Microsatellite variability in European anchovy (Engraulis encrasicolus) calls for further investigation of its genetic structure and biogeography

Abstract

Zarraonaindia, I., Pardo, M. A., Iriondo, M., Manzano, C., and Estonba, A. 2009. Microsatellite variability in European anchovy (Engraulis encrasicolus) calls for further investigation of its genetic structure and biogeography. – ICES Journal of Marine Science, 66: 2176–2182.

Attention to genetic and ecological perspectives can enhance strategies for using fishery resources sustainably. A potentially important application is the use of molecular markers to assess the genetic stock structure of a harvested species. In this study, seven microsatellite markers were analysed in anchovy samples from the Bay of Biscay, the Gulf of Cádiz, and the Gulf of Lions to assess the genetic structure of anchovy (Engraulis encrasicolus) populations in the Bay of Biscay and to infer the biogeographic origin of these populations. All samples showed a deficit of heterozygotes that could be explained by non-random mating, Wahlund's effect, and especially by the presence of null alleles. Global F_ST and R_ST values, uncorrected and corrected for null alleles, were significant. There was significant genetic heterogeneity between two populations in the Bay of Biscay, suggesting that anchovy there may not be panmictic. Moreover, the results reinforce the hypothesis of a recent common ancestor shared by Bay of Biscay and western Mediterranean anchovy. These results, together with those of earlier studies, suggest merit in further investigating spatio-temporal genetic variation among anchovy populations in the Northeastern Atlantic.

Introduction

European anchovy (Engraulis encrasicolus) in the Bay of Biscay support a commercially valuable fishery (Uriarte et al., 1996). The depletion of the fishery since 2002 has made explicit the need to manage the population on a well-founded scientific basis, and various research surveys are being conducted in an effort to monitor and scientifically assess this fishery. These surveys provide information on population biomass, reproduction, and ecology of the stock in the region, but one important requirement for managing a fishery is to understand how populations are partitioned: as a single unit or as several genetically distinctive groups (Beaumont and Hoare, 2003).

Studies conducted on other species of anchovy such as E. capensis (Grant, 1985), E. mordax (Hedgecock et al., 1989; Lecomte et al., 2004), and E. japonicus (Yu et al., 2005; Liu et al., 2006) have detected little genetic structuring among populations. Nevertheless, Yu et al. (2002) and Funamoto and Aoki (2002) reported significant genetic structuring among populations of E. japonicus. Several authors have found geographical structuring in the Mediterranean populations of European anchovy using allozymes, mitochondrial DNA (mtDNA) and morphological data (Spanakis et al., 1989; Bembo et al., 1996; Magoulas et al., 1996, 2006; Borsa, 2002; Borsa et al., 2004). Fewer studies have been conducted on Atlantic populations of European anchovy. Studies of populations in the Bay of Biscay reveal divergent results. Junquera and Pérez-Gándaras (1993) found some reproductive isolation between two groups in the region: (i) anchovy from Subdivision VIIIb and the eastern part of VIIIc, and (ii) those from the western part of VIIIc and from Subdivision IX (Figure 1). Prouzet and Metuzals-Sebedio (1994) also described two or three morphological groups in the region, but could not differentiate among them by analysing 17 polymorphic enzymes. Studies of the biology and the population dynamics of Bay of Biscay anchovy (Motos et al., 1996; Uriarte et al., 1996) did not support the notion of complete isolated groups. However, in a recent study using 27 allozyme markers, Sanz et al. (2008) showed that anchovy collected at two spawning areas in the eastern Bay of Biscay were genetically distinguishable from each other. Despite these indications of heterogeneity among populations, Bay of Biscay anchovy populations are currently managed as a single unit (ICES, 1996).

Figure 1.

The sampling locations.

Previous results of genetic surveys indicate that anchovy populations in the Bay of Biscay may be more closely related to western Mediterranean populations than to other Northeast Atlantic populations (Magoulas et al., 1996, 2006). A reconstruction of the biogeographic origins of anchovy populations in the Bay of Biscay to explain this unexpected distribution needs to consider the likely response of anchovy populations to extreme climate shifts during the Pleistocene ice-age oscillations. First, European anchovy inhabit an area with a complex coastline, unlike the populations of southern Africa and California. Long north–south coastlines provide an opportunity for anchovy populations to remain intact in response to climate change (Grant and Bowen, 2006). A more complex coastline, however, such as that in the NE Atlantic and western Mediterranean, may lead to isolation in refugia and to a more complex genetic structure among contemporary populations as climates warm. Second, the last glacial maximum (LGM) was ∼18 000 years ago, so marine populations in the NE Atlantic and Bay of Biscay were driven to extinction or were displaced to warmer waters of West Africa (Grant, 2005). Contemporary populations of anchovy could only have arisen when the NE Atlantic reached its present climatic regime 10 000 years ago.

Molecular markers can be used both to estimate contemporary levels of gene flow between populations and to reconstruct biogeographic scenarios to explain present-day distributions of intraspecific lineages. Ideally, different classes of markers should be used to resolve different facets of population structure. Although the use of mtDNA may provide insights into relationships among evolutionary lineages through maternal inheritance, these inferences are based on a single locus, which represents a single realization of a genealogy in a population. Evidence from additional classes of molecular markers, including nuclear microsatellite DNA or introns, may be needed to provide the resolution of shallow population differences expected in a pelagic species with high levels of gene flow (Waples, 1998).

Here, we present the results of a preliminary genetic study examining the genetic structure and origin of anchovy populations in the Bay of Biscay. The distribution of genetic diversity within and among populations was estimated with seven microsatellite loci in four samples around the Iberian Peninsula, including the Bay of Biscay, (BI8B, BI8C), the Gulf of Cádiz, and the Gulf of Lions (Mediterranean). Our goal is to provide genetic information for anchovy populations to complement conventional methods of assessing anchovy stocks. A genetic-ecological perspective would improve the development of management strategies to implement sustainable harvesting of anchovy in the Bay of Biscay.

Material and methods

Two samples (BI8B, n = 95; BI8C, n = 96) from ICES Divisions VIIIb and VIIIc in the Bay of Biscay were collected in 2005 (Figure 1) by commercial fishers. BI8C was caught near San Sebastian and BI8B near the mouth of the Garona River. Additional samples were collected in 2005 from the Gulfs of Cádiz (n = 96) and Lions (n = 96) to infer relationships with other populations. Samples were stored at −80°C.

Laboratory analyses

Genomic DNA was extracted from 200 mg of muscle tissue with ABI PRISM 6100. The amount of DNA from each sample was subsequently quantified in a GENQUANT PRO spectrophotometer. Seven microsatellite loci were amplified in three independent polymerase chain reactions (PCR): 1-plex amplifications included EJ2, EJ27.1, EJ35, and EJ9 microsatellites (Chiu et al., 2002; T_a = 58°C); in the 2-plex amplifications, EE2 and EE10 microsatellites (Landi et al., 2005) were co-amplified (T_a = 56°C), and EJ41.1 (Chiu et al., 2002) was amplified independently (T_a = 55°C). PCR reactions were performed in a volume of 10 µl containing 25–100 ng DNA, 1.5 mM MgCl₂ (Promega), 0.3–0.5 µM of each primer, 1× Buffer (Promega), 0.2–0.25 µM dNTP, and 1–1.5 U of GOTaq flexi DNA polymerase (Promega). PCR amplifications were carried out with an initial denaturation at 95°C for 3 min, followed by 35 cycles at 95°C for 30 s, an optimal annealing temperature (specified above for each PCR reaction) for 1 min, and 72°C for 1 min in a GeneAmp PCR system 9700 thermocycler. A final extension was carried out at 72°C for 3 min. Two independent automatic runs were performed in an ABI-PRISM 3100 AVANT unit (Applied Biosystems): first run for EJ2, EJ27.1, EJ35, and EJ9 microsatellites, and a second run for EE2, EE10, and EJ41.1. LIZ 500 (Applied Biosystems) was used as an internal standard. Raw data were processed with Genescan 3.7 and Genotyper 2.5 software (Applied Biosystems).

Statistical analysis

For each population, expected heterozygosity (H_e) and observed heterozygosity (H_o), mean number of alleles (MNA), and allelic richness (AR) were calculated using FSTAT v2.9.3 (Goudet, 2001). The Fisher exact test was used to evaluate each locus and population for departures from Hardy–Weinberg expectation (HWE), as implemented in GENEPOP v4.0 (Rousset, 2007). F_IS (Weir and Cockerham, 1984) was calculated using FSTAT v1.2 (Goudet, 1995) to evaluate deficit or excess of heterozygotes for each locus and sample. Standard deviations and significances were obtained by jackknifing and permuting individuals, respectively. When appropriate, the Bonferroni correction was applied (Weir, 1996). The Beaumont–Nichols test was performed to detect the influence of natural selection on loci using Fdist v2 software (Beaumont and Nichols, 1996). The presence and frequency of null alleles was tested using the expectation maximization (EM) algorithm of Dempster et al. (1977) implemented in FreeNA (Chapuis and Estoup, 2007). The bottleneck effect was tested with the program BOTTLENECK v1.2.02 (Cornuet and Luikart, 1996), using the two-phase model of mutation (TPM), because most microsatellite loci fit the TPM better than the infinite allele model or the stepwise mutation model (Di Rienzo et al., 1994). The Wilcoxon sign-rank test was used to determine whether a population exhibits a significant number of loci with gene diversity excess.

Genetic differences between samples were tested by two measures based on allelic identity (F_ST) and on allelic size (R_ST) using FSTAT v1.2 (Goudet, 1995) and SPAGeDi v1.2 (Hardy and Vekemans, 2002), respectively, and significances were obtained by permuting individuals. The results of Gaggiotti et al. (1999) indicate that overall F_ST estimates are more reliable than R_ST when fewer than 20 microsatellites are used. To elucidate the role of mutation in the genetic data, the F_ST distribution (pR_ST) with 95% confidence intervals (CIs; Hardy et al., 2003) was constructed using SPAGeDi v1.2 (Hardy and Vekemans, 2002). As null alleles may affect the estimates of genetic differentiation, F_ST values were recalculated with FreeNA software, using the ENA method as described in Chapuis and Estoup (2007). A hierarchical analysis of molecular variance (AMOVA; Excoffier et al., 1992) was used to partition the genetic variance between subsamples within groups (F_SC) and among groups (F_CT) using ARLEQUIN v3.0 (Excoffier et al., 2005) with the objective of maximizing the variance among groups (F_CT) and minimizing the within-group variance (F_SC). Finally, G′_ST (Hedrick, 2005) allows comparison of the genetic differentiation between loci with different levels of genetic variation and different mutation rates, such as allozymes and microsatellite loci.

Results

Successful PCR amplifications varied among the seven microsatellite markers and ranged from 64 to 89 fish for BI8B (n = 95), from 82 to 90 for BI8C (n = 96), from 77 to 94 for CÁDIZ (n = 96), and from 81 to 93 for LIONS (n = 96). Results from this study (Table 1) and others using the same microsatellites on E. encrasicolus (Landi et al., 2005) and E. japonicus (Chiu et al., 2002; Yu et al., 2002) revealed high levels of heterozygosity and a large mean number of alleles. The number of distinct alleles at each locus varied considerably. EJ35 was the least variable locus (25 distinct alleles, MNA = 16.5 ± 2.6) and EJ9 the most variable (54 distinct alleles, MNA = 42.3 ± 3.2). AR values ranged from 16.2 to 41.5. The dinucleotide microsatellites with the largest number of repeats, EJ2 (CT)₄₃, EJ27.1 (GA)₃₆, and EJ9 (TC)₃₉, showed the largest number of alleles (>43). All the markers showed large values of H_e, ranging from 0.803 to 0.970. Significant departures from HWE were observed for five of the seven loci (Table 1). The Beaumont–Nichols test did not detect the effect of natural selection at any locus. Mean null allele frequencies >0.10 appeared at EJ9, EE2, and EJ41.1, whereas the remaining loci showed frequencies <0.05 (Table 1).

Table 1.

Descriptive statistics for seven microsatellite loci over the four European anchovy samples analysed: BI8B, BI8C, CÁDIZ, and LIONS.

Marker	n	k	MNA ± s.e.	AR	Ho ± s.e.	He ± s.e.	HWE	FIS ± s.e.	Null alleles ± s.e.
EJ2	360	43	30.0 ± 03.4	27.11	0.914 ± 0.022	0.943 ± 0.007	n.s.	0.031 ± 0.015*	0.012 ± 0.011
EJ27.1	345	49	36.5 ± 02.6	35.08	0.872 ± 0.054	0.952 ± 0.008	Significant	0.080 ± 0.026*	0.037 ± 0.025
EJ35	335	25	16.5 ± 02.4	16.18	0.824 ± 0.017	0.874 ± 0.033	n.s.	0.059 ± 0.011*	0.027 ± 0.011
EJ9	349	54	42.3 ± 03.2	41.47	0.683 ± 0.126	0.970 ± 0.002	Significant	0.295 ± 0.061*	0.144 ± 0.063
EE10	338	34	24.0 ± 03.2	24.28	0.727 ± 0.058	0.814 ± 0.044	Significant	0.098 ± 0.057*	0.035 ± 0.023
EE2	327	63	37.0 ± 11.7	33.72	0.618 ± 0.094	0.850 ± 0.065	Significant	0.268 ± 0.057*	0.118 ± 0.061
EJ41.1	326	41	21.8 ± 06.9	20.79	0.569 ± 0.085	0.803 ± 0.046	Significant	0.288 ± 0.038*	0.121 ± 0.043

n, number of individuals typed; k, allele number; MNA, mean number of alleles; AR, allelic richness; H_o, observed heterozygosity; H_e, expected heterozygosity; HWE, significance for the Hardy–Weinberg equilibrium; F_IS, null allele frequencies; s.e., standard error.

*Significant values after Bonferroni correction.

All the samples showed large MNA and H_e values (Table 2), with similar values of genetic variability in LIONS (H_e = 0.875 ± 0.084) and in the two samples from the Bay of Biscay (H_e = 0.869 ± 0.087, and H_e = 0.889 ± 0.079). No evidence of a recent bottleneck in population size was detected (p > 0.05). Significant deviations from HWE (deficit of heterozygotes) were found in all samples (Table 2).

Table 2.

Descriptive statistics for each population over all loci.

Population	n	MNA ± s.e.	AR	Ho ± s.e.	He ± s.e.	HWE	FIS ± s.e.	Null alleles ± s.e.
BI8B	95	25.7 ± 9.8	24.32	0.712 ± 0.192	0.869 ± 0.087	Significant	0.180 ± 0.059*	0.081 ± 0.072
BI8C	94	28.3 ± 11.6	25.58	0.762 ± 0.147	0.889 ± 0.079	Significant	0.143 ± 0.045*	0.064 ± 0.050
CÁDIZ	95	33.7 ± 12.2	30.43	0.762 ± 0.092	0.905 ± 0.056	Significant	0.159 ± 0.044*	0.068 ± 0.051
LIONS	96	30.4 ± 7.0	27.08	0.731 ± 0.153	0.875 ± 0.084	Significant	0.165 ± 0.067*	0.070 ± 0.084

n, sample size; MNA, mean number of alleles; AR, allelic richness; H_o, observed heterozygosity; H_e, expected heterozygosity; HWE, significance for the Hardy–Weinberg equilibrium; F_IS, mean null allele frequencies; s.e., standard error.

*Significant values after Bonferroni correction.

Overall, F_ST (0.009) and R_ST (0.0321) were significant. This multilocus R_ST value exceeded the 95% CI of pR_ST values (pR_ST = 0.0083; CI: −0.00021 to 0.020), supporting a mutational component to differentiation. Nevertheless, it must be stressed that it was only one locus, EE2, that showed a significant value (Table 3). The two geographically more distant populations, BI8B and LIONS, showed the smallest and only non-significant F_ST value (after the Bonferroni correction), when both raw data and corrected data for null alleles were analysed. Overall, values of F_ST ≤ 0.006 were found in comparisons among BI8B, BI8C, and LIONS, and comparisons with CÁDIZ showed the largest F_ST values (F_ST > 0.013). In addition, significant differences were found between the two samples from the Bay of Biscay (BI8B vs. BI8C; F_ST = 0.005 ± 0.004), which remained significant after a correction for null alleles (F_ST = 0.005 ± 0.002; Table 4). In the AMOVA, the two samples from the Bay of Biscay were placed in one group and the samples LIONS and CÁDIZ in another group. In this model of population structure, the within-group variance accounted for 100% of F_ST (F_ST = 0.009, p < 0.0001; F_CT = 0.000, p = 0.4748; F_SC = 0.009, p < 0.0001). The maximum F_CT relative to F_ST (F_CT/F_ST = 63%) was obtained when BI8B, BI8C, and LIONS were grouped, leaving CÁDIZ alone in the second group.

Table 3.

Genetic differentiation among samples.

Marker	FST	RST	PRST (95% range)
EJ2	0.000	−0.0005	−3.15e−005 (−0.0053–0.0105)
EJ27.1	0.003a	0.0116	0.00301 (0.0056–0.0209)
EJ35	0.017a	−0.0039	0.0163 (−0.00455–0.0714)
EJ9	0.0012	0.0127	0.00124 (−0.00687–0.0188)
EE10	0.018a	0.0053	0.0185 (−0.00392–0.0727)
EE2	0.018a	0.0810b	0.0164 (−0.0049–0.0498)
EJ41.1	0.006	0.0201	0.0075 (−0.00721–0.0481)
Multilocus	0.009a	0.0315b	0.00832 (−0.00021–0.0208)

F_ST, R_ST, and pR_ST values for the seven microsatellites over the four European anchovy samples analysed: BI8B, BI8C, CÁDIZ, and LIONS.

^aSignificant values after Bonferroni correction.

^bSignificant p-values of allele size permutation test.

Table 4.

Genetic structure of European anchovy: diversity distribution between the four samples analysed in terms of mean pairwise population F_ST values and its standard error, s.e., using raw data and data corrected for null alleles over seven microsatellite loci.

Populations	Raw data (FST ± s.e.)	Null allele corrected (FST ± s.e.)
BI8B vs. BI8C	0.005 ± 0.004*	0.005 ± 0.002*
BI8B vs. CÁDIZ	0.013 ± 0.005*	0.015 ± 0.005*
BI8B vs. LIONS	0.003 ± 0.002	0.002 ± 0.002
BI8C vs. CÁDIZ	0.013 ± 0.005*	0.013 ± 0.005*
BI8C vs. LIONS	0.006 ± 0.003*	0.006 ± 0.003*
CÁDIZ vs. LIONS	0.013 ± 0.006*	0.014 ± 0.006*

*Significant after Bonferroni correction.

Discussion

The main objective of this study was to estimate the levels of divergence among anchovy populations around the Iberian Peninsula and to infer the origin of the anchovy populations in the Bay of Biscay. To that end, seven microsatellite markers were analysed in four samples: two ICES Divisions from the Bay of Biscay and samples from CÁDIZ and LIONS for the purposes of comparison.

Microsatellite variability

The high heterozygosity and large values of MNA in these samples are typical of the microsatellite diversity found in other teleosts, especially small pelagic fish (DeWoody and Avise, 2000). The large number of alleles per locus may be caused by large population sizes in these marine species as well as the large mutation rates at these highly polymorphic loci. Another factor may be a positive correlation between the number of nucleotides in the core repeat unit and the level of allelic variation (Weber, 1990; Yu et al., 2002).

Each of the four samples of anchovy examined here showed a significant deficit of heterozygotes. These deficits may potentially reflect natural selection, non-random mating (inbreeding), Wahlund's effect (inclusion in a sample of individuals from genetically different populations), the presence of null alleles, or a combination of these factors. It is unlikely that natural selection is acting in the same way on seven, presumably unlinked microsatellite loci that are distributed throughout the anchovy genome. Furthermore, the Beaumont–Nichols test did not support a model of divergent selection, which can lead to a deficit of heterozygotes. The test indicated that the levels of divergence for any locus, as estimated with F_ST, were not greater than those expected by chance for a given level of heterozygosity. These results indicate that divergent selection is unlikely to be operating on these loci and is therefore an unlikely explanation for the heterozygote deficits.

Another possible explanation for the heterozygote deficits is some form of assortative mating. Although census sizes of anchovy populations can be large, the ratio of effective population size to census size can be small. Effective population sizes in some marine fish can be 3–4 orders of magnitude smaller than census size (Hauser et al., 2002; Turner et al., 2002). These low effective population sizes in combination with social behaviour and heavy fishing may make some marine species exceptionally vulnerable to inbreeding (Hoarau et al., 2005). The ratio of N_e to census size (N_e/N) is unknown in populations of anchovy in the Bay of Biscay, but these populations declined precipitously in 2005 owing to commercial exploitation. This reduction in census size would have also resulted in a corresponding reduction in effective population size. Further, a large variance in family size from sweepstakes recruitment would reduce effective population size (Cushing, 1990; Hedgecock, 1994; Chikhi et al., 1998; Li and Hedgecock, 1998). Hence, the heterozygote deficits observed in this study may have arisen in part from inbreeding.

Heterozygote deficiencies can also arise from Wahlund's effect. The classic source of Wahlund's effect is the inclusion of individuals in a sample that have originated from genetically differentiated populations. The results of the present study and of many other studies (Spanakis et al., 1989; Bembo et al., 1996; Magoulas et al., 1996, 2006; Borsa, 2002; Borsa et al., 2004) indicate that despite the general lack of dispersal barriers along a coast, anchovy populations can be genetically subdivided. Individuals from these populations may mix at various times in their lives. At a smaller spatial scale, the mixing of individuals from small inbreeding groups arising from sweepstakes recruitment could also give rise to Wahlund's effect. Recruitment variability among years can lead to age-structured groups in a population that, because of sweepstakes recruitment, may show genetic variability between age classes. Each of these mechanisms may have contributed in part to the heterozygote deficits observed in the present study.

Finally, the presence of null alleles may also have contributed to the heterozygote deficits (DeWoody and Avise, 2000; O'Reilly et al., 2004). The most common source of null alleles is the failure of PCR to amplify some alleles. Some of the primers used in the present study were designed for microsatellites in E. japonicus (Chapuis and Estoup, 2007), and many cross-amplifications may not have been successful. In fact, the possible presence of null alleles at several loci was indicated by the positive results of tests using the EM algorithm (Dempster et al., 1977), which performs better than other methods (Chapuis and Estoup, 2007). Nevertheless, the presence of null alleles does not greatly impact inferences about the genetic structure of populations because the results based on raw and corrected data show the same pattern.

Genetic population structure

The paradigm of an association between high dispersal and low genetic differentiation in highly mobile, exploited species has been called into question as more suitable sampling methods and more sensitive genetic markers have revealed structuring in populations previously considered to be panmictic (Hutchinson et al., 2001; Wirth and Bernatchez, 2001; Knutsen et al., 2003; O'Reilly et al., 2004; Von der Heyden et al., 2006). In the present study, multilocus values of F_ST and R_ST among samples were significant and support the conclusion that European anchovy populations in this area are genetically subdivided. This conclusion is likely to remain unaltered, despite the presence of null alleles, because F_ST recalculated with the ENA method yielded similar results (Table 4).

An important result of the present study is that significant microsatellite frequency differences appeared between two samples, BI8B and BI8C, collected from the Bay of Biscay. These differences remained significant even after the presence of null alleles was taken into account. These results are similar to those of Sanz et al. (2008), who found significant allozyme-frequency differences between samples collected from the same spawning areas of the Bay of Biscay in 1996. A comparison of these earlier data with those in the present study provides an opportunity to detect shifts in genetic population structure over three or four generations of anchovy (1996–2005). The results of the present study, based on samples collected in 2005, showed a larger value of G′_ST (0.027) than the results of the earlier allozyme study of samples collected in 1996 (G′_ST = 0.015). Together, these results suggest an increase in the genetic fragmentation among populations over this period. This increase in population structure may well be related to the biomass depletion since 2002. In years with low spawning biomass, the distributions of spawning anchovy in the Bay of Biscay coalesce into discrete discontinuous centres (Motos et al., 1996), and this may result in an increase in subpopulation isolation, which leads to greater levels of genetic differentiation.

Historical phylogeography

Most phylogeographic studies of European anchovy have been based on the analysis of mtDNA sequence variation. However, the analysis of microsatellite variability is increasingly replacing or complementing mtDNA markers, because microsatellites may provide greater resolution to some aspects of population structure. The results of earlier studies demonstrate the importance of historical events in shaping genetic patterns among contemporary populations. Unquestionably, Pleistocene climate–ocean oscillations in temperature greatly influenced the distributions of marine species, and genetic imprints of these events are evident in many pelagic species (Lecomte et al., 2004; Viñas et al., 2004; Pampoulie et al., 2007), including European anchovy (Magoulas et al., 1996, 2006). In this sense, microsatellite variation coincides with earlier results based on mtDNA markers with regard to the origin of Bay of Biscay anchovy. Both markers show that this population is genetically more closely related to populations in the northwestern Mediterranean than to the geographically closer population of CÁDIZ. Bay of Biscay and Mediterranean populations show low F_ST values on seven independent nuclear markers, and BI8B and LIONS samples show a non-significant F_ST value. A recent common ancestor shared by these populations seems to be the most plausible explanation for the observed genetic similarity. Moreover, in accord with the results for mtDNA (Magoulas et al., 2006), the seven nuclear microsatellite markers showed a lack of reduced diversity in the Bay of Biscay, indicating colonizations of this region by large numbers of migrants (Grant, 2005). In all, these data would support the idea of the same migrant population giving rise to Mediterranean and Bay of Biscay anchovy. In agreement with the genetic data, Díaz et al. (2008) reported that these populations showed striking similarities in the relationship between larval dry weight and standard length. Many features in the early life-history stage, such as larval development, are determined genetically, so these similarities may also reflect a genetic link between Bay of Biscay and western Mediterranean anchovy.

In summary, this preliminary study using microsatellite loci has reinforced the previous hypothesis based on mtDNA polymorphism that proposes a recent common ancestry for Bay of Biscay and western Mediterranean anchovy populations. It also detects departures from the classical panmictic population in Bay of Biscay anchovy, suggesting the existence of genetically structured populations in the region. These results question some assumptions currently used for fishery management, and call for further investigation on the genetic features of anchovy populations in the Bay of Biscay. That is why we are now undertaking a second-step analysis of the geographical and temporal variation in European anchovy applying SNPs, genotyping large samples from the Bay of Biscay and from other locations in the distribution of European anchovy. We believe that additional genetic data can provide insights that will improve the sustainable management of anchovy fisheries in the Bay of Biscay.

Acknowledgements

We are hugely indebted to W. Stewart Grant for his thoughtful suggestions, encouragement, and assistance in reviewing, then helping us finalize our revision of this manuscript. We also acknowledge the helpful comments of an anonymous referee. We further thank F. Rendo and the Sequencing and Genotyping Service of the University of the Basque Country for their technical assistance, and Andres Uriarte and Xavier Irigoien (AZTI-Tecnalia) for providing anchovy samples and advice on ecological issues. The work was supported by Research Grant (n° BFI06.346) from the Education, Universities and Investigation Department of the Basque Government, the INIA (RTA2006-00068-C02-02 project), the Industry, Tourism and Commerce department (FOODBASK project), and the Agriculture and Fisheries Department (ECOANCHOA project) of the Basque Government.