SEARCH

SEARCH BY CITATION

Keywords:

  • allozymes;
  • gastropoda;
  • hybrid zone;
  • ITS1;
  • phylogeography;
  • reproductive isolation;
  • secondary contact;
  • taxonomy;
  • terrestrial slugs;
  • 16S rDNA

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Appendices

Extremely high levels of intraspecific mtDNA differences in pulmonate gastropods have been reported repeatedly and several hypotheses to explain them have been postulated. We studied the phylogeny and phylogeography of 51 populations (n = 843) of the highly polymorphic terrestrial slug Arion subfuscus (Draparnaud, 1805) across its native distribution range in Western Europe. By combining the analysis of single stranded conformation polymorphisms (SSCP) and nucleotide sequencing, we obtained individual sequence data for a fragment of the mitochondrial 16S rDNA and a fragment of the nuclear ITS1. Additionally, five polymorphic allozyme loci were scored. Based on the 16S rDNA phylogeny, five monophyletic haplotype groups with sequence divergences of 9–21% were found. Despite this deep mitochondrial divergence, the haplotype groups were not monophyletic for the nuclear ITS1 fragment and haplotype group-specific allozyme alleles were not found. Although there is evidence for an accelerated mtDNA clock, the divergence among the haplotype groups is older than the Pleistocene and their current allopatric ranges probably reflect allopatric divergence and glacial survival in separate refugia from which different post-glacial colonization routes were established. A range-overlap of two mtDNA groups (S1 and S2, 21% sequence divergence) stretched from Central France and Belgium up to the North of the British Isles. The nuclear data suggest that this secondary contact resulted in hybridization between the allopatrically diverged groups. Therefore, it seems that, at least for two of the groups, the deep mtDNA divergence was only partially accompanied by the formation of reproductive isolation.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Appendices

The phylogeny of a taxon, seen in time and space, provides invaluable insights on the process of speciation (Avise, 2000; Barraclough & Vögler, 2000; Barraclough & Nee, 2001; Hewitt, 2001; Shaw, 2001). Wiens (2004) argues that the study of speciation should focus on (i) the splitting of evolutionary lineages and (ii) the divergence of such lineages. In pulmonate gastropods, extremely high degrees of mtDNA variation (e.g. up to 20%; Thomaz et al., 1996) raise questions on the relative importance of different evolutionary processes and their taxonomic implications. Thomaz et al. (1996) give four mutually nonexclusive hypotheses explaining the extreme intra-specific mtDNA divergences observed in the pulmonate snail Cepaea nemoralis: (i) mtDNA evolution in pulmonate snails is exceptionally fast, (ii) haplotypes differentiated in isolated ‘refuges’ and subsequently came together, (iii) natural selection has acted to preserve variation, and (iv) the population structure of pulmonates favours the persistence of ancient haplotypes. Various studies have attempted to assess the relative importance of each of these hypotheses in pulmonate gastropods: (i) evidence for an accelerated mtDNA evolution was found in some taxa (e.g. Chiba, 1999; Hayashi & Chiba, 2000; Thacker & Hadfield, 2000), but not in others (e.g. Murray et al., 1991; Douris et al., 1998; Pfenninger & Posada, 2002; Guiller et al., 2001), (ii) several studies showed that a phylogeographic history with allopatric divergence and secondary contact was the most parsimonious way to explain sympatry of diverged haplotypes (e.g. Davison & Clarke, 2000; Shimizu & Ueshima, 2000; Guiller et al., 2001; Pfenninger & Posada, 2002; Teshima et al., 2003), (iii) the influence of natural selection on the molecular genetic population structure of pulmonates was never shown, but the historical explanations did not exclude selection (e.g. Davison & Clarke, 2000; Davison, 2002; Goodacre, 2002) and, (iv) demographic factors, such as highly structured populations and limited active dispersal capacities, are important in the small-scale conservation of polymorphisms (e.g. Davison & Clarke, 2000; Arnaud et al., 2001; Goodacre, 2001; Arnaud & Laval, 2004) and this may be reflected in the long term and large-scale structuring of populations as well (e.g. Thacker & Hadfield, 2000; Guiller et al., 2001; Davison, 2002).

Particularly the first two hypotheses may have a profound influence on the taxonomic interpretation of strongly diverged mtDNA groups within ‘classical, well-known’ pulmonate species. A case in point is the terrestrial slug A. subfuscus (Draparnaud, 1805), a widespread and highly polymorphic species, whose taxonomic status has been debated for many years (e.g. Forcart, 1966; Wiktor, 1983; Backeljau et al., 1996). Recently, Pinceel et al. (2004) showed that the species consists of a complex of cryptic taxa, involving at least two biological species in NW Europe. The two species, A. fuscus (Müller, 1774) and A. subfuscus (Draparnaud, 1805) have distinct gonad morphologies, show strong allozyme and mtDNA differentiation and do not interbreed in sympatry (Pinceel et al., 2004). Within A. subfuscus, several highly diverged (11 up to 21% sequence divergence) and largely allopatric mtDNA groups were detected. Yet, the groups were indistinguishable on the basis of gonad morphology and allozymes (Pinceel et al., 2004). Therefore, and because a high degree of sequence divergence does not necessarily imply the presence of a reproductive barrier (Ferguson, 2002; Lee, 2003; Wiens, 2004), the A. subfuscus mtDNA groups were provisionally treated as a single species (Pinceel et al., 2004).

In this study, we attempt to unravel the evolutionary history of the strongly diverged mtDNA groups in A. subfuscus. We evaluate whether the high mtDNA divergences are due to an accelerated evolution or a deep divergence produced by historical and demographic factors, and discuss the taxonomic implications of these alternatives. To this end, we sampled A. subfuscus across its native distribution range (Belgium, France, W Germany, The Netherlands and Great Britain) (Pinceel et al., 2005) and assessed mitochondrial (16S rDNA) and nuclear [ITS1 (first ribosomal internal transcribed spacer) and five polymorphic allozymes] variation in individuals from 51 locations (n = 843). Using these data, we (1) reconstructed the phylogenetic relationships among the mtDNA groups, (2) estimated their age, (3) established their distributions and the processes that may account for their distributions, and, (4) assessed their nuclear gene-pool (GP) in order to infer ancient or recent gene flow and reproductive isolation or hybridization among the mtDNA groups.

Our results suggest that the five mtDNA groups have separate phylogeographic histories and have evolved with a fast mtDNA clock (>5% divergence per MY). The deep allopatric divergence among the groups was partially reflected in the nuclear GP but, (i) none of the groups had specific allozyme alleles, (ii) only one group (S4) was monophyletic for the ITS1 fragment and, (iii) at least two groups (S1 and S2) appeared capable of hybridization, although the mixing of their nuclear gene-pools (GPs) did not seem to be unconstrained, suggesting that hybridization was either recent and/or that certain mechanisms act against it. In terms of taxonomy, we suggest the mtDNA groups may be treated as evolutionary species.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Appendices

Sample collection

Slugs of all life stages, identified as A. subfuscus on the basis of gonad morphology (Pinceel et al., 2004) were collected by handpicking in locations across the species’ range (Pinceel et al., 2005) (51 populations, n = 843; Table 1 and Fig. 1). Although gonad morphology proved a reliable character, even in small individuals (approximately 10 mm), specific alleles at the Pgm allozyme locus and 16S rDNA sequences were analysed to confirm identifications (Pinceel et al., 2004). All samples were included for phylogenetic analyses, but only samples with 15 or more individuals were used for phylogeographic and population genetic analyses that did not take sample size into account. Animals were stored at −80°C until sample preparation for molecular analysis. The DNA samples, 16S rDNA sequences and allozyme data of 19 of the 51 populations (n = 212) were taken from a previous study (Pinceel et al., 2004) (Table 1). Two specimens of A. fuscus (Genbank accession numbers AJ518037 and AJ786722) and one specimen of A. hortensis (Férussac, 1819) (Genbank accession number AJ518061) were added to the phylogenetic analyses as reference material.

Table 1.  List of sampling locations with indication of population code (ID), location, number of individuals sampled (n), geographic position (latitude and longitude) and 16S rDNA haplotypes for which the number of individuals is given in parentheses.
IDLocationnLatitudeLongitudeHaplotype
  1. For populations indicated with an asterisk, mtDNA and allozyme data were taken from Pinceel et al. (2004).

  2. NT, neotype of A. subfuscus.

  3. TT, topotype of A. subfuscus.

Belgium
 BE1*Kasterlee2751°14′N04°57′ES2-1 (22), S2-5 (5)
 BE2*Mortsel 12251°09′N04°28′ES1-1 (22)
 BE3*Mortsel 22851°09′N04°28′ES1-1 (28)
 BE4*Kortrijk3350°49′N03°16′ES1-11 (33)
 BE5*Halle1350°43′N04°15′ES1-1 (13)
 BE6*Plombières750°43′N05°57′ES2-1 (4), S2-10 (2), S2-11 (1)
 BE7*Clermont1250°39′N05°54′ES2-1 (11), S2-28 (1)
 BE8*Gomzé350°32′N05°41′ES2-1 (1), S2-3 (2)
 BE9*Hoyoux150°23′N05°18′ES2-2 (1)
 BE10Heppenbach2350°21′N06°14′ES2-1 (19), S2-24 (1), S2-25 (1), S2-26 (2)
 BE11*Maffe850°20′N05°19′ES2-1 (2), S2-4 (6)
 BE12*Vaux-Chavanne650°17′N05°42′ES2-2 (4), S2-8 (2)
 BE13*Soy-Fissenne2150°16′N05°31′ES2-1 (15), S2-5 (5), S2-9 (1)
France
 FR1*Boulogne-s-Mer150°43′N01°37′ES1-2 (1)
 FR2Berneuil1850°05′N02°10′ES2-2 (17), S2-12 (1)
 FR3Le Landin1549°23′N00°48′ES2-7 (14), S2-13 (1)
 FR4Rambouillet2248°39′N01°49′ES2-2 (1), S2-7 (18), S2-14 (3)
 FR5Neufchâtel848°22′N00°14′ES2-7 (6), S2-22 (1), S2-23 (1)
 FR6La Salle3048°19′N06°50′ES3-4 (2), S3-5 (19), S3-6 (4), S3-7 (1), S3-8 (1), S3-9 (1), S3-10 (2)
 FR7Servon-s-Vilaine648°07′N01°26′WS2-2 (4), S2-7 (2)
 FR8Mézières-l-Cléry2947°48′N01°47′ES2-2 (3), S2-7 (9), S2-16 (17)
 FR9Ingrandes447°23′N00°53′WS1-4 (4)
 FR10Les Essards3047°20′N00°18′ES2-7 (17), S2-18 (2), S2-19 (1), S2-20 (10)
 FR11Roulans2047°18′N06°13′ES3-4 (20)
 FR12Villemont-Baubiat846°11′N01°16′ES5-6 (1), S5-7 (6), S5-8 (1)
 FR13Chabanais2045°52′N00°43′ES2-1 (2), S2-27 (18)
 FR14Fix-St. Geneys3045°08′N03°40′ES2-1 (30)
 FR15Oulès443°43′N02°35′ES4-3 (1), S4-4 (3)
 FR16Mazamet743°30′N02°24′ES4-5 (7)
 FR17Les Martys443°24′N02°19′ES4-5 (4)
 FR18*Montagne Noire NT143°24′N02°19′ES4-1 (1)
 FR19*Montagne Noire TT143°24′N02°19′ES4-2 (1)
Germany
 DE1*Sennenstadt751°57′N08°34′ES3-2 (7)
 DE2*Blankenheim150°25′N06°38′ES3-1 (1)
 DE3Buchholz950°12′N07°32′ES3-3 (9)
Ireland
 IR1Garrettstown851°54′N08°28′WS1-1 (4), S2-7 (4)
The Netherlands
 NL1*Emmen1552°47′N06°54′ES2-1 (11), S2-6 (4)
 NL2*Berg en Dal551°49′N05°54′ES2-1 (5)
United Kingdom
 GB1Innerleithen3155°37′N03°04′WS1-1 (4), S1-5 (27)
 GB2Ae village3055°10′N03°33′WS1-1 (30)
 GB3Burnopfield2354°54′N01°43′WS1-1 (23)
 GB4Penrith1054°39′N02°46′WS2-7 (10)
 GB5Leeds3053°50′N01°32′WS1-9 (12), S1-10 (18)
 GB6Kingsley3053°16′N02°40′WS1-1 (30)
 GB7Rugeley2652°44′N01°55′WS1-1 (25), S2-7 (1)
 GB8Cranham3051°48′N02°11′WS1-1 (30)
 GB9Loughton2751°39′N00°02′ES1-1 (27)
 GB10Bilsington3051°04′N00°54′ES2-7 (30)
 GB11Friehead2351°02′N02°52′WS1-1 (23)
 GB12Lower Beeding3051°01′N00°15′WS1-1 (7), S2-7 (9), S2-15 (7), S2-17 (6), S2-21 (1)
 GB13Lyndhurst1850°52′N01°34′WS1-1 (18)
image

Figure 1. Sampling locations of A. subfuscus with frequencies of the five mitochondrial 16S rDNA groups indicated.

Download figure to PowerPoint

DNA analysis

A portion of foot tissue was used for total genomic DNA extraction. Tissue was digested for 4 h up to overnight in 500 μL of CTAB buffer (2% CTAB, 1.4 m NaCl, 20 mm EDTA and 100 mm Tris/HCl) with 0.1 mg proteinase K at 65°C. The DNA was extracted following the protocol of Winnepenninckx et al. (1993).

For mtDNA analysis, all individuals were PCR–SSCP analysed for an approximately 170 bp segment of the mitochondrial 16S rDNA gene and all detected haplotypes were sequenced for a 388–407 bp stretch comprising the SSCP fragment. All procedures were as in Pinceel et al. (2004). The 16S rDNA sequences are available at Genbank under accession numbers AY860672860724.

For nDNA analysis, all individuals were PCR–SSCP analysed for an approximately 160 bp fragment of the ribosomal first internal transcribed spacer (ITS1). To amplify this fragment, Arion specific primers FITSar2 (5′-GCCTCAGCCTTCTCTGTCC-3′) and RITSar2 (5′-CCGCACAAAGTTATCATTTGC-3′) were developed. The PCR conditions were: (1) 5 min at 95°C, (2) 30 cycles of 45 s at 95°C, 45 s at 52°C and 90 s at 72°C and (3) 5 min at 72°C. The PCR reaction volumes (10 μL) contained 50 mm KCl, 10 mm Tris/HCl (pH 9.0 at 25°C), 1% Triton® X-100 (Promega Inc., Madison, WI, USA), 200 μm of each dNTP, 0.3 μm of each primer, 0.5 units of Taq DNA polymerase (Promega Inc.), 1.5 mm MgCl2, and 0.5 μL of template DNA. The SSCP analysis of this fragment was performed on polyacrylamide gels (T = 14%, C = 2%) with addition of 5% glycerol. Other SSCP procedures followed Pinceel et al. (2004). All detected ITS1 patterns were sequenced for a 561–574 bp fragment, comprising the SSCP fragment, using universal primers ITS1 (5′-TCCGTAGGTGAACCTGCGGAAGGAT-3′) (White et al., 1990) and 5.8c (5′-TGCGTTCAAGATATCGATGTTCAA-3′) modified from Hillis & Dixon (1991). The PCR conditions were: (1) 3 min at 97°C, (2) 35 cycles of 1 min at 95°C, 1 min at 55°C and 2 min at 72°C and (3) 5 min at 72°C. The PCR reaction volumes (25 μL) contained 50 mm KCl, 10 mm Tris/HCl (pH 9.0 at 25°C), 1% Triton® X-100 (Promega Inc.), 200 μm of each dNTP, 0.2 μm of each primer, 1.2 units of Taq DNA polymerase (Promega Inc.), 1.5 mm MgCl2 and 1.2 μL of template DNA. The PCR products were purified using the GFX PCR DNA purification kit (Amersham Biosciences, Little Chalfont, UK) and diluted in 25 μL of sterile water. Purified PCR products were sequenced using the ABI PRISM® BigDyeTM Terminator cycle sequencing kit (Applied Biosystems, Foster City, CA, USA) and run on a capillary Applied Biosystems 3730 DNA Analyser. Several individuals of each SSCP pattern were sequenced in order to check the reliability of the SSCP for detecting nucleotide substitutions. The ITS1 sequences were deposited at Genbank under accession numbers AY860725860735.

Allozyme analysis

Procedures for the preparation of digestive gland homogenates and vertical polyacrylamide gel electrophoresis (PAGE) followed Backeljau et al. (1987) and enzyme stainings were modified from Harris & Hopkinson (1976). After dissection, specimens were stored in 70% ethanol.

Five polymorphic loci (Pinceel et al., 2004) were scored in 827 specimens: α-amylase (α-Amy, EC 3.2.1.1), fumarate hydratase (Fumh, EC 4.2.1.2), glycerol-3-phosphate dehydrogenase (α-Gpd, EC 1.1.1.8), isocitrate dehydrogenase (NADP+) (Idh, E.C. 1.1.1.42) and phosphoglucomutase (Pgm, EC 5.4.2.2).

Allele frequencies were calculated in Popgene Version 1.32 (Yeh et al., 1997). The Fis values were estimated according to Weir & Cockerham (1984) and Hardy–Weinberg equilibrium (HWE) was tested for each locus with the exact probability test implemented by GENEPOP Version 3.3 (Raymond & Rousset, 1995). This latter was done in all populations with at least 15 individuals (29 locations, n = 711). The sequential Bonferroni procedure was applied to correct for multiple testing (Rice, 1989). As the SSCP and sequence analyses did not show evidence for intragenomic ITS1 variation other than heterozygosity, allele frequencies, Fis values and HWE tests were also calculated for the ITS1 locus.

DNA polymorphism and phylogenetic analysis

Sequence chromatograms were inspected visually in Chromas Version 1.45 (McCarthy, 1996). In heterozygous ITS1 genotypes, sequences established for the respective homozygotes were subtracted manually from the ambiguous sequences in order to determine both alleles. This could be done unambiguously as only a few alleles occurred for the ITS1 fragment. Sequences were aligned in ClustalX Version 1.81 (Thompson et al., 1997) with default gap opening/gap extension costs of 15/6.66. Alignments were checked for ‘unstable hence unreliable’ alignment blocks with SOAP Version 1.1 (Löytynoja & Milinkovitch, 2001), but no such regions were detected. Standard DNA polymorphism measures were calculated in DnaSP Version 4.00 (Rozas et al., 2003).

Phylogenetic relationships among 16S haplotypes and ITS1 alleles were inferred with A. fuscus as ingroup and A. hortensis as outgroup. Three different tree-reconstruction methods were used. A neighbour-joining tree (NJ) (Saitou & Nei, 1987) was constructed on the basis of uncorrected P-distances calculated with complete deletion of sites with alignment gaps in MEGA Version 2.1 (Kumar et al., 2001) and a maximum-parsimony (MP) tree with gaps treated as a fifth state was constructed in PAUP* Version 4.0b10 (Swofford, 1998). For both trees, relative branch support was evaluated with 1000 bootstrap replicates (Felsenstein, 1985). A Bayesian analysis was run for 1 000 000 generations, sampling the Markov chain every 10 generations with a burn-in length of 10 000 generations in Mr Bayes Version 3.0 (Huelsenbeck et al., 2001). Four independent chains, which converged rapidly from their random starting points, were run. The MODELTEST Version 3.06 (Posada & Crandall, 1998) was used to select the most appropriate likelihood settings for Bayesian phylogenetic analysis. The molecular clock hypothesis was tested in the complete dataset, as well as in each haplotype group separately, using the likelihood ratio-test implemented in PAUP* (Swofford, 1998). Mean percentages of sequence divergence within and among haplotype groups were calculated with MEGA Version 2.1 (Kumar et al., 2001) using the P-distance model.

Phylogeographic mtDNA analysis

In order to separate population structure from population history (Templeton, 1998), a nested clade analysis (NCA) was performed. A statistical parsimony network was constructed with the haplotypes of each group in TCS Version 1.18 (Clement et al., 2000). A hierarchical set of nested clades was constructed by applying the rules of Templeton et al. (1987) and Templeton & Sing (1993) on the statistical parsimony networks. The higher level nesting of separate mtDNA groups was based on the phylogenetic trees. The random association between geography and genetic variation was tested by comparing the clade distance (Dc) and nested clade distance (Dn). The frequency of haplotypes and clades, and the sample size per location were used to simulate and test the null hypothesis of random distribution of haplotypes and clades in GeoDis Version 2.2 (Posada et al., 2000). In all cases where the null hypothesis was rejected (H0 = random geographical association of clades), the possible biological cause for association between geography and genetic variation was inferred using the key provided by Templeton et al. (1995) (updated version at http://inbio.byu.edu/Faculty/kac/crandall_lab/geodis.htm). This inference key has subsequently been altered to minimize the chance of making false inferences (Templeton, 2004).

Population divergence times can easily be overestimated because coalescence times are often lengthened by genetic drift or polymorphisms in the ancestral population (Knowles & Maddison, 2002). However, under certain demographic circumstances, it appears reasonable to assess demographic parameters from single gene genealogies (Lascoux et al., 2003; Wakeley & Takahashi, 2003). When the geographic range of a population is extremely reduced, it is likely that genetic variation is lost. When such a population subsequently expands, it is expected that genetic drift and ancient polymorphisms are relatively unimportant in the demographic structuring of the expanding population. In order to estimate the timing of such population expansions and the 16S rDNA clock rate, we selected a number of clades (see results) for which range expansions were inferred in the NCA. Clades at low nesting levels and with a wide, contiguous range was selected because such clades probably lack ancient polymorphisms by which coalescence times could be overestimated. The coalescent null model (growth of an initial population under a model of sudden expansion; Rogers & Harpending, 1992) was tested using Tajima's (1989)D and Fu & Li (1993)Fs statistics. The frequency distributions of pair-wise sequence differences were analysed by mismatch analysis, modelled and fitted to the observed distribution under a sudden expansion model using a parametric bootstrap approach (1000 replicates) with the sum of squared deviations (SSD) (Rogers & Harpending, 1992). The time of population divergence (τ) and its confidence interval were estimated from the model distribution by a general nonlinear least-squares approach with 1000 parametric bootstraps. All analyses were performed in Arlequin Version 2.0 (Schneider et al., 2000). The estimates of τ were used in the relationship τ = 2ut, in which u is the mutation rate of the 16S rDNA fragment per year and t is the number of generations since the expansion (Rogers & Harpending, 1992). As the sudden expansions in the analysed clades were most likely triggered by one of the glacial terminations (Petit et al., 1999), we calculated the clock rates accounting for the estimated τ values for each of the three last glacial terminations. The Tajima (1989)D and Fu & Li (1993)Fs statistics were also applied as an indication for mutation/drift balance in the complete dataset.

Comparison of mitochondrial and nuclear GPs

The congruence between the nuclear data and the mtDNA data was evaluated with a model based clustering analysis performed in Structure Version 2 (Pritchard et al., 2000). The analysis assigns individual nuclear multi-locus genotypes (MLGs) of all individuals (n = 828) probabilistically to one of K (K is user-defined) GPs. Three independent chains of 1 000 000 iterations after a burn-in of 100 000 were run for all possible K values from 1 to 20. We did not use prior information on mtDNA group or geographic origin of individuals. The admixture model was used with individual admixture (α) constant in all GPs, and allele frequencies were kept independent among GPs. Average posterior ancestry probabilities (i.e. the probability with which an individual is assigned to each of K GPs) were calculated for all individuals and averaged per population and per mtDNA group. Additionally, the assignment probabilities per mtDNA group were summed separately for individuals from continental Europe and the British Isles. In this way, it was possible to infer (i) whether the mtDNA groups are reflected in the structure of the nuclear GP, (ii) whether the nuclear GPs of the mtDNA groups mix, and (iii) whether there is a directionality or geographic pattern in the mixing.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Appendices

Molecular polymorphism and phylogeny

A total of 53 mtDNA haplotypes were found. The aligned haplotype sequences (reference taxa excluded) had a length of 411 bp with 149 polymorphic positions of which 41 were gaps, 96 were parsimony informative and 12 were singletons. We found 11 different ITS1 alleles which had an aligned length of 577 bp with 28 polymorphic sites of which 17 were gaps, four were parsimony informative and seven were singletons. The ITS1-gaps represented five INDEL events of which three were parsimony informative. The reliability of the SSCP was high since sequencing detected only little additional variation that was not reflected in the SSCP patterns (none of 75 16S rDNA sequences and 7 of 90 ITS1 sequences). The identical SSCP patterns yielding different ITS1 sequences corresponded to alleles ITS1-03, ITS1-03a, ITS1-03b and ITS1-03c which all differed by a single point mutation. Therefore, these alleles were lumped into a single allele for population genetic analyses. The representativity of the SSCP stretch for variation in the DNA stretches outside the SSCP fragment was high, since only five of the 53 16S rDNA haplotypes and two of the 11 ITS1 alleles varied outside the SSCP stretch only. For population genetic analyses, the ITS1 alleles that differed outside the SSCP fragment only (ITS1-05 and 05a and ITS1-06 and 06a) were lumped into a single allele (ITS1-05 and ITS1-06, respectively).

The NJ, MP and Bayesian phylogenetic analyses of the 16S rDNA data yielded highly congruent results (Fig. 2). Five haplotype groups with strong branch support and deep divergence were distinguished (S1 to S5) and formed a monophyletic group with respect to the ingroup and outgroup taxa. The deepest split within A. subfuscus separated groups S1, S4 and S5 from groups S2 and S3. The molecular clock hypothesis for the complete dataset was rejected by the likelihood ratio-test, but within groups S2 and S3 a clock was not rejected. Mean sequence divergence ranged from 0 to 2.3% within groups, from 9 to 21% among the groups and from 18 to 22% between the groups and A. fuscus (Table 2). The NJ and Bayesian analyses of the ITS1 alignment yielded weakly resolved trees, as these methods do not take INDELs into account. The MP tree showed that A. subfuscus ITS1 sequences were monophyletic with respect to A. fuscus and A. hortensis, but the relationships among the A. subfuscus alleles remained unresolved (results not shown). The mean pairwise sequence divergence among A. subfuscus ITS1 alleles was 0.3% and differed on average 1.3% from the A. fuscus ITS1 alleles. Interestingly, the two most frequent ITS1 alleles (ITS1-02 and ITS1-03) were shared by the two most frequent mtDNA groups (S1 and S2), such that ITS1-03 was common in S1 (80%), but rare in S2 (3%), whereas ITS1-02 was common in S2 (97%), but less frequent in S1 (18%) (Fig. 2).

image

Figure 2. Phylogenetic tree with indication of the five major groups of 53 A. subfuscus 16S rDNA haplotypes, reference haplotypes of two A. fuscus and outgroup haplotype of an A. hortensis specimen. Branch lengths were calculated on the basis of P-distances. Relative branch support indices are given as Bayesian probability/maximum parsimony bootstrap/NJ bootstrap values. (a) Proportion of assignment of each mtDNA group to each of the five GPs as defined by the model-based clustering of the nuclear MLGs (Pritchard et al., 2000): GP1 inline image, GP2 inline image, GP3 inline image, GP4 inline image and GP5 inline image. (b) Frequencies of the ITS1 alleles in each mtDNA group: ITS1-02 inline image, ITS1-03 inline image, ITS1-04 inline image, ITS1-05 inline image, ITS1-06 inline image and ITS1-07 inline image.

Download figure to PowerPoint

Table 2.  Sequence divergences for A. fuscus (A. f.) and the five mtDNA groups in A. subfuscus. Mean within group 16S rDNA sequence divergences (diagonal) and mean between group divergences (± SE) calculated in MEGA (Kumar et al., 2001) on the basis of the P-distance model.
 A. fuscusS1S2S3S4S5
A. f.2.9 ± 0.8     
S120.6 ± 2.51.5 ± 0.4    
S220.5 ± 2.319.8 ± 1.90.6 ± 0.2   
S318.2 ± 2.218.2 ± 1.813.0 ± 1.60.7 ± 0.3  
S420.6 ± 2.413.6 ± 1.920.1 ± 2.518.3 ± 2.22.3 ± 0.5 
S521.9 ± 2.69.3 ± 1.420.9 ± 2.019.0 ± 1.914.0 ± 2.00.0 ± 0.0

Allele frequencies per population for the allozyme and ITS1 loci are given in Appendix 1. The Fis estimates for populations with n ≥ 15 ranged from −0.422 to 1.000. There were eight (of 59 tests) significant deviations from HWE (three after sequential Bonferroni correction), but there was no general pattern according to locus, geography or mtDNA group (Appendix 2).

Phylogeography

The distribution of the mtDNA haplotypes and haplotype groups is given in Table 1 and Fig. 1. Groups S1, S2 and S3 were widespread and the ranges of S1 and S2 overlapped in the British Isles, NW France and Belgium. In three locations in the British Isles (IR1, GB7 and GB12) S1 and S2 even co-occurred. Group S4 was only found in the type locality of A. subfuscus (Montagne Noire, SE France) and in adjacent areas (FR15-19), and group S5 was only found in one locality in the Central-Massif (FR12). Although we sampled thoroughly further to the east (Eastern France, Central Germany and NW Italy) and to the south (Southern France and Iberia), none of the five mtDNA groups was found in any of these regions.

The statistical parsimony networks of mtDNA haplotypes yielded six hierarchical sets of clades (Fig. 3). At the higher nesting levels, S1 (clade 4-1), S4 (clade 4-4) and S5 (clade 4-5) formed clade 5-1 and S2 (clade 4-2) and S3 (clade 4-3) formed clade 5-2. The allopatric distribution of the mtDNA groups was explained by the NCA inference of allopatric fragmentation (AF). At the lower nesting levels, several range expansions (RE) were inferred. For clade 2-1 (the only widespread clade in S1) range expansion with some long distance colonization and subsequent fragmentation was inferred (Fig. 4a). For the level-2 clades of S2 and S3, contiguous range expansions (CRE) were suggested, which covered parallel areas with a wide north-south range (Fig. 4b). The overlapping ranges of S1 and S2 in the British Isles, NW France and Belgium are the result of the overlap of clades 2-1, 2-2 and 2-3 (S1) with clade 2-9 (S2). There were more clade 2-1 haplotypes (S1) in the British Isles (n = 4), than in continental Europe (n = 1). In addition, clade 2-1 was widespread in the British Isles (12 locations), whereas it was geographically restricted in continental Europe (three locations in Belgium). Clades 2-2 and 2-3 (S1) were only found in continental Europe. There were more clade 2-9 haplotypes (S2) in continental Europe (n = 11), than in the British Isles (n = 4).

image

Figure 3. Statistical parsimony networks of the 16S rDNA haplotypes constructed in TCS (Clement et al., 2000) with indication of the hierarchical nesting for NCA. Haplotypes are represented relatively to the number of populations in which they occurred. Small open circles are missing haplotypes and each connection represents one mutational step. Dashed connections were excluded from the network according to the rules presented in Posada & Crandall (2001). The level-three nesting is not indicated as it corresponds with the level-four nesting, except in clade 4-1.

Download figure to PowerPoint

image

Figure 4. Distribution of nested clades and significant biological interpretations as inferred by the updated key of Templeton et al. (1995): AF, allopatric fragmentation; CRE, contiguous range expansion; RE w LDC/suF, range expansion with long distance colonization and subsequent fragmentation; i.o., inconclusive outcome. (a) Distribution of mtDNA groups S1, S4 and S5 (nested clade 5-1). (b) Distribution of mtDNA groups S2 and S3 (nested clade 5-2).

Download figure to PowerPoint

The mismatch distributions of the single clades fitted a model of sudden expansion, but only one Tajima (1989)D value was significantly negative (clade 2-9), which is required to accept the coalescent null model. The test of Fu & Li (1993) showed the same results. Assuming that the expansion occurred after one of the glacial terminations, the inferred τ value for clade 2-9 (τ = 0.61, Table 3) yielded nucleotide divergence rates of 5.4, 0.6 or 0.3% per MY for expansion 14, 130 and 240 kyBP, respectively. In clades for which the coalescent null model was rejected, τ values were not used because they are probably overestimated due to the presence of ancestral polymorphisms. Such overestimates may be substantial in recent divergences (Wakeley, 2000; Knowles, 2004). The analysis of the complete dataset yielded significantly positive Tajima (1989)D and Fu & Li's Fs (1993) values, indicating that A. subfuscus is not in mutation/drift balance (Table 3).

Table 3.  Results of the mismatch analysis (Rogers & Harpending, 1992): clade analysed, number of individuals in the analysis (n), estimated τ with confidence interval (C.I.), probability that the observed data do not fit a sudden expansion model [P(SSDs ≥ SSDo)], Tajima (1989)D value (D), probability that D is different from zero [P(Dr < Do)], probability test of Fu & Li (1993) [P(Fss≤ Fso)] and sequence divergence rate per million years (μ) for expansion after the three last glacial terminations.
Cladenτ (C.I.)P(SSDs ≥ SSDo)DP(Dr < Do)P(Fss≤ Fso)μ (14 kyBP) (%)μ (130 kyBP) (%)μ (240 kyBP) (%)
  1. P-values indicated with an asterisk are significant.

  2. NA, not applicable.

2–13413.00 (0.42–4.62)0.47−0.110.490.4726.82.91.6
2–61383.00 (0.39–4.31)0.17−1.120.120.1526.82.91.6
2–7402.81 (0.30–6.31)0.20−0.340.420.5025.12.71.5
2–91760.61 (0.00–1.94)0.95−1.680.02*0.00*5.40.60.3
All84315.19 (3.38–34.93)0.014.330.00*1.00NANANA

Relationship between mtDNA groups and the nuclear GP

The model-based clustering performed by Structure (Pritchard et al., 2000) yielded many very similar log-likelihood values for different K ≥ 5. In such a case it is advised to use the smallest K after the log-likelihood has reached its upper platform (Rosenberg et al., 2002; Heuertz et al., 2004). Therefore, we looked into detail at the results for K = 5.

The mean individual assignments to each of the five GPs per population are visualized in Fig. 5. The GPs were partially good predictors of the mtDNA groups, as individuals assigned to GP1 and GP2 belonged almost exclusively to S1, whereas individuals assigned to GP3, GP4 and GP5 belonged almost exclusively to S2, S3, S4 or S5 (Table 4 and Fig. 2). Yet, the probability of S1 individuals to be assigned to GP1 was larger in continental Europe (72%), than in the British Isles (43%), and smaller to GP2 in continental Europe (16%), than in the British Isles (47%). Continental European S2 individuals were nearly all assigned to GP3, GP4 and GP5 (29, 29 and 38%), whereas S2 individuals from the British Isles were more often assigned to GP3 (73%) and GP2 (15%) (Table 4). The overlapping ranges of S1 and S2 in the British Isles and NW France and Belgium seem to result in a mixing of the nuclear GPs of the two mtDNA groups.

image

Figure 5. Mean proportion of ancestry of each population in five GPs defined by the model-based clustering analysis (Pritchard et al., 2000). Data were not available for DE2 and FR18 because allozymes were not scored in individuals of these populations.

Download figure to PowerPoint

Table 4.  Posterior assignment probabilities of individuals of the five mtDNA groups (S1-5) to five nuclear GPs (GP1-5) as inferred by Structure (Pritchard et al., 2000); in all populations (ALL), in continental European populations (CEU, italics) and in populations from the British Isles (BRI, bold).
 GP1GP2GP3GP4GP5
ALLCEUBRIALLCEUBRIALLCEUBRIALLCEUBRIALLCEUBRI
S10.510.720.430.390.160.470.040.020.050.030.040.020.040.070.03
S20.020.020.010.050.020.150.370.290.730.240.290.040.320.380.07
S30.010.010.010.010.020.020.460.460.500.50
S40.010.010.010.010.010.010.020.020.950.95
S50.010.010.010.010.020.020.740.740.220.22

Because the majority of S1 individuals were almost completely assigned to GP1 + GP2 and the majority of S2 individuals almost completely to GP3 + GP4 + GP5, we treated individuals which were assigned with a probability of 5% or more to both GP1 + GP2 and to GP3 + GP4 + GP5 as individuals with a mixed nuclear genome. This not necessarily implies recent hybridization events, because the mixing can also be the result of past hybridization or back-crosses. Individuals with a mixed nuclear genome were found in 26 locations: four in Belgium (BE4, BE5, BE7 and BE13), eight in Northern France (FR2, FR4, FR5, FR7–FR10 and FR13) and all 14 in the British Isles. In total, 212 individuals had a mixed nuclear genome. In continental Europe, these were 17% of all S1 and S2 individuals, representing 34% of all S1 individuals and 12% of all S2 individuals. In the British Isles, 41% of all S1 and S2 individuals had a mixed nuclear genome, representing 44% of all S1 individuals and 28% of all S2 individuals.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Appendices

Evolutionary rates of mtDNA and nDNA in pulmonates

We found two magnitudes of sequence divergence among the 16S rDNA haplotypes: i.e. a strong divergence among the five major groups (S1 to S5: 9–21%) and a low divergence within the groups (≤2.3%). The test statistics of Tajima (1989) and Fu & Li (1993) confirmed that the nodes among the mtDNA groups are deeper than expected in populations in mutation/drift balance. This observation was probably caused by expansion/contraction events in different time periods. First, the mtDNA groups seem not to share a common phylogeographic history and have diverged allopatrically. Second, the small within-group divergences and the wide, relatively northern distribution of groups S1, S2 and S3 suggest a recent genetic impoverishment of the groups, followed by RE. The NCA confirmed the AF of the ranges of the groups and the range expansion of the most common, widespread lower-level clades (Fig. 4).

The divergence rates of the 16S rDNA (based on clade 2-9) range from 5.4 to 0.3% divergence per MY, depending on whether the post-glacial expansion started 14, 130 or 240 kyBP. Given the wide, relatively northern distribution of clade 2-9 (Fig. 4b), it is most plausible that the expansion occurred after the last glacial maximum, so that the divergence rate would be around 5.4% per MY. This is faster than the general divergence rate of invertebrate 16S mtDNA (0.5–2% per MY) (Cunningham et al., 1992; Romano & Palumbi, 1997), but not unusual for pulmonates in which divergence rates of up to 10% have been reported (Chiba, 1999; Thacker & Hadfield, 2000; Watanabe & Chiba, 2001).

The small divergences among the ITS1 alleles (mean = 0.3%) and the weak signal of reciprocal monophyly between mtDNA and nDNA (Fig. 2) are in contrast with the deep divergences among the mtDNA groups. Although the evolutionary rate of ITS genes is poorly understood and influenced by many factors such as functional constraints, internal repetition and intragenomic variation (Schlötterer et al., 1994; Harris & Crandall, 2000; Dejong et al., 2001), clock rates of invertebrate ITS are estimated from 0.4 to 1.2% sequence divergence per MY (Schlötterer et al., 1994; Bargues et al., 2000; Wares, 2001). With this divergence rate, the split between the A. subfuscus ITS1 sequences would not be older than approximately 0.25–1.00 MY, whereas the age of the mtDNA groups probably ranges from 2.00–4.00 MY. The nDNA and mtDNA divergences among A. subfuscus and A. fuscus are more congruent: 1.00–4.00 MY for ITS1 and approximately 4.00 MY for 16S rDNA. According to the ‘three-times’ rule (Palumbi et al., 2001), the majority of neutral nuclear markers in a taxon will be monophyletic when the mitochondrial divergence to the most recent common ancestor of that taxon is approximately three times higher than the divergence within the taxon. Although the applicability of this rule has been severely criticized because of the stochasticity of the mtDNA coalescent process (Hudson & Turelli, 2003), it is useful to compare divergence and reciprocal monophyly of mitochondrial and nuclear markers. The three times rule is for instance applicable in the pulmonate snails Trochoidea geyeri and for sister taxa of Candidula (Pfenninger et al., 2003). In A. subfuscus however, it is not, since the branch lengths to each haplotype group were 3–18 times longer than the intra-group branches, yet, this did not result in reciprocal monophyly. Hence, the mtDNA divergences reported here are probably of the highest known in the absence of reciprocal monophyly at a supposedly fast evolving nuclear locus like ITS1 (Page & Holmes, 1998). The most likely explanation for this discrepancy is a lack of reproductive isolation between the mtDNA groups. The ITS1 and allozyme data indeed suggest that at least the co-occurring A. subfuscus mtDNA groups S1 and S2 hybridise (Fig. 2 and Table 4).

Allopatric divergence with secondary contact and/or ancestral polymorphisms?

Although there is evidence for an accelerated mtDNA evolution, divergences among A. subfuscus mtDNA groups (9–21%) are still too high to have evolved recently and hence the groups must be older than the Pleistocene. It is not obvious to unravel the history of such deep divergences, but (i) the current allopatric distribution of the mtDNA groups (except for the overlap between S1 and S2) and (ii) the AF inferred by the NCA, both suggest a long-term allopatric divergence. The sympatric occurrence of groups S1 and S2 would then be best explained by a secondary contact after recolonization from different refugia. The level-two clades of S1 (2-1, 2-2 and 2-3) were probably fragmented due to range contraction/expansion events (Fig. 4a), which restricted clade 2-1 to the British Isles and a few locations in Belgium, and clades 2-2 and 2-3 to continental Europe. Of the five level-two clades of S2, only one was found in the British Isles, which had a wide distribution in continental Europe as well (Fig. 4b). It was probably the RE of clade 2-1 (S1) and clade 2-9 (S2) that resulted in the current overlapping range of S1 and S2. If these clades would had expanded from a single ancestral population, one would expect that their resulting expanded ranges would overlap more (compare Fig. 4a with Fig. 4b).

Although the explanation of secondary contact between allopatrically diverged mtDNA groups seems the most parsimonious hypothesis, it cannot be excluded that the co-occurrence of S1 and S2 is an ancestral situation. On the basis of the nuclear data, however, we should be able to discriminate between these two scenarios. Besides the range overlap, individuals of group S1 and S2 also co-occurred in three locations in the British Isles (i.e. IR1, GB7 and GB12). The Structure analysis of the nuclear data suggested a strong correspondence between the mtDNA group and the nuclear GP to which individuals belong. Such a clear correspondence between nuclear and mitochondrial genes would not be expected if the co-occurrence S1 and S2 would be an ancestral situation. Hence, S1 and S2 individuals in which the nuclear GP does not correspond with the mtDNA group are probably hybrids resulting from interbreeding events after secondary contact. This was confirmed by the fact that such individuals are most frequent in regions where the ranges of S1 and S2 overlap. Hybridization between the two mtDNA groups seem to occur in both directions and the majority of populations and loci did not deviate from HWE (Appendix 2), suggesting that the deep, allopatric mtDNA divergence was not accompanied by reproductive isolation. However, if hybridization would be unconstrained for many generations, one would expect a much stronger homogenization of the nuclear GP across the mtDNA groups. As this is not the case, the sympatric distribution of the groups must either be recent and/or particular mechanisms must prevent complete panmixis. For instance, the ecological characteristics of the A. subfuscus groups, which remain to be studied, may result in differential adaptation and the formation of (narrow) hybrid zones (Barton, 1983; Barton & Hewitt, 1989; Vines et al., 2003). Moreover, the patchy structure of pulmonate populations and their low active dispersal capacities may also help to maintain the molecular differences among the mtDNA groups (Thomaz et al., 1996; Arnaud et al., 2001; Arnaud & Laval, 2004). That most populations comprise only one mtDNA group (except IR1, GB7 and GB12), while 26 populations have mixed nuclear GPs, is probably due to the stronger effect of genetic drift on the mtDNA than on the nuclear markers because the effective population size for mtDNA is only one fourth of that of nuclear genes (Avise, 2000).

In Europe, secondary contact of post-glacially recolonizing clades and the formation of hybrid zones have been observed in several taxa (reviewed in Hewitt, 1999) and may play an important role in speciation processes (Barton & Hewitt, 1989; Barton, 2001; Hewitt, 2001; Rundle et al., 2001; Seehausen, 2004). Although, strictly speaking, a hybrid zone is a narrow area of contact, the overlap between S1 and S2 is large, which could mean that several hybrid zones are involved and that our sampling design did not allow discriminating among them. It could also be that the secondary contact follows a patchy pattern (e.g. if the distribution is influenced by human activities). In Britain, in the pulmonate snail Cepaea nemoralis, a narrow contact zone between an eastern and a western mtDNA group was observed and explained by historical factors such as post-glacial recolonization and secondary contact (Davison, 2000; Davison & Clarke, 2000). Unfortunately, further comparison of the large-scale phylogeographic patterns are currently impossible because only few continental European Cepaea populations included in these studies.

Taxonomic implications

In an earlier phylogenetic study of A. subfuscus in NW Europe (Pinceel et al., 2004), four strongly differentiated mtDNA groups (11–21%) were observed. In the present study, we not only discovered a fifth group, but also showed that each group has a separate phylogeographic history of deep allopatric divergence. Such AF and divergence of mtDNA groups are suggested to play an important role in speciation processes (Wiens, 2004). In strict allopatry (e.g. S3, S4 and S5), the biological species concept cannot be implemented, and where it can be applied in our study (S1 and S2), it would not recognize the mtDNA groups as species because they hybridise. Alternatively, the application of the phylogenetic species concept is inappropriate in pulmonates that show deeply diverged mtDNA groups whose monophyly is not reciprocated by nuclear genes. As such, the A. subfuscus groups are neither phylogenetic species, nor evolutionarily significant units (Moritz, 1995; Mayden, 1997). Therefore, we argue that the evolutionary species concept (Wiley, 1978) may be the best alternative for the taxonomic interpretations of the A. subfuscus mtDNA groups. The hybridization of individuals of groups S1 and S2 showed that the development of an intrinsic barrier to gene-flow (e.g. reproductive isolation) as a by-product of the allopatric divergence between the groups (Rice & Hostert, 1993) is not complete. Although mixed nuclear genomes seem to occur frequently, the nuclear GPs of S1 and S2 remain relatively differentiated (Table 4) and the partially allopatric ranges seem to be maintained (Fig. 1). Possibly, incomplete reproductive isolation between the mtDNA groups is also the cause of the absence of reciprocal monophyly for mtDNA and nDNA (Moritz, 1995; Palumbi et al., 2001). Because (i) each mtDNA group has a separate phylogeographic history and (ii) certain mechanisms seem to prevent the complete mixing of the nuclear GPs of the hybridizing mtDNA groups, we argue that each group is a separate and functional unit of evolution (Wiens, 2004). The groups may therefore be regarded as different evolutionary species, and at least provide a framework to test morphological and ecological differentiation. Depending on the outcome, our evolutionary species may then be translated into a valuable taxonomic and nomenclatorial framework.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Appendices

In this study we reconstructed the phylogeographic history of strongly diverged mtDNA haplotypes in the terrestrial slug A. subfuscus. By means of a range wide analysis of combined mtDNA, nDNA and allozyme data, we show that the diverged haplotypes belong to five mtDNA groups, each with an independent phylogeographic history. The deep divergence among the mtDNA groups (up to 21% sequence divergence) did not result in reciprocal monophyletic groups for the ITS1 gene or in specific alleles at any of the five polymorphic allozyme loci. Our data suggest that this incongruence is due to the combined effect of (i) an accelerated mtDNA evolution, which is not unusual in pulmonates (Chiba, 1999), and (ii) the absence of a complete reproductive barrier to gene flow among at least two of the mtDNA groups. Where other studies only suggested the importance of different post-glacial recolonization routes in explaining extreme mtDNA diversities within pulmonate populations (e.g. Thomaz et al., 1996), we suggest that the current range overlap of two of the allopatrically diverged mtDNA groups (S1 and S2) is the result of a secondary contact. Taxonomically, we interpret each of the five mtDNA groups as functional evolutionary units, which may be treated as valid species under the evolutionary species concept (Wiley, 1978; Wiens, 2004). Further research of the morphological and anatomical differentiation among the mtDNA groups is needed to further stabilize the taxonomy of this species complex. In addition, there is a need to study the ecology and the fine-scale genetic structure of populations in which groups S1 and S2 hybridize, because our data suggest that hybridization is either recent or limited.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Appendices

We are grateful to K. De Gelas, S. Geenen, M. Pfenninger, J. Van Houdt, N. Van Houtte and many others for their help with the collection of slugs, the laboratory analyses and the data analyses. We are indebted to two anonymous referees who provided improvements to this paper. JP thanks S. Smets and the family Pinceel for their help and support in course of this research. This work was funded by a BOF-NOI 44.84 project and RAFO-fund JORKKP02 from which KJ is holder and FWO project G.0003.02 and OSTC projects MO/36/003 and MO/36/008 to TB.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Appendices

Appendices

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  10. Appendices

Appendix 1

Allele frequencies at ITS1 and five polymorphic allozyme loci in populations of A. subfuscus. Alleles ITS1-03/03a/03b/03c were pooled into one allele ITS1-03 because the resolution of the SSCP technique did not allow to separate the alleles unambiguously. (a) ITS1; (b) α-Amy; (c) Fumh; (d) α-Gpd; (e) Idh; (f) Pgm.

inline image

Appendix 2

Table 5.  Results of exact tests for HWE at five allozyme loci and the nDNA ITS1 locus in A. subfuscus populations with at least 15 individuals.
IDn  α-AmyPgmα-GpdFumhIdhITS1
HWEFisHWEFisHWEFisHWEFisHWEFisHWEFis
  1. Population code (ID), number of individuals (n), probability for deviation from HWE and Fis estimates are given. NS, not significant and *:significant after sequential Bonferroni correction.

BE127NS0.085
BE2220.000*1.000NS−0.050
BE3280.0181.000
BE4330.000*0.7000.0170.468NS−0.103NS0.115
BE1023NS0.254
BE13210.0021.000
FR218NS−0.063NS0.128
FR315
FR422NS0.468
FR630NS−0.261NS0.477
FR8290.0060.591NS−0.143
FR1030NS−0.074NS−0.018NS−0.063
FR1120NS0.216NS0.000
FR13200.0010.789
FR1430NS−0.191
NL115
GB131NS−0.064NS−0.034NS−0.082NS0.277
GB230NS−0.094NS−0.036NS0.045
GB323NS−0.048NS0.347NS−0.023
GB530NS−0.036NS−0.115NS0.392
GB630NS0.659NS−0.018NS−0.137
GB7260.000*0.875NS0.096NS0.296
GB830NS−0.018NS−0.086NS0.133NS0.389
GB927NS−0.026NS−0.020NS0.185NS0.204
GB1030NS0.2750.044−0.422
GB1123NS−0.023NS−0.106NS−0.048
GB1230NS0.477NS0.477NS0.151
GB1318NS0.197NS−0.015NS0.112