The Spanish coast of the Cantabrian Sea (Fig. 1A) is characterized by the alternation of cliffs, beaches, and estuaries (Díez et al. 2000). In the same area, there are major urban and industrial centres that hold trading ports. Upwelling processes which take place on the Cantabrian coast lead to great fishing activity in this zone, with major fisheries such as sardine, mackerel and anchovy.

The Ria of Vigo (Fig. 1B) is an ecosystem subjected to a strong human impact (Fernández et al. 2016). Its surroundings are highly urbanized, with a total population of ~428000 inhabitants. The urban pressure is especially great in the central zone of the south seashore, where the city of Vigo is situated (~295000 inhabitants). Around the Ria of Vigo there is a great presence of industry, which stimulates the traffic of goods through the Port of Vigo, which is also one of the main fishing ports in the world. Urban and industrial activities coexist with culture of molluscs (mainly mussels) in the inner part of the Ria. In spite of the generalized impact, some areas of the Ria of Vigo have remained little affected and therefore possess a high environmental value.

Mussels were collected at 14 sampling stations located in the Ria of Vigo and the Spanish coast of the Cantabrian Sea (Fig. 1) throughout autumn 2016 in the framework of the Spanish Marine Environmental Monitoring programme carried out by the IEO (Instituto Español de Oceanografía). No distinction was made during sampling between both mussel species present along the north coast of Spain (Mytilus edulis Linnaeus, 1758 and Mytilus galloprovincialis Lamarck, 1819). Consequently, the studied mussels were covered by the denomination Mytilus spp.

Table 1 provides a brief description of the sampling stations. For this study, only mussels with a shell length of between 35 and 65 mm were selected. The shells of mussels were discarded and the soft tissues were placed in aluminium vessels and preserved at 20°C. Prior to each digestion, the sample was defrosted in a refrigerator (~3°C) for a minimum time of 3 hours.

For extraction of microplastics, the mussels were digested with 65% HNO₃ (acid digestion) and with 10% KOH (alkaline digestion). The digestive procedures performed with both digestive agents are described below.

The digestive procedure was carried out according to Claessens et al. (2013). Three replicates with mussel soft tissue were carried out for each station. Each mussel was weighed and individually treated. Then 10 mL of 65% HNO₃ per gram of mussel soft tissue were added. Because of the destructive effect on microplastics caused by excessively high temperatures (Munno et al. 2018), the digestive procedure was carried out at room temperature for 24 hours (instead of carrying out a digestion at 100°C for 2 hours, as proposed by Claessens et al. 2013). The beakers were not stirred.

The digestive procedure was carried out as described by Foekema et al. (2013). Three replicates with mussel soft tissue per station were carried out. Each mussel was weighed and individually treated. Then 20 mL of 10% KOH per gram of mussel soft tissue were added. Soft tissues of mussels were digested at room temperature for 21 days. The beakers were submitted to vigorous manual stirring on a daily basis.

Once the digestive procedure had finished, the resulting extract of each replicate was diluted in water at room temperature in the proportion 1:10 (v/v). It was then filtered with a Millipore vacuum pump through a glass-fibre filter (GF/C Whatman) with a 0.7 µm pore diameter. Once the filtration procedure was finished, each filter was individually preserved in a Petri dish until further analysis.

Microplastics were identified and counted with a Leica Zeiss AxioCam ERc 5s stereomicroscope. The procedure used for identifying suspected particles as plastic were 1) absence of cellular or organic structures, 2) having a constant thickness along their length, 3) having a uniform colour, and 4) consistency when poked with a punch (Hidalgo-Ruz et al. 2012).

An image of each microplastic was taken by means of ZEN software. Observed microplastic particles were classified according to their morphology, colour and length, as suggested by Galgani et al. (2013).

To avoid contamination of samples, several measures were taken, such as carrying out the digestive procedure in a laminar flow cabinet, avoiding the use of plastic (e.g. nitrile globes, glass beakers and metal spatulas), and covering the samples with clock glass (acid digestion) or aluminium foil (alkaline digestion and filtration procedure).

Moreover, in order to estimate contamination, in each experiment a procedural blank (a replicate without mussel tissue) and a control blank (a Petri dish with a GF/C filter open in the laboratory) were carried out in parallel to three replicates. The procedural blanks contained the same volume of digestive material as that which would be used to digest 1 g of mussel tissue and comprised the whole analytical method, while the control blanks only took into account airborne contamination during the count. The microplastic quantity observed in each individual mussel (hereinafter, MPs/individual) was corrected by excluding from counting those items that were similar in morphotype and colour to the ones present in the procedural blanks (similar to Digka et al. 2018).

In the statistical analysis concentration of microplastics (microplastics per gram of wet weight, hereinafter, MPs g^–1 WW) was used as variable. Normality of data was tested by means of the Shapiro-Wilk normality test. Homogeneity of variances was tested by means of the Bartlett test. Because of the lack of normality of data, the existence of significant differences between digestive treatments and between geographical regions was tested by means of the Mann-Whitney test (equivalent to the Wilcoxon rank sum test). The existence of significant differences among sampling stations was tested by means of the Kruskal-Wallis test. Pairwise comparisons were carried out by means of the Conover test (a non-parametric equivalent to the Bonferroni test). The existence of significant differences in the proportion in which each microplastic type was present in the two digestive treatments and the two study regions was tested by means of a chi-squared test. All statistical analysis was carried out using R software (R Core Team 2017).

The wet weight of the studied mussels ranged between 0.59 and 4.22 g, with an average of 1.64±0.82 g. In the mussel replicates, observed microplastic quantity per filter ranged between 0 and 14 MPs filter^–1, with an average of 3.05±2.67 MPs filter^–1.

In the procedural blanks, contamination ranged between 0 and 19 MPs filter^–1, with an average of 4.29±4.68 MPs filter^–1 in the digestive treatment with KOH and 1.93±1.77 MPs filter^–1 in the one with HNO₃. Both in the procedural blanks of the digestive treatment with KOH and in those of the one with HNO₃ a high proportion of fibres was observed (68% and 63% of total microplastics, respectively), while fragments were the second most abundant morphotype (24% and 30%, respectively). Regarding the colour, blue microplastics were the most abundant both in the alkaline and acid procedural blanks (37% and 30%, respectively), followed by yellow (15%) in the alkaline procedural blanks and by white (19%) in the acid procedural blanks. In the control blanks, the number of microplastics observed in each blank ranged between 1 and 11 MPs filter^–1, with an average of 4.79±4.06 MPs filter^–1. The most abundant morphotype were fibres (78%), followed by fragments (18%). The most frequent colour was blue (40%), followed by yellow (22%).

Average concentrations of microplastics observed in mussels digested with each digestive agent are shown in Figure 2. The average concentration of microplastics in mussels digested with HNO₃ was 1.22±1.42 MPs g^–1 WW. In mussels digested with KOH, the average concentration of microplastics was 2.07±2.21 MPs g^–1 WW. The microplastic concentration observed in mussels digested with HNO₃ was significantly lower than that observed in those digested with KOH (p<0.05).

The proportion of fibres in the digestive treatment with HNO₃ (52%) was higher than the one observed in the digestive treatment with KOH (44%). However, pellets were present in a greater proportion in the digestive treatment with KOH (23%) than in the digestive treatment with HNO₃ (4%). The differences in the proportions in which each morphotype was present in the two digestive treatments were significant (p<0.05).

Regarding microplastic colour, no yellow microplastics were observed in the digestive treatment with KOH, while green and orange microplastics (tending-to-yellow colours) accounted for 5% and 2% of total microplastics, respectively. In the digestive treatment with HNO₃, yellow microplastics were present in low proportion (1%), similar to that for green and orange microplastics (3% and 1%). A higher proportion of black microplastics was seen in the digestive treatment with HNO₃ (48%) than in the digestive treatment with KOH (11%). The differences between the two digestive treatments with regard to the proportion in which microplastics of each colour were present were significant (p<0.05).

Similar proportions of microplastics in each length range were seen in the two digestive treatments, the most frequent length range of microplastics being 200-500 µm both in the one with KOH and the one with HNO₃ (38% and 31% of observed microplastics, respectively). The highest length range (1000-5000 µm) and the lowest one were present in a lower proportion in the digestive treatment with HNO₃ than in the one with KOH. However, the differences between the two digestive treatments with regard to the proportion in which each length range was found were not significant (p>0.05).

To analyse the geographical pattern of pollution, only the results obtained from the digestive treatment with KOH were used. Microplastics were observed in 88% of the studied mussels. Microplastic quantity individually observed ranged between 0 and 10 MPs ind.^–1, with an average value of 2.19±1.57 MPs ind.^–1 in the Ria of Vigo and 2.81±2.80 MPs ind.^–1 in the Cantabrian Sea. The microplastic concentration observed in each mussel ranged between 0 and 8.90 MPs g^–1 WW. The average concentration of microplastics in the Ria of Vigo was 1.59±1.28 MPs g^–1 WW, while in the Cantabrian Sea it was 2.55±2.80 MPs g^–1 WW (Fig. 3). The difference between the two regions with regard to microplastic concentration was not significant (p>0.05).

The variation of microplastic concentration among the sampling stations of the Ria of Vigo is shown in Figure 4A, while the variation among stations of the Cantabrian Sea is shown in Figure 4B. In the Ria of Vigo, the highest average concentration of microplastics was observed at A Guía, with 2.48±1.00 MPs g^–1 WW, while the lowest was observed at San Adrián, with 0.42±0.14 MPs g^–1 WW. No significant differences among stations were observed in the Ria of Vigo (p>0.05). In the Cantabrian Sea, the highest average concentration of microplastics was observed at Santander, with 7.57±1.16 MPs g^–1 WW, while the lowest concentration was observed at Avilés, with 0.15±0.16 MPs g^–1 WW. In the Cantabrian Sea, significant differences were only found between these two stations (p<0.05).

As in the previous case, to analyse the physical characteristics of observed microplastics in each region, only the results from mussels digested with KOH were used. In the Ria of Vigo the most common morphotype (Fig. 5A) were fibres (56% of observed microplastics), followed by fragments (33%), and pellets (9%). In the Cantabrian Sea (Fig. 5B), fibres and pellets were the two main morphotypes (34% each one), followed by fragments (30%).

Regarding the colour of observed microplastics (Fig. 6), the most frequent ones in the Ria of Vigo were both white and grey (accounting for 22% of total microplastics each), followed by blue and red (15% each). In the Cantabrian Sea, white was also the most frequent microplastic colour (36%), followed by blue (17%), and grey (13%).

The proportions in which microplastics of each length range were observed in mussels from the two regions are shown in Figure 7. The most abundant length range of extracted microplastics in both the Ria of Vigo and the Cantabrian Sea was 200-500 µm (37% and 39%, respectively). The second most frequent length range was 500-1000 µm in the Ria of Vigo (24%) and 100-200 µm in the Cantabrian Sea (19%).

The microplastic quantity observed in the procedural blanks (4.29±4.68 MPs filter^–1 for the digestive treatment with KOH and 1.93±1.77 MPs filter^–1 for the one with HNO₃) was lower than the that observed in the procedural blanks by Catarino et al. (2018), who observed 6.5±0.95 particles filter^–1, and higher than that observed in other studies, in which 1.02 particles per filter (Bråte et al. 2018), 1 particle filter^–1 (Karlsson et al. 2017) and 0.67±0.82 particles filter^–1 (Li et al. 2016) were observed. Therefore, blank contamination is in the range of that observed in previous studies.

In the present study a lower concentration of microplastics was observed in mussels digested with HNO₃ than in those digested with KOH (Fig. 2). In previous studies it has been reported that HNO₃ exerts a destructive effect on microplastics when they are exposed to it in the laboratory, either directly (Claessens et al. 2013, Dehaut et al. 2016) or spiked in biological samples (Karami et al. 2017). Consequently, it has been suggested that the use of HNO₃ as a digestive agent induces an underestimation of the microplastic concentration observed in biota samples. The results obtained in the present study confirm this hypothesis.

In a study on recovering microplastics previously spiked in mussel tissue, it was observed that fibres were more sensitive to the destructive effect of HNO₃ than other morphotypes such as pellets due to their higher surface/volume ratio (Claessens et al. 2013). Contrarily to expected, the proportion of fibres in mussels digested with HNO₃ was higher than that in mussels digested with KOH, and the opposite situation happened in the case of pellets. Additionally to their higher surface/volume ratio, Claessens et al. (2013) attributed the greater destructive effect of HNO₃ on fibres observed in their study to the polymer plastic type that composed the tested morphotypes (mainly nylon in the case of fibres and polypropylene in that of pellets), coming to consider it a most probable reason. As in the present study analytical determination of plastic type was not carried out, it is not possible to draw conclusions about the reason why the proportions of fibres and pellets differed from the expected ones.

It has also been observed in the laboratory that HNO₃ induces a yellowing effect in microplastic colour (Claessens et al. 2013). No yellow microplastics were seen in the digestive treatment with KOH, but they were observed in the one with HNO₃, where they accounted for 1% of total microplastics. By contrast, green and orange microplastics were present in a higher proportion in the alkaline treatment (5% and 2%, respectively) than in the acid treatment (3% and 1%, respectively). Unexpectedly, a very high proportion of black microplastics (48%) was observed in the digestive treatment with HNO₃. Since in the acid procedural blanks black microplastics were only 4% of total microplastics, this finding cannot be attributed to contamination of samples.

The average microplastic concentration observed in the Cantabrian Sea (2.55±2.80 MPs g^–1 WW) was higher than that observed in the Ria of Vigo (1.59±1.28 MPs g^–1 WW), as shown in Figure 3. In a previous study carried out on the north coast of Spain it was reported that average levels of pollutants in mussels were higher in the Cantabrian Sea than in the northwestern Spanish coast, where the Ria of Vigo is placed (Bellas et al. 2014). Even so, the difference between the two regions observed in the present study was not significant. The lower intensity of upwelling processes in the Cantabrian Sea than on the western Galician coast (Díez et al. 2000), where they tend to induce the exit of surface waters from the Rias and their replacement by deep-waters from the open ocean, could explain why microplastic levels were higher in the Cantabrian Sea than in the Ria of Vigo.

A review of microplastic levels (concentration and individual quantity) reported in geographical areas worldwide is provided in Table 2. Average microplastic concentrations observed in both the Ria of Vigo and the Cantabrian Sea are in the range of those observed in the United Kingdom, both on the west coast of Scotland (Catarino et al. 2018) and along the coasts of England and Wales (Li et al. 2018). They are also similar to those observed along the Chinese coast (Li et al. 2016, Qu et al. 2018). Comparison with studies carried out in the Netherlands is harder, because their results are presented as a function of dry weight. Nevertheless, the concentrations reported there seem higher than the ones from both the Ria of Vigo and the Cantabrian Sea (Karlsson et al. 2017, Leslie et al. 2017). In contrast, lower levels than in the present study were reported on the French Atlantic coast (Phuong et al. 2018), along the coast of Norway (Bråte et al. 2018), and in the southwest of Finland (Railo et al. 2018). Microplastic concentrations from both the Ria of Vigo and the Cantabrian Sea are higher than the ones observed on the Belgian coast (De Witte et al. 2014, Van Cauwenberghe et al. 2015), in some parts of the Mediterranean Sea (in the Ebro estuary and in the Po estuary) and in the Tagus Estuary, in Portugal (Vandermeersch et al. 2015). In the case of these studies, the low levels observed can be attributed, at least partially, to the use of acid compounds as digestive agents, which induces an underestimation in microplastic levels, as reported in the present study.

As in other studies, a high variability in microplastic concentrations among individuals was observed, which is attributable to the effect of biological factors influencing the individual accumulation of pollutants, such as body size, sex and lipid content (Newman 2010). In order to minimize the effect of natural variability in biological factors, OSPAR (2012) recommends analysing a pool of 20 individuals by replicate. Thus, the inter-individual variability observed in this study could be avoided in future studies by analysing a higher number of individuals.

Regarding microplastic quantity per individual (see Table 2), the ones observed in this study (2.19±1.57 MPs ind.^–1 in the Ria of Vigo and 2.81±2.80 MPs ind.^–1 in the Cantabrian Sea) were lower than those observed on the west coast of Scotland (Catarino et al. 2018) and higher than those observed on the Greek coast of the Ionian Sea (Digka et al. 2018), on the Norwegian coast (Bråte et al. 2018) and in the southwest of Finland (Railo et al. 2018). Comparison with other results is more complicated because the above studies inferred microplastic quantity per individual from that observed in a pool of various mussels. Microplastic quantity per individual observed both in the Ria of Vigo and the Cantabrian Sea coast is also in the range observed on the coast of China (Li et al. 2016, Qu et al. 2018). Quantities in the range of this study were also reported on the coast of England and Wales (Li et al. 2018). However, the ones observed in our study are higher than those reported on the French Atlantic coast (Phuong et al. 2018).