**Hg Levels in Marine Porifera of Montecristo and Giglio Islands (Tuscan Archipelago, Italy)**

**Camilla Roveta \*, Daniela Pica, Barbara Calcinai, Federico Girolametti, Cristina Truzzi, Silvia Illuminati, Anna Annibaldi \*and Stefania Puce**

Department of Life and Environmental Sciences, Polytechnic University of Marche, Via Brecce Bianche, 60131 Ancona, Italy; daniela.pica@gmail.com (D.P.); b.calcinai@staff.univpm.it (B.C.); f.girolametti@pm.univpm.it (F.G.); c.truzzi@staff.univpm.it (C.T.); s.illuminati@staff.univpm.it (S.I.); s.puce@staff.univpm.it (S.P.)

**\*** Correspondence: c.roveta@pm.univpm.it (C.R.); a.annibaldi@staff.univpm.it (A.A.); Tel.: +39-0712204649 (C.R.); +39-0712204981 (A.A.)

Received: 29 May 2020; Accepted: 23 June 2020; Published: 24 June 2020

**Abstract:** Porifera are filter-feeding organisms known to bioaccumulate different contaminants in their tissues. The presence of mercury (Hg) has been reported in different Mediterranean species, mainly collected in the southern coast of France. In the present study, mercury concentrations in the tissue of the sponges of Montecristo and Giglio, two islands of Tuscany Archipelago National Park (TANP), are presented for the first time. Analyses of total mercury content were performed by Direct Mercury Analyzer. Statistical differences have been reported in the Hg concentrations of species collected in both islands, but they do not appear related to the anthropic impacts of the islands. Among the collected species, a high intra- and inter-variability have been recorded, with *Cliona viridis* showing the lowest concentration (0.0167–0.033 mg·kg−<sup>1</sup> dry weight), and *Chondrosia reniformis* and *Sarcotragus spinosulus* the highest (0.57 <sup>±</sup> 0.15 and 0.64 <sup>±</sup> 0.01 mg·kg−<sup>1</sup> dry weight, respectively). The variability of Hg measured did not allow us to identify sponges as bioindicators of toxic elements. Anyway, these results improve knowledge on the ecosystem of the TANP, underlining the species-specificity of metal concentrations for Porifera, and providing additional data to address the main input of the Marine Strategy guidelines to protect coasts, seas and oceans.

**Keywords:** toxic element; Mediterranean Sea; sponges; biomonitoring

#### **1. Introduction**

Mercury (Hg) is considered to be one of the most toxic heavy metals due to its persistence in the environment, bioaccumulation in organisms and biomagnification in the trophic chain [1,2]. Atmospheric inputs of Hg have tripled during the last 150 years, being two-thirds of its actual concentration from anthropogenic sources [2,3]. Traces of mercury have been found in all the compartments of the ecosphere (atmosphere, hydrosphere, lithosphere and biosphere) [4–6]. In aquatic environments, it is transformed by chemical and biological reactions in organomercury compounds, as methylmercury (MeHg), the most toxic mercury species, which can be bioaccumulated more than other trace elements along the trophic chain [6]. Mercury can have many different effects on a wide range of organisms, both vertebrates, e.g., References [7–9], and invertebrates, e.g., References [10–12]. For all these reasons, Hg is listed in the European Water Framework Directive (WFD 2000/60/EC) as a priority substance representing a risk for the good chemical status for the aquatic environment, with a Maximum Allowable Concentration (MAC) of 0.07 μg. L<sup>−</sup>1.

The presence of mercury in the Mediterranean Sea has been documented since the 1970s [13,14]. The principal mercury input (94%) was recognized in rivers' dischargements, while only 5.5% was related to direct industrial wastewater and 0.5% to domestic sewage [15]. The Mediterranean area

presents a high number of natural deposits of mercury distributed along the coasts of many countries, containing about 65% of the world's cinnabar (HgS) deposits [15,16]. Particularly, in Tuscany (Italy), the levels of Hg in different environmental matrices are derived from both a natural contribution of the mineralization and the pollution caused by the huge exploitation of the area of Mount Amiata [17]. In fact, Mount Amiata, located in the south of Tuscany, is part of the geologic anomaly of the Mediterranean basin and it is characterized by a large cinnabar deposit [16,18]. Moreover, it is well known from ARPAT (acronym for Regional Agency for Environmental Protection of Tuscany) and ISPRA (acronym for Higher Institute for Environmental Protection and Research) technical reports, and scientific papers [19], the presence of mercury in the waters and organisms of the Tuscany coast and island. From 2012 to 2017, high concentrations of mercury have been reported in water, sediment and biota (as *Posidonia oceanica*, *Mytilus galloprovincialis* and different fish species) in the Tuscan Archipelago islands, e.g., References [20–24].

Many studies have been conducted worldwide on the bioaccumulation and the effects of mercury and its organic compounds in edible species, such as bivalves, cephalopods, decapods crustaceans and fish [6,25–27]. Although, other filter-feeding organisms, such as polychaetes, tunicates, sponges and barnacles, have been proposed as bioindicators in shallow waters [28], but these taxa are still little used for this purpose. Especially, sponges satisfy all the characteristics listed in References [29,30] for a suitable bioindicator and have been recommended by many authors, e.g., References [31,32], and by the WFD as possible monitors for heavy metals. Being filter-feeders, sponges can filter a large amount of water, and they can collect and accumulate many different contaminants (such as hydrocarbons, organochlorinated compounds, heavy metals, etc.) in their tissues, which are present in the water column both in the soluble and particulate phases [10,28]. Moreover, the level of the bioaccumulation in their tissues is a function of the contaminants' concentration in the water [33]. The presence of heavy metals in sponges can affect their physiology and survival [28,34,35]. Only a few authors have investigated the presence of Hg in sponges and most of the studies have been conducted on samples, mainly on the genus *Spongia,* collected in the coast of Marseille [10,36–38].

The current paper presents, for the first time, total mercury content in the tissue of the sponges of Montecristo and Giglio, two islands of Tuscany Archipelago National Park (TANP). The study addresses two hypotheses: (1) sponges collected show detectable total mercury contents, and (2) there are any differences in Hg concentrations between the sponge species collected in both studied islands. Our results point out that sponges have detectable Hg concentrations, showing a high inter- and intra-specific variability. This variability could also be responsible of the lack of trend in the Hg concentrations between the specimens collected from the more anthropic Giglio and the Integral Reserve of Montecristo. Anyway, the results obtained with statistical analyses give new important insights for the area on the differences among Hg concentration in sponges.

#### **2. Materials and Methods**

#### *2.1. Study Area*

The study was carried out in June 2019 in Montecristo and Giglio islands (Tuscany, Italy) (Figure 1). The Montecristo Island (42.3317◦ N, 10.3083◦ E) (Figure 1B) is an uninhabited and isolated island, sub-circular in shape, fourth in size after Elba, Giglio and Capraia islands, and completely mountainous, reaching a height of 645 m above sea-level with an almost constant slope of 25◦ [39–42]. Its history and geography distinguish the island from the others: during the Quaternary period, Montecristo remained in contact with the Tuscany's littoral for a shorter period than the other islands, ending up located 63 km from the mainland, on the limit of the continental shelf, closer to Corsica than to the mainland [39].

**Figure 1.** Map of the Tuscan Archipelago National Park (TANP) (**A**), Montecristo Island (**B**) and Giglio Island (**C**). The sampling sites are indicated in the islands' maps.

In 1971, Montecristo was established as an Integral Nature Reserve, and in 1988, was declared a biogenetic nature reserve by the Council of Europe. Moreover, the island has the status of Special Protection Area (SPA, Directive 79/409/EEC). The land and the adjacent waters, up to 1 km offshore, are controlled by the Coast Guard and the Carabinieri Corps [41,43]. Bathing, diving, fishing, mooring and circumnavigating are forbidden, while landing, berthing and scientific activities are allowed, only under specific conditions and with the permission of the Territorial Office for Biodiversity of the Carabinieri Corps of Follonica [41]. Due to the high protection which Montecristo has undergone, the island is considered one of the most pristine and best-preserved sites of the Mediterranean Sea, being previously identified as a reference site for the ecological quality assessment of the western Mediterranean benthic assemblages on rocky bottoms [44,45].

The Giglio Island (42.3603◦ N, 10.9229◦ E) (Figure 1C), being the second in size after Elba, is included in the southern group of the Archipelago, only 14 km from the Italian coast [42,46]. It is an oval-shaped island, characterized by a mountain chain along the north–south axis, and it reaches a height of 496 m above the sea-level. Contrary to Montecristo, the island hosts 1500 residents, spread out among the villages of Giglio Castello, Giglio Porto and Campese [42]. Only 40% of the island is included inside the protected area, while adjacent waters and many terrestrial areas do not have any protection (DPR 22 July 1996) [47]. The areas included in the legislation of the TANP are divided into four zones: (1) zone A of strict reserve, including five rocks (Cappa, Corvo, Mezzo Franco, Pietra Bona, and Le Scole), (2) zone B of general reserve, (3) zone C of general protection and (4) zone D of socio-economic promotion [42]. In general, Giglio is considered to have undergone anthropogenic pressures [47], due to the high flow of tourists during Summer [42], and to be more exposed to potential stress deriving from the mainland.

#### *2.2. Samples Collection and Identification*

Surveys were conducted by SCUBA diving and eighteen sponge samples were collected between 5 and 40 m, nine on the hard bottom assemblages of Punta del Diavolo (42◦21 02.46" N; 10◦17 55.86" E) in the Montecristo Island (TANP permission #00068010) (Figure 1B), and nine between Punta del Fenaio (42◦23 21.54" N; 10◦52 48.18" E) and Scoglio del Corvo (42◦20 17.76" N; 10◦53 21.36" E) in the Giglio Island (Figure 1C). The low number of samples for this preliminary study is strictly subordinated to protect and respect the island ecosystem, which needs to be protected even in the case of scientific research.

Samples were immediately frozen on dry ice, and then stored at −20 ◦C until analysis.

Among the samples, fourteen species of Demospongiae have been identified: *Agelas oroides* (Schmidt, 1864)*, Axinella damicornis* (Esper, 1794)*, Cliona viridis* (Schmidt, 1862)*, Haliclona* (*Halichoclona*) *fulva* (Topsent, 1893)*, Haliclona* (*Soestella*) *mucosa* (Griessinger, 1971)*, Penares euastrum* (Schmidt, 1868) and *Sarcotragus spinosulus* Schmidt, 1862 in Montecristo, and *Chondrosia reniformis* Nardo, 1847*, C. viridis, Crambe crambe* (Schmidt, 1862)*, H.* (*H.*) *fulva, Hemimycale columella* (Bowerbank, 1874)*, Hymedesmia* (*Hymedesmia*) *baculifera* (Topsent, 1901) and *Petrosia* (*Petrosia*) *ficiformis* (Poiret, 1789) in Giglio. For the species *P. euastrum* and *C. reniformis*, three samples each were collected; therefore, we denominated them with the name of the species followed by the number 1, 2 or 3 between brackets.

#### *2.3. Samples Treatment and Mercury Analysis*

A clean room laboratory ISO 14644-1 Class 6, with areas at ISO Class 5 under laminar flow, was used for all laboratory activities. After the identification, samples were weighted (laboratory analytical balance, AT261 Mettler Toledo Greifensee, Switzerland, readability 0.01 mg, repeatability standard deviation (SD) = 0.015 mg) and cleaned with ultrapure water (A10 Milli-Q system, Merk Millipore, Bedford, MA, USA). The acid-cleaning procedures, used for all the laboratory materials, were performed as described in References [48,49].

After cutting samples into small pieces, sponges were lyophilized (Edwards EF4 modulyo, Crawley, Sussex, England), minced, homogenized and divided in aliquots of about 0.02 g each. Analyses of total mercury content (THg) were performed by Direct Mercury Analyzer (DMA-1 Milestone, Sorisole (BG), Italy), as described in Reference [50]. Briefly, the total mercury content was quantified by thermal decomposition amalgamation atomic absorption spectrometry at 253.7 nm. The calibration curve method was used for the quantification of Hg content. All measurements were replicated at least 4 times.

#### *2.4. Accuracy*

Quality assurance and quality control were assessed by processing blank samples and certified reference material (dogfish muscle DORM-2, NRCC; Ottawa, ON, Canada). The experimental values obtained for Hg in blanks are negligible compared with the metal content in sponge tissue (<1%). For DORM-2 analysis (n = 8), Hg content (4.58 <sup>±</sup> 0.10 mg·kg−1) is in agreement with the certified value (4.43 <sup>±</sup> 0.05 mg·kg<sup>−</sup>1) and no statistically significant differences were observed (*p*-value > 0.05, Student's T test, STATGRAPHICS 18 Centurion, 2018).

#### *2.5. Data Analyses*

Data are expressed as arithmetic mean ± standard deviation (SD) of the performed replications. Statistical analyses of differences within organisms were performed using the analysis of variance (one-way ANOVA) after testing the homogeneity of the variance with Levene's test [51]. In case of heteroscedasticity, we applied the non-parametric Kruskal–Wallis analysis of variance. Depending on the resulting statistics, post-hoc comparison was eventually performed with the Bonferroni correction, always considering a significant level of 0.05. All graphs and statistical analyses were performed using STATGRAPHICS (STATGRAPHICS Centurion 2018, Statgraphics Technologies Inc., The Plains, VA, USA).

#### **3. Results**

THg, expressed on a dry weight (dw) basis, in sponges collected in the area of Montecristo and Giglio islands, are reported in Figure 2A,B, respectively.

**Figure 2.** Mercury levels (± standard deviation (SD)) in the sponges collected in Montecristo (**A**) and Giglio (**B**) islands. In each graph, different letters (a, b, c) indicate statistically significant differences among sponges (Kruskal–Wallis test, *p* < 0.05). Ad = *Axinella damicornis*; Ao = *Agelas oroides*; Cv = *Cliona viridis*; Hf = *Haliclona* (*Halichoclona*) *fulva*; Hm = *Haliclona* (*Soestella*) *mucosa*; Pe = *Penares euastrum*; Ss = *Sarcotragus spinosulus*; Cc = *Crambe crambe*; Cr = *Chondrosia reniformis*; Hb = *Hymedesmia* (*Hymedesmia*) *baculifera*; Hc = *Hemimycale columella*; Pf = *Petrosia* (*Petrosia*) *ficiformis*.

Total mercury showed high variability in sponges coming from the marine area of Montecristo island, ranging from 0.033 to 0.64 mg·kg−1, with the minimum content in *Cliona viridis* (0.033 <sup>±</sup> 0.001 mg·kg<sup>−</sup>1) and the maximum (20-fold higher) in *Sarcotragus spinosulus* (0.64 <sup>±</sup> 0.01 mg·kg<sup>−</sup>1). Values ranging from about 0.11 to 0.29 mg·kg−<sup>1</sup> were recorded in *Agelas oroides, Axinella damicornis, Haliclona (Halichoclona)* *fulva* and *H. (Soestella) mucosa*, whereas slightly higher values were found for the three samples of *Penares euastrum* (1), (2) and (3) (0.35 <sup>±</sup> 0.03 mg·kg<sup>−</sup>1)*,* that showed similar concentration among them. Statistical analysis of the data showed a significant difference (Kruskal–Wallis test, *p* < 0.05) between *C. viridis* vs. *P. euastrum* (3) and *S. spinosulus*, *H.* (*S.*) *mucosa* vs. *P. euastrum* (3) and *S. spinosulus* (Figure 2A).

A high variability of THg content was also found in sponges collected in the marine area of Giglio Island: total Hg ranged from 0.017 to 0.73 mg·kg−<sup>1</sup> dw. Even in this case, the minimum content was measured in *C. viridis* (0.0167 <sup>±</sup> 0.0003 mg·kg−1), which showed values from 10 to 70 times lower than other species. On the other hand, the three samples of *Chondrosia reniformis* (1), (2) and (3) showed the highest concentrations (ranging from 0.39 to 0.73 mg·kg−1). In the other sponges, the THg content ranged from <sup>≈</sup>0.1 (*Crambe crambe, Hemimycale columella*) to about 0.22 mg·kg−<sup>1</sup> in *H. (H.) fulva, Hymedesmia baculifera* and *Petrosia (P.) ficiformis.* The Kruskal–Wallis test highlighted significant differences (*p* < 0.05) between *C. reniformis* (1) vs. *C. crambe*, *C. viridis* vs. *C. reniformis* (1) and *C. reniformis* (3) (Figure 2B).

Since samples of *C. viridis* and *H. (H.) fulva* were found in both islands, differences in the THg content between the two islands have been investigated, to compare the possible influence of different sites. *C. viridis* showed a higher THg in Montecristo island (Kruskal–Wallis test, *p* < 0.05), while the concentration of mercury in *H. (H.) fulva* was higher in Giglio (Kruskal–Wallis test, *p* < 0.05).

#### **4. Discussion**

Samples show the presence of mercury both in Montecristo and Giglio. Although statistical differences in the Hg concentrations have been detected in the sponge species (*Cliona viridis* and *Haliclona (H.) fulva*) collected in both islands of the Tuscan Archipelago, the lack of a trend excludes a possible influence of the site. In fact, even though since 1971 Montecristo has been an Integral Nature Reserve considered a pristine area not interested by mechanic impacts, such as anchoring and diving disturbances, fishing, etc., its water can still be affected by other different impacts, such as climate change and alien species, e.g., References [52,53], and the presence of different pollutants [20–23] due to water circulation [54].

Currently, no law limit is defined for Hg in sponges and few data are available on Hg concentration in this phylum, most of which regard species and genera different from those collected in this study, e.g., References [28,55,56]. THg values of *Sarcotragus spinosulus,* measured by us, are comparable with the concentrations found in other species belonging to the subclass Keratosa, characterized by a skeleton of spongin fibers [57] such as *Scalarispongia scalaris* (Schmidt, 1862) (cited as *Cacospongia scalaris*) [37], *Spongia (Spongia) lamella* (Schulze, 1879) (cited as *S. agaricina*) [36,37], *S. (S.) nitens* (Schmidt, 1862) (cited as *S. nitens*) [36] and *S. (S.) o*ffi*cinalis* (Linnaeus, 1759) (cited as *S. o*ffi*cinalis*) [10,36–38] (Table 1). THg values of *Chondrosia reniformis* recorded in specimens from Montecristo and Giglio are similar to the ones recorded in Reference [37] for the same species collected in different localities near Marseille (Table 1). On the other hand, results obtained for *Agelas oroides* and *Cliona viridis* in Reference [37] are higher than our THg values for both species (Table 1). Perez et al. [37] collected samples of *Cliona viridis* in polluted and non-polluted areas, and he did not find any differences in the Hg concentrations. Therefore, also the difference reported between our specimen and the one of Reference [37] could be related not only to the species-specificity but also to the individual specificity of sponges, with a high intra- and inter-specific variability [58,59].

Although it is known that some heavy metals (e.g., copper, lead and vanadium) have effects on sponges, increasing their fission frequency, inducing changes in cellular aggregation and reducing growth and filtration rates [34,35], no information is available for Hg. However, some studies showed, from laboratory experiments, that mercury can cause death, inhibition of gemmule formation and malformation in gemmoscleres in the freshwater sponge *Ephydatia fluviatilis* (Linnaeus, 1759) (0.001 to 1.000 mg·kg<sup>−</sup>1) [60], and it can arrest movement of single sponge cells of *Scopalina lophyropoda*

(Schmidt, 1862) (1 and 5 <sup>μ</sup>g·kg<sup>−</sup>1), which tended to be rounded without pseudopodia [61], while a high concentration of MeHg (0.6 mg·kg<sup>−</sup>1) induces apoptosis in tissue of *Geodia cydonium* (Linnaeus, 1767) [62].


**Table 1.** Selection of literature data for mercury concentrations in sponges.

The highest THg values were found in *Chondrosia reniformis* and *Sarcotragus spinosulus*, collected in Giglio and Montecristo islands, respectively. These two species share the absence of a mineral skeleton, showing *C. reniformis*, a dispersal fibrillary collagen, and *S. spinosulus*, an organic skeleton made of spongin [57]. Three papers on Antarctic and Mediterranean species [59,63,64] showed that spicules accumulate only a small part of heavy metals compared to the sponge tissues; while in *Spongia* spp., the skeletal spongin fibers can trap and consequently concentrate metals, such as Fe, Pb, Cr, Zn and V [37,65]. These studies suggest that a collagenous skeleton instead of a mineral one can bioaccumulate a higher concentration of heavy metals. On the other hand, the mercury concentrations recorded for *C. viridis*, which is the only boring species and symbiotic with zooxanthellae [66], were the lowest among all the species collected. It is possible that the association with zooxanthellae influences the accumulation capacity of the sponge, as also pointed out in Reference [37]. It has been demonstrated that *C. viridis* can vary its filtration rates depending on the photosynthetic activities of its zooxanthellae [67], and this could partially explain why our samples present low concentrations of THg in their tissues.

The aforementioned intra- and inter-specific variability suggested by our observations and other authors [37,38] (Table 1) appear to not always be in agreement with the distance from the main source of pollutants. This can be a result of physiological and skeletal differences among sponge species. For example, the diversity of morphology of the aquiferous system could influence the size range of particles that can be filtered as well as the filtration rates [37]. Moreover, sponges are long-living organisms, with variable rate of growth, and it is very difficult to age each individual [36], therefore it is possible that low concentrations of heavy metals found in one organism could be related to its young

age. In different sponge species, *Clathria (Clathria) prolifera* (Ellis and Solander, 1786) (cited as *Microciona prolifera*) [68], *Suberites domuncula* (Olivi, 1792) [69] and *Spongia (Spongia) o*ffi*cinalis* [10], authors found the presence of metallothioneins (MT) and metallothionein-like proteins (MTLPs), which are known to be used in the sequestration of metals in some invertebrates [70,71], and a positive correlation between MTLP concentrations and some heavy metals' (Cu, Zn and Hg) concentrations has been established [10].

Moreover, sponges can host complex communities of microorganisms, including bacteria, cyanobacteria and fungi, in their tissues [72]. The relationships between sponges and their microbiota can be defined as a symbiosis, in which microorganism communities can take part in the metabolic cycles and exchange different metabolites with their host [73]. It has been demonstrated that many bacteria, which can contribute up to 40% of the sponge biomass, are resistant to different antibiotics and pollutants, including Persistent Organic Pollutants (POPs) and heavy metals, such as Cu, Pb, Co, Cd, Zn, Ni, Hg and their organic compounds, e.g., References [72,74–76]. Moreover, some studies asserted that sponges and the associated bacteria can potentially be applied in the bioremediation of aquatic environments contaminated by mercury [74,75]. Therefore, the microbiota also seems to play an important role in the bioaccumulation of heavy metals in sponge tissues, suggesting the necessity of deepening the investigation towards this aspect.

#### **5. Conclusions**

In conclusion, our results are the first data about total Hg concentration in sponges from the TANP and suggest that these metazoa could accumulate toxic elements in coastal waters. On the other hand, the high variability of concentrations in THg measured in all specimens in both islands did not allow us to identify sponges as bioindicators of toxic elements. However, further studies with a higher number of sponge samples are needed to understand in which compartment (skeleton, tissue or microbiota) they accumulate the heaviest metals, analyzing the main effects on the histology and physiology of this group. Anyway, these results improve the knowledge on the ecosystem of the TANP, pointing out the species/individual-specificity of metal concentrations for Porifera and the key role of the organic skeleton, tissue and microbiota. Moreover, these data provide additional environmental information on the Tuscany Archipelago to address the main input of international guidelines on the Marine Strategy to protect and clean up coasts, seas and oceans.

**Author Contributions:** Conceptualization, A.A. and S.P.; Formal analysis, A.A.; Funding acquisition, S.P.; Investigation, C.R. and F.G.; Project administration, S.P.; Resources, C.R. and D.P.; Supervision, B.C.; Writing—original draft, C.R. and A.A.; Writing—review and editing, D.P., B.C., C.T., S.I. and S.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study has been conducted with the financial support of PADI FOUNDATION (grant number #32694) and Università Politecnica delle Marche.

**Acknowledgments:** The authors are grateful to Isla Negra diving, for the logistic assistance during samplings, and to the Tuscan Archipelago National Park, for the authorization of sampling activities.

**Conflicts of Interest:** The authors declare they have no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Methylmercury and Polycyclic Aromatic Hydrocarbons in Mediterranean Seafood: A Molecular Anthropological Perspective**

**Andrea De Giovanni 1,2,\*, Cristina Giuliani 3, Mauro Marini 2,4 and Donata Luiselli 1,2**


**Abstract:** Eating seafood has numerous health benefits; however, it constitutes one of the main sources of exposure to several harmful environmental pollutants, both of anthropogenic and natural origin. Among these, methylmercury and polycyclic aromatic hydrocarbons give rise to concerns related to their possible effects on human biology. In the present review, we summarize the results of epidemiological investigations on the genetic component of individual susceptibility to methylmercury and polycyclic aromatic hydrocarbons exposure in humans, and on the effects that these two pollutants have on human epigenetic profiles (DNA methylation). Then, we provide evidence that Mediterranean coastal communities represent an informative case study to investigate the potential impact of methylmercury and polycyclic aromatic hydrocarbons on the human genome and epigenome, since they are characterized by a traditionally high local seafood consumption, and given the characteristics that render the Mediterranean Sea particularly polluted. Finally, we discuss the challenges of a molecular anthropological approach to this topic.

**Keywords:**review; DNA methylation; genetic polymorphisms; ecogenetics; anthropology; environmental pollutants; methylmercury; polycyclic aromatic hydrocarbons; seafood

#### **1. Introduction**

Despite being usually considered a healthy food [1], seafood carries several contaminants that can negatively affect human health [2]. It is recognized that the benefits of fish intake exceed the potential risks, but here we address how contaminants levels in seafood are significantly affected by biological and ecological factors [3–8]. Moreover, seafood habitual intake is a crucial factor in determining contaminants exposure [9,10].

In the present study, we first give a glimpse into the latest findings on the genetic diversity underlying differences in human response to mercury (Hg) and polycyclic aromatic hydrocarbons (PAHs) exposure, and on the impact of these pollutants on human DNA methylation patterns. We decided to include only epidemiological investigations assessing environmental chemical exposures using biomarkers (such as hair and blood mercury, and PAHs urine metabolites). We excluded studies addressing occupational exposure because we were interested in the potential effects on human molecular variability of Hg and PAHs from seafood. As detailed before, Hg in aquatic organisms is mostly found in the form of methylmercury (MeHg), while in occupational exposure, elemental Hg vapor is the major contributor of Hg load in the human body [11,12]. Concerning PAHs, occupational

**Citation:** De Giovanni, A.; Giuliani, C.; Marini, M.; Luiselli, D. Methylmercury and Polycyclic Aromatic Hydrocarbons in Mediterranean Seafood: A Molecular Anthropological Perspective. *Appl. Sci.* **2021**, *11*, 11179. https://doi.org/ 10.3390/app112311179

Academic Editor: Ibrahim M. Banat

Received: 13 October 2021 Accepted: 23 November 2021 Published: 25 November 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

settings are associated with exposure levels much higher than those resulting from diet [13]. Then, we address the ecological evidence that makes Mediterranean coastal communities a potential informative case study to explore this topic.

We decided to focus on MeHg and PAHs because they are two of the most concerning and widespread seafood contaminants, and because of their high levels in Mediterranean seafood.

#### **2. Seafood Contaminants**

#### *2.1. MeHg*

Hg is a heavy metal found naturally in the Earth's crust. From here, mercury is released into the atmosphere via natural phenomena such as volcanic activity and forest fires, and human activities, such as the burning of coal, oil and wood, and mining. In particular, artisanal and small-scale gold mining in developing countries has recently replaced coal combustion as the largest anthropogenic mercury emission source globally [14]. Once released into the environment, it starts to circulate following what is known as the global mercury cycle, which can last up to 3000 years [15]. When mercury passes into water, it is readily transformed by bacteria in its organic form, methylmercury, which can interact with biological components and eventually biomagnify along aquatic food chains [16]. Many studies have suggested that climate change will increase mercury inputs and methylmercury production and bioaccumulation in aquatic ecosystems [14]. Seafood is recognized as the main source of mercury in the general population, and MeHg accounts for the majority (70–100%) of Hg found in muscle tissue of fishes, molluscs and crustaceans [17].

MeHg is a well-established neurotoxicant, and exposure to MeHg has been associated with nervous system damage in adults and impaired neurological development in infants and children [18]. Decrements in memory, attention, language, and visual–motor skills in childhood have been associated with MeHg biomarkers at birth in populations with moderate MeHg exposure from regular seafood consumption [19]. Even low mercury levels (i.e., levels lower than 4 μg/g in hair; 20 μg/L in cord blood, or approximately 12 μg/L in adult blood) can negatively affect fetal and infant growth and cause neurologic outcomes [20]. Urinary levels of Hg are frequently used to estimate the level of exposure to Hg vapours or inorganic Hg (IHg), whereas blood, hair and toenail [11] Hg predicts MeHg exposure.

Because of the threat that mercury poses to human health, the EU set a maximum level of mercury in seafood of 1 or 0.5 mg/kg, depending on the species, after which seafood shall not be placed on the market [21], while the Joint FAO/WHO Expert Committee on Food Additives (JECFA) established a tolerable intake of 1.6 μg/kg bodyweight per week for methylmercury in order to protect the developing fetus from neurotoxic effects [22]. On the basis of multiple epidemiological studies [23,24] that observed adverse effects in children as consequences of maternal exposures, the European Food Safety Authority (EFSA) eventually decreased this limit to 1.3 μg/kg bodyweight per week [25], corresponding to a Hg level of ~11.5 mg/kg and ~46 μg/L in hair and blood, respectively. This threshold value has been adopted for all classes of consumers, even though adults may be less sensitive to the adverse effects of MeHg [26]. Furthermore, the US-EPA established an oral reference dose (RfD) for MeHg—that is, the maximum acceptable oral dose for this contaminant—of <sup>1</sup> × <sup>10</sup>−<sup>4</sup> mg/kg day−<sup>1</sup> (US-EPA, 2010).

#### *2.2. PAHs*

PAHs are a class of organic compounds consisting of two or more fused benzene rings, deriving from the incomplete combustion or pyrolysis of organic materials. Natural sources of PAHs include volcanoes, forest fires and petroleum seeps, while the combustion of fossil fuels, oil and wood are among the main anthropogenic sources [27,28]. Due to their physicochemical properties, PAHs are persistent pollutants, in that they can stay in the environment for long periods [29]. They represent the largest share among the main organic contaminants present in the marine environment, due to marine traffic and possible accidents involving oil tankers [30]. Although most PAHs are metabolized a short time after uptake, thanks to their lipophilic nature, a fraction accumulates in lipidcontaining tissues such as liver, eggs and muscle [8]. The most important non-occupational source of human exposure to PAHs is the consumption of contaminated food, including seafood [31], especially mollusks and crustaceans [32]. Sixteen PAHs are categorized as priority environmental pollutants, and some of them are deemed to be probable human carcinogens by the US Environmental Protection Agency (US-EPA), with benzo(a)pyrene (B(a)P) arousing more concern because of being the most carcinogenic, teratogenic and toxic compound [33]. The most used biomarkers of PAH exposure are metabolites of PAHs, particularly 1-hydroxypyrene (1-OHP), and PAH–DNA or protein adducts. 1-OHP is the principal product of pyrene metabolism [34], and its urinary excretion has been attributed mainly to the ingestion of PAHs through the diet [35]. Rather, PAH–DNA adducts, which are the products of the Phase I metabolism of PAHs, are deemed a biomarker that integrates multiple BB(a)P exposure routes (including inhalation, dermal absorption, and ingestion) and reflects a biologically effective dose [36]. PAH–DNA adduct formation is significantly influenced by individual susceptibility, which is linked to specific genetic polymorphisms [36,37]. Urine PAHs metabolites and, to a less extent, PAH–DNA adducts are also related to parent air PAH exposures, both at elevated exposures in occupational cohorts, and at low levels of air pollution [34,38].

The EU set a maximum level of B(a)P in seafood to be sold ranging from 2 μg/kg wet weight, for muscle meat of fish (other than smoked fish), to 10 μg/kg for bivalve mollusks [21]. Concerning human exposure, the US-EPA set an RfD for several PAH compounds, including anthracene (0.3 mg/kg day−1), acenaphthene (0.06 mg/kg day−1), fluorene (0.04 mg/kg day−1), fluoranthene (0.04 mg/kg day−1), pyrene (0.03 mg/kg day−1), naphthalene (0.02 mg/kg day<sup>−</sup>1) and B(a)P (0.0003 mg/kg day−1).

Several findings point to an important role of genetic diversity in shaping individual susceptibility to Hg [39] and PAHs [40] exposure, and consequently some authors claim the urgent need to include this factor in risk assessment and decision making [41]. Moreover, Hg and PAHs, similar to several other environmental toxicants [42], can impact the human epigenome through different mechanisms, and some of the epigenetic alterations driven by these two substances were shown to be associated with adverse health effects [43,44].

Being a semi-enclosed sea, delimited by highly industrialized countries and characterized by large deposits of cinnabar (HgS), the Mediterranean Sea is at a high-risk for contamination by toxic compounds [45], and evidence exists that significant anthropogenic chemical inputs into the Mediterranean began in prehistoric times [46]. In line with this, several studies have shown higher levels of contaminants in marine organisms from the Mediterranean Sea compared to those from other geographic areas [47]. At the same time, the European countries bordering the Mediterranean are among the world's highest seafood consumers, with Spain, Italy and France accounting for more than half of the European expenditure on fish and fishery products, despite having only around a third of the EU's population (EUROSTAT, 2014). Accordingly, high Hg concentrations in the blood and hair of several Mediterranean communities [6] have been found. Given such a traditionally high consumption of contaminated seafood, in our view, it is important to gain insights into the molecular diversity underpinning potential differences in susceptibility to MeHg and PAHs exposure in these communities. Moreover, Mediterranean populations might represent an interesting case study to investigate the potential impact of Hg and PAHs on the human genome and epigenome.

Modern technologies allow us to explore human genomic and epigenomic variability in a cost- and time-efficient way, enabling us, for example, to portray molecular diversity at the populational level [48], and to detect natural selection footprints in genomic regions [49].

#### **3. Human Genetic Diversity**

Epidemiological investigations are showing the role of genetics in shaping individual susceptibility to MeHg and PAHs. Through a literature search, we identified 18 (Tables 1 and S1 for further details) and 3 (Tables 2 and S2 for further details) epidemiological studies addressing the role of genetic polymorphisms in MeHg and PAHs toxicokinetics, respectively. Below, we describe some of the main findings of the above studies. Please refer to the tables for the full list of retrieved publications.

**Table 1.** List of epidemiological studies investigating the influence of genetic polymorphisms on MeHg toxicokinetic. Genes in which the above polymorphisms were identified, biomarkers affected, and samples studied are shown for each study.


#### *3.1. MeHg Exposure and Human Genetic Diversity*

The majority of ingested MeHg passes into the bloodstream, by which route it reaches all tissues. Here, MeHg enters cells thanks to its ability to form water-soluble complexes with the amino acid cysteine. After forming a complex with reduced glutathione (GSH) (Figure 1), MeHg is excreted by the liver cells into the bile. At this point, the glutathione is hydrolyzed, leading to the release of the methylmercury–cysteine complex. The latter is mostly secreted into the intestine tract, where MeHg is demethylated by intestine microflora. The resulting inorganic Hg is then eliminated via the feces [70].


**Table 2.** List of epidemiological studies investigating the influence of genetic polymorphisms on PAHs toxicokinetics. Genes in which the above polymorphisms were identified, biomarkers affected, and samples studied are shown for each study.

B(a)P, Benzo(a)pyrene.

**Figure 1.** Schematic representation of the role played by several genes in MeHg elimination from the body. (**a**) The GCLC and GCLM genes encode the catalytic and modifier subunits of the glutamate–cysteine ligase (GCL) enzyme, respectively. The GCL is the first rate-limiting enzyme of glutathione (GSH) synthesis. The GSS gene encodes the glutathione synthetase (GSS), another enzyme involved in GSH synthesis. (**b**) The GSTM1 and GSTP1 genes encode the glutathione S-transferases (GST) Mu1 and Pi1, respectively, which catalyze the conjugation of GSH to MeHg.

> In recent years, epidemiological studies on populations exposed to MeHg have been showing that several genes mediating the toxicokinetics of Hg are polymorphic in humans, and may influence inter-individual variability in Hg exposure biomarker values and health outcomes (Tables 1 and S1 for further details). In line with this, as demonstrated by kinetic studies, MeHg half-life, which is a direct determinant of the Hg body burden [71], can vary widely in humans, which may be due also to a naturally occurring biological basis for the variation in MeHg toxicokinetics.

> Goodrich and colleagues [55] analyzed Single-Nucleotide Polymorphisms (SNPs; genetic variants due to a base substitution or the insertion or deletion of a single base) variability in a cohort of dental professionals exposed to inorganic Hg via dental amalgams and to MeHg via seafood consumption, in order to investigate potential associations with Hg levels in hair and urine. In this study, fish consumption as estimated by a selfadministered survey was the best predictor of measured hair Hg level, and two SNPs were associated with this biomarker. In particular, SEPP1 3- UTR (rs7579) T allele was associated with lower hair Hg per unit of intake from fish consumption, while the GSS 5- (rs3761144) minor allele (i.e., the less common allele of a SNP) (G) was associated with increasing

hair Hg concentration per unit of fish Hg. SEPP1 encodes a selenoprotein, which combats the oxidative stress created by Hg by binding the toxicant directly via a selenocysteine residue. The latter is an amino acid unique to selenoproteins that can bind Hg–selenium conjugates or MeHg. Interestingly, as demonstrated by previous studies, the 3- UTR T allele is linked to greater SEPP1 expression and Hg-binding capacity. The GSS gene encodes for an enzyme, glutathione synthetase (Figure 1a), that is involved in the synthesis of GSH, to which Hg is conjugated before being eliminated (Figure 1b). The association of the minor allele with increasing hair Hg concentration may be ascribable to a decreased expression of GSS and, thus, to decreased GSH synthesis, which in turn could impact the body's ability to eliminate MeHg as a GSH conjugate, with the higher body burden reflected in hair Hg levels.

The study of de Oliveira and colleagues [60] was the first to investigate the genetic predisposition to mercury accumulation in the plasma, where this pollutant is more bioavailable and therefore potentially harmful to human health. In this study, authors focused on riverside communities of the Brazilian Amazon, for which the only source of Hg exposure was the intake of contaminated fish. They genotyped two glutathione-related genes, GSTM1 and GCLC. The first encodes a glutathione S-transferase, an enzyme that catalyzes the conjugation of GSH to MeHg (Figure 1b), while the second encodes the catalytic subunit of the glutamate-cysteine ligase (GCL), the first rate-limiting enzyme of glutathione synthesis (Figure 1a). What the study found is that null homozygotes for GSTM1, that is, individuals that possess two copies of a non-functional allele for this gene, showed higher plasmatic MeHg levels (MeHgP) compared to subjects with functional GSTM1, which may be related to their lower MeHg-conjugating activity, lower MeHg excretion, and a higher MeHg retention. Moreover, individuals carrying at least one T allele for GCLC (rs17883901) also had significantly higher MeHgP.

As recent findings suggest, apart from being associated with hair mercury level, the SNPs in glutathione-related genes can influence the impact of methylmercury exposure on early child neurodevelopment. Wahlberg and colleagues [66] analyzed GSH-related gene variability in mothers with a diet rich in fish coming from the population of Seychellois. Genotypes of these mothers were analyzed in association with maternal hair and blood Hg, cord blood Hg, and children's mental and motor development, as expressed by the Mental Developmental Index (MDI) and the Psychomotor Developmental Index (PDI), respectively. The authors genotyped SNPs within three genes: GCLC, whose function have been described above; GCLM, encoding the modifier subunit of the GCL (Figure 1a), and GSTP1, which encodes a glutathione S-transferase (Figure 1b). What they found is that individuals with GCLC rs761142 TT genotype showed higher mean maternal hair Hg than AG and GG. Moreover, individuals carrying the combination of GCLC rs761142-TT and GCLM rs41303970-CC genotypes showed higher hair Hg than G plus T carriers. Finally, increasing Hg in maternal and cord blood was associated with lower PDI among GCLC rs761142 TT carriers, while increasing Hg in hair was associated with lower MDI among GSTP1 rs1695 GG carriers.

Another recent study carried out on children from Valentia [67] showed that hair Hg levels were associated with worse neurobehavioral development, and that several SNPs located in the GSTP1 (rs1695) and BDNF (rs1519480, rs7934165, rs7103411) genes modified the association between Hg levels in children's hair samples and two indexes of neurobehavioral function. The brain-derived neurotrophic factor (BDNF), in particular, is a protein that promotes neural survival in adult brains, and is poorly expressed in several diseases, such as Alzheimer's and Parkinson's.

Two recent reviews [39,72] focusing on this topic collectively listed thirty-two genes whose variation is related to Hg body burden and susceptibility to Hg toxicity, and, in particular, twelve of these genes are related to hair Hg level and/or to MeHg exposure outcomes.

#### *3.2. PAHs Exposure and Human Genetic Diversity*

PAHs metabolism is a complex process consisting of two major phases (Figure 2): In the first phase, following ingestion or inhalation, the xenobiotic compound is epoxidated by enzymes belonging to the cytochromes P450 family, with the formation of diols and dihydrodiols [31]. Dihydrodiols can thus bind to the DNA, to give rise to DNA adducts, starting the mutagenic processes and eventually leading to cancer [73]. Then, in the second phase, the intermediate diols conjugate with glutathione, thanks to glutathione s-transferase enzymes. This leads to the formation of polar compounds, which can be easily excreted by renal or biliary routes [74]. The liver is the major site of the metabolism of PAHs. However, in the case of ingestion, gut micro flora and intestinal cytochrome P450 enzymes can also contribute to the process [31].

**Figure 2.** Schematic representation of the metabolic phases following PAHs ingestion. In the first phase, following ingestion or inhalation, PAHs are epoxidated by enzymes belonging to the cytochromes P450 family, with the formation of diols and dihydrodiols. Dihydrodiols can thus bind to the DNA, to give rise to DNA adducts. Then, in the second phase, the intermediate diols conjugate with glutathione, thanks to glutathione s-transferase enzymes. This leads to the formation of polar compounds, which can be easily excreted.

Epidemiological studies showed that polymorphisms at several genes influence levels of biomarkers of exposure to PAHs in several human populations (Tables 2 and S2 for further details).

As suggested by much evidence, PAH–DNA adducts may be a potential source of heritable prezygotic DNA damage in spermatozoa. Ji and colleagues [68] detected PAH– DNA adducts in ejaculated sperm of infertile adults environmentally exposed to low levels of PAHs, showing that the consumption of PAH-rich meals at least three times a week contributed significantly to an increase in DNA adduct formation. Moreover, the authors demonstrated an association between specific XRCC1 polymorphisms and an increase in sperm adduct levels. XRCC1 encodes a protein that is essential to providing an efficient repair of the DNA, and thus polymorphisms at this gene may be useful to identify individuals susceptible to DNA damage resulting from PAHs exposure.

Myeloperoxidase (MPO), an enzyme central to the microbicidal activity of neutrophils, and *N*-acetyltransferase 2 (NAT2), which functions to both activate and deactivate drugs and carcinogens, are involved in Phase I and Phase II of PAH metabolism, respectively, while ERCC5 is a single-strand-specific DNA endonuclease that participates in DNA excision repair. The SNPs in these genes have been shown to affect the PAH-driven formation of the DNA adduct. In a study [69] of more than one hundred healthy female non-smokers from Golestan Province, a region in north-eastern Iran, characterized by very high levels of exposure to PAH probably due to diet and methods of food preparation, the DNA adduct level in blood was significantly lower in homozygotes for NAT2 slow alleles, which are responsible for a less efficient detoxification of carcinogen-reactive metabolites, and the ERCC5 (rs1047768) non-risk-allele genotype. In contrast, DNA adduct level was higher in the MPO (rs2333227) homozygote risk-allele genotype.

In a cohort of non-smoking Polish mothers and newborns, Iyer and colleagues [36] observed a significant interaction between maternal exposure to airborne PAHs, measured by personal air monitoring, and SNPs in selected B(a)P metabolism genes on cord blood B(a)P–DNA adducts. These genes included: maternal CYP1A1 and GSTT2, and newborn CYP1A1 and CYP1B1. CYP1A1 and CYP1B1 are involved in metabolizing the parent B(a)P compound to the reactive B(a)P 7,8-diol-9,10-epoxide (BPDE) metabolite, which is involved in the formation of B(a)P–DNA adducts. In contrast, GSTT2 is involved in shifting B(a)P metabolism so as to prevent the formation of the reactive BPDE. In particular, the authors concluded that the T allele at the GSTT2 SNP position in mothers is protective with regard to cord B(a)P–DNA adduct formation, while the maternal and newborn G allele at the CYP1A1 SNP position, and the newborn G allele at the CYP1B1 SNP position, are not.

#### **4. The Epigenetic Impact of Seafood Contaminants**

Several studies have demonstrated the role of the environment in shaping human molecular variability at the epigenetic level in different populations [48,75–77]. Epigenetic changes are defined as any stable changes in the chromatin structure that are heritable from cell to cell, and can result in alteration of gene expression without altering DNA sequences [75]. The epigenome functions as an interface between the inherited genome and the dynamism imposed by the environment [78], and, as such, can be affected by the latter.

DNA methylation is among the most frequently studied epigenetic modifications. It consists in the covalent addition of a methyl group from a methyl group donor, the coenzyme S-adenosylmethionine (SAM), to the fifth carbon atom of a cytosine ring, and it is catalyzed by the DNA methyltransferase (DNMT) enzyme family. In mammals and insects, cytosine methylation is found almost exclusively in the context of CpG dinucleotides [75].

Changes in DNA methylation status influence genes' accessibility, thus altering gene expression, and aberrant DNA methylation has been discovered in a wide range of pathophysiological conditions [79].

Environmental chemicals, such as Hg and PAHs, can interfere with the one-carbon and citric acid metabolism pathways, resulting in anomalous DNA methylation status all over the genome [42]. Hg and PAHs can alter DNA methylation profiles in specific genes [80].

Three reviews address this topic [42,80,81]. Briefly, Ruiz-Hernandez and colleagues retrieve and discuss two and three epidemiologic studies investigating the association between DNA methylation and Hg [82,83] and PAHs [84–86], respectively, in adults. Considering all the strengths and weaknesses of the various studies, e.g., the lack of adjustments for potential confounding, such as sex, age, smoking status and tissue cell heterogeneity, the authors' conclusion is that the evidence they accrued supports the importance of environmental exposures in modulating the epigenome, but is insufficient to support causality because of the heterogeneity among epidemiologic studies in addressing the residual confounding of the associations, differences in DNA methylation assessment methods, and random error. The review of Culbreth and Aschner concludes that, despite some inconsistencies across different studies, dependent on the tissue or species examined, MeHg undoubtedly induces epigenetic modifications, and these modifications can potentially mediate its toxicity. In particular, regarding DNA methylation changes, controlled exposure studies on human and animal in vitro and animal in vivo models reveal that MeHg can lead to hypomethylation of the DNA in brain-derived tissue, but not in the liver, while selected individual genes show exposure-driven DNA hypermethylation.

Through a literature search, we identified seven (Tables 3 and S3 for further details) and eight (Tables 4 and S4 for further details) epidemiological studies assessing the impact of MeHg and PAHs on DNA methylation. Below, we describe some of the main findings of the above studies. Please refer to the tables for the full list of retrieved publications.

**Table 3.** List of epidemiological studies investigating the impact of MeHg exposure on DNA methylation. Genes in which the differentially methylated CpG dinucleotides were identified, biomarkers measured, tissues from which DNA was extracted, and samples studied are shown for each study.


**Table 4.** List of epidemiological studies investigating the impact of PAHs exposure on DNA methylation. Genes in which the differentially methylated CpG dinucleotides were identified, biomarkers measured, tissues from which DNA was extracted, and samples studied are shown for each study.


1-OH-Pyr, 1-hydroxypyrene urinary metabolite; 2-OH-Nap, 9-hydroxynaphthalene urinary metabolite; 9-OH-Phe, 9-hydroxyphenanthrene urinary metabolite; Ace, acenaphthene; Ant, anthracene; B(a)P, benzo(a)pyrene; Fl, fluorine; Nap, naphthalene; NTD, neural tube defects; PAM, personal air monitor; Phe, phenanthrene; PMA, phenylmercuric acetate; Pyr, pyrene; VAT, visceral adipose tissue; ΣH\_PAHs, sum of high-molecular weight PAHs including pyrene, benz[a]anthracene, chrysene, benzo[b]fluoranthene; ΣOH-PAHs, total urinary monohydroxy-PAH metabolites.

#### *4.1. The Impact of the Exposure to MeHg on DNA Methylation*

Epidemiological investigations on different populations have demonstrated the ability of Hg to impact the DNA methylation pattern in several genes (Tables 3 and S3 for further details).

A study [82] showed increased DNA methylation of the GSTM5 promoter in women with higher Hg levels in whole blood. This gene is a member of th eGSTM gene family, which encodes for enzymes that are involved in the metabolism of several environmental agents.

Hg can also influence the DNA methylation status of genes that are involved in the protection against chemical toxicity. As previously mentioned, the SEPP1 gene encodes a protein known to bind Hg that has antioxidant properties. Its promoter shows a trend of DNA hypomethylation with increasing hair Hg levels, which was predicted by estimated Hg from fish consumption [83].

Some of the neurologic outcomes of the exposure to Hg were associated with DNA methylation changes [43]. The suppressive effect that MeHg exposure has on the expression of the BDNF gene, which is poorly expressed in depressed patients, seem to be mediated also by the hypermethylation of the DNA [99]. As demonstrated by Maccani and colleagues [89], after crossing the placenta, MeHg can disrupt placental DNA methylation patterns, leading to the DNA hypomethylation of the EMID2 gene, and likely to adverse neurobehavioral outcome in infants. Cardenas and collaborators [90] found that, in male children, maternal prenatal blood mercury levels were associated with DNA hypomethylation of the Paraoxonase 1 gene (PON1), a gene involved in drug and fatty acids metabolism, and the DNA methylation pattern of this gene predicted lower cognitive test scores during early childhood. It is important to note that cord blood Hg level has previously been demonstrated to be a more accurate measure of prenatal MeHg exposure than maternal hair Hg level, and so DNA methylation changes associated with this biomarker potentially reflect MeHg effects more accurately [81]. In another very recent study [91], carried out on 406 mother–child pairs from a population who consume large amounts of fish and who are characterized by hair Hg levels that are higher than those in European and US populations, the authors found a positive association between prenatal MeHg exposure and DNA methylation in two nervous system-related genes, GRIN2B and NR3C1, measured in children's saliva. GRIN2B encodes a subunit of receptors that are important for the regulation of neural morphology, learning and memory, while NR3C1 is a receptor that is crucial to the stress responses in the brain. As stated by the authors, the observed DNA methylation changes associated with MeHg prenatal exposure at these two genes are predicted to lead to lower gene expression, and are likely to influence neurodevelopment and mental health.

#### *4.2. The Impact of the Exposure to PAHs on DNA Methylation*

Both in vitro and in vivo analyses have revealed the ability of these substances to disrupt human DNA methylation patterns [42] (Tables 4 and S4 for further details).

The developing fetus is particularly susceptible to PAH-induced DNA damage, and studies support the hypothesis that this may be due also to epigenetic dysregulations caused by these chemicals. In a study of non-smoking African-American and Dominican women from New York City [93], the authors found that prenatal exposure to PAHs measured using a personal air monitor was associated with lower global DNA methylation levels measured in umbilical cord blood DNA.

Kim and colleagues [94] analyzed samples of visceral adipose tissue of non-smoking female patients with myoma. They showed that the DNA methylation level of IRS2 gene increased as the concentrations of PAHs in adipose tissue increased. Interestingly, the IRS2 gene mediates the effects of insulin on various cellular processes, and it has been associated with several diseases, such as type 2 diabetes. Furthermore, promoter methylation of the IRS2 gene turned out to mediate the transcriptional silencing of this gene in the same study, and this led the authors to suggest that exposure to PAHs might contribute to the pathogenesis of insulin resistance through the methylation-mediated suppression of IRS2.

PAHs can accelerate human aging through epigenetic modifications. In their study of Chinese and Caucasian populations, Li and collaborators [96] first developed a DNA methylation age predictor based on the methylation status of many CpG sites across the genome, and then defined two aging indicators: Δage, defined as methylation age minus chronological age; and aging rate, defined as the ratio between methylation age and chronological age. Evaluating the association of PAHs exposure biomarkers with the above-defined aging indicators, the authors found that the increase in several urine PAHs metabolites was associated with an increase in both Δage and aging rate.

Possible hints of PAHs-mediated DNA methylation changes that may affect neurodevelopment emerged also from a study of pregnant women living close to a coal-fired power plant in China [95]. In that study, the authors analyzed cord blood samples for PAH–DNA adducts and assessed global DNA methylation by measuring genomic long interspersed nuclear elements (LINE1) methylation. LINE1 is one of the transposable repetitive elements, repetitive DNA sequences scattered across the genome and found in most eukaryotic organisms, which can change their position. Changes in LINE1 methylation can disrupt gene expression, and have been associated with birth defects, such as NTDs. In Lee and collaborators' study, a significant inverse relationship was observed between PAH–DNA adducts and LINE1 DNA methylation. Interestingly, the latter was a positive predictor of IQ (Intelligence Quotient) scores at 5 years of age in women enrolled before the closure of the power plant.

Neural tube defects (NTDs) are common and severe congenital malformations that arise from a failed or disordered closure of the neural tube during embryogenesis. Studies have linked NTDs to abnormal genome-wide DNA methylation. Authors found that PAX3, a gene encoding a transcription factor involved in development, is hypermethylated in NTD cases, and that the mean DNA methylation level of this gene in fetal neural tissue is positively correlated with median concentrations of PAHs in maternal serum [97]. Moreover, mean DNA methylation levels in the promoter region and 5- UTR of ZIC4 gene tended to be inversely associated with levels of HMW-PAHs in the livers of NTD fetuses in a recent survey [44]. ZIC4 encodes a zinc finger protein whose absence can hamper cerebellum development in both humans and mice.

Concerning potential mechanisms underlying DNA methylation changes driven by Hg and PAHs, several findings support different hypotheses. Evidence exists that MeHg exposure is associated with the reduced expression or biochemical activity of DNMT, but Hg may also affect the methionine cycle, thus influencing the availability of SAM for DNA methylation [41]. Moreover, various studies support the hypothesis that oxidative stress mediates the effects of PAHs exposure on DNA methylation, via both the suppression of DNMT and excessive SAM consumption [44].

Even if not exhaustive, given that they report all the evidence on the subject beyond the scope of this paper, the above-described results demonstrate the ability of these important seafood pollutants to impact the human epigenome and, in particular, the DNA methylation profiles.

#### **5. An Anthropological Perspective**

Human populations that traditionally consume seafood are at an increased risk of MeHg exposure and bioaccumulation. This is supported by recent data that demonstrate that populations consuming more fish or marine mammals have greater blood MeHg values than those consuming marine foods less than once a week [17]. Moreover, evidence exists that fish-eating populations tend to show the typical symptoms associated with Hg exposure at a high rate [19]. One of the first studies addressing this topic was carried out on a cohort of 1022 consecutive singleton births from the Faroe Islands [100], where maternal exposure to MeHg is derived from the consumption of pilot whale meat. This study found a statistically significant relationship between higher prenatal Hg exposure and poorer scores on tests of neurologic function [101]. In a cross-sectional study conducted on the adults of six fishing villages of the Pantanal region of Brazil, Hg exposures associated with fish consumption, as measured by hair mercury levels, were associated with detectable alterations in performance in tests of fine motor speed, dexterity, and concentration, and the magnitude of the effects increased with hair mercury concentration, consistent with a dose-dependent effect [102].

At present, unlike the case of MeHg, there is no direct evidence that populations that consume high amounts of seafood are more exposed to PAHs. Nonetheless, evidence exists that traditional fish smoking methods can introduce potentially harmful combustion by-products into the smoked fillets, leading to concentrations of PAHs that pose a threat to human health [103,104]. Moreover, human exposure to PAHs in seafood may date back to ancient times: with fish, shellfish and sea mammals being rich in fats, and considering the high lipophilicity of PAHs, these foods may have had absorbed substantial amount of PAHs from the bitumen used for prehistoric container production [105]. Finally, as detailed below, several investigations revealed high levels of PAHs in several commercially relevant marine species, with concentrations sometimes exceeding legal limits.

Modern technologies allow us to explore human molecular variation, both at a locusspecific and at a genome-wide level, enabling us to answer several questions about population evolutionary history and the relationship between environment and human biodiversity.

The first evidence of human adaptation to a toxic chemical was reported in arsenicexposed women from the northern Argentinean Andes [106]. The inhabitants of this region, which is characterized by elevated arsenic concentrations in available drinking water, show a uniquely efficient arsenic metabolism. Accordingly, the authors found that the AS3MT gene, which encodes the arsenite methyltransferase and functions as the major gene for arsenic metabolism in humans, strongly differentiates the Argentinean Andes population from a highly related Peruvian population much less exposed to this environmental toxicant. Then, they confirmed that SNPs mapping in that gene was positively selected.

Similar results were obtained from investigations on another Andean community. Through analyses of ancient human remains from the Camarones Valley, it has been shown that the inhabitants of the area have been exposed to arsenic-contaminated drinking water for the last 7000 years [107]. Interestingly, a decreasing trend has been detected in the average hair and bone arsenic levels, starting from Archaic hunter–gatherers and leading to the current populations, and this evidence has been interpreted as the potential result of an adaptive increasingly efficient metabolic detoxification [108]. In support of the above scenario, analyses carried out through polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP), targeting SNPs strongly associated with arsenic metabolization, showed that, contrary to alleles associated with increased toxicity risk, protective variants are much more frequent in exposed populations compared to a southern Chilean community [109].

Potential hints of human adaptation to PAHs exposure come from the comparison between the exome sequence (i.e., the sequence of nucleotides that make up the proteincoding portion of the genome) of the aryl hydrocarbon receptor (AHR) gene in Neanderthal, Denisovan and modern human individuals [110]. Once activated by endogenous or exogenous ligands, such as diet-derived metabolites and PAHs, respectively, the complex made up of AHR and other proteins passes from the cytoplasm to the nucleus. Here, following further biochemical mechanisms, the AHR regulates the CYP1A1/1A2/1B1 genes expression, thus initiating PAHs metabolism, with the resultant production of PAHs' reactive metabolites and DNA adducts. Hubbard and colleagues found that modern humans carry the same allele at a codon of the AHR gene that is unique to our species, and which is associated with a reduced AHR activation by PAHs, specifically 2,3,7,8-tetrachlorodibenzofuran (TCDF), B(a)P and benz(a)anthracene, compared to Neanderthal and other primates receptors. On the other hand, modern humans and Neanderthal AHR showed similar levels of activation by endogenous ligands. Based on the above results, the authors postulate that exposure to potentially toxic environmental AHR ligands, such as PAHs derived from controlled fires in caves, may have driven the selection of genetic variants conferring a reduced sensitivity to AHR exogenous ligands, and thus a lower DNA adduct synthesis [110].

As demonstrated also by the above-mentioned studies, molecular anthropologists are only starting to depict the role that environmental toxicants could have had in human evolution, and this is thanks to modern genomic technologies. Moreover, environmental toxicants can impact human biological variability at multiple levels, as shown by abovementioned studies on the DNA methylation.

Environmental pressures can shape DNA methylation variability across human groups, and methods have been developed to explore the epigenetic side of human diversity at different levels. In a study analyzing the DNA methylation profiles of three human populations at 450,000 CpG sites, the authors found that DNA methylation differences contribute to the phenotypic variability of these populations, and that 68% of differentially methylated CpG sites were significantly related to underlying genetic variation, while the remaining 32% was probably related to external stimuli able to induce epigenetic changes with an impact on subsequent generations (e.g., toxic xenobiotics, differences in dietary or hormone exposure, or stress response) [111].

Many toxicants are sources of differences within human populations [80]. In Giuliani and colleagues' [112] investigation, for example, the authors compared the DNA methylation of individuals living in areas that were heavily sprayed with Agent Orange during the Vietnam War with that of individuals from non-contaminated areas. What they found is that past exposure to dioxin, the main ingredient of Agent Orange, led to DNA methylation changes among Vietnamese individuals from areas heavily sprayed, and in those whose parents participated in the war in sprayed zones.

#### **6. Mediterranean Coastal Communities as an Informative Case Study**

Certain Mediterranean communities may be more exposed to the negative effects of seafood contaminants, for both environmental and cultural reasons.

The Mediterranean Sea is a semi-enclosed basin, surrounded by countries highly industrialized and with high agricultural development, and, as such, is of particular concern with respect to contamination by toxic compounds. The Mediterranean Sea is generally considered a geological hot spot for mercury [26], as it is characterized by large deposits of HgS that account for about 65% of the global mercury reserves [113]. Moreover, it should not be surprising that several studies have shown higher levels of contaminants in marine organisms from the Mediterranean Sea compared to those from other geographic areas [6,47,114–116], with levels of PAHs and especially Hg often exceeding recommended limits for human consumption [6,30,117,118]

Concerning Hg, in particular, the highest concentrations in Europe tend to be found in fish caught in the Mediterranean Sea [16]. Data showed a more marked Hg bioavailability in the Tyrrhenian and the Adriatic coastal waters compared to the rest of the Mediterranean [119], and MeHg levels higher than the legal limit have been discovered in seafood caught in both areas [117,118,120–124], as well as in the Ionian Sea (Sidimar 2018), and in different classes of marine organisms, including fishes, crustaceans, and mollusks. MeHg contamination hotspots are represented by the Trieste gulf [125], the coastal waters between Cattolica and Rimini, in the Central Adriatic Sea [124], and those between Anzio and Civitavecchia, in the Central Tyrrhenian Sea [126]. Turning to PAHs, potential contamination hotspots are generally deemed to be located near Taranto, Trieste [125], Naples [127,128], Genoa and Palermo [127]. It is also worth reporting that some studies revealed possible risks to human health arising from the consumption of PAHs-contaminated seafood from the Mediterranean. The above conclusions were drawn from the calculation of risk indexes, such as the Excess Lifetime Cancer Risk (ELCR) and the Target Hazard Quotient (THQ) [30,73,129] The authors calculated the ELCR according to the following equation:

ELCR = EF (day/y) <sup>×</sup> ED (y) <sup>×</sup> IR (Kg/day) <sup>×</sup> CSF (mg Kg−<sup>1</sup> day) <sup>÷</sup> BW(Kg) <sup>×</sup> AT

where EF is the exposure frequency (365 days/year), ED is the exposure duration (y), IR is the ingestion rate, which is equal to PAH concentration times the mean ingestion rate of the species, CSF is the Cancer Slope Factor for each of the analyzed PAHs (OEHHA, 2009), BW is the body weight, and AT is the averaging time, which is equal to EF × ED.

A CR above the acceptable lifetime risk (ALR) of 10−<sup>5</sup> [130] indicates a probability greater than 1 in 100,000 of developing cancer [30].

THQ indicates the ratio between exposure and the reference dose, and is calculated through the formula:

THQ = EF (day/y) <sup>×</sup> ED (y) <sup>×</sup> IR (Kg/day) <sup>×</sup> C (mg Kg <sup>−</sup>1) <sup>÷</sup> RfD (mg kg <sup>−</sup><sup>1</sup> day) <sup>×</sup> BW(Kg) <sup>×</sup> AT

In Europe, coastal populations consume greater amounts of seafood compared to inland populations [11]. Moreover, Mediterranean coastal populations are characterized by food habits based on local seafood consumption [26], and European countries bordering the Mediterranean are among the world's highest seafood consumers, with Spain, Italy and France accounting for more than half of the European expenditure on fish and fishery products, despite having only around a third of the EU's population (EUROSTAT, 2014). In Italy, apparent consumption of fish and seafood products amounted to 28.4 kg per capita, a share significantly higher than the EU average (EUROFISH, 2015). The traditionally high consumption of local seafood can lead to high MeHg and PAHs exposure levels among Mediterranean communities.

The above scenario is supported by several lines of evidence. Assuming an exposure frequency of 365 days a year, an exposure duration of 80 years (equivalent to the average lifetime in Italy in 2011), an ingestion rate of 18 g per day, and a body weight of 70 kg, after measuring Hg levels in various species of demersal fish commonly consumed in Italy, Storelli and Barone calculated a high target hazard quotient (THQ) and estimated weekly intake (EWI) for larger fish specimens caught in the Adriatic Sea [123]. In a study taking into account various commercially relevant marine species from the Ionian Sea, assuming a consumption rate greater than once per week, the authors found a possible risk for chronic systemic effects derived from Hg content [131]. Similar findings were derived from several other studies on seafood from various parts of the Mediterranean Sea, focusing on levels of Hg [124,132] and PAHs [129].

Given all the above, and considering the high average seafood consumption in the Mediterranean regions, we highlight the need for policymakers to take into account evidence of potentially high exposure to seafood contaminants among Mediterranean communities, especially as regards MeHg, and to consider if it is reasonable to revise law limits and/or recommendations.

In line with the above-mentioned evidence, studies on newborns and preschool children from Mediterranean populations have shown high Hg concentrations in blood and hair [6]. Analyzing data collected from over 200 cross-sectional studies measuring Hg biomarkers in human populations, Basu and collaborators [12] found geographic differences in Hg exposure, with pooled central median blood mercury concentrations being higher in general background populations (i.e., those with no particular or significant exposure to mercury) living in certain geographic areas, including the Eastern Mediterranean. They also found that subpopulations that consume high amounts of seafood are approximately four times more exposed than the general background population. In particular, exposures were higher in Indigenous people in many regions of the world, in populations living in proximity to water bodies or associated with marine ecosystems, among which were populations living along the Mediterranean Sea.

Višnjevec and colleagues [11] compared the results of several investigations on Hg exposure in Europe countries, and found that the highest hair Hg levels were found in Madeira fishermen, habitual tuna consumers in Sardinia, and Greenlandic children and mothers.

A study on the adult population of Naples, Italy, found a strong correlation between total mercury concentrations (THg) in hair and fish consumption, while almost no association was found between THg and number, surface and area of dental amalgam fillings, another possible source of MeHg in the general population, thus confirming previous findings of the major role of seafood consumption in human exposure to this pollutant [133]. Moreover, the same study found THg levels higher than the reference dose adopted by the U.S. Environmental Protection Agency (0.1 μg per kg body weight) in 5.9% of the samples.

Finally, it is interesting to note that two of the three Mediterranean countries taken into consideration in the EU-funded human biomonitoring study named DEMOCOPHES [134], i.e., Spain and Cyprus, with the fourth being Slovenia, are respectively the first and the third countries as regards mercury levels in the hair of mothers compared to all the others, with Spain being also the second European country for per capita seafood consumption in 2016 [135], as well as the Mediterranean country with the second highest share of domestic wild capture consumption compared to imported seafood (FAOSTAT, 2018), after Croatia, which was not included in the DEMOCOPHES study.

Some subgroups of the Mediterranean coastal population may be more exposed than others to the harmful action of seafood pollutants. As shown by the above-described review from Basu and colleagues [12], coastal communities are more exposed to MeHg. A study [136] of mother–infant pairs from Croatia found higher levels of Hg and selenium (Se) in hair, blood, placenta and cord blood of mothers from the coast compared to those living in continental areas, due to higher fish consumption. Interestingly, in this study, the authors also evaluated the relationship between Hg and Se levels and a polymorphism (rs28366003) in the MT2A-5 gene, which encodes a protein that plays a role in the detoxification of heavy metals, but they did not find any association.

Fishermen, in particular, have shown a tendency for a greater accumulation of Hg derived from fish, and this is related to their higher mean consumption of this food compared to the general population. Evidence in this sense comes from a fishing community on the Mediterranean coast of Morocco. In this community, researchers measured hair Hg levels, and found that these were closely related to fish intake, and that fishermen and their families were the most exposed population subgroup [137]. Additionally, in Sicily, greater Hg accumulation has been proven in fishermen, as they showed significantly higher mean hair Hg levels (6.45 ± 7.03 <sup>μ</sup>g g−<sup>1</sup> vs. 0.23 ± 0.4 <sup>μ</sup>g g−<sup>1</sup> in the control group) [138]. Similar results have been derived from investigations carried out on Italian coastal communities from the northern Adriatic Sea [139].

What may emerge from all the above is that Mediterranean fishing communities could represent an informative case study to gain insight into the potential impact of Hg and PAHs on the human genome and epigenome.

Additionally, to fulfil the need for an "ecogenetic approach" to the study of the health effects of environmental chemicals stressed by Basu and colleagues [41], what we suggest is to extend the research on Mediterranean seafood contamination by Hg and PAHs by including information about the genomic and epigenomic backgrounds of the exposed communities. Such an approach would involve the following main steps: identification of communities that are particularly exposed to seafood contaminants, in terms of both cultural (i.e., traditional high consumption of local seafood) and environmental factors (i.e., subsisting on resources caught from pollution hotspots), and the selection of communities that would represent the control group, for example inland communities, characterized by a very low fish intake; simultaneous collection of biological samples (e.g., buccal mucosa cells) and information on the family history, diet and lifestyle of participants (e.g., through a Food Frequency Questionnaire); analysis of the genetic variability and DNA methylation profiles of those genes implicated, for example, in fatty acids metabolism and in susceptibility to environmental chemicals. Such an approach would enable us to answer different questions, such as, are there any biological differences between fishing and non-fishing communities that could have been caused by different seafood intakes? Are there any differences in the biological predisposition of Mediterranean communities to the health effects of seafood intake?

#### **7. Challenges**

The main challenges in such a study refer to the epigenetic investigation, and this is for several reasons. First of all, DNA methylation may be influenced by several factors [80], including many dietary components (e.g., folate, vitamin B6, vitamin B12, betaine, methionine and choline) [75], other environmental chemicals [42], pathogens load, various

environmental and climatic conditions [140], sex and age [42], socioeconomic status [141], and genetic background [48,78]. Consequently, it is difficult to tell which is the actual correlative factor underlying the observed patterns, even when an association between a given factor or biomarker has been detected.

The simultaneous analysis of genetic and epigenetic data, coupled with information on eating habits and lifestyle and personal details of participants, would allow us to account for several potential confounders, possibly distorting the association between estimated exposure and epigenetic changes. To this end, it is important to gather information that is crucial to depict the whole set of environmental stimuli affecting the individual's methylome (e.g., smoking status, diet, occupation), as well as to identify, sample and to compare populations that differ markedly when it comes to seafood consumption rate and/or levels of Hg and PAHs in fish consumed.

However, it is worth noting that the potential simultaneous exposure of individuals to a plethora of chemicals constitutes one of the main challenges in the field.

Seafood, along with other food items, contains several nutrients and contaminants, which can impact human biology at a molecular level, and this may confound the association between MeHg and/or PAHs exposure and DNA methylation.

Among seafood contaminants, the heavy metals arsenic (As), cadmium (Cd) and lead (Pb) also constitute an emerging issue due to their concentrations often exceeding regulatory limits [123,142,143] and studies have demonstrated their ability to elicit DNA methylation changes [80]. This is quite expected, as comparisons of the mechanisms of action reported similar biological pathways of these metals inducing toxicity, such as ROS generation, weakening of the antioxidant defense, enzyme inactivation, and oxidative stress (for a detailed review see [144]).

Moreover, epidemiological studies showed a general hypomethylation of LINE-1 elements after PAHs, As, Cd and Pb exposure [145–149].

In vitro and animal studies performed under rigorous experimental conditions constitute a powerful method to identify the impact of single chemicals on DNA methylation. In this respect, molecular anthropological investigations could help to make a list of candidate genes to be tested through functional studies, or vice versa, could constitute a method to evaluate the real effect.

Another crucial aspect to consider is the tissue-specificity of DNA methylation [42]. Most of the retrieved epidemiological studies on MeHg and PAHs epigenetic effects measured DNA methylation in blood, with only one study using buccal mucosa [83], another study sampling saliva [91], one study measuring adipose tissue DNA methylation [94], and two studies using neural tissue [44,97]. Molecular anthropologists, on the other hand, often collect saliva or buccal mucosa cells as an alternative DNA source, because whole blood is difficult to collect during fieldwork [150,151]. It is also important to note that, despite the risk of discordant results due to potential tissue-specific DNA methylation changes (and especially to heterogeneity in cell composition), several studies demonstrated high correlations between the DNA methylation profiles of blood and saliva [152–154], pointing to the suitability of saliva as a source for genomic DNA in cohort studies. In the same way, recent investigations have shown that DNA methylation also correlates well between saliva and the brain [155]. Additionally, buccal cells also offer potential advantages to human epigenetic studies, as they represent a better surrogate tissue for brain tissues, with both being ectodermal tissues, and because they can be collected via a non-invasive method (buccal swabs) [156]. Finally, it should be noted that statistical methods to account for cell composition in DNA methylation assays are implemented and available [157].

Seafood is not the only source of exposure to Hg and PAHs. Working with dental amalgam fillings and working or residing among artisanal and small-scale gold mining sites result in elevated exposures to elemental and inorganic Hg [12], which may lead to DNA methylation changes [83,158]. In the same way, several occupations, such as coke oven manufacturing, chimney sweeping, paving and roofing, entail relevant exposure to PAH mixtures [159,160], with consequent impacts on the DNA methylation status [86,161]. Asking participants about their occupation is therefore of fundamental importance in order to address potential confounders of the association between exposure level to MeHg or PAHs via seafood consumption and DNA methylation changes.

As regards the influence of genetics on the individual's biological response to MeHg and PAHs exposure, it should be considered that, sometimes, the same genes are involved in the toxicokinetic of and/or susceptibility to different substances. This, obviously, complicates the detection of genes that may be subject to natural selection driven by a specific chemical. This is the case of GSTP1, MT4 and ALAD genes, whose variation can influence the toxicokinetic of Hg (Tables 1 and S1), but also of Cd [162] and Pb [163–165]. The same is true for NAT2 gene polymorphisms, which influence the toxicokinetics of both Hg [65] and PAHs [69] (Tables 3 and 4).

A further level of complexity is related to the fact that concentrations of these chemicals in edible tissues of aquatic organisms are influenced by several factors.

Several studies have highlighted the role of trophic level, habitat and size of the organism in determining the level of MeHg in sea animals. In particular, despite discordant evidence [121,166], MeHg concentration in fishes often increases with trophic level [126,167], size and age [47], with MeHg uptake being a process of bioaccumulation during the whole life [118], and other studies show that MeHg concentrations are affected also by changes in feeding habits during fish lifespan [132]. The relationship between size and MeHg concentration in marine invertebrates is rather less clear: while some results point to a positive correlation between these two variables in bivalves and crustaceans [124,168], others show a negative [120,169] or no significant correlation [124] in the same taxonomic classes, or even in the same species. As regards PAHs, only few studies have tried to assess the influence of biological factors on their accumulation in aquatic organisms [8], and these led to discordant results. Several studies found no significant correlation between PAHs concentrations and fish size or age [170–173], while Frapiccini and colleagues [28] found a negative correlation between body size and PAH concentrations in the liver and gills of common soles caught in the Po Delta and off Chioggia. Additionally, the sex [170,173,174] and the reproductive stage [8] of the organism seem to affect PAHs accumulation and metabolism in fish.

MeHg and PAHs concentration in fish also vary with seasons [8,171,175–178] and with the geographic origin of the fished specimen, with some areas of the Mediterranean being more polluted than others.

Addressing the above factors through a questionnaire is not easy, if not impossible, which means that a proper estimate of the habitual seafood intake or, more generally, of the eating habits of the sampled individuals, does not always correspond to a precise estimate of their habitual MeHg and/or PAHs intake [179].

To overcome these limitations, the most effective solution is the use of biomarkers of exposition. In particular, as already mentioned, hair mercury level has proven to predict exposition to MeHg [136,180], whose primary route in the general population is seafood consumption [181], while urinary excretion of the metabolite 1-OHP has been attributed mainly to the ingestion of PAHs through the diet [35,182].

However, it is important to note that, unlike MeHg, seafood is not always the main dietary contributor to PAHs intake, and that the concentrations of PAHs in food are also influenced by cooking procedures [183]; as a consequence, it would be difficult to tell whether eventual epigenetic modifications correlated with 1-OHP urinary level are actually driven by PAHs in seafood, unless a very detailed questionnaire on eating habits is collected.

As regards the uncertainties on the geographic origin of seafood consumed, sampling fishermen could help trace the origin of the fish they eat, as fishermen tend to consume their own catch (unpublished data). Moreover, as already mentioned, fishermen represent an interesting case study, given their exposure to high levels of MeHg and, potentially, PAHs, due to their traditionally high seafood consumption. Studying fishermen, however, implies the inclusion of a potential further modifier of the effect of MeHg and PAHs on

DNA methylation, that is, the typical lifestyle of fishermen. Fishing is strongly demanding, both physically and psychologically, implying, among the several challenges, working long hours, frequent night shifts, the unpredictability of the sea, and prolonged separation from the family [184]. Moreover, several investigations have linked the above factors to health conditions and harmful habits that are common among fishermen from different parts of the world, including the Mediterranean Sea [185], which comprise tobacco smoking, alcohol abuse, sleep deprivation, chronic stress, and so on [186,187]. Such habits are known to impact human DNA methylation [188–191], and hence must be taken into consideration when asking sampled individuals about their daily life.

#### **8. Conclusions**

The Mediterranean Sea is considered a pollution hotspot for both natural and anthropogenic factors. As a consequence, Mediterranean communities may be particularly exposed to MeHg and PAHs through ingestion, due to their traditional high consumption rate of local seafood, and much evidence supports the above scenario. MeHg and PAHs can impact DNA methylation patterns in humans, even at low doses. Moreover, some of the epigenetic changes associated with MeHg and PAHs exposure are in turn associated with their known health outcomes. Finally, increasing evidence points to a significant contribution of human genetic variability in determining individual susceptibility to the chronic exposure to these chemicals, which, in certain cases, may be the results of population adaptation to certain ecological settings. In this framework, and also considering the growing concern about MeHg pollution due to climate change, we highlighted the benefit of an integrated approach, including molecular anthropologists and environmental and marine chemists, to the investigation of the relationship between the molecular diversity of Mediterranean communities and the exposure to MeHg and PAHs through seafood intake. Such an approach will help us to cope with uncertainties when it comes to risk assessment and decision-making about contaminant limits in seafood.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/ 10.3390/app112311179/s1, Table S1: List of epidemiological studies investigating the influence of genetic polymorphisms on MeHg toxicokinetic. Genes in which above polymorphisms were identified, polymorphisms, alleles/genotypes effect on biomarkers, and samples studied are shown for each study. Table S2: List of epidemiological studies investigating the impact of MeHg exposure on DNA methylation. Genes in which differentially methylated CpG dinucleotides were identified, technology used for methylation assay, biomarkers measured, tissues in which biomarkers were measured, tissues from which DNA was extracted, and samples (with sample sizes) studied are shown for each study. Table S3: List of epidemiological studies investigating the influence of genetic polymorphisms on PAHs toxicokinetic. Genes in which above polymorphisms were identified, polymorphisms, alleles/genotypes effect on biomarkers, and samples studied are shown for each study. Table S4: List of epidemiological studies investigating the impact of PAHs exposure on DNA methylation. Genes in which differentially methylated CpG dinucleotides were identified, technology used for methylation assay, biomarkers measured, tissues in which biomarkers were measured, tissues from which DNA was extracted, and samples (with sample sizes) studied are shown for each study.

**Author Contributions:** Conceptualization, A.D.G., C.G., M.M. and D.L.; methodology, A.D.G., C.G. and D.L.; formal analysis, A.D.G. and C.G.; investigation, A.D.G.; resources, D.L.; data curation, A.D.G. and C.G.; writing—original draft preparation, A.D.G., C.G., M.M. and D.L.; writing—review and editing, A.D.G. and C.G.; visualization, A.D.G. and C.G.; supervision, A.D.G., C.G., M.M. and D.L.; project administration, D.L. and M.M.; funding acquisition, D.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** The research leading to these results was conceived under the collaboration between the University of Bologna and the National Research Council for the implementation of the International PhD Program "Innovative Technologies and Sustainable Use of Mediterranean Sea Fishery and Biological Resources" (www.FishMed-PhD.org).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Meta-Analysis of a New Georeferenced Database on Polycyclic Aromatic Hydrocarbons in Western and Central Mediterranean Seafood**

**Andrea De Giovanni 1,2,\*, Paolo Abondio 1,3, Emanuela Frapiccini 4, Donata Luiselli 1,2 and Mauro Marini 2,4**


**Featured Application: The database featured in the present work aims to support researchers and decisionmakers in planning future investigations and in evaluating current knowledge on polycyclic aromatic hydrocarbons when it comes to fixing the limits to PAH levels in fishery products.**

**Abstract:** The aim of this work was to collect and harmonize the results of several studies achieved over the years, in order to obtain a database of georeferenced observations on polycyclic aromatic hydrocarbons (PAHs) in Western and Central Mediterranean seafood. For each observation, some information on the taxonomy and the ecology of the sampled species are reported, as well as details on the investigated hydrocarbon, and spatial and temporal information on sampling. Moreover, two health risk indexes were calculated for each record and included in the database. Through several statistical methods, we conducted a meta-analysis of the data on some of the species in this database, identifying trends that could be related to the biology of the investigated organisms, as well as to the physico-chemical properties of each hydrocarbon and to the oceanographic characteristic of this part of the Mediterranean. The analysis of the data showed that, at a consumption rate like the one typical of the Italian population, seafood caught from the area considered in the present work seems to pose a minimal risk to health. However, we also found evidence of an increasing trend of PAH concentrations in Mediterranean mussels, pointing to the need for constant monitoring.

**Keywords:** environmental pollutants; polycyclic aromatic hydrocarbons; seafood; Mediterranean Sea; Adriatic Sea; Ionian Sea; Tyrrhenian Sea; database; meta-analysis; contaminants

#### **1. Introduction**

Polycyclic aromatic hydrocarbons (PAHs) are a large group of toxic [1,2] organic compounds, made up of a variable number of fused aromatic rings of carbon and hydrogen [3,4].

Depending on their origin, PAHs can be classified as petrogenic and pyrolytic (or pyrogenic). Pyrolytic PAHs result from the incomplete combustion of organic matter, such as the combustion of wood, oil, vehicular and industrial emissions and forest fires; petrogenic PAHs, instead, derive from fossil fuels, and can occur as a result of oil spills and petroleum production [5,6].

As regards chemical structure, low molecular weight (LMW) PAHs comprise two to three aromatic rings, while PAHs comprising four aromatic rings are defined as middle molecular weight (MMW) PAHs, and those made of five or more rings are referred to

**Citation:** De Giovanni, A.; Abondio, P.; Frapiccini, E.; Luiselli, D.; Marini, M. Meta-Analysis of a New Georeferenced Database on Polycyclic Aromatic Hydrocarbons in Western and Central Mediterranean Seafood. *Appl. Sci.* **2022**, *12*, 2776. https://doi.org/10.3390/ app12062776

Academic Editors: Raffaele Marotta and Anna Annibaldi

Received: 17 February 2022 Accepted: 3 March 2022 Published: 8 March 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

as high molecular weight (HMW) PAHs [3]. In the marine environment, MMW- and HMW PAHs are mostly pyrolytic in their origin, while LMW PAHs are from petrogenic sources [7,8].

From a chemical point of view, LMW- and HMW PAHs behave differently, with MMW PAHs showing intermediate behavior. In particular, HMW PAHs are much less insoluble in water, and so they tend to bind organic particulate matter, being less bioavailable for their uptake from the water. Despite this, they can still be absorbed by aquatic organisms from sediments and particulate matter present in the water column [3,9].

For non-smokers, the major source of exposure to PAHs is diet; seafood, along with cereals, is a major contributor to the dietary intake of these compounds in Europe [10]. In 2011, the EU set the maximum allowable levels of benzo(a)pyrene (BaP) and of the sum of four HMW PAHs (PAH4; i.e., benzo(a)anthracene BaA, chrysene Chr, benzo(b)fluoranthene BbF, and BaP) in several fishery products (Table 1) [11].

**Table 1.** Maximum levels for PAHs in seafood, set by EU in 2011 [11]. Concentrations are expressed in mg/kg.


<sup>1</sup> *PAH4* sum of BaA, Chr, BbF and BaP.

Several factors could affect PAHs concentrations in seafood. As already mentioned, physico-chemical properties of PAHs influence their bioavailability, HMW PAHs being less available for uptake from the water, compared to LMW and MMW PAHs. Moreover, in fish, HMW PAHs are readily excreted, thanks to their fast metabolism, whose rate is higher compared to that of lighter PAHs [9,12]. PAHs levels in seafood also depend on the nutritional condition of the organism [13] and on the season [14], likely as an effect of changes in the pollutant environmental inputs [12], seasonal variation of the pollutant elimination rate [15], hydrodynamic processes [16], and/or because of factors pertaining to the reproductive cycle of the species [17]. Even the age of the fish is related to PAH accumulation, with earlier stages of life being more prone to accumulating higher amounts of contaminants because of the immaturity of detoxification pathways [15]. Finally, the level of PAHs varies with the species taken into account, and this is also because of differences in the PAH-metabolizing capability [18]. Most notably, filter feeding organisms show the slowest PAHs elimination rates [14]. Accordingly, metabolic efficiency is greater in fish, intermediate in crustaceans and lowest in mollusks [19].

Over the years, several studies have investigated the levels of PAHs in seafood caught in the Adriatic [12,20–35], Ionian [9,20,34,36–39], and Tyrrhenian Seas [7,14,16,17,19,30,40–50], with the aims of monitoring the environmental status of particularly impacted areas [39,48], improving the knowledge of factors influencing PAHs levels in marine species [12,35], and, ultimately, informing decisionmakers when it comes to fixing the limits to PAH levels in fishery products [9].

The aim of this work was to collect and harmonize the results of those studies, in order to obtain a database of georeferenced observations on PAHs in seafood caught in the Western and Central Mediterranean Sea. For each observation, information on the taxonomy and the ecology of the investigated species are reported, as well as any details on the hydrocarbon, temporal and spatial information on sampling. Additionally, two health risk indexes were calculated for each record and included in the database. Finally, we conducted a meta-analysis of the data on Mediterranean mussel, Manila clam, red mullet and common sole, identifying trends that could be related to the biology of the investigated organisms, as well as to the physico-chemical properties of each hydrocarbon and to the water circulation in this part of the Mediterranean.

#### **2. Materials and Methods**

The literature search was carried out using Scopus and PubMed, aiming for the retrieval of publications reporting PAHs levels in seafood, i.e., in marine species included in D.M. 19105, 22 September 2017, Annex 1, which details commercially relevant fish species in Italy, or having "commercial", "minor commercial", "subsistence fisheries", or "of potential interest" in the "Human uses" section on FishBase (https://www.fishbase.se/search.php, accessed on 15 January 2022), or whose specimens were recovered from the fish market. Moreover, publications had to report PAH concentrations measured in marine species caught in the Adriatic (FAO geographical subareas 17 and 18), Ionian (FAO geographical subareas 13–16, 19–21) or Tyrrhenian Seas (FAO geographical subareas 8–10, 11.2, 12).

From each publication, we extrapolated several data, including the sample location and date, species and biological tissue investigated, sample length (cm) and weight (g), sample size, and PAH concentrations detected (mg/kg) (Table S1 for the full list and description of variables included in the database). Moreover, the database comprises a column specifying whether the record comes from wild, farmed or transplanted animals.

Each record was georeferenced, providing latitude and longitude in decimal degrees of the sampling location. Following [51], a geographic precision code was assigned to each record (Table S2), based on whether the study provided the exact coordinates or a more or less precise description of the sample location, from which geographical coordinates were inferred.

The trophic level of each species was obtained from online resources (Table S3) and included in the database.

Abbreviations for the biological tissue where PAHs concentration was measured, as well as for each PAH and molecular weight class (i.e., LMW, MMW or HMW PAHs), are reported in the Supplementary Materials (Tables S4–S6, respectively).

A column specifying if the reported PAH concentration is expressed in fresh (FW), wet (WW) or dry weight (DW) was added to the database (Table S7). Moreover, whenever a study provided concentration in DW, we converted that measure in WW, reporting both the original and the inferred value in separate columns. In the analyses, FW measurements were considered WW. For statistical analyses, when the measured concentration was below the limit of quantification (LOQ) or limit of detection (LOD), and the authors specified those limits, we assigned to that record a value equal to half the LOQ or LOD.

For conversion from DW to WW, we used the following formula:

$$\mathbb{C}\text{ (mg/kg w.w.)} = \text{((100 } - \% \text{ of water)} \div 100) \times \mathbb{C}\text{ (mg/kg d.w.)}\tag{1}$$

where C is the PAH concentration, and % of water is the percentage of water in the analyzed tissue. The percentage of water in Mediterranean mussel (*Mytilus galloprovincialis*) and Manila clam (*Ruditapes philippinarum*) was assumed to be equal to 85%, based on personal data not shown. The percentage of water in common sole (*Solea solea*) liver, gills and muscle was assumed to be equal to 72.75%, 70% and 74.4%, respectively, based on [33]. Finally, the percentage of water in red mullet (*Mullus barbatus*) fillet was assumed to be equal to 80%, following [51].

Two health risk indexes (Table S8), excess lifetime cancer risk (ELCR) and target hazard quotient (THQ), were calculated for each record and included in the database. Specifically, ELCR was calculated according to the following equation:

ELCR = EF (day/yr) × ED (yr) × IR (kg/day) × CSF (mg kg−1 day) ÷ BW (kg) × AT (2)

where EF is the exposure frequency (365 days/year); ED is the exposure duration, which was assumed to be equal to the Italian mean life expectancy (83,226 yr) (ISTAT, 2019); IR is the ingestion rate, which is equal to the PAH concentration times the mean ingestion rate of the species (FAOSTAT, 2018); CSF is the cancer slope factor for each of the analyzed PAH (OEHHA, 2009); BW is the body weight, which was assumed to be equal to the mean body weight of the Italian population (67 Kg) (ANSA, 2013); and AT is the averaging time, which is equal to EF×ED. An ELCR above 10−5, which is the acceptable lifetime risk (ALR) [52], indicates a probability greater than 1 chance over 100,000 of developing cancer [46].

THQ, which indicates the ratio between exposure and the reference dose, was calculated according to the following equation:

THQ = EF (day/yr) × ED (yr) × IR (kg/day) × C (mg/kg) ÷ RFD (mg kg−1 day) × BW (kg) × AT (3)

where C is the PAH concentration, and RFD the oral reference dose for PAH. When THQ risk is above 1, it means that THQ is higher than the reference dose, and systemic effects may occur [46].

Graphs and statistics were produced using RStudio version3.6.1. All the analyses were carried out after grouping records according to PAHs molecular weight, i.e., analyses were conducted separately on LMW, MMW and HMW PAHs in each selected species.

After checking the assumption of normality of residual distributions through the Shapiro–Wilk test (R function: shapiro\_test), we used the Wilcoxon rank sum test (R function: wilcox.test) with data on Mediterranean mussel, Manila clam, red mullet and common sole to calculate the statistical significance of differences between the mean concentrations of LMW, MMW and HMW PAHs in those species.

To look for seasonal trends, we used the Wilcoxon rank sum test to calculate the statistical significance of differences between the mean concentrations of each class of PAHs in cold (October–March) and warm (April–September) months. Additionally, to control for potential confounding factors, we repeated the above analysis using non-parametric ANCOVA (R function: ancova.np), with a sampling depth and sampling year as covariates. In the present work, we used the same month clustering as in [12,53], which is based on sea water temperature: January–March (winter), April–June (spring), July–September (summer) and October–December (autumn).

Finally, to see if there is a relationship between the latitude, sampling depth and sampling year and the PAH concentrations in Mediterranean mussel caught along the Italian coast of the Adriatic and the Tyrrhenian Seas, we used Kendall's rank correlation (R function: pcor; method = "kendall").

#### **3. Results**

#### *3.1. Database Description*

Of the 10,704 records included in the database, 5790 were extracted from a database on contaminants in Mediterranean biota available at https://www.emodnet-chemistry.eu/ data (accessed on 1 December 2021), while the other 4914 are from 38 scientific publications in peer-reviewed journals. The database was compiled along the lines of the work of Cinnirella and colleagues (https://doi.org/10.1594/PANGAEA.899723) on mercury concentration in Mediterranean biota [51].

The geographical distribution of sampling sites is shown in Figure 1.

In total, 1025 records are from the Ionian Sea, 4469 from the Adriatic Sea, and 5210 from the Tyrrhenian Sea (Figure 2).

Mediterranean mussel (*Mytilus galloprovincialis*, code: Mytgal) is the most studied species, with 7710 records from 22 sources, followed by Manila clam (*Ruditapes philippinarum*, code: Rudphi), with 982 records from two sources, and common sole (*Solea solea*, code: Solsol), with 344 records from four sources. All the other species are present in this database with fewer than 250 observations (Figure 3).

**Figure 1.** Geographical distribution of sampling sites (red dots). FAO geographical subareas are reported and colored based whether they belong to Adriatic (cadet blue), Ionian (royal blue) or Tyrrhenian Seas (turquoise).

**Figure 2.** Bar plot showing the number of observations (x axis) by sea (i.e., Ionian, Adriatic, and Tyrrhenian Seas) in this database (y axis), with colors indicating the season in which sampling took place.

Year of publication goes from 1990 to 2021, while sampling year goes from 1981 to 2019.

**Figure 3.** Bar plot showing the number of observations (x axis) for each marine species present in this database (y axis). The bars are arranged in ascending order of number of observations, by taxonomic class. On top of each bar, in brackets, is the number of sources from which the above observations were obtained. Each color corresponds to the taxonomic class of the species. Species codes are on the X-axis and are made of the first three letter of genus and species. The graph was realized using ggbreak library (version 0.0.7) in R [54].

#### *3.2. PAHs Concentration by Molecular Weight and by Season*

We compared the concentrations of LMW, MMW and HMW PAHs in Mediterranean mussel, Manila clam, common sole and red mullet, being that these species are the ones on which more records are available in this database. Summary statistics for each class of PAHs in each species are reported in Table 2. Statistics and *p*-values for each test are reported in Supplementary Materials (Tables S9–S12).

**Table 2.** Summary statistics on LMW (low molecular weight), MMW (middle molecular weight) and HMW (high molecular weight) PAHs in Mediterranean mussel, Manila clam, common sole and red mullet caught in the Adriatic, Ionian and Tyrrhenian Seas. Concentrations are in mg/kg wet weight.


<sup>1</sup> *PAHs class*—class of PAHs based on molecular weight, <sup>2</sup> *Cmean*—mean concentration, <sup>3</sup> *Cmin*—minimum concentration, <sup>4</sup> *Cmax*—maximum concentration.

In all four species, MMW PAHs show the highest mean concentration, followed by LMW PAHs and, finally, HMW PAHs, which show the lowest mean concentration (Figure 4). In all cases, the difference between the mean concentration of each PAHs class was statistically significant (*p* < 0.05), except for the difference between MMW and LMW PAHs in red mullet.

**Figure 4.** PAHs concentration (mg/kg wet weight) by molecular weight in (**a**) Mediterranean mussel, (**b**) Manila clam, (**c**) common sole, and (**d**) red mullet. Sample sizes are in brackets; outliers not shown.

For each class of PAHs, we compared mean concentrations measured in cold and warm months in Mediterranean mussel, Manila clam, and red mullet, controlling for the effect of sampling depth and sampling year (Figure 5). We excluded common sole from the analysis because all samples for which sampling date was specified in the original sources were collected in autumn (October–December). Summary statistics on each variable included in the present analysis are reported in Table 3.

In Mediterranean mussel, the Wilcoxon rank sum test reveals that both MMW- and HMW PAHs are present at significantly (*p* < 0.05) higher concentrations in cold months (Table S13). Moreover, ANCOVA shows that the same trend is statistically significant, even after controlling for the sampling depth and sampling year (Table S14).

In Manila clam, the Wilcoxon rank sum test reveals that all PAHs classes (i.e., LMW-, MMW- and HMW PAHs) are present at significantly higher concentrations in cold months (Table S15). However, as shown by ANCOVA, the above trend remains statistically significant after controlling for sampling year only in LMW- and MMW PAHs (Table S16). Data on sampling depth for Manila clam were not sufficient to use this variable as covariate in ANCOVA.

**Figure 5.** Comparison of PAHs concentrations (mg/kg wet weight) in cold and warm months in (**a**) Mediterranean mussel, (**b**) Manila clam, and (**c**) red mullet; sample sizes are in brackets; outliers not shown.

**Table 3.** Summary statistics on LMW (low molecular weight), MMW (middle molecular weight) and HMW (high molecular weight) PAHs in Mediterranean mussel, Manila clam and red mullet caught in different seasons in Adriatic, Ionian and Tyrrhenian Seas. Concentrations are in mg/kg wet weight; sampling depth is in meters.


<sup>1</sup> *PAHs class*—class of PAHs based on molecular weight, <sup>2</sup> *Cmean*—mean concentration, <sup>3</sup> *Cmin*—minimum concentration, <sup>4</sup> *Cmax*—maximum concentration, <sup>5</sup> *Depthmin*—minimum sampling depth, <sup>6</sup> *Depthmax*—maximum sampling depth, <sup>7</sup> *Yearmin*—year of the first sampling campaign, <sup>8</sup> *Yearmax*—year of the last sampling campaign.

Finally, in red mullet, both the Wilcoxon rank sum test and ANCOVA show that for any of the PAH classes, concentrations in warm months are not significantly different from those in cold months (Tables S17 and S18). Additionally, in the case of red mullet, ANCOVA was carried out with only sampling year as a covariate, as data on sampling depth were not sufficient.

#### *3.3. Latitude, Depth and Sampling Year Effect in Mediterranean Mussel*

We tested the presence of a relationship between the latitude, sampling depth and sampling year and the PAH concentrations in Mediterranean mussel caught along the Italian coast of the Adriatic and the Tyrrhenian Seas. Kendall's rank correlation coefficients (τb) and *p*-values are reported in the Supplementary Materials (Tables S19–S30), along with maps showing the geographical distributions of sampling sites from where data used in each analysis derive (Figures S1–S4).

#### 3.3.1. Adriatic

Summary statistics on each variable included in the correlation analysis on data from the Adriatic Sea are reported in Table 4.

**Table 4.** Summary statistics on LMW (low molecular weight), MMW (middle molecular weight) and HMW (high molecular weight) PAHs in Mediterranean mussels caught in different seasons along the Italian coast of the Adriatic Sea. Concentrations are in mg/kg wet weight; sampling depth is in meters; latitude are in decimal degrees. Reported statistics were calculated after missing-data removal from concentration, sampling depth, sampling year and latitude columns.


<sup>1</sup> *PAHs class*—class of PAHs based on molecular weight, <sup>2</sup> *Cmean*—mean concentration, <sup>3</sup> *Cmin*—minimum concentration, <sup>4</sup> *Cmax*—maximum concentration, <sup>5</sup> *Depthmin*—minimum sampling depth, <sup>6</sup> *Depthmax*—maximum sampling depth, <sup>7</sup> *Yearmin*—year of the first sampling campaign, <sup>8</sup> *Yearmax*—year of the last sampling campaign, <sup>9</sup> *Latmin*—minimum latitude of sampling sites, <sup>10</sup> *Latmax*—maximum of sampling sites.

As Kendall's rank correlation revealed, after removing the effect of the sampling depth and sampling year, concentrations of LMW-, MMW- and HMW PAHs in Mediterranean mussel caught along the Italian coast of the Adriatic Sea turned out to be negatively correlated with latitude in warm months, while the correlation becomes positive in cold months. Results are all statistically significant (*p* < 0.05), except for MMW PAHs in both periods of the year.

In Mediterranean mussel caught along the Italian coast of the Adriatic Sea, after removing the effect of the latitude and sampling year, concentrations of LMW-, MMWand HMW PAHs are always negatively correlated with the sampling depth. Results are all statistically significant (*p* < 0.05), except in the case of MMW PAHs in warm months.

Finally, concentrations of all three classes of PAHs in warm months increase over the years, while concentrations of all three classes of PAHs in cold months decrease over the years, after removing the effect of latitude and sampling depth. Results are all statistically significant (*p* < 0.05), except for those on MMW PAHs in cold months.

#### 3.3.2. Tyrrhenian

Summary statistics on each variable included in the correlation analysis on data from the Tyrrhenian Sea are reported in Table 5.

As Kendall's rank correlation revealed, after removing the effect of the sampling depth and sampling year, concentrations of LMW-, MMW- and HMW PAHs in Mediterranean mussel caught in the Tyrrhenian Sea increase with latitude in both cold and warm periods, and this increase is statistically significant (*p* < 0.05) for every PAH class and period of the year, excluding MMW PAHs.

**Table 5.** Summary statistics on LMW (low molecular weight), MMW (middle molecular weight) and HMW (high molecular weight) PAHs in Mediterranean mussels caught in different seasons in the Tyrrhenian Sea. Concentrations are in mg/kg wet weight; sampling depth is in meters; latitude are in decimal degrees. Reported statistics were calculated after missing data removal from concentration, sampling depth, sampling year and latitude columns.


<sup>1</sup> *PAHs class*—class of PAHs based on molecular weight, <sup>2</sup> *Cmean*—mean concentration, <sup>3</sup> *Cmin*—minimum concentration, <sup>4</sup> *Cmax*—maximum concentration, <sup>5</sup> *Depthmin*—minimum sampling depth, <sup>6</sup> *Depthmax*—maximum sampling depth, <sup>7</sup> *Yearmin*—year of the first sampling campaign, <sup>8</sup> *Yearmax*—year of the last sampling campaign, <sup>9</sup> *Latmin*—minimum latitude of sampling sites, <sup>10</sup> *Latmax*—maximum of sampling sites.

In Mediterranean mussel caught in the Tyrrhenian Sea, after removing the effect of the latitude and sampling year, concentrations of LMW in warm months and of MMW in both periods of the year are negatively correlated with the sampling depth. These results are always statistically significant (*p* < 0.05), apart from those on MMW in cold months. On the contrary, in all other cases (i.e., LMW PAHs in cold months, and HMW PAHs in both periods of the year), the PAH concentrations and sampling depth turned out to be positively correlated, always reaching statistical significance, except for HMW PAHs in warm months.

Finally, concentrations of LMW-, MMW- and HMW PAHs in Mediterranean mussels caught in the Tyrrhenian Sea, in both cold and warm months, increase over the years, and this trend is statistically significant (*p* < 0.05) for LMW- and MMW PAHs in warm months, and for HMW PAHs in both periods of the year.

#### *3.4. Human Health Risks Assessment*

In total, 92 out of 651 (~14%) records of BaP in bivalves exceed the limit of 0.005 mg/kg set by the EU [11] (Table 1), with most of these records (57) being on Mediterranean mussels caught in the Tyrrhenian Sea. Conversely, none of the 26 records of PAH4 in bivalves exceed the limit of 0.03 mg/kg set by the EU [11] (Table 1).

In this database, assuming a consumption rate equal to the average per capita consumption in Italy (FAOSTAT, 2018), ELCR values range from a minimum of 1.36 × <sup>10</sup>−<sup>10</sup> to a maximum of 4.52 × <sup>10</sup>−4, which is reached by a record of dibenzo(a,i)pyrene (DaiP) in Mediterranean mussel. A total of 324 out of 5275 records (~6%) exceeds the threshold value of 10−<sup>5</sup> for ELCR. Records exceeding ELCR threshold value are mostly on Mediterranean mussel (225), while dibenzo(a,h)anthracene (DahA) is the compound most frequently associated with such high ELCR values.

THQ values range from a minimum of 2.2 × <sup>10</sup>−<sup>9</sup> to a maximum of ~0.13, which is reached by a record of BaP in Mediterranean mussel. Therefore, none of the records in this database exceeds the threshold value of 1 for THQ. THQ distributions for some of the most consumed species in Italy [55] are reported in Figure 6. As can be seen, among those included in the graph, European anchovy (*Engraulis encrasicolus*), blue mussel (*Mytilus edulis*) and European hake (*Merluccius merluccius*) are the first three species showing the highest median THQ.

**Figure 6.** Boxplots showing the THQs distribution in several marine species included in this database. Species codes are on the X-axis and are made of the first three letter of genus and species. Boxplots are in ascending order from left to right, based on median values; outliers not shown.

#### **4. Discussion**

The development of this database on PAHs in seafood caught in the Western and Central Mediterranean Sea showed that most of the studies carried out from 1990 to 2021 focused on the Tyrrhenian and Adriatic Seas, with fewer records coming from the Ionian Sea. Moreover, Mediterranean mussel was by far the most studied species, accounting for more than half of records in the database. This is not surprising, given that, along with red mullet, Mediterranean mussel is considered one of the most suitable organisms to be used in biomonitoring studies, because of the widespread distribution, the ability to accumulate contaminants to a degree proportional to their bioavailability, as well as the ease of sampling [41,56].

In this database, MMW PAHs and LMW PAHs showed higher concentrations than HMW PAHs in Mediterranean mussel, Manila clam, common sole and red mullet. This is in line with the expectations, given the greater solubility and bioavailability of lighter PAHs and the faster metabolism of the heavier ones [12].

In a recent investigation carried out using large-scale monitoring data on PAHs in sediments of the Mediterranean Sea [57], the authors found that the two most prevalent PAHs in the Western Mediterranean basin were fluoranthene (Flu) and phenanthrene (Phe), with the former being the most abundant also in the Adriatic Sea and in the Central Mediterranean basin. In view of this, it is interesting to note that, in our meta-analysis, Flu and Phe are also the most abundant PAHs in both Mediterranean mussel and Manila clam (Figures S5 and S6), which are benthic filter-feeding bivalves, and so may be particularly prone to absorbing PAHs accumulated in bottom sediments, after their remobilization.

The analysis on seasonality shows that PAH concentrations tend to be higher in specimens sampled in cold months (October–March), and this trend is statistically significant in both Mediterranean mussel and Manila clam, confirming what several studies, including those in this database, found independently [8,16,19,29,32,41].

As suggested by previous investigators, the reason for such a seasonal pattern may lie both in the biology of these species and in changes in the emission and mobilization of PAHs in the environment. PAHs are lipophilic compounds [49], and as such, they accumulate preferentially in lipid-rich tissues [50]. Given that the lipid content of tissues of marine species can vary under the effect of, for example, nutritional [13] and reproductive status [35,58], one might speculate that seasonal fluctuations in such parameters may be reflected in PAH concentration changes. However, this should not be the case with Mediterranean mussel, since several investigations carried out in the Mediterranean Sea found higher lipid content in mussel sampled during summer [59,60], and this is likely due to the depletion of lipids that takes place after spawning [58,60]. Another possibility is that higher PAHs concentrations in winter are the result of increased PAH emission through, for example, domestic heat [12,17,61,62] and industrial activities [63]. Moreover, in summer, the degradation of PAHs, especially of the lighter ones (i.e., LMW PAHs) [63,64], is enhanced by UV radiation, high temperatures, and ozone [65]. On the other hand, low seawater temperature in winter can inhibit the microbial degradation of PAHs [66,67]. Additionally, the resuspension of particulate matter that takes place during winter via sea storms [19] and water mixing [12] could foster PAHs availability and accumulation in filter feeding organisms, such as mussels, and this is particularly true for heavier and more recalcitrant PAHs [12,68].

In red mullet, lighter (i.e., LMW) and heavier (i.e., MMW and HMW) PAHs show opposite trends, with the former being present at greater concentrations during warm months, and the latter during cold months. Frapiccini and colleagues [35] suggested that higher LMW PAHs in summer months might be related to an increase in maritime traffic during this period of the year, while the increase in heavier PAHs concentrations observed in winter might be caused by a reduced expression in detoxification enzymes in red mullet [69], along with the environmental factors that are hypothesized to determine the same pattern in Mediterranean mussel. Nevertheless, in our meta-analysis, the difference between PAHs concentrations in warm and cold months was not statistically significant. Moreover, it is important to note that, analyzing the liver of red mullets caught in the Northern Adriatic Sea, Guerranti and colleagues [30] found higher levels of PAHs in autumn than in spring, but they found the opposite trend in the liver of red mullets caught in the Southern Adriatic and Tyrrhenian Seas.

In this database, concentrations of PAHs in Mediterranean mussel caught along the Italian coast of the Adriatic Sea are negatively correlated with latitude in warm months, (average τb ∼= − 0.2). On the contrary, concentrations of PAHs and latitude show a positive correlation (average τb ∼= 0.2) in cold months. We tentatively attribute the above results to the water masses circulation that characterizes the Adriatic Sea, and/or to the seasonal changes in the river discharge from the Italian coast. The major riverine freshwater input in the Adriatic Sea [70], the Po River, flows into the northern part of the basin, with greater discharge in spring and autumn, and lower discharge in summer [53]. In winter, the Po plume flows mostly southward [53] along the Italian coast, forming the Western Adriatic Current, a buoyant fresher water layer more than 50 m deep [71]. Moreover, in this period of the year, the cold Bora wind that affects the region causes the cooling of the sea surface layer, resulting in the complete vertical mixing of the sea water. This in turn leads to the resuspension of the bottom sediments and, therefore, to the remobilization of contaminants accumulated therein [53]. On the other hand, during spring and summer, the strong thermal stratification of the water results in the offshore propagation of the Po plume, which reaches the center of the basin [53].

Considering all the above, we hypothesize that the positive correlation between the PAH concentration and latitude observed in cold months may be due to, or at least favored by, the increased input of pollutants from the Po River, coupled with the resuspension of contaminants determined by the cooling action of the Bora wind in the northern Adriatic Sea. This latitudinal trend may be disrupted in warm months because of a greater contribution of pollutants from local rivers in the middle Adriatic, given the weaker influence of the Po plume on the regions south of the Po delta.

Our results seem to agree with a previous study analyzing nutrients transport along the Western Adriatic coast [70]. Indeed, what that study found is that, in winter, nutrient concentrations in water decreased southward from the Po River, while in spring, concentrations off Pescara were higher than concentrations in the Po area.

Although above explanation might sound plausible, our results should be interpreted with caution, considering the high data heterogeneity and the different geographical distribution of observations during warm and cold months, the latter being more discontinuous (Figure S2).

The negative correlation (average τb ∼= − 0.2) observed between the sampling depth and PAH concentrations in Mediterranean mussel caught along the Italian coast of the Adriatic Sea is in accordance with several past investigations. A study carried out in the Western Mediterranean Sea revealed a higher PAH content in suspended particulate matter collected near the sea surface compared to that sampled from deep water [72], while another study found similar results in water samples collected in the Baltic Sea [66]. Moreover, in [7], concentrations of PAHs in Mediterranean mussels collected in the Gulf of Naples were negatively correlated with sampling depth. We hypothesize that the observed bathymetric gradient may be due, at least in the Adriatic Sea, to a greater impact of river discharge at a low depth.

Concerning the discrepancy between results of the correlation analyses on the PAH concentration and sampling year, with concentrations increasing over the years in warm months and decreasing over the years in cold months, we were not able to find a likely explanation based on information retrieved from the literature. Overall, we cannot exclude that our results are an artefact stemming from the high degree of heterogeneity of the data collected in the database. Moreover, it is notable that data on PAH concentrations in Mediterranean mussel caught along the Western Adriatic coast in cold months are from the periods 2006–2009 and 2005–2009, for LMW PAHs and for MMW- and HMW PAHs, respectively, while data on mussels caught in warm months cover a longer period of time (2006–2011 and 2005–2017 for LMW PAHs and for MMW- and HMW PAHs, respectively) (Table 4); therefore, we cannot rule out the possibility of an actual declining trend in PAH concentrations in the period covered by data on cold months, whose signal is lost in the broader period that the data on warm months span. In this regard, it is interesting to note that an oscillating temporal trend of PAHs concentrations was observed in Adriatic sediments by Rizzi and colleagues [57], with several PAHs showing a decrease in concentrations until the years between 2005 and 2010, when a new increase took place (Figure S2 of [57]).

Turning to mussels from the Tyrrhenian Sea, we observed a latitudinal trend, consistent across the three PAHs classes, in both periods of the year, that is, an increase in PAH concentrations from the southern to the northern part of the basin (average τb ∼= 0.1). Interpreting the above results is challenging, given the low number of sampling sites, especially in cold months, and their uneven geographical distribution along the Italian coast. Therefore, we limit ourselves to observe that, in the Tyrrhenian Sea, the salinity of the shallower water mass (0–150 m) follows the same latitudinal trend that we observed for the PAH concentration in mussels, passing from 36.2 psu in the southern region to 38.4 in the northern [73]; salinity superior or equal to 37 psu has been shown to slow down the degradation of PAH molecules [64]. Moreover, water circulation in the Tyrrhenian Sea is dominated by a wide cyclonic path that enters the basin through the Sardinia Channel and flows along the Sicilian and Italian coasts [73]. Thereby, it could be that mussels at more northerly latitudes are affected not only by local input of contaminants, but also by the substances that currents catch along their path. Finally, it is worth noting that, as can be seen from the marine traffic density map [74] of the European Atlas of the Seas [75], the Tyrrhenian coast of Calabria, in the southern part of the basin, is the least affected by vessel traffic.

The inconsistency between the results of the correlation analysis on PAH concentrations and sampling depth in the Tyrrhenian Sea is not easy to interpret, and again, it may represent a simple artefact of data heterogeneity. The statistically significant negative correlation between sampling depth and LMW- and MMW PAHs concentrations in warm months is in line with the results on LMW- and HMW PAHs in the Adriatic Sea and may be attributable to a greater impact of river discharge at low depth. On the

contrary, we found that LMW- and HMW PAHs measured in the Tyrrhenian Sea in cold months significantly increase with sampling depth. Although we cannot find a convincing explanation to such a pattern, it is noteworthy that the Tyrrhenian Sea is rich in volcanic submarine structures [76,77], which can be important contributors of pyrolytic PAHs in the environment [78,79]. As such, we hypothesize that Tyrrhenian mussels that inhabit greater depths may be more susceptible to contamination driven by submarine volcanism, being at the same time less affected by river discharge.

The increasing trend of PAH concentrations in Mediterranean mussels from Tyrrhenian Sea is in line with recent findings suggesting that, in recent years, concentrations of these compounds in Mediterranean Sea sediments have increased, especially in the western part of the basin, probably as an effect of a parallel rise in PAH emissions from forest fires [57].

Both human and animal studies point to PAH exposure as being detrimental for the health, due to, for example, their carcinogenicity, teratogenicity and endocrine-disrupting effects [2]. Accordingly, the EU set a maximum level of several PAHs in fresh and smoked seafood to be sold [11] (Table 1). None of the records included in this database exceed those limits. Moreover, based on FAOSTAT data on per capita seafood consumption in Italy (FAOSTAT, 2018), none of the records in this database exceed the threshold value of 1 for THQ, pointing to a minimum risk of running into systemic effects at a consumption rate like the one typical of the Italian population. Conversely, in around 6% of cases, samples exceed the threshold value of 10−<sup>5</sup> for ELCR, pointing to a probability of greater than 1 chance over 100,000 of developing cancer [46]. Ultimately, at a consumption rate like the one typical of the Italian population, seafood caught from the area considered in the present work seems to pose a minimal risk to health. However, it should be considered that the seafood ingestion rate is variable among the population [80], and that an individual is simultaneously exposed to several PAH sources, from both ingestion and other routes. Furthermore, it is vital to note the emerging role of genetics in shaping individual susceptibility to PAHs [81].

#### **5. Conclusions**

Gathering the results of several investigations, we produced a database on PAHs in seafood from the Western and Central Mediterranean Sea. A clear imbalance in favor of studies addressing PAHs in bivalve mollusks emerged.

The meta-analysis carried out on the database led us to obtain potential hints on factors (e.g., reproductive status, water masses circulation, and river discharge seasonal variability) that could determine differences in the PAH contamination of marine species.

The assessment of human health risks posed by PAH seafood contamination showed that, at a consumption rate like the one typical of the Italian population, seafood caught from the study areas seems to pose a minimal risk to health. Despite this, concerns may arise considering the individual susceptibility to PAHs exposure as well as the apparent increasing trend of PAHs levels observed in both environmental matrices and sea animals.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/app12062776/s1. Table S1: Columns of the database Table S2: Geographic precision code used in the database. Table S3: Sources of the trophic levels of the organisms included in the database. Table S4: Tissues codes used in the database. Table S5: PAHs included in the database. Table S6: PAHs molecular weight abbreviations used in the database. Table S7: Water content abbreviations used in the database. Table S8: Indices and formulas included in the database. Table S9: Pairwise comparisons between the three PAH classes in Mediterranean mussel, using Wilcoxon rank sum test. *P*-value adjustment method: Benjamini–Hochberg. Table S10: Pairwise comparisons between the three PAHs classes in Manila clam, using Wilcoxon rank sum test. *P*-value adjustment method: Benjamini–Hochberg. Table S11: pairwise comparisons between the three PAHs classes in common sole, using Wilcoxon rank sum test. *P*-value adjustment method: Benjamini–Hochberg. Table S12: Pairwise comparisons between the three PAHs classes in red mullet, using Wilcoxon rank sum test. *P*-value adjustment method: Benjamini–Hochberg. Table S13: Comparison between mean concentrations measured in cold and warm months in Mediterranean mussel, using Wilcoxon rank sum test with continuity correction. Table S14: Comparison between mean concentrations measured

in cold and warm months in Mediterranean mussel, controlling for the effect of sampling depth and sampling year with ANCOVA. Table S15: Comparison between mean concentrations measured in cold and warm months in Manila clam, using Wilcoxon rank sum test with continuity correction. Table S16: Comparison between mean concentrations measured in cold and warm months in Manila clam, controlling for the effect of sampling year with ANCOVA. Table S17: Comparison between mean concentrations measured in cold and warm months in red mullet, using Wilcoxon rank sum test with continuity correction. Table S18: Comparison between mean concentrations measured in cold and warm months in red mullet, controlling for the effect of sampling year with ANCOVA. Table S19: Pairwise correlation coefficients with *p*-values (among brackets) between latitude, sampling depth and sampling year and LMW PAHs concentrations in Mediterranean mussel caught along the Italian coast of the Adriatic Sea in warm months, Table S20: Pairwise correlation coefficients with *p*-values (among brackets) between latitude, sampling depth and sampling year and LMW PAHs concentrations in Mediterranean mussel caught along the Italian coast of the Adriatic Sea in cold months. Table S21: Pairwise correlation coefficients with *p*-values (among brackets) between latitude, sampling depth and sampling year and MMW PAHs concentrations in Mediterranean mussel caught along the Italian coast of the Adriatic Sea in warm months. Table S22: Pairwise correlation coefficients with *p*-values (among brackets) between latitude, sampling depth and sampling year and MMW PAHs concentrations in Mediterranean mussel caught along the Italian coast of the Adriatic Sea in cold months. Table S23: Pairwise correlation coefficients with *p*-values (among brackets) between latitude, sampling depth and sampling year and HMW PAHs concentrations in Mediterranean mussel caught along the Italian coast of the Adriatic Sea in warm months. Table S24: Pairwise correlation coefficients with *p*-values (among brackets) between latitude, sampling depth and sampling year and HMW PAHs concentrations in Mediterranean mussel caught along the Italian coast of the Adriatic Sea in cold months. Table S25: Pairwise correlation coefficients with *p*-values (among brackets) between latitude, sampling depth and sampling year and LMW PAHs concentrations in Mediterranean mussel caught in the Tyrrhenian Sea in warm months. Table S26: Pairwise correlation coefficients with *p*-values (among brackets) between latitude, sampling depth and sampling year and LMW PAHs concentrations in Mediterranean mussel caught in the Tyrrhenian Sea in cold months. Table S27: Pairwise correlation coefficients with *p*-values (among brackets) between latitude, sampling depth and sampling year and MMW PAHs concentrations in Mediterranean mussel caught in the Tyrrhenian Sea in warm months. Table S28: Pairwise correlation coefficients with *p*-values (among brackets) between latitude, sampling depth and sampling year and MMW PAHs concentrations in Mediterranean mussel caught in the Tyrrhenian Sea in cold months. Table S29: Pairwise correlation coefficients with *p*-values (among brackets) between latitude, sampling depth and sampling year and HMW PAHs concentrations in Mediterranean mussel caught in the Tyrrhenian Sea in warm months. Table S30: Pairwise correlation coefficients with *p*-values (among brackets) between latitude, sampling depth and sampling year and HMW PAHs concentrations in Mediterranean mussel caught in the Tyrrhenian Sea in cold months. Figure S1: Geographical distributions of sampling sites of Mediterranean mussel caught along the Italian coast of the Adriatic Sea in warm months. Figure S2: Geographical distributions of sampling sites of Mediterranean mussel caught along the Italian coast of the Adriatic Sea in cold months. Figure S3: Geographical distributions of sampling sites of Mediterranean mussel caught in the Tyrrhenian Sea in warm months. Figure S4: Geographical distributions of sampling sites of Mediterranean mussels caught in the Tyrrhenian Sea in cold months. Figure S5: Boxplots of the concentration of single PAHs in Mediterranean mussel. The color of the boxes indicates the molecular weight of the given PAH. Boxplots are arranged in ascending order by median. Figure S6: Boxplots of the concentration of single PAHs in Manila clam. The color of the boxes indicates the molecular weight of the given PAH. Boxplots are arranged in ascending order by median.

**Author Contributions:** Conceptualization, A.D.G. and M.M.; methodology, A.D.G. and M.M.; formal analysis, A.D.G., P.A.; data curation, A.D.G.; writing—original draft preparation, A.D.G.; writing review and editing, P.A., E.F., D.L., M.M.; supervision, E.F., D.L., M.M.; project administration, D.L., M.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The comma-separated values file of the database presented in this study is in the Supplementary Materials.

**Acknowledgments:** The research leading to these results was conceived under the collaboration between the University of Bologna and the National Research Council for the implementation of the International PhD Program "Innovative Technologies and Sustainable Use of Mediterranean Sea Fishery and Biological Resources" (www.FishMed-PhD.org, accessed on 15 January 2022).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Polycyclic Aromatic Hydrocarbons and Polychlorinated Biphenyls in Seawater, Sediment and Biota of Neritic Ecosystems: Occurrence and Partition Study in Southern Ligurian Sea**

**Luca Rivoira 1,\*, Michele Castiglioni 1,\*, Nicola Nurra 2,3, Marco Battuello 2,3, Rocco Mussat Sartor 2,3, Livio Favaro <sup>2</sup> and Maria Concetta Bruzzoniti 1,\***


**Abstract:** The Mediterranean Sea is subjected to a high anthropic pressure, which determines direct or indirect discharges of persistent organic pollutants deriving from intensive industrial activities. These compounds could easily enter and contaminate the whole marine compartment, with possible transfers (and contamination) among water, sediment and biota. Based on the above-mentioned assumptions, in this work we studied the presence of 16 polycyclic aromatic hydrocarbons (PAHs) and 14 dioxin and non-dioxin-like polychlorinated biphenyls (PCBs) in the neritic protected marine area of the Southern Ligurian Sea, affected by the impact of human activities. The study was focused on the possible partition of micropollutants within seawater, sediment and zooplankton. Results showed that both seasonal and anthropic causes strongly affect contaminant transfer behaviors, with summertime periods more impacted by PAH and PCB contamination. Regarding the PAH contamination, low molecular weight congeners were mainly detected in the target matrices, revealing concentrations up to 1 μg/L in seawater (anthracene), 250 μg/Kg in sediments (benzo[b]fluoranthene) and 2.3 mg/Kg in carnivorous copepods. Concerning PCBs, only few congeners were detected in the matrices studied. To better understand the occurrence of preferential bioaccumulation pathways in zooplankton, partition studies were also performed in several taxa (hyperbenthic Isopoda, holoplanktonic crustacean copepods and ichthyoplankton) through the calculation of BAF values, observing that both living and feeding habits could influence the bioaccumulation process.

**Keywords:** PAH; PCB; neritic environment; seawater; sediment; biota; partition; contamination

#### **1. Introduction**

Neritic environments are peculiar marine areas acting as interface between the atmosphere, the sea and the continental masses [1]. Due to their proximity to land and to sunlight infiltration, they are rich in nutrients and, consequently, in biologic activities, showing a remarkable biodiversity and biomass of algae, seagrasses and animal organisms inhabiting the coastal ecosystems.

Zooplankton, along with phytoplankton, representing the most abundant form of life in terms of biomass and biodiversity in neritic ecosystems, is comprised of heterotrophic microscopic, unicellular or multicellular organisms with size classes ranging from a few microns (picozooplankton) to a millimeter or more (mesozooplankton), such as the gelatinous zooplankton [2]. The marine zooplankton can be divided into two ecological categories: holoplankton, spending their entire lifecycle in the water column (e.g., crustaceans such as

**Citation:** Rivoira, L.; Castiglioni, M.; Nurra, N.; Battuello, M.; Sartor, R.M.; Favaro, L.; Bruzzoniti, M.C. Polycyclic Aromatic Hydrocarbons and Polychlorinated Biphenyls in Seawater, Sediment and Biota of Neritic Ecosystems: Occurrence and Partition Study in Southern Ligurian Sea. *Appl. Sci.* **2022**, *12*, 2564. https://doi.org/10.3390/ app12052564

Academic Editors: Mauro Marini and Anna Annibaldi

Received: 30 December 2021 Accepted: 25 February 2022 Published: 1 March 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

copepods and krill) and meroplankton, spending only the larval stages of their lifecycle as zooplankton, such as decapod crustaceans, echinoderms and fish larvae that, once they mature, adopt a benthic or nektonic lifestyle.

Taxa can be further distinguished depending on their typical living habits (permanently in the water column or diel vertical migration from benthos to sea surface such as hyperbenthic organisms e.g., Isopoda) and on their feeding habits and behavior (herbivorous, carnivorous and omnivorous).

Zooplankton plays a critical ecological role in marine food webs, both in the neritic and pelagic ecosystems, since it is involved in the conservation of energy from primary producers (phytoplankton) to higher trophic levels [3], in biogeochemistry cycles and in supporting the ocean's biological pump of carbon export. Hence, an alteration of this category, i.e., through chemical contaminations, can lead to disruptions up to the highest trophic levels.

Due to their peculiar conformation, neritic areas are strongly vulnerable to harmful effects of several anthropogenic pollutants derived from human activities, even when located far away from the pollution sources [4]. Such compounds could easily enter the whole marine ecosystem, with possible transfer (and contamination) along its main compartments, namely the water column, the sediment and the biota [5–7]. Transfer and contamination pathways mainly depend on the physicochemical properties of the pollutant molecules [8] or on the living habits of biota. Moreover, it is worth noting that the fate of pollutants is strongly influenced by the characteristics of the sea, such as water exchange capacity, as the enclosed or semi-enclosed basins (i.e., the Mediterranean Sea) are more susceptible to the accumulation of pollutants in marine sediments than open basins [9]. Among the organic micropollutants frequently detected in marine waters, are polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs) [10,11]. PAHs represent pollutants of natural and anthropogenic origins (e.g., volcanic eruptions, combustion) while PCBs mainly derive from use in industrial activities (e.g., dielectric fluids, motor oils).

Seawater contamination from PAHs and PCBs derives from the not negligible impact of ports, industries and touristic activities located on coastal areas [12,13], as well as from wastewater treatment plants [14] and river effluents [15]. Atmospheric deposition presents an additional source of marine water contamination [16]. Since both PAHs and PCBs are mutagenic/carcinogenic recalcitrant pollutants [17], there are serious concerns regarding their presence in the environment [18–20], particularly in neritic compartments, and also with consideration of their tendency to partition between water and sediments and from water and sediments to biota, due to their high octanol/water partition coefficients (logKOW ranging from 2.96 to 5.60 for PAHs and from 5.41 to 7.83 for PCBs) [21].

In Europe, the environmental status of water compartments (including neritic areas) is regulated by the 2013/39/EU directive [22], adopted in Italy by the Legislative Decree 172/2015 [23], in which a list of priority substances (including selected PAHs and PCBs compounds) to be monitored in water, sediments and biota is reported. For individual congeners of PAHs, Environmental Quality Standards (EQS) are fixed for water, sediments and biota, defining annual average values (AA-EQS) and maximum allowable concentrations (MAC-EQS). Conversely, AA-EQS for PCBs in biota refer only to dioxin-like congeners, while with regards to sediments the same dioxin-like compounds, together with other eight congeners, are considered. It should be highlighted that, to date, the limits for PCBs in marine waters remain unregulated.

Due to the strategic importance of coastal areas in the marine ecosystems, as previously presented, several studies investigated the contamination from PAHs and PCBs in neritic environments, as well as their partition between water columns, sediments and biota [24–26]. Within this literature, an extensive bibliography is dedicated to the Mediterranean Sea which, due to its morphological peculiarities, is strongly subjected to anthropic impact. In particular, studies on PAH and PCB contamination of the Venice Lagoon [27], Tyrrhenian Sea [28], and Ionian Sea [29] are reported. On the other hand, few studies

were focused on Ligurian Sea [30–32], even though its pollution causes serious concern, due to intense anthropic activities (e.g., ports, industries, tourism, etc.) [32]. Furthermore, studies investigating the possible distribution pathways of PAHs and PCBs within neritic compartments of Ligurian Sea are, as far as we know, absent thus far, despite the natural (marine protected area) and economic (shipping, industries, harbors) strategical importance of this area.

Based on the above considerations, the aim of this work is the study of the contamination of PAHs and PCBs (both dioxin and non-dioxin-like congeners) in the neritic environment of the Southern Ligurian Sea, which hosts a protected marine area called "Cetacean Sanctuary". The study focuses comprehensively on the possible partition of micropollutants among the environmental compartments of the marine ecosystem (seawater, sediment and zooplankton), and on the factors that could influences these transfer behaviors (i.e., seasonality). With this aim, analytical protocols for the extraction and quantitation of PAHs and PCBs from seawater, sediment and zooplankton were developed and validated. Furthermore, to better understand the presence of preferential pathways of bioaccumulation in zooplankton, partition studies were also carried out in four taxa, chosen on the basis of dietary habits (herbivorous or carnivorous) and living environment: hyperbenthic Isopoda, herbivores and carnivorous holoplanktonic crustacean copepods and fish larvae (ichthyoplankton). To the best of our knowledge, this study is the first to describe the partition of PAHs and PCBs in this protected area of the Ligurian Sea, with unique natural characteristics, providing insights on the contamination of the whole bioma.

#### **2. Materials and Methods**

#### *2.1. Sampling Area and Sample Pre-Treatments*

Waters, sediments and biota were sampled in the Ligurian Sea, in an offshore area (12.5 nautical miles off the Italian coast), above the continental shelf, at the border with Northern Tyrrhenian Sea, the same study area of previous research [5] and included the following stations: (i) 43◦29- 40-- N–10◦01- 45-- E, (ii) 43◦28- 10-- N, 10◦01- 55-- E and (iii) 43◦27- 10-- N, 10◦03- 00-- E (see Figure S1 of Supplementary Material). The area was considered of great interest for several reasons. In fact, it is characterized by one of the highest levels of shipping in the Mediterranean basin; moreover, a strong anthropogenic influence and impact are present due to several commercial, industrial and harbor activities. The investigated area hosts the "Cetacean Sanctuary", an extensive marine protected area where the number of cetaceans is at least double that of any other part of the Mediterranean. Finally, the area is subjected to a high dynamic water current equilibrium, since both perpetual and seasonal currents influence the possible dispersion and partition of micropollutants originating from the highly developed coastline of Italy.

Two samplings were carried out for both water column and marine biota during summer 2017 and winter 2018, in order to assess the possible contribution of seasonality. For the sediment, a single sample was collected in summer 2017, due to the low effect of the seasonality on the circalittoral mud sediments.

Sampling methods were performed as follows: (i) water column, 5 L collection for each sampling using a PVC Niskin bottle water sampler (collected at 5 m depth). Subsequently, the samples were filtered via qualitative paper and refrigerated until analysis; (ii) sediments, sampled using a 18 L Van Veen grab sampler on the seabed at 110 m depth. Moisture evaporation (at 60 ◦C for 12 h) was performed prior to micropollutant extraction and analysis; (iii) marine biota (mesozooplankton and ichthyoplankton) was collected in a surface haul, using a WP-2 standard net (200 μm mesh size and diameter 57 cm). The horizontal sampling time was approximately 15 min (2 knots vessel cruising speed). The net was fitted with a flow-meter (KC Denmark model 23.090) to measure volume of water filtered. After collection, each zooplankton sample was washed, on board first, with filtered seawater from the sampling site, in order to remove terrigenous or inorganic particles, and hence washed with distilled water. After the washing procedure, samples were immediately fixed in 70% ethanol and seawater and stored in the dark.

In the laboratory, a qualitative–quantitative analysis of the marine biota was performed with a Leica stereomicroscope and microscope, which allowed the identification and regrouping of the following four taxa: hyperbenthic Isopoda, holoplanktonic crustaceans' copepods both herbivores and carnivorous and fish larvae (ichthyoplankton). Samples were dried at 80 ◦C for 12 h before extraction and analysis.

#### *2.2. Reagents*

Acetone ≥ 99.8%, dichloromethane ≥ 99.9%, 2-propanol ≥ 99.8%, methanol ≥ 99.9%, sodium hydroxide ≥ 98.0%, sulfuric acid 96–97%, magnesium sulphate anhydrous ≥ 99.5% and NaCl ≥ 99.5% were obtained from Honeywell Riedel-de-Haën, Fisher Scientific Italia, Rodano, MI (Italy). Cyclohexane 99.5% and dichloromethane were obtained from VWR International (Radnor, PA, USA).

High-purity water (18.2 MΩ cm resistivity at 25 ◦C), produced by an Elix-Milli Q Academic system (Millipore-Billerica, MA, USA) was used.

The d-SPE sorbent used was Primary Secondary Amine (PSA) from Agilent Technologies (Santa Clara, CA, USA).

The 16 PAHs studied, i.e., naphthalene (NaPh), acenaphthylene (AcPY), acenaphthene (AcPh), fluorene (Flu), phenanthrene (Phe), anthracene (Ant), fluoranthene (Flth), pyrene (Pyr), benzo[a]anthracene (BaA), chrysene (Chr), benzo[b]fluoranthene (BbFl), benzo[k]fluoranthene (BkFl), benzo[a]pyrene (BaP), indeno[1,2,3-cd]pyrene (Ind), dibenzo[a,h]anthracene (DBA) and benzo[ghi]perylene (BP), were the compounds listed by the United States Environmental Protection Agency (US-EPA) and were purchased from Wellington Laboratories (Guelph, ON, Canada).

The 14 PCBs studied were purchased from Chemical Research 2000 (Rome, Italy). They were chosen according to the results of the main monitoring campaigns and included 3,3- -dichlorobiphenyl (PCB 11), 4,4- -dichlorobiphenyl (PCB 15), 2,4,4- -trichlorobiphenyl (PCB 28), 2,2- ,5,5- -tetrachlorobiphenyl (PCB 52), 2,2- ,4,5,5- -pentachlorobiphenyl (PCB 101), 2,2- ,3,4,4- ,5-hexachlorobiphenyl (PCB 138), 2,2- ,4, 4- ,5,5- -hexachlorobiphenyl (PCB 153) and 2,2- ,3,4,4- ,5,5- -heptachlorobiphenyl (PCB 180). Furthermore, the following dioxin-like PCBs were included 3,4,4- ,5-tetrachlorobiphenyl (PCB 81), 2,3- ,4,4- ,5-pentachlorobiphenyl (PCB 118), 2- ,3,4,4- , 5-penta-chlorobiphenyl (PCB 123), and 2,3- ,4,4- ,5, 5- -hexachlorobiphenyl (PCB 167), 3,3- ,4,4- ,5,5- -hexachlorobiphenyl (PCB 169) and 2,3,3- ,4,4- ,5,5- -heptachlorobiphenyl (PCB 189)

Isotope labelled compounds for PAHs (5 mg/L) and for PCBs (2 mg/L), both purchased from Wellington Laboratories, were employed as internal standards and surrogates to obtain calibration curves and to calculate extraction recoveries. The deuterated PAH surrogate solution included the following compounds: benzo[a]anthracene-d12 (BaA-d12), chrisene-d12 (Chr-d12), benzo[b]fluoranthene-d12 (BbFl-d12), benzo[k]fluoranthene-d12 (BkFl-d12), benzo[a]pyrene-d12 (BaP-d12), indeno[1,2,3-cd]pyrene-d12 (Ind-d12), dibenzoanthracene-d14 (DBA-d14), benzoperylene-d12 (BP-d12). The 13C-PCB surrogate solution included the following congeners: 13C12-PCB28, 13C12-PCB52, 13C12-PCB118, 13C12-PCB153, and 13C12-PCB180. 2H-anthracene and 13C12-PCB70 were used as internal standards.

#### *2.3. Chromatographic Analysis*

PAHs and PCBs, extracted from neritic matrices as detailed in the following paragraphs, were analyzed by gas chromatography coupled with mass spectrometry (GC–MS) using an Agilent 6980 series gas chromatograph and an Agilent 5973 Network MS detector controlled by Agilent ChemStation software. The gas chromatograph was provided with an autosampler of the Agilent 7683 Series.

The GC column was a (5%-Phenyl)-methylpolysiloxane column (DB-5 ms, 30 m × 0.25 mm × 25 μm; Agilent). Helium was employed as gas carrier (1 mL/min). MS detection was performed in Single Ion Monitoring (SIM) mode at proper *m*/*z* ratio (*m*/*z* ratio available upon request). Injections (2 μL) were performed by the pulsed splitless

mode (pressure at 40 psi for 2.5 min, injector temperature 280 ◦C). The oven ramp for CH3CN and CH2Cl2 was set as follows: starting temperature: 80 ◦C, hold for 2 min; ramp to 176 ◦C, 12 ◦C/min rate; ramp to 196 ◦C, 5 ◦C/min rate, hold for 3 min; ramp to 224 ◦C, 12 ◦C/min rate; ramp to 244 ◦C, 12 ◦C/min rate, hold for 3 min; ramp to 270 ◦C, 7 ◦C/min rate, hold for 3 min; final ramp to 300 ◦C, 5 ◦C/min, hold for 10 min to completely clean and restore the GC column. The complete separation of the 13 PAHs and 14 PCBs was obtained within 49 min.

#### *2.4. Extraction of PAHs and PCBs from Neritic Matrices*

All the extraction procedures used for water, sediment and biota were specifically optimized for this study and are hereafter detailed. Aliquots of all the matrixes were extracted in triplicate. Blank samples were run in parallel to evaluate possible ambient contamination.

#### 2.4.1. Water Column

Extraction was performed via solid-phase extraction (SPE) using an SPE Vacuum manifold and a polymeric reversed-phase (RP) cartridge (STRATA-XL-100 μm; Phenomenex, Torrance, CA, USA), The cartridge was conditioned (20 psi) with 5 mL CH2Cl2, 5 mL 2-propanol, and 5 mL H2O. Then, 200 mL water sample added with 20 mL 2-propanol was loaded (50 psi) on the SPE cartridge. The water sample container was subsequently washed with 20 mL 2-propanol–water solution (10 + 90, *v*/*v*). After loading, the cartridge was washed (20 psi) with 5 mL H2O, and 5 mL 2-propanol–water solution (85 + 15, *v*/*v*). The cartridge was dried for 10 min, and the analytes were finally eluted with two aliquots of 1.0 mL CH2Cl2. The eluted extract was spiked with the internal standard solution of PAHs and PCBs to achieve a final concentration of 5 μg/L and injected for GC-MS analysis. To evaluate the extraction recoveries of PAHs and PCBs, before extraction, the water samples were spiked with surrogates (as detailed in "Reagents" section) to achieve a final concentration of 2 μg/L.

#### 2.4.2. Sediment

Sediments were extracted through a QuEChERS procedure. In detail, 5 g sediment (previously spiked with surrogates to achieve a final concentration of 2 μg/L in the extract solvent) were extracted in 10 mL CH2Cl2 with the addition of 1 g NaCl and 0.4 g MgSO4. The extraction mixture was shaken for 5 min and subsequently centrifuged for 10 min (1534× *g*). For the d-SPE purification step, 6 mL of the supernatant was transferred into a 15 mL tube containing 50 mg of PSA and 150 mg MgSO4. The mixture was again shaken for 5 min and centrifuged for 10 min (7871× *g*). A 2 mL aliquot of the supernatant was spiked with internal standards to achieve a final concentration of 5 μg/L and injected for GCMS analysis.

#### 2.4.3. Biota

Each taxon was separately extracted to study possible partition pathways. A total of 100 mg of each taxon were weighted inside a 50 mL tube and spiked with surrogates to achieve a final concentration of 2 μg/L in the extract solvent. Subsequently, 16 mL 4 M NaOH, 4 mL methanol and 10 mL CH2Cl2 were added. The extraction mixture was shaken for 1 min, sonicated for 15 min and refrigerated at 4 ◦C for 2 h. Finally, the tube was centrifuged for 10 min (7871× *g*) and a 2 mL aliquot of the supernatant was spiked with internal standards to achieve a final concentration of 5 μg/L and injected for GCMS analysis.

#### *2.5. Recovery Evaluation*

To evaluate the apparent recoveries of each extraction method developed, each sample was spiked, before extraction, with surrogate solutions of PAHs and PCBs to achieve a final concentration of 2 μg/L (*CS)*. After extraction, the concentrations were calculated by using an external standard calibration curve prepared in dichloromethane and the extraction yield (*E%*) was calculated according to the following equation:

$$E\% = \frac{C\_c}{C\_s} \ast 100$$

where *Ce* is the calculated concentration of the surrogate after extraction expressed as μg/L.

#### *2.6. Validation of Analytical Protocols*

The protocols developed for the extraction and quantitation of PAHs and PCBs from water, sediment and biota were validated through the evaluation of linearity, limits of detection (MDL) and quantitation (MQL). Linearity was evaluated over ten concentration levels, within a concentration range included between 0.87 μg/L and 14 μg/L for PAHs and between 0.42 μg/L and 6.75 μg/L for PCBs in CH2Cl2. The values of MDL and MQL for the 30 target compounds were calculated by means of the response error and the slope of the calibration curve, using the expression MDL = 3.3 Sy/m, and MQL = 10 Sy/m, where Sy = response error; m = slope of the calibration.

#### *2.7. Bioaccumulation Factor*

Bioaccumulation factor (BAF) is defined as the ratio of the concentration of a micropollutant in an organism to the concentration of the same compound in water [33]. Logarithmic BAFs were individually estimated for all the analytes detected both in biota matrices, as follows:

$$\log{BAF} = \log{\frac{\mathcal{L}\_{biota}}{\mathcal{C}\_{water}}}\tag{1}$$

Finally, for both herbivorous and carnivorous zooplankton, the average of *logBAF* values was calculated and results were compared and discussed to understand possible effects of feeding habits in the bioaccumulation process.

#### **3. Results and Discussion**

#### *3.1. Optimization of Analytical Protocols*

To verify the effectiveness of the analytical methods developed for the analysis of PAH and PCB occurrence and partition in water, sediment and biota, extraction yields, methods detection (MDL) and quantitation limits (MQL) were assessed.

For each matrix, the extraction recoveries were determined using surrogates and were identified in the following ranges, water: from 52% (Ind-d12) to 88% (BaA-d12) for PAHs and from 54% (13C12-PCB118) to 77% (13C12-PCB28) for PCBs, with RSD% lower than 10% for all the analytes; sediment: from 54% (Ind-d12) to 96% (BaA-d12) for PAHs and from 56% (13C12-PCB52) to 86% (13C12-PCB28) for PCBs, with RSD% lower than 7% for all the analytes; biota: from 80% (DBA-d14) to 93% (BaA-d12) for PAHs and from 79% (13C12- PCB28) to 93% (13C12-PCB118) for PCBs, with RSD% lower than 8% for all the analytes. Extraction yields are reported in Figures S2–S4 of Supplementary Materials for both PAHs and PCBs in the three matrices.

The results here obtained for all the matrices are in the same range (or higher) than other studies devoted to the analysis of PAHs and PCBs in marine compartments [25,34,35]. Differently from the procedures typically applied to sediments, the proposed QuEChERS protocol is green and sustainable since it does not require high volumes of organic solvents. Furthermore, the method does not require a final evaporation step (which increases the risk of losing target analytes) [25].

MDL and MQL for water, sediment and biota are reported in Tables S1 and S2, together with regulation limits (if present). MQL were in the following ranges, water: from 4.6 ng/L (NaPh) to 27 ng/L (DBA) for PAHs and from 6 ng/L (PCB11) to 32 ng/L (PCB123) for PCBs; sediment: from 3 μg/Kg (DBA) to 20 μg/Kg (AcPY) for PAHs and from 1 μg/Kg (PCB153) to 3 μg/Kg (PCB123) for PCBs; biota: from 10 μg/Kg (BaP and Ind) to 21 μg/Kg (BP) for PAHs and from 15 μg/Kg (PCB138) to 35 μg/Kg (PCB189) for PCBs. These values fully satisfy the limits fixed by the Italian and European regulations (previously described).

#### *3.2. Chemical Characterization of Seawater*

PAHs detected in seawater (both in summer and winter samples) are summarized in Figure 1.

**Figure 1.** Concentration of PAHs detected in seawater samples, both in summer (orange) and winter (green) sampling. Analytical conditions are detailed in the Experimental section.

It is interesting to observe that PAHs with a higher number of aromatic rings (i.e., BbFl, BaP, DBA, BP and Ind) were not detected in seawater, regardless of the season, in accordance with their higher hydrophobicity behavior (and hence lower solubility in water).

In addition, an inter-seasonal variability of PAH concentrations (among summer and winter sampling campaigns) was also clearly present for all the detected analytes. Indeed, a higher number of congeners was detected in summer (8), rather than in winter (2), with a total PAH concentration about three times higher in summer than winter (2.2 μg/L and 0.8 μg/L, respectively).

To explain these differences, both climatological and anthropogenic effects should be considered. To elucidate, one of the main aspects to consider is the seasonal variation of thermocline. Thermocline refers to the water layer below the surface, characterized by a massive temperature gradient which decreases up to 4 ◦C (or lower for deep waters). During summertime, surface waters increase in temperature while the deeper waters remain cold. Stratification causes rapid vertical changes in the density of water, which largely prevents the mixing of the water column [36]. This phenomenon could affect the sinking and the accumulation of micropollutants [37]. Conversely, during winter, temperatures of surface water can undergo those of deeper layers, which tends to rise due to its lower density (homothermic conditions). Consequently, the movement of water masses promotes a mixing and, therefore, a diffusion of dissolved compounds, such as pollutants, thus justifying their lower concentrations.

In addition to the thermocline effect, the higher concentration of PAHs in the summer season could be addressed to the effects of the Northern current, the northern part of the cyclonic surface circulation of the North-Western Mediterranean Sea. Indeed, the Northern current exhibits its weakest intensities during summertime [38], thus promoting a possible accumulation of land and air derived pollutants, while during winter its higher intensities boosts their dispersion into the sea waters.

Finally, the intense touristic activities that characterize the Ligurian Sea during summertime (with private boat traffic, cruises, etc.) with respect to winter, could act as a possible source of PAHs, which can be released by thermic engines. Evidence of the effects of maritime traffic on the increment of PAH contamination were also investigated elsewhere [39]. In addition, the origin of PAHs (evaluated by studying the ration between congeners [40]) confirmed this hypothesis. In fact, both the ratio between the concentration of low (NaPh, AcPh, AcPY, Flu) and high molecular weight (Phe, Ant, FlTh, BkFl) PAHs (0.92) and the ratio between phenanthrene and anthracene (0.30) are lower than 1, thus suggesting a dominant pyrogenic origin of the detected PAHs (i.e., transport activities, use of petroleum derived products, accumulation of small emissions of gasoline, engine oils, incomplete combustion of organic substances) [41,42].

It should be noted that PAH concentrations detected in sea water within this study were in the same level of other Mediterranean areas [43], with a pollution classification from moderate to high, according to the indications proposed by Baumard and colleagues [44]. Concerning regulation limits, benzo[k]fluoranthene was the only congener detected among the compounds included in the regulated list, with an average concentration of 45 ng/L, which exceeds the maximum allowable concentrations (17 ng/L) [22].

Concerning PCBs, only congeners 15 and 138 were detected, with concentrations below the quantitation limits.

#### *3.3. Chemical Characterization of Sediment*

As discussed in the previous section, the presence of PAHs and PCBs was observed in seawater. Hence, the analysis of sediment belonging to the same sea area is justified to assess whether a partition took place from the waters. Results are reported in Figure 2.

**Figure 2.** Concentration of PAHs and PCBs detected in sediment. Analytical conditions are detailed in the Experimental section.

Results showed that, as for seawater, PAHs containing the highest number of aromatic rings (i.e., DBA, BP and Ind) were not present (Figure 2), except for BbFl and BaP that were not previously detected in the water column.

The total amount of PAHs in sediment was assessed to reach about 700 μg/Kg, with most of the contribution ascribed to higher molecular weight compounds (i.e., BaA (14 μg/Kg), Chr (73 μg/kg), BbFl (250 μg/Kg), BkFl (99 μg/Kg) and BaP (230 μg/Kg)). The

concentration of these compounds accounted for almost 95% of the total PAH concentration. Upon comparing these latter results (Figure 2) with those previously presented for the water column (Figure 1), it was observed that the contribution of the compounds with higher molecular weight to the total PAH content measured higher in sediment rather than in water. Such differences remain in good agreement with the physicochemical properties of PAHs and may be ascribed to the different partitioning attitudes of the individual compounds. Indeed, the higher the molecular weight, the higher the octanol–water partition coefficients, thus enhancing the sorption on the organic matter present in the sediment [45]. In addition, it should be mentioned that neritic areas are highly subjected to sediment resuspension [46] which causes organic matter to rise temporarily in water and, hence, favoring the sorption of PAHs dissolved in water.

As for seawater, the ratios between PAH congeners (see "Chemical characterization of seawater" section) confirm a pyrogenic origin of the PAHs detected in sediment (ΣLowPAHs/ΣHighPAHs = 0.04 and BaP/(BaP + Chr) = 0.76). Hence, detected contamination should be tentatively ascribed to industrial activities, as well as to maritime transport.

Concerning PCBs, detected concentrations were about two orders of magnitude lower in comparison to PAHs (Figure 2), ranging from 1.2 μg/Kg (PCB167) to 3.3 μg/Kg (PCB101). Even if no congeners were detected in seawater, their presence in sediment at quantifiable concentrations could be ascribed to the sorption properties of sediment towards microorganic pollutants, as described for PAHs [47]. A similar behavior was also confirmed by studies performed in other marine areas, where PCBs were detected at fractions of ng/L in water and at μg/Kg in sediments [48].

It should be mentioned that PAH concentrations detected in sediment exceeded the limit imposed by Italian and European regulations for BbFl (limit: 40 μg/Kg), BkFl (limit: 40 μg/Kg) and BaP (limit: 40 μg/Kg). As for the sum of PCB congeners (10.5 μg/Kg), this value partially exceeded the limit of 8 μg/Kg. However, similar values are not uncommonly found in the Mediterranean sea, due to the high impact of anthropic activities, as shown by other studies [44,49].

#### *3.4. Chemical Characterization of Biota and Partition Studies*

PAH and PCB contamination was measured in four zooplankton taxa, namely *hyperbenthic Isopoda*, herbivores and carnivorous *holoplanktonic crustaceans copepods* and fish larvae-*ichthyoplankton,* previously sorted after summer and winter sampling campaigns. The results obtained are hereafter discussed.

#### 3.4.1. Copepods

Regarding *crustaceans* copepods, results are summarized in Figure 3A (herbivorous) and Figure 3B (carnivorous).

Data showed that PAHs were detected in both taxa at higher concentrations than PCBs, with a total concentration from 270 μg/Kg (winter) to 1.3 mg/Kg (summer) for PAHs witn respect to 43 μg/Kg for PCBs (detected only in summer campaign) for herbivorous copepods. For carnivorous copepods, PAHs concentrations from 230 μg/Kg (winter) to 4.3 mg/Kg (summer) were measured, with no detection of PCBs. The measured concentrations are in the same range of those obtained by Cailleaud and colleagues in estuarine water (in which seasonal variability was also assessed) [50] or by Tiano and colleagues in Marseille Bay (Mediterranean Sea) [51].

Since copepods tend to spend their entire lives mostly in the water column [52], comparison with PAH and PCB contamination in this matrix seems to be fully justified. Concerning PAHs, the same congeners (NaPh, AcPY, Flu, Phe, Ant, FlTh and BkFl) detected in seawater were detected in both herbivorous and carnivorous copepods (Figure 1 vs. Figure 3A,B). Furthermore, the same seasonal trend observed for water (i.e., higher number of congeners and concentrations in summer sampling, rather than in winter) was observed for copepods. This behavior suggests that a possible transfer of micropollutants from water to copepods had occurred. To support this hypothesis, it is important to note that Phe

and Ant, which were found to be almost the most abundant PAH compounds in water, confirmed this trend also in herbivorous copepods.

(**A**)

**Figure 3.** Concentration of PAHs and PCBs detected in herbivorous (**A**) and carnivorous (**B**) copepods. Both summer (orange) and winter (green) sampling are reported. Analytical conditions are detailed in the Experimental section.

In contrast, Pyr, BaA and Chr, which were detected in the herbivorous copepods, were not present in carnivorous copepods and in seawater. The presence of these congeners could be explained taking into account feeding habits, since it was recently demonstrated that herbivorous copepods could be deceived in their feeding by microplastics and nanoplastics, upon which algae and phytoplankton grow [53]. Such micropolymers have been shown to adsorb organic and inorganic pollutants (such as PAHs and PCBs) [54], thus suggesting that they could be an additional source of PAH congeners once ingested by organisms [55].

Moreover, the same hypothesis could be formulated to explain the detection of PCB11 and 101 in herbivorous copepods (Figure 3A), which were not previously detected in the water column and that were below detection limits in carnivorous organisms. The observed PCB contamination does not exceed limits imposed by European and Italian regulation, since both congeners are not dioxin-like and limits on biota are referred only to dioxin-like compounds.

#### 3.4.2. Hyperbenthic Isopoda

PAH and PCB contamination detected in *hyperbenthic Isopoda* was observed to be almost absent, apart from NaPh, detected at 280 ± 20 μg/Kg, and PCB101, detected at 140 μg/Kg (summer-winter average concentrations). RSD% between summer and winter measured below 8% for both analytes, indicating that, for this taxon, a seasonal variance is not present.

Moreover, It is interesting to note that PCB101, the most abundant pollutant detected in sediment (see "Chemical characterization of sediment" paragraph), is the only congener detected in *hyperbenthic Isopoda*. Since *hyperbenthic Isopoda* was assessed to live half of its life in contact with sediment [56], a possible transfer could not be excluded.

No comparison of contamination level with previous studies could be performed (as for copepods), since, to the best of our knowledge, this is the first study investigating PAH and PCB occurrence in *Hyperbenthic Isopoda.*

#### 3.4.3. Ichthyoplankton

Finally, the PAH and PCB occurrence was also determined for the fourth taxon, fish larvae (*ichthyoplankton*). Results revealed, regarding PAHs, only NaPh and AcPy were detected, with concentrations below quantitation limits, while PCB congeners were not detected.

Comparing to previous taxa, *ichthyoplankton* showed the lowest contamination; such behavior could be ascribed to fish larvae feeding habits. Indeed, *ichthyoplankton* feeding is typically based on copepods larvae that are characterized by a short lifetime and, hence, a short exposure to pollution agents dissolved in water. The transfer of pollutants is, therefore, further hindered.

Finally, it should be mentioned that PAH concentrations detected in the four target taxa did not exceed regulation limits (5 μg/Kg of BaP).

#### *3.5. Bioaccumulation Factors*

The BAF (Bioaccumulation Factor) represents one of the most used models to predict the partitioning between an exposure medium (such as seawater) and biota [57]. Two possible mechanisms could influence the bioaccumulation of copepods (and hence the BAF values), namely the ingestion of contaminated dietary sources and the diffusive sorption of pollutants from water [58].

In the present study, BAF values were calculated for the PAH congeners detected both in seawater and in copepods (NaPh, AcPY, Flu, Phe, Ant, FlTh and BkFl). To evaluate a possible effect of feeding habits in the bioaccumulation, BAF values were calculated for herbivorous and carnivorous copepods. PCBs were not included since they were seldom detected in all the sea compartments investigated. The logarithms of BAF values are summarized in Table 1.


**Table 1.** Logarithmic Bioaccumulation Factors (BAF) from seawater calculated for both herbivorous and carnivorous copepods.

The logBAFs obtained revealed positive values for both herbivorous and carnivorous copepods, thus highlighting that, for both taxa, a bioaccumulation from seawater occurred. Average values, 2.8 and 3.4 for herbivorous and carnivorous copepods, respectively, are in good agreement with those discovered by Arias et al. in Pseudodiaptomus marinus copepods [59].

LogBAF average values of herbivores and carnivores were shown to be statistically different (t-test, N = 7, *p* = 0.001), meaning that higher values observed for carnivorous copepods corresponded to a higher bioaccumulation. Even if it is not possible to univocally explain this behavior, this higher value could represent a first biomagnification effect, since carnivorous copepods feed mainly on herbivorous ones. Hence, an increase of pollutant concentrations along the trophic chain is not unexpected.

The plotting of logBAFs against logKow (Figure 4) exhibited a sound correlation for both taxa (R2 = 0.92 for herbivorous and R2 = 0.81 for carnivorous), supporting the hypothesis that the lipophilicity of the molecule provides the main factor driving the type of PAH congeners in copepods.

**Figure 4.** Linear modelling of LogBAFs against LogKow of PAHs for both herbivorous (black squared symbols) and carnivorous (red circled symbols) copepods.

#### **4. Conclusions**

This work represents the first study devoted to the comprehensive investigation of 16 PAHs and 14 PCBs (including dioxin-like congeners) in the seawater, sediment and biota of the Ligurian Sea.

For both PAHs and PCBs, the overall results indicate that the contamination levels are equivalent to those typically found in other areas of the Mediterranean Sea characterized by moderate to high pollution. Moreover, this study reveals that PAHs origin is mainly ascribed to pyrogenic activities (combustion, transport, etc). A seasonal trend was observed for the presence of PAHs and PCBs in seawater, since micropollutant concentrations measured in summer were higher than those measured in winter, in agreement with increased touristic and harbor activities and with the higher stratification of water layers.

Micropollutant partition between water and sediment was confirmed to be influenced by the hydrophobicity of the pollutant molecules. To elucidate, higher molecular weight pollutants were preferentially detected in sediment, with organic matter acting as adsorbent. Bioaccumulation on biota was also assessed for copepods.

In addition, partition pathways were investigated within four different taxa (hyperbenthic Isopoda, herbivores and carnivorous holoplanktonic crustaceans copepods and fish larvae-ichthyoplankton), by the calculation of BAF values. The effects of living habits were shown to strongly influence bioaccumulation, as the congeners detected in sediment were preferentially discovered in taxa that spent most of their lives there. Moreover, feeding habits were found to influence the bioaccumulation of PAHs and PCBs, and a positive correlation between BAF values and hydrophobicity was confirmed. This study performed in the Southern Ligurian Sea, provides a novel insight into PAH and PCB distribution in marine zooplankton, which represents the main food constituent of most fish of the coastal waters, thus playing an important role in the transfer of pollutants through the food web. The experimental approach here followed can be transposed to other geographical areas of strategic importance to understand peculiar distributions patterns.

**Supplementary Materials:** The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/app12052564/s1, Figure S1. Ligurian Sea (Western Mediterranean): study area and location of sampling stations; Figure S2: average (*n* = 3) PAH and PCB recoveries from seawater samples. Recoveries were calculated using surrogates as described in the Materials and Methods Section; Figure S3: average (*n* = 3) PAH and PCB recoveries from sediment samples. Recoveries were calculated using surrogates as described in the Materials and Methods Section; Figure S4: average (*n* = 3) PAH and PCB recoveries from biota samples. Recoveries were calculated using surrogates as described in the Materials and Methods Section; Table S1: MDL and MQL of PAHs in seawater, sediment and biota, using optimized analytical protocols, compared with their regulation limits reported by 2013/39/EU normative and Italian Legislative Decree 172/2015; Table S2: MDL and MQL of PCBs in seawater, sediment and biota, using optimized analytical protocols, compared with their regulation limits reported by 2013/39/EU normative and Italian Legislative Decree 172/2015.

**Author Contributions:** Conceptualization, L.R., M.C., N.N. and M.C.B.; Formal analysis, M.C.; Funding acquisition, M.C.B.; Investigation, L.R., N.N., M.B., R.M.S. and M.C.B.; Methodology, L.R., M.C., N.N. and M.C.B.; Project administration, M.C.B.; Supervision, M.C.B.; Validation, L.R. and M.C.; Visualization, M.B., L.F. and M.C.B.; Writing—original draft, L.R.; Writing—review and editing, L.R., N.N., R.M.S. and M.C.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Ministero dell'Università e della Ricerca, Ex60%.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** The authors would like to thank Andrea Dalla Libera and Lamberto Giusti for their kind assistance in laboratory experiments. Financial support from Ministero della Ricerca e dell'Università (Ex-60%) is gratefully acknowledged.

#### **Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

