The lack of physical barriers in the ocean and large populations sizes generally results in low genetic differentiation in marine fishes (Ward et al., 1994; Bradbury et al., 2008), which renders the task of inferring demographic patterns in this environment particularly difficult. Yet, over the last few years, an increasing number of studies based on high-throughput genetic data have provided evidences of fine scale population differentiation (e.g., Hess et al., 2013; Benestan et al., 2015; Rodríguez-Ezpeleta et al., 2016), challenging previous assumptions based on traditional methods (Hauser and Carvalho, 2008). In particular, the advent of the restriction site associated DNA sequencing (RAD-seq) (Baird et al., 2008) and related methods (Davey et al., 2011) has revolutionized the field of marine conservation genomics (Narum et al., 2013). This approach consists on subsampling putative homologous regions from the genome in several individuals with the aim of identifying and genotyping single nucleotide polymorphisms (SNPs). Interestingly, the method can be applied to organisms for which no prior genomic resources are available, and it is suitable to study both, neutral population structure and local adaptation (Allendorf et al., 2010). In the context of marine management, RAD-seq is particularly relevant for poorly studied widely distributed species, as it can provide quick estimates of genetic diversity, population connectivity and adaptation more cost-effectively than relying on genome or transcriptome sequencing, or on non-sequencing based microsatellite or SNP typing.

Among the poorly studied marine organisms are mesopelagic fish, i.e., those inhabiting the deep scattering layer, that represent the largest fish biomass in the ocean (Kaartvedt et al., 2012; Irigoien et al., 2014) and play important roles in the marine ecosystem and global biogeochemical cycles (Klevjer et al., 2016). The large abundance of these fish make them particularly attractive for exploitation (St. John et al., 2016). Yet, the mesopelagic ecosystem is still largely unknown (Sutton, 2013), and the potential effects of newly introduced anthropogenic pressures in this realm should be anticipated so that sustainable management strategies for this valuable resource can be developed (St. John et al., 2016). Thus, there is an urgent need for establishing focused scientific surveys, developing appropriate sampling gear and generating additional biological data to booster knowledge in these species. In particular, developing genomic tools is foremost, as they can provide insights to understand population persistence, productivity and response to exploitation (Palumbi, 2003; Cowen et al., 2006).

Several studies have focused on deciphering species boundaries and population connectivity within mesopelagic fishes using genetic variants, but they are all based on allozymes (Gunawickrama and Naevdal, 2001; Kristoffersen and Salvanes, 2009), single mitochondrial markers (Kojima et al., 2009; Habib et al., 2012; Gordeeva, 2014) or a few microsatellites (Gordeeva, 2011; Van de Putte et al., 2012). To our knowledge, no studies based on genomic-wide SNP data have been published. Yet, simulated and empirical data based evidences support that high-throughput SNP data analyses provide more accurate population structure inferences than single or a few polymorphic marker based analyses (Haasl and Payseur, 2011; Rašić et al., 2014; Benestan et al., 2015; Szulkin et al., 2016). Here, in a first attempt to introduce genome-wide SNP discovery and genotyping to study mesopelagic fishes, we have focused on the Mueller’s pearlside, Maurolicus muelleri.

Although initially thought to be distributed worldwide, previous reports of M. muelleri in the Northeast Atlantic (Gjøsætr and Kawaguchi, 1980), the South Atlantic (Hulley and Prosch, 1987), the Southeast Pacific (Robertson, 1976), and the coasts of Japan (Okiyama, 1971) are now believed to belong to different species of the genus Maurolicus. The genus was split into fifteen species based on morphological characters and geographic distribution (Parin and Kobyliansky, 1993), but recent studies have suggested an overestimate of species diversity in Maurolicus (Kim et al., 2008; Rees et al., 2017), illustrating a need for more in-depth phylogeographical analyses. In response to these needs, we have discovered and genotyped thousands of SNP markers in M. muelleri, which constitute the first genomic resource of a mesopelagic fish.

Restriction-site associated DNA sequencing constitutes an unprecedented opportunity for performing demographic inferences in species for which no prior genetic resources are available (Corander et al., 2013; Hess et al., 2013). Here, we have discovered and genotyped thousands of RAD-seq derived SNP markers in M. muelleri in a first attempt to use genome wide data to study diversity and connectivity in this mesopelagic species, which is particularly relevant in view of the imminent exploitation of this till now pristine marine resource.

We have demonstrated that the Sbf I restriction enzyme is as a good candidate for RAD-sequencing based SNP discovery in M. muelleri. Interestingly, we found that, although the average number of RAD-tags per individual is similar to that found in other species (Amores et al., 2011; Catchen et al., 2013b; Puebla et al., 2014; Diaz-Arce et al., 2016; Rodríguez-Ezpeleta et al., 2016), the number of RAD-tags obtained per individual increases with the number of sequencing reads produced; this suggests that most likely the genome of M. muelleri could have as many as 50 000 SbfI cut sites, which is higher than most fish species studied so far. The presence of a large number of cut sites, and therefore of loci, could suggest lower coverage per locus. Yet, although lower coverage than in other species is obtained, all individuals had more than 6× coverage and average was of 11×. Multiplexing less individuals per lane or sequencing the same pool in more than one lane would increase coverage accordingly. Yet, the large number of loci shared among 75% of the individuals and the above one thousand SNP markers obtained after filtering suggest that this increased sequencing would not be necessary.

In our analyses, observed heterozygosity is lower than expected. This could be indicator of (i) population differentiation within the area of study, referred to as the Wahlund effect (Wahlund, 1928), (ii) preferential mating with close relatives or inbreeding (Wright, 1921) or (iii) non-random sampling of members from a limited number of families (Robertson, 1965). Due to the large biomass of mesopelagic fish (Irigoien et al., 2014), it would seem unlikely that our samples, composed by a few individuals per location, contain relatives. Similarly, preferential mating with close relatives or inbreeding would be difficult also to explain. Thus, the natural explanation for the low observed heterozygosity obtained would be population differentiation within the area of study. Yet, we do not find clear evidences of population stratification except for catch 9018, which appears genetically differentiated in the STRUCTURE plots, but not in the PCA.

The fact that when separating individuals into catches, observed heterozygosity is also lower than expected, suggests that one of the three possible explanations should be also acting at the catch level. In this sense, the possibilities of inbreeding and non-random sampling members of the same family should also be considered. Indeed, the fact that size distributions are quite homogeneous within catches (Figure 1), supports previous observations that individuals are grouped according to their age (Staby et al., 2011), and claims for more studies linking the reproductive behavior of the species and the genetic diversity results obtained here, which should be further confirmed with additional data. Similarly, no explanation to the different genetic diversity estimates obtained per catch could be found. These differences could be simply due to a low sample size per catch or have a more biologically meaningful information, and thus, more analyses are needed to confirm either one.

Overall, we have validated the RAD-seq based SNP discovery method using the SbfI restriction enzyme for M. muelleri, which has resulted in the first genomic resources for this species that are moreover made available for future studies (raw sequences are available at NCBI SRA Bioproject PRJNA417219). This data will contribute to shed light on phylogeographic patterns within the species (Rodríguez-Ezpeleta et al., 2016), detect genetic adaptation (Van Wyngaarden et al., 2017), and to resolve the phylogeny within the genus for species delimitation (Diaz-Arce et al., 2016).

NR-E designed the study, performed the analyses, interpreted the data, and wrote the manuscript. PÁ collected the samples, contributed to the interpretation of the data and revised the manuscript. XI contributed to the interpretation of the data and revised the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.