Mytilus

We aimed to examine mussel populations in ports, following the discovery of mussels with unexpected Med. M. galloprovincialis ancestry in the port of Cherbourg (France), as sampled in 2003 (Simon et al., 2019). Besides a new sampling in Cherbourg, we sampled seven additional ports and neighbouring natural populations. We also aimed to compare the admixture patterns observed in the ports to other admixed populations, involving different lineages of the same species. The sampling focused on populations where we had a priori expectations of admixture. Therefore, it should not be confused with a representative sample of the M. edulis complex, where populations are usually much closer to the reference parental populations. Most of the port sites were sampled between 2015 and 2017, and older samples were used as references or for temporal information. We either received samples from collaborators or directly sampled in the areas of interest (see Figure S1 and Table S1 for full details).
As part of our sampling process, we re‐genotyped samples from several previous studies that reported the presence of M. galloprovincialis alleles, but had not assigned the samples to the Atl. or Med. M. galloprovincialis lineages. In particular, we used previously extracted DNA from the following studies: (a) Mathiesen et al. (2016) who studied the genetics of Mytilus spp. in the sub‐Arctic and Arctic using 81 randomly ascertained SNPs. They identified M. galloprovincialis and putative hybrids with M. edulis in the Lofoten Islands, Svalbard and Greenland. Their parental reference samples included only the Atl. M. galloprovincialis lineage (Galicia, Spain). Our aim was to further assess the origin of the M. galloprovincialis ancestry. (b) Coolen (2017) studied connectivity between offshore energy installations in the North Sea, characterising samples with 6 microsatellite markers and the locus Me15/16. He identified populations containing individuals with M. galloprovincialis ancestry, using an Atl. M. galloprovincialis reference as well (Lisbon, Portugal).
Samples originating from another oil platform from the Norwegian Sea (Murchison oil station, MCH) and one Norwegian sample (Gåseid, GAS) were also included. We note that the MCH oil rig was free of settled mussels at the time of deployment.
These natural samples were compared to laboratory crosses between M. edulis and Med. M. galloprovincialis, produced in Bierne, Bonhomme, Boudry, Szulkin, and David (2006), and genotyped in Simon, Bierne, and Welch (2018). Briefly, F1 hybrids were first produced by crossing five males and five females of M. edulis from the North Sea (Grand‐Fort‐Philippe, France) and M. galloprovincialis from the western Mediterranean Sea (Thau lagoon, France). F2s were produced by crossing one F1 female and five F1 males. Additionally, sex‐reciprocal backcrosses to M. galloprovincialis were made, they are named BCG when the females were M. galloprovincialis and BCF1 when the female was F1 (Table 1). Production of crosses is described in full detail in Bierne, David, Boudry, and Bonhomme (2002), Bierne et al. (2006) and Simon et al. (2018).
We collected gill, mantle or haemolymph tissues from mussels either fixed in 96% ethanol or freshly collected for DNA extraction. We used the NucleoMag™ 96 Tissue Kit (Macherey‐Nagel) in combination with a Kingfisher Flex (serial number 711‐920, Thermo Fisher Scientific) extraction robot to extract DNA. We followed the kit protocol with modified volumes for the following reagents: 2 × diluted magnetic beads, 200 μl of MB3 and MB4, 300 μl of MB5 and 100 μl of MB6. The extraction program is presented in Figure S2.
Genotyping was subcontracted to LGC genomics (Hoddesdon, UK) and performed with the KASP™ array method (Semagn, Babu, Hearne, & Olsen, 2014). We used a set of ancestry‐informative SNPs developed previously (Simon et al., 2018; Simon et al., 2019). For cost reduction, we used a subset of SNPs that were sufficient for species and population delineation. Multiple experiments of genotyping were performed. The results were pooled to obtain a data set of 81 common markers.

For sequencing, we extracted 4μg of DNA from muscle tissue from a single mussel extracted from the Ria of Vigo, Spain. Using this DNA, three sequencing libraries with insert sizes of 180, 500 and 800 bp were constructed and sequenced at BGI (Beijing Genomics Institute—China). These libraries were sequenced with the Illumina HiSeq2000 high-throughput platform using paired-end sequencing (100-bp reads). To clean the initial set of reads, we filtered out raw reads if they fulfilled any of these conditions: a) >5% ambiguous bases (represented by the letter N); b) poly-A structures; c) > = 20 bases with low quality scores; d) adapter contamination: reads with more than 10 bp aligned to the adapter sequence (no more than 3-bp mismatch allowed); or e) small insert-size reads in which paired reads overlapped more than or equal to 10 bp (10% mismatch allowed).
We used Jellyfish [26 (link)] for counting k-mers and obtaining their frequency distributions. With these data, we drew frequency plots using k-mer lengths of 15, 17, 19 and 21. To assign the “true” coverage peak, we compared these plots to identify the peak that changed in height (“heterozygous peak”) and the one that did not (“coverage peak”). The latter was then used to calculate the genome size as the total k-mer number divided by the coverage-peak depth [27 (link)]. Finally, we assembled de novo the reads resulting from the quality filtering step using SOAPdenovo v1.05 [27 (link)] with parameters -K 31 -d 1 -M 1 -F–R. Then, we ran the Assemblathon 2 script [28 (link)] to obtain assembly statistics. Using this script, we compared the genome assemblies of M. galloprovincialis with those of A. californica, L. gigantea, P. fucata, and C. gigas (S1 File). Genome surveys of other molluscs with scarce sequencing depth [22 (link)] were not included in these comparisons. We confirmed the identification of the studied mussel as M. galloprovincialis by scanning the assembled sequences with two Mytilus genetic markers, Glu-5’ [29 (link)] and EFbis [30 (link)], using BLASTN [31 (link)] and Geneious version 6.1.8 [32 (link)].

The Portuguese shellfish safety monitoring programme was implemented in 1986 [9 (link),51 ]. Currently, the Portuguese monitoring programme comprises 40 classified shellfish-producing areas divided into 13 offshore production areas and 27 estuarine and lagoonar areas [31 ]. Data collected between 2011 and 2020 from Aveiro (RIAV1 and RIAV2) and Óbidos Lagoons (LOB), and the offshore areas L5, L6, L7, L8 and L9 were selected for this study in view of their relevance to shellfish production and impact of HABs (Figure 7).
Several shellfish species are produced in the selected production areas, and different sampling frequencies are used according to the three types of shellfish contamination: microbiological, metals and marine biotoxins, in order to assess the parameters/contaminant levels with regards to regulatory limits (RL) in EU shellfish hygiene regulations, as shown in Table 1. Therefore, different approaches were taken to improve the analysis of the data available regarding each type of contamination.
Regarding microbiological and metals contamination, the data from the different shellfish species were grouped, by location and year, independently of the species. The implemented sampling periodicity and the species availability decreased the number of samples per species, so a global approach was selected for these parameters. The species used were mainly clams (Venerupiscorrugata; Ruditapesdecussatus; Ruditapesphilippinarum; Callista chione; Spisulasolida; Donax sp.), cockles (Cerastodermaedule), oysters (Crassostrea angulata; Crassostrea gigas; Ostrea edulis) and mussels (Mytilus galloprovincialis).
As established in the rules of EU legislation for marine biotoxins monitoring programs, the species with the highest toxin accumulation rate can be used as an indicator for the group of species growing in the same production area [32 ]. Since 2002, the Portuguese monitoring programme has been using the concept of indicator species for the control of biotoxin contamination in shellfish production areas [9 (link)]. In this way, biotoxins data collected for the indicator species for each shellfish production area from 2011 to 2020were used for the present study. For the estuaries/coastal lagoon areas, RIAV1, RIAV2, LOB and littoral areas of the west coast L5, L6, and L7, data from mussels (Mytilus galloprovincialis) toxicity were used. For the littoral areas of the south coast L8 and L9, characterised by vast expanses of sandy beaches where natural banks of clams are regularly exploited, the indicator species selected was the donax clam (Donaxsp.).

Mussels Mytilus galloprovincialis of similar size (shell length 45–55 mm) were collected from Bizerte lagoon, Tunisia (37°13′19.26″ N 9°55′46.24″ E). Upon return to the laboratory, the external side of the bivalves was polished. They were thereafter distributed in glass tanks (3 Liter volume) for an acclimatization under environmental condition for a week prior to TCS contamination. During the experiment, mussels lived in natural seawater which was renewed every two days.
Due to the fact that dimethyl-sulfoxide (DMSO) had no discernible impact on the biomarker responses, stock TCS solutions (purity 98%) were prepared by first dissolving TCS in DMSO, and then added to tanks [20 (link)]. Additionally, it has been demonstrated that the marine bivalves’ metabolic profiles were unaffected by low concentration of DMSO [21 (link)]. DMSO and TCS (purity ≥ 96%) were purchased from Sigma-Aldrich, Co., St. Louis, MO, USA.
Firstly, we evaluated the physiological impact of TCS measuring the filtration and respirations capacities of mussels. To this aim, mussels (three replicates of 10 individuals per condition) were treated during 14 days to 0, 0 + DMSO, TCS1 = 50 µg·L⁻¹ and TCS2 = 100 µg·L⁻¹. The TCS concentrations were obtained after dissolution of 20 mg of TCS in 1 mL of DMSO. Exposures were performed under semi-static conditions, with daily changes of the entire volume of water and the addition of TCS stock solution in order to yield the final test nominal concentrations. A DMSO-treated group (15 µL/3 L = 5 µg·L⁻¹) was established to examine the effects of the solvent and this represents the highest volume of DMSO used in the present study to allow a concentration 100 µg·L⁻¹ TCS/seawater (TCS2). In parallel, to measure oxidative stress, lipid peroxidation, and neurotoxicity induced by TCS, three replicates of 10 individuals per condition were considered. During the experimental period, salinity, temperature, dissolved oxygen, and pH were measured daily with a thermo-salinity meter (LF196; WTW, Weilheim, Germany), an oximeter (OXI 330/SET, WTW), and a pH meter (pH 330/SET-1, WTW), respectively. The temperature was maintained at 19 ± 2 °C, oxygen at 6.2 mg/L, and the salinity was 32‰. After exposure to TCS, tissues were excised, gills and digestive glands were then stored at −80 °C.