Large‐scale genetic panmixia in the blue shark (Prionace glauca): A single worldwide population, or a genetic lag‐time effect of the “grey zone” of differentiation?

Abstract The blue shark Prionace glauca, among the most common and widely studied pelagic sharks, is a top predator, exhibiting the widest distribution range. However, little is known about its population structure and spatial dynamics. With an estimated removal of 10–20 million individuals per year by fisheries, the species is classified as “Near Threatened” by International Union for Conservation of Nature. We lack the knowledge to forecast the long‐term consequences of such a huge removal on this top predator itself and on its trophic network. The genetic analysis of more than 200 samples collected at broad scale (from Mediterranean Sea, North Atlantic and Pacific Oceans) using mtDNA and nine microsatellite markers allowed to detect signatures of genetic bottlenecks but a nearly complete genetic homogeneity across the entire studied range. This apparent panmixia could be explained by a genetic lag‐time effect illustrated by simulations of demographic changes that were not detectable through standard genetic analysis before a long transitional phase here introduced as the “population grey zone.” The results presented here can thus encompass distinct explanatory scenarios spanning from a single demographic population to several independent populations. This limitation prevents the genetic‐based delineation of stocks and thus the ability to anticipate the consequences of severe depletions at all scales. More information is required for the conservation of population(s) and management of stocks, which may be provided by large‐scale sampling not only of individuals worldwide, but also of loci genomewide.

In other words, the genetic connectivity and the demographic connectivity exhibit, in some conditions, a phase difference that prevents the former from being a good proxy of the latter. Common situations involving the homogeneous distribution of genetic polymorphism can thus derive from a wide range of very distinct demographic situations, depending on the relative weight of population size and effective dispersal. These demographic scenarios range from a rate of migratory exchange high enough to lead to both genetic homogeneity and strong demographic interdependency among (sub)populations, even with limited effective sizes, to nearly negligible rates of migratory exchange (m) among populations exhibiting large effective sizes (N e ).
The incomplete genetic sorting of populations could be considered as the homologous version, at the intraspecific level of the "grey zone" from De Queiroz (2007). This "species grey zone" represents the lag during which, lineage sorting being incomplete, species delimitation is not possible based solely on the genetic information in hand. Here, we introduce the concept of "population grey zone." In contrast to the "species grey zone" defined by De Queiroz, where an agreement on the actual number of species can be reached before the split or after a reasonable time of divergence, a consensus on the level of demographic connectivity (or interdependency) of populations estimated through the analysis of genetic connectivity can be reached only after the "population grey zone" (Figure 1).
The blue shark is by far the most abundant shark species caught by fisheries (Beerkircher, Cortes, & Shivji, 2002;Buencuerpo, Rios, & Morón, 1998;Campana, Marks, Joyce, & Kohler, 2006;García-Cortés & Mejuto, 2001;Rogan & Mackey, 2007). Most of the time, blue sharks are taken either as bycatches, that is, undesirable nontargeted species, by tuna longlines and swordfish fisheries (Carvalho et al., 2015) or as the targeted species, depending on the fishing location and the nationality of the fisheries. Indeed, Oliver, Braccini, Newman, and Harvey (2015) reported that "approximately 50% of the global shark production is composed of sharks caught as bycatch in the high seas pelagic longline fisheries" and Clarke, Harley, Hoyle, and Rice (2013) found that blue sharks caught in Pacific were either discarded (50%) or finned (42%). Although abundance studies have delivered mixed conclusions, the recent estimates most often converge towards a rather sharp decline in population during recent decades. The blue shark populations' trends could be estimated by CPUE (catch per unit effort), which is an indirect measure of the abundance of the species; CPUE variations signify changes to the species' real abundance. The blue shark CPUE F I G U R E 1 The "grey zone" of population differentiation. Analogy to De Queiroz speciation grey zone: inside the "grey zone", it is impossible to discriminate populations based on genetic data alone. N, population size; m, migration rate  (Nakano & Clarke, 2005), the Indian (Nakano, 1996in Nakano & Stevens, 2008 and the Pacific Oceans (Matsunaga & Nakano, 1999). In contrast, Simpfendorfer, Hueter, Bergman, and Connett (2002) estimated an 80% male decline during 1980-1990 (females results lacked significance) in the North Atlantic, Baum et al. (2003) a 60% decline in the North Atlantic from CPUE and Ferretti, Myers, Serena, and Lotze (2008) a 97% decline in abundance in the Mediterranean Sea during the mid-20th century. In the Pacific, blue sharks decline was estimated to have reached 57% during the period spanning from 1950 to 1990 (Ward & Myers, 2005) and 1995to 2003(Clarke et al., 2013. With an estimate of 10 million (Clarke et al., 2006) to 20 million individuals removed per year (Stevens, 2009) (Barnett, Abrantes, Stevens, & Semmens, 2011;Feldheim et al., 2014;Hueter, Heupel, Heist, & Keeney, 2004;Jorgensen et al., 2010;Pardini et al., 2001). Blue shark mating, pupping and nursery sites are suspected in the Atlantic (Azores, Brazil and South Africa as nurseries, Aires-da-Silva, Ferreira, & Pereira, 2008;Vandeperre et al., 2014;Verissimo et al., 2017) and Pacific (California as nursery, Carrera-Fernández et al., 2010;Caldera as pupping and nursery, Bustamante & Bennett, 2013) and Mediterranean Sea (as mating area and nursery, Megalofonou, Damalas, & de Metrio, 2009). Thus, despite long-range migration, philopatry, together with recent reductions in population sizes, may lead to expect some level of genetic differentiation across the species range.
Specific challenges are associated with the study of pelagic and migrating sharks, particularly the estimation of population size and the delimitation of stocks of a worldwide species exploited by several nations and in international waters (Heist, 2008). The stock concept was initially proposed to address the sustainability of fishing activity (Carvalho & Hauser, 1994). Stock includes a broad range of definitions, depending on the management aims, and reconciling them is not always easy (Carvalho & Hauser, 1994). Fishery stock (Smith, Jamieson, & Birley, 1990) usually refers to a group of fishes exploited in a specific area or using specific gear, whereas biological stock (Ihssen et al., 1981) is defined as "an intraspecific group of randomly mating individuals with temporal and spatial integrity," in line with the definition of a demographic population. The genetic stock, according to Ovenden (1990), is defined as "the largest group of animals that can be shown to be genetically connected through time." In this study, we analysed blue shark samples from Mediterranean Sea, North Atlantic and Pacific Oceans (Figure 2), using mitochondrial and nuclear DNA, to provide an initial assessment of genetic stocks across the range of the species. We then performed population simulations to generate in silico data illustrating the properties of the "population grey zone" and discuss their potential similarities with the in vivo results as obtained for blue sharks.

| Sample collections and DNA extraction
DNA was isolated from fin tissue or muscle samples collected between 2010 and 2014 in three oceanic basins (Table S1, Figure 2), mostly as bycatches from collaborative fisheries. Representative samples were collected in the Atlantic Ocean (Vigo, Spain and Azores islands), the Mediterranean Sea (Gulf of Lion including Grau du Roi and Corsica from France, Malta, and Greece) and the Pacific Ocean (Hawaii, F I G U R E 2 Sampling sites across blue shark distribution area. The distribution area is drawn in blue, and sampling sites are represented by blue dots for the Mediterranean Sea (Gulf of Lion, Malta and Greece), green for the Atlantic Ocean (Spain and Azores) and orange for the Pacific Ocean (Australia, New Zealand and Hawaii) Australia and New Zealand), the Indian Ocean could unfortunately not be sampled for this study. Samples were preserved in 96% ethanol and stored at room temperature.
DNA was extracted using CTAB (Doyle & Doyle, 1987) to which we added a digestion step using proteinase K.

| Markers selection and PCR amplification
Genetic variation at two kinds of putatively neutral genetic markers (mitochondrial DNA/mtDNA and microsatellites) was used to test the hypothesis of population differentiation and bottlenecks. A fragment of the mitochondrial cytochrome b (cytb) gene, 1,000 bp on average, was amplified with Glu14314L and Thr15546H (5′-TCTTCGACTTACAAGGTC-3′). Reaction volumes were 50 μl,  The loci D0MST presented critical amplification failures and was eventually removed from data, leaving nine loci for further data analysis.
We used the program POWSIM 4.1 from Ryman and Palm (2006) to determine the statistical power of our markers to detect genetic differentiation given differences in their levels of allelic diversity and sample sizes. As POWSIM computes Nei's F ST (Nei, 1987;Nei & Chesser, 1983) "modified to be independent of the number of subpopulations" (POWSIM manual), we used GENETIX were fixed to 10,000, 1,000 and 10,000 for, respectively, dememorizations (burn-ins), batches and iterations per run, for a total of 1,000 runs each.

| Clustering analysis
Population structure was investigated using two statistical approaches.
First, STRUCTURE 2.3.4 (Pritchard, Stephens, & Donnelly, 2000) and STRUCTURE HARVESTER 0.6.94 (Earl & vonHoldt, 2012) were used to reveal the clustering, if existing, in the data set. Ten independent runs were performed on STRUCTURE for each assumed number of population(s) K = 1-8 under an admixture model. This model is consistent with the fact that blue sharks are able to migrate over long distances. All runs were executed with 50,000 burn-in periods and 200,000 MCMC (Markov chain Monte Carlo) repetitions, using the eight regions defined earlier as prior information. STRUCTURE HARVESTER displays results allowing to assess K, the number of genetic populations that best fit the data, based on maximum likelihood (Evanno, Regnaut, & Goudet, 2005).
In addition, a discriminant analysis of principal components (DACP, Jombart, Devillard, & Balloux, 2010) was performed. DAPC is a multivariate analysis that integrates principal component analysis (PCA) with discriminant analysis to summarize genetic differentiation between groups (Jombart, 2008; (adegenet package v2.0.1, Jombart, 2008). Sampling location was used as prior. While STRUCTURE forms genetic clusters of individuals by minimizing departure from Hardy-Weinberg and linkage disequilibria, DAPC maximizes genetic separation among groups and minimizes variation within groups (Jombart et al., 2010), which may constitute a more accurate approach for species exhibiting potentially high gene flow.
A haplotype network aiming to construct the shortest possible tree of haplotypes was computed using Network 4.6.1.4 (Fluxus Technology Limited 2010). We chose the median-joining (MJ) network algorithm and epsilon (the weighted genetic distance to the known sequences in the data set) set to 0 (Bandelt, Forster, & Röhl, 1999).

| Effective population sizes and bottleneck tests
Fu and Li's D (Fu & Li, 1993) and Fu's Fs (Fu, 1997) were used to test for the occurrence of demographic changes such as expansion or bottlenecks.
Tests for evidence of genetic bottlenecks were performed using BOTTLENECK 1.2.02 (Cornuet & Luikart, 1996) in every region defined and in the global sample. Wilcoxon signed rank tests (most appropriate and statistically powerful for data set with a limited number of polymorphic loci, Maudet et al., 2002) were used to investigate microsatellite heterozygote excess and the allele frequency distribution test. Genetic bottlenecks reduce allelic diversity faster than heterozygosity (Nei, 1989). Consequently, populations exhibit an excess of heterozygosity with a greater number of microsatellite loci than predicted by chance until mutation-drift equilibrium is established (Cornuet & Luikart, 1996). The genetic differentiation between populations would be greater if these populations had suffered from a recent bottleneck. The test was performed under two assumptions for the mutation model: the pure stepwise-mutation model (SMM) and two-phase mutation (TPM, Di Rienzo et al., 1994) with 70% proportion of SMM in TPM, 30% variance for TPM and 1,000 replicates.
We used NeEstimator 2.01 (Do et al., 2014) to compute the N e estimates for every population and for the whole population with the parameters suggested for low sample size (linkage disequilibrium method, with p-value of .02 and 95% confidence interval estimation by parametric method).

| Simulations
Simulations were performed to characterize the genetic differentiation among two increasingly diverging populations, in order to illustrate the number of generations required to overcome the "population grey zone," depending on the variation of the two driving demographic parameters: effective population size and migration. We used simuPOP 1.1.7 (Peng & Amos, 2008)   From left to right: the number of sequences obtained (cytb samples), the number of unique haplotypes per location (haplotypes), the haplotype diversity (h), the nucleotide diversity (π), D of Fu and Li (1993) and F s of Fu (1997). Statistical significance: * for p-values < .05, ** for p-values < .01 and *** for p-values < .001.
T A B L E 1 Genetic diversity within each sampling location based on mitochondrial DNA

| Genetic variability
From a total of 201 sequences (Table 1)  for the Gulf of Lion (Table 2). Once standardized to consider the minimum number of samples taken (13 entire genotypes, Australia), the allelic richness did not exhibit differences among geographic regions (Table 2). Similarly, the expected and observed heterozygosity were comparable, ranging between 0.69 and 0.76 (Greece excluded). No F IS values departed significantly from 0 ( Table 2).

| Population differentiation
The haplotype network ( Figure 3) showed three widely distributed major haplotypes and no spatial segregation of haplotypes. Similarly, the AMOVA showed no significant partition of the variance among the oceanic basins or among populations within oceanic basins.
Similar results were obtained through F ST estimates based on both cytb (Table 3a) and microsatellite data ( From left to right: the number of genotypes obtained ("μsat samples" stands for microsatellite samples), the mean number of alleles per location (mean alleles), the allelic richness after rarefaction for the smallest sample size (Ar (3)) and for the second smallest sample size (Ar (13)), the expected (H e ) and observed heterozygosity (H o ) and the inbreeding coefficient (F is ).
F I G U R E 3 Haplotype network of blue shark individuals. Each circle represents a unique haplotype, and their sizes are proportional to the number of individuals sharing this haplotype. The colour inside each circle indicates the sampling site origin of the individual. The lengths of the branches joining the circles are proportional to the number of differences between the haplotypes estimates (Table S2). For cytb (Table 3a), only F ST between Australia and the Gulf of Lion was significant with a very low FDR (p-value correction for multiple tests, q-value = 5.84·10 −307 ). As for microsatellites ( to detect correctly F ST values from approximately 0.01 (Figure 4a).
GENETIX computed a Nei's F ST of 0.0048 for microsatellite data.
POWSIM's results indicated the microsatellite data set has the power to detect correctly F ST values from approximately 0.0026 (Figure 4b).
Results from POWSIM suggest data sets have the power to detect the corresponding genetic differentiations.
In line with those results, whatever the k tested, STRUCTURE and STRUCTURE HARVESTER pictured a lack of structure ( Figure 5).
Comparable results were obtained with the DAPC analysis ( Figure S1), where no distinct group emerged. In fact, no stark separation between locations was evident ( Figure S1b) and only the three samples from Greece gave hints of difference along the first principal component ( Figure S1a).

| Effective population size and bottlenecks
Fu and Li's D values were highly negative but not supported (

| Simulations
Whatever the effective size (N e ) or the number of migrants exchanged (N e m), the F ST and the sub-F ST values are very similar (see Figure 6).
The F ST and sub-F ST values decrease with increasing number of migrants and effective sizes.
F I G U R E 5 Bayesian clustering of blue shark individuals from STRUCTURE analysis. a: Within barplots for K from 2 to 5. Each individual is represented by a vertical bar partitioned into coloured sub-bars whose lengths are proportional to its estimated probability of membership for the K clusters. b: Plot of the mean of estimated "log probability of data" for each value of K. c: Delta K of Evanno's method based on the rate of change in the log probability of data. d: Evanno table output   For an effective size of 10,000 individuals, the number of migrants exchanged has nearly no influence on the "grey zone" range: it takes an average of 200 generations to obtain a detection capacity of a significant sub-F ST of 95%. For an effective size of 100,000 individuals, the number of generations necessary ranges from at least 1,000 (N e m = 0), to 1,400 (N e m = 1), and to 1,600 (N e m = 10).
As for the bottleneck results (Figure 7), the number of generations necessary to detect a significant change with the F ST index ranges from at least 160 (bottleneck of 99%) to 2,200 generations (bottleneck of 90%).
Adding simulations with larger effective population sizes would be extremely computationally time-consuming for an iterative process.
The estimates given here are thus only lower bound in terms of generation time required for the divergence to be detected. The comparison of results obtained when switching effective population size thus delivers rather conservative estimates of the number of generations during which the "grey zone" effect applies. These results suggested such "grey zone" effect for effective population size of hundred millions individuals is likely even more pervasive then illustrated in Figure   S2.

| DISCUSSION
Widespread genetic homogeneity of one of the most widely dis-  also may form several populations having shared the same or a similar environmental pressure(s) in the past) or (ii) the genetic bottleneck is recent (and thus, blue shark population is likely panmictic, suffering rather uniformly a present-day reduction in population size) or (iii) a combination of both resulting in bottlenecks detection without population differentiation. The capacity of next-generation sequencing to access high-density genome scanning is expected to allow improved inference of parentage or kinship through coalescent analyses, to expand analyses based on linkage disequilibrium (Hellberg, 2009) and to refine both the dating of bottlenecks and the interpretation of patterns of genetic differentiation (Waples, Seeb, & Seeb, 2016 (Queiroz et al., 2005(Queiroz et al., , 2016Vandeperre et al., 2014) or very rare (Kohler & Turner, 2008) shark movements were detected between the Atlantic and Mediterranean basins using conventional tagging. This result may be due to the low statistical power and representativeness of tags on a limited number of specimens but also could reflect a real demographic independence not revealed in the genetic estimates of differentiation.
Indeed, the particular life-history traits of several marine species (large population sizes and high dispersal potential) often lead to weak or no genetic differentiation (Hedgecock et al., 2007;Waples, 1998), which could be explained by an homologous version at the intraspecific level (Figure 1) of the "species grey zone" concept proposed by De Queiroz (2007), that would be here the "populations grey zone." We represented this concept through simulations ( Figure 6). Figure 6  1,000 2,000 3,000 4,000 5,000 0 1,000 2,000 3 ,000 4,000 5 ,000 0 1 ,000 2,000 3 ,000 4,000 5 ,000 0 1,000 2,000 3,000 4,000 5,000 0 1,000 2,000 3 ,000 4,000 5 ,000 0 1 ,000 2,000 3 ,000 4,000 5 ,000 N m = 1 0% 20% 40% 60% 80% 100% of migrants (m) ensuring genetic homogeneity may be largely insufficient to lead to demographic interdependency or to ensure a rescue effect (Gagnaire et al., 2015;Waples & Gaggiotti, 2006). Within the "population grey zone," the pace of drift and thus the accumulation of detectable genetic differentiation depend strongly on the population size and number of migrants ( Figure 6). Thus, despite migration rates low enough to ensure demographic interindependence and contribute to the accumulation of divergence of populations in terms of allelic frequencies, the detection of population differentiation may not be possible for many (thousands of) generations. Only after the sufficient number of generation elapsed to exit the "population grey zone" will the systematic rejection of panmixia allow the safe conclusion of demographic independence.
During the last glacial episode, colder water conditions may have caused a cessation of between-ocean gene flow for the blue shark, a temperate species. Verissimo et al. (2017) reviewed the cosmopolitan coastal pelagic carcharhinoids and oceanic epipelagic sharks for which isolation and lineage divergence have been shown between Atlantic and Indo-Pacific, probably due to colder water conditions around the tip of South Africa and to the cold Benguela current. These authors concluded that the current apparent genetic homogenization of the species is due to extensive interbasin gene flow since the last glacial period, which may apply the blue shark. However, according to simulations presented here, apparent panmixia is also compatible with the opposite scenario of a limitation to gene flow following the reorganization of the distribution range, nursery and feeding grounds after the end of the last glaciations. Considering the estimated generation time of blue sharks (8.1 years, IOCT 2007), the time required to escape the "grey zone" of detection corresponds to 12,960 years with N e = 100,000 ( Figure 6). Such effective population size is rather conservative considering the estimated 20 million individuals extirpated yearly by fishing for this species, although recent genetic-based estimates suggested effective population sizes of only several thousand individuals in the Atlantic (Verissimo et al., 2017) and Pacific . In fact, within the hypothesis of the "grey zone," even a number of migrants per generation of 10 with effective population size of 100,000 individuals are enough to hide until nowadays a genetic divergence initiated about 11,500 years ago.
The simulations performed here to test for the effect of a variation in N e m driven by a reduction in N e rather than a modification of m also showed an extensive time lag (Figure 7). The removal of 99% of a population of high effective size results in significant F ST only 160 generations at least after the occurrence of the demographic event.
Expectedly, less severe bottleneck expands further the genetic lag effect (Figure 7). These results imply that the 97% decline in abundance of the blue shark in the Mediterranean Sea during the mid-20th century (Ferretti et al., 2008) is fully compatible with the lack of structure detected by our analysis, as it may only result in significant F ST after 1,300-17,820 years depending on the strength of the bottleneck suffered.
F I G U R E 7 Illustration of the impact of huge bottlenecks with data from simulations of splits among populations. Simulated population separation process with initial N e = 1,000,000 and m = 0.0001. After 500 generations (first dashed red line), each population was reduced by 90% and 99%. The detection of significant sub-F ST above 95% is indicated by the second dashed red line. For each plot, the x-axis represents the number of generations since the divergence, the right y-axis the F ST values (blue lines, full for the median value and dashed for the 95% envelope) and the left y-axis the percentage of significant F ST values (green line) Bottleneck 99% Bottleneck 90% This "population grey zone" can also be suspected and illustrated with other species, including at least the wahoo (Acanthocybium solandri, Theisen, Bowen, Lanier, & Baldwin, 2008) and the white marlin (Kajikia albida, Mamoozadeh, McDowell, Rooker, & Graves, 2017).
The "population grey zone" may well explain as well the rather elevated number of studies where the null hypothesis of panmixia could not be rejected despite other data supporting the existence of distinct stocks.
For example, the lack of genetic differentiation between dolphinfish and found basinwide panmixia.
The effective management of fisheries requires a clear definition and understanding of the stock structure of the target species (Begg & Waldman, 1999), and this difference between the genetic and demographic concept of populations parallels the problem of the multiple existing definitions of stocks. Booke (1981, in Waples & Gaggiotti, 2006 defined a stock as a species, group or population of fish that maintains and sustains itself over time in a definable area. Although informative from a fishery perspective, this definition does not encompass the concept of the interdependence or independence of groups of individuals and thus does not provide useful information for the management and sharing of common resources. Depending on whether the perspective and scale at stake are genetic and evolutionary, demographic and ecological or management, two more operational and complementary definitions of stock may be considered, each corresponding to one of the two main concepts of connectivity described above. The genetic stock, according to Ovenden (1990), defines a reproductively isolated unit, which is genetically different from other stocks (genetic connectivity), while the harvest stock (Gauldie, 1991) designates a "locally accessible fish resource in which fishing pressure on one resource has no effect on the abundance of fish in another contiguous resource" (demographic connectivity). A stock thus might be an aggregate of biologically homogeneous (according to Booke's terms) but genetically different groups (Ovenden, 1990) or "substocks" (Altukhov, 1981), characterized by differentiated gene pools and/or demographically independent (according to Gauldie's definition) groups. From a stock management perspective, Gauldie's definition would be the most accurate, whereas from a longer-term conservation perspective, recognizing and preserving potentially differentiated gene pools is also essential (Laikre, 2010). It requires, however, a rigorous sampling strategy whenever possible, and a clear account for the "population grey zone" concept.

| CONCLUSIONS
Elasmobranch recovery is possible but requires in situ management actions (Ward-Paige, Keith, Worm, & Lotze, 2012). Nursery protection has been proposed (Beck et al., 2001), as juvenile survival seems to be key for shark conservation (Cortés, 2002). However, the benefits of marine-protected areas for mobile species are questioned (Grüss, Kaplan, Guénette, Roberts, & Botsford, 2011), even more so when fishery fleets would not respect them (Baum et al., 2003;Botsford, Castilla, & Peterson, 1997). As putative nurseries in the Atlantic, Mediterranean and Pacific overlap with tuna vessels and swordfish fleets, creating protected areas may be unrealistic and/or unproductive. Most of the time, blue sharks are not even targets but mere bycatch. A recent study (Queiroz et al., 2016) showed an 80% overlap between pelagic sharks' hot spots, including blue sharks, and longline fleets in the North Atlantic. In the absence of prospects for successful marine-protected areas, the management of fish stocks is essential and requires a good knowledge of their delineation. As a long-living and slow-growing species, the blue shark is vulnerable to fisheries, which could rely on stock production models defined on traditional short-living species and the recovery time estimated with fishery management plan greatly underestimated (Musick, 1999). Effective mitigation measures exist to reduce elasmobranch bycatch mortality for fisheries, if employed (Poisson et al., 2016).
The case study used here to illustrate the "population grey zone" does not reject the possibility that the blue sharks all belong to a single worldwide population. On the one hand, such global panmixia may fit a very large global population size, providing robustness to the species in cases of local depletion. On the other hand, depending on the spatial dynamics of the species, it may also imply that any serious impact in any part of the world might have serious consequences at a global scale. In both cases and as suggested by Theisen et al. (2008) who reported similar results and wonders on the pelagic wahoo, such a global population call for a global management. The wide-scale decline or extirpation of such a top predator as blue sharks is a serious concern (Ferretti, Worm, Britten, Heithaus, & Lotze, 2010;Lewison, Crowder, Read, & Freeman, 2004;Myers, Baum, Shepherd, Powers, & Peterson, 2007;Rogers & Ellis, 2000;Stevens, Bonfil, Dulvy, & Walker, 2000).
A conservative, concerted and global management and conservation strategy is thus required until in-depth analysis allows the confirmation of homogeneity or delineation of differentiated demographic groups and/or stocks. The complex relationship among populations could be resolved by more powerful high-density genome scan analysis. The scoring of thousands of genetic markers, allowing the identification of outlier loci, has proven useful for delineating local stock and defining conservation units (Nielsen, Hemmer-Hansen, Larsen, & Bekkevold, 2009), as well as for the reconstruction of pedigrees to estimate stock size and number (Bravington, Grewe, & Davies, 2014). All emerging methods allowing the high-density coverage of the genome are prospects to overcome the "population grey zone" problem likely responsible for the frequent mismatch between high expectations and inconclusive results obtained thus far when applying population genetics to detect differentiated stocks of fisheries' targets.

ACKNOWLEDGEMENTS
We thanks the editor and the anonymous reviewers for their very helpful comments and suggestions. We thank the associations