed in the North Atlantic Ocean. Hence, molecular tools support previous conclusions based on morphometric measurements of the bill (Hémery and d’Elbée 1985) and inferences based on body size (Catalisano et al. 1988). Therefore our approach provides a robust tool for further identifications, without constraints linked to the maturity of the specimens.

Phylogeographic structure of H. pelagicus

The phylogeographic pattern observed for H. pelagicus is a commonly encountered situation that reflects the type-1 mtDNA and geographic assemblage previously defined by Avise et al. (1987). Although similar patterns have been described for aquatic organisms, few studies have attempted to address this question for marine birds that breed both in the Atlantic and in the Mediterranean. Heidrich et al. (1998) used cytochrome b mtDNA variation to establish the phylogenetic relationships between Mediterranean and North Atlantic shearwaters and suggested the recognition of three Puffinus species: P. puffinus from the Atlantic and P. yelkouan and P. mauretanicus from the Mediterranean. These species were previously considered as three distinct subspecies. From the same study, a situation similar to the one we described for H. pelagicus can also be suspected for Cory’s shearwater between Atlantic (Calonectris diomedea borealis) and Mediterranean (C.d. diomedea) populations. However, the sample size is too small to be confident with the pattern revealed for this species. Conversely, the study of Liebers et al. (2001) that investigated population genetic structure within and among nine gull taxa including Larus atlantis from the Atlantic and L. michaellis from the Mediterranean showed that most michaellis haplotypes are either identical or closely related to those in the northern atlantis populations. Their approach based on HVR1 sequence variation of mtDNA suggests ongoing gene flow between the birds of the two regions. This is in contrast to our data for H. pelagicus and to the data of Heidrich et al. (1998) for Puffinus. Ecological differences between Larus and both Hydrobates and Puffinus, such as habitat preferences, philopatry, colonization abilities, or capacity of birds to cross land mass, might account in part for these contrasted phylogeographic patterns.

Hence, in this geographical context (Atlantic vs Mediterranean), we showed clear reciprocal monophyly based on mitochondrial DNA variation for H. pelagicus. This result confirms the pattern found previously, with a smaller sample size, for other tube-nosed seabirds (Puffinus, Calonectris).

Pleistocene imprinting in the phylogeographic structure of H. pelagicus

The existence of some barrier to gene flow between the Atlantic Ocean and Mediterranean Sea is suggested by the mtDNA phylogeographic pattern of H. pelagicus. This finding is surprising, given the high dispersal capability of marine birds (Palumbi 1994). Some of the previous studies on aquatic species have credited separation of the lineages under study to restriction of migration through the Strait of Gibraltar (Perez-Losada et al. 1999; Roldan et al. 1998). Indeed, the paleogeography of this region has been intermittently modified, straightening and partially blocking the route, as a result of glacio-eustatic processes during the Pleistocene epoch (Quaternary; Collina-Girard 2001). These modifications might have constituted a biogeographic barrier precluding colonization or gene flow between the two basins. Applying the provisional molecular clock of 1.29% per million years to the corrected net K2P distance, Atlantic and Mediterranean lineages of H. pelagicus coalesced approximately 550,000 years ago (or 350,000 years for an evolutionary rate of 2% per million years). Moreover, fixed nucleotide differences are observed between the two groups, consistent with long-term geographic isolation between birds of the two basins. The molecular clock estimate gives good congruence with Pleistocene glaciation events and associated climatic phenomena both in the Atlantic and Mediterranean regions. Based on fossil records, a Northern Atlantic origin has been suggested previously for most of the extant Mediterranean seabirds (Zotier et al. 1999). Hence, the inferences made using fossil records and molecular data both suggest similar dates for vicariance events.

Genetic structure of storm petrel populations within each basin

Our results suggest that two contrasting patterns of differentiation between the Atlantic and Mediterranean basins could exist. Indeed, strong genetic structure is observed in the Atlantic suggesting that these populations could be isolated from one another. This result could be explained by a philopatric behavior of the Atlantic population. This hypothesis is consistent with field experiments reporting that no bird exchange took place among populations or colonies (Hémery et al. 1986). In contrast, although strong site fidelity has been observed in the Mediterranean, the low differentiation noticed in this basin could be supported by evidence of contemporary gene flow between populations (Lo Valvo and Massa 2000).

Phylogeographic trends in Atlantic and Mediterranean seabirds

The Atlantic–Mediterranean differentiation noticed across seabird studies and observed in H. pelagicus indicates that the Strait of Gibraltar could have acted as a dividing zone for seabirds species. Note that the separation of lineages in the different seabird species investigated apparently took place at different geological time scales: after the “Messinian crisis” (Miocene) for Puffinus and eventually Calonectris or during the Pleistocene for H. pelagicus.

The use of the mitochondrial cytochrome b sequence allowed us to support previously proposed subspecies distinctions in H. pelagicus. To go a step further in the analysis it would be interesting to estimate past demographic events and factors leading to the structuring of each subspecies. It would be necessary to generate data with higher levels of polymorphism. These objectives are important in a conservation context of the species. Indeed, H. pelagicus is listed on Annex 1 of the EC Birds Directive and its reproductive sites are registered in the “Habitat Directive” of the EC Birds Directive (2nd circular), which define “special zones of conservation”. Moreover, the species has been described previously as vulnerable and endangered (Evans 1986). Hence, since declines have been noticed in the southern part of the range distribution (Hémery 2004; Lloyd et al. 1991) and since lack of gene flow has been revealed among the two subspecies, we suggest that the storm petrel needs to be managed as two independent units.

Acknowledgements

We are greatly indebted to the persons who participated in sample collection: Jean-Claude Thibault (Corsica), Yann Lalanne (Marseille and Biarritz), Jacques Nisser, Bernard Cadiou and Gilles Bentz (Brittany), Frank d’Amico and Jean d’Elbée (Biarritz). We are grateful to Jon Fjeldså and the Zoological Museum at the University of Copenhagen for providing us samples from Faeroe. We thank Solange Karama for technical support and Barbara Mable, Valérie Laporte, and John O’Halloran for comments on the manuscript. Three anonymous reviewers and Associate Editor S. Poulet greatly helped to improve the quality of the manuscript. B. Lauga was supported by a grant from the ‘Société de Secours des Amis des Sciences’. The capture of French birds was under permit from the Centre de Recherche sur la Biologie des Populations d’Oiseaux (CRBPO), and blood collection was approved by the Ministère de l’Environnement et de l’Aménagement du Territoire.

Copyright information

© Springer-Verlag 2004