The patterns of elemental concentration ( Ca , Na , Sr , Mg , Mn , Ba , Cu , Pb , V , Y , U and Cd ) in shells of invertebrates representing different CaCO 3 polymorphs : a case study from the brackish Gulf of Gdańsk ( the Baltic Sea ) 5

The shells of calcitic arthropod Amphibalanus improvisus, aragonitic bivalve Cerastoderma glaucum, Limecola balthica, Mya arenaria and bimineralic bivalve Mytilus trossulus were collected in the brackish waters 15 of the southern Baltic Sea in order to study patterns of bulk elemental concentration (Ca, Na, Sr, Mg, Ba, Mn, Cu, Pb, V, Y, U and Cd) in shells composed of different crystal lattices (calcite and aragonite). The factors controlling the elemental composition of shells are discussed in the context of crystal lattice properties, size classes of organisms and potential environmental differences between locations. Clams that precipitate fully aragonitic shells have a clear predominance of Sr over Mg in shells, contrary to dominant Mg accumulation over Sr in calcitic shells 20 of barnacles. However, the barnacle calcite shell contains higher Sr concentration than bivalve aragonite. The elemental variability between size-grouped shells is different for each studied species and the elemental concentrations tend to be lower in the large size classes compared to the smaller size classes. Biological differences between and within species, such as growth rate, feeding strategy including feeding rate and assimilation efficiency or composition and contribution of organic material seem to be important factors determining the elemental 25 accumulation in shells. Because specimens used in this study were obtained from different sampling sites within the gulf, the impact of local environmental factors associated with sampling location, such as sediment type, cannot be excluded.

equator (Kuklinski and Taylor, 2009;Loxton et al., 2014;Taylor et al., 2014). Thermodynamics predicts that aragonite is the stable phase at pressures higher than 5000 hPa (roughly 40 m depth), and calcite is the stable phase at lower pressures. However, aragonite is still the major constituent of shells or pearls, indicating its metastable formation in shallow waters (Sunagawa et al., 2007). The incorporation of Sr was suggested to play a significant 80 role in the biomineralogical precipitation of aragonite (Allison et al., 2001). Many studies have demonstrated a clear correlation between the concentration of Sr in the hard parts and precipitation of the aragonite layer (Iglikowska et al., 2016;Reeder, 1983). The ionic radius of Sr is larger than that of Ca; thus, Sr is more likely to form 9-fold coordination, which triggers metastable aragonite nucleation (Sunagawa et al., 2007).
Many studies have demonstrated that the biological control of shell composition is often more important 85 than the environmental control (Carré et al., 2006;Freitas et al., 2005Freitas et al., , 2006Gillikin et al., 2005b). Elements incorporated into skeletons originate from the environment, yet as some of them are the components of enzymes or body fluids, the trace element pathway to the skeleton can be altered by the biological processes (Cubadda et al., 2001;Luoma and Rainbow, 2008), which can affect the relationship between the concentrations of a given trace element between skeleton and environment. In biologically controlled mineralization, the organism drives 90 the process of nucleation and growth of the minerals in a way that is not entirely dependent on the environmental conditions. Endogenous factors manifest themselves through co-regulation of all the structures and functions of the organism, including its sex, growth rate, metabolism, and feeding strategy (Lowenstam and Weiner, 1989).
The main physiological processes involved in element accumulation are ingestion, assimilation, elimination and growth (Wang and Fisher, 1997). Throughout the lifespan, the biological system experiences ontogenetic trends 95 and seasonal variations in physiology, determining metabolic expenses based on life's needs. Biological effects have been repeatedly used to explain shifts of elemental concentrations in shells from a theoretical equilibrium (Davis et al., 2000;Roger et al., 2017;Watson et al., 1995). Ontogenetic fluctuations of the growth rate and metabolic activity affect the intensity of the element uptake (Lee et al., 1998). Vander Putten et al. (2000) concluded that the seasonality of the accumulation of Mg, Sr and Pb in Mytilus edulis shells shows significant 100 similarity across individuals, with a maximum during spring and early summer, and the elemental profiles cannot be explained by seasonal variations in the seawater composition. Carré et al. (2006) developed a model of ion transport in bivalve shells that shows that Ca 2+ channels are less ion-selective when Ca 2+ fluxes are higher. Other studies have found that the rate of trace element uptake increases as mussel filtration rate increases (Janssen and primary production and freshwater inflow) determine the element sources and drive the physiological processes of living organisms (Urey et al., 1951).

Study area 120
The study area is located in the Gulf of Gdańsk in the southern Baltic Sea, more precisely, in the outer Puck Bay and the central Gulf of Gdańsk (Fig. 1). The north-western part of the gulf is separated by the Hel Peninsula and in the west and south the coastline stretches (Kruk-Dowgiałło and Szaniawska, 2008;Rainbow et al., 2004). This location makes the seawater the most turbulent in January and the calmest in June, with weak bottom currents and minimal tidal amplitudes. The hydrophysical parameters of the gulf are mostly driven by the temperate climate 125 and the following seasonal changes. Differences in air temperature and water mixing cause seasonal fluctuations of the surface water temperature, ranging from approximately 4 to 22°C (Uścinowicz, 2011). The Gulf of Gdańsk is a low-salinity system under the influence of brackish water from the open southern Baltic Sea and fresh waters from rivers, mainly the Vistula River; the Vistula is the largest river in Poland and has an average annual inflow in the estuary of 1080 m³ s -1 , which varies seasonally from 250 to 8000 m³ s -1 and has a maximum in spring 130 (Cyberski et al., 2006). Thus, the average water salinity in the gulf is 7, varying from approximately 5.5 in summer to 8.4 in winter (Bulnheim and Gosling, 1988;Szefer, 2002).
The Gulf of Gdańsk is an area highly influenced by human activities. This is due to intensive usage of its resources and to anthropogenic emissions originating from various coastal sources, river inflows and atmospheric deposition. The most significant input of industrial and municipal pollution into the gulf is derived from the Vistula 135 River, which transports pollutants from a catchment area of 194,000 km 2 (Pruszak et al., 2005). Both the water discharge and sediment load into the gulf are strongly seasonally dependent. Because of the local conditions, mainly the limited water exchange, river-borne contaminants remain in the ecosystem for decades, accumulating in the sediments and in living organisms (Glasby et al., 2004;Szumiło-Pilarska et al., 2016).
Most regions of the Baltic Sea have lower salinity and alkalinity (that is, lower Ca 2+ and CO3 2-140 concentrations) than oceanic surface waters (Beldowski et al., 2010;Cai et al., 2010;Findlay et al., 2008). Due to the seasonality of temperature and biogeochemical cycle, the amplitude of CaCO3 saturation state ( Ω = 84. C. glaucum can tolerate habitats with wide range of temperatures, from periodically freezing to above 30°C. It is a filter feeder that actively lives near the sediment surface, acting as a biodiffuser (Urban-Malinga et al., 2013). 155 The clam is surrounded by a ribbed aragonite shell, which is externally yellowish to greenish brown (Jelnes et al., 1971). In the brackish environment of the Gulf of Gdańsk, C. glaucum spawns in May-July and typically lives up to 4 years, achieving a height of 27 mm (Żmudziński, 1990).
The soft-shell clam Mya arenaria (Mollusca, Bivalvia; Fig. 2.2) is a marine invasive species introduced into European waters from the Atlantic coasts of North America (Behrends et al., 2005). It has a wide global 160 distribution, mainly due to its adaptability to varying environments with salinities between 4 and 35 and temperatures between -2 and 28°C (Gofas, 2004;Strasser et al., 1999). M. arenaria is a filter feeder, filtering organic particles and microinvertabrates using long fused siphons, and a deposit feeder. In the Gulf of Gdańsk, M.
arenaria is a common inhabitant of shallow waters down to a depth of 30 m. It spawns once or twice a year in spring or summer, at temperatures of 10 -15°C. Individuals live 10 -12 years. They have aragonitic shells and 165 grow up to 70 mm (Żmudziński, 1990 (Strelkov et al., 2007). It is a euryhaline clam capable of living in a wide range of water salinities from 3 to 40 and at temperatures from -2°C to above 30°C (Sartori and Gofas, 2016). L. balthica is a 170 filter feeder and deposit feeder, has a semi-sessile lifestyle, with the ability to undertake periodic migrations (Hiddink et al., 2002). In the Baltic Sea, L. balthica lives at depths down to 40 m and grows to 24 mm. Adults reproduce in spring when the water temperature reaches 10°C and live 12 years (Żmudziński, 1990). They have aragonitic shells varying in colour between individuals and locations, mainly exhibiting white, pink, yellow and orange (Sartori and Gofas, 2016). 175 purity) for lighter elements (V, Mn, Cu, Y and Cd) to minimize the molecular interference from plasma and solution components and Ca from the samples.
The accuracy and reproducibility were checked by analyses of JLs-1 and JDo-1 before and after every 235 batch of samples. The results obtained for all elements were within ± 2.5 SD of the recommended values (Imai et al. 1996). Accuracy of Pb determination cannot be checked using these CRMs because of the large spread of reference values probably due to insufficient homogeneity of Pb distribution in these samples. Based on the analyses of CRMs and matrix-matched solutions, the maximum analytical error for the typical range of concentrations in the shells can be estimated (in relative percentage) as 1.5% for Ca, Mg and Sr; 3% for Ba; 20% 240 for Cu and U; and 10% for all other elements. More details on method validation were reported previously (Piwoni-Piórewicz et al., 2017).

Statistical analyses
To evaluate the effect of the shell size (ontogenesis stage) on the concentrations of trace elements in calcareous parts of A. improvisus, C. glaucum, M. arenaria L. balthica and M. trossulus, the concentrations of Ca, Na, Sr, 245 Mg, Mn, Ba, Cu, Pb, V, Y, U and Cd were examined in the four size classes separately for each species. The data were not normally distributed (Shapiro-Wilk test); therefore, significant differences between the mean concentrations of the selected trace elements in the size classes were identified by one-way Kruskal-Wallis nonparametric ANOVA (p-value = 0.05) and post-hoc Dunn's tests for multiple independent groups. Statistical computing and graphical visualizations were performed in RStudio. 250

Results
The concentrations of trace elements in studied individuals can be found in Table S1 in the Supplement. The mean concentrations in all studied organisms decreased, being the highest for Ca and the lowest for Cd, in the following order Ca>Na>Sr>Mg>Mn>Ba>Cu>Pb>V>Y>U>Cd. However, the concentrations of given elements were different for Cerastoderma glaucum, Mya arenaria, Limecola balthica, Mytilus trossulus and Amphibalanus improvisus, showing high variability both between and within species (Table 3, Fig. 3).
There is a pattern of the highest concentration of Ca in all individuals, that ranges from 316 ± 26 mg g -1 (mean ± standard deviation) in A. improvisus to 363 ± 18 mg g -1 in M. arenaria. The rest of elements had mean concentrations in shells below 4.0 mg g -1 . In mussels most concentrated were Na>Sr>Mg, while in barnacles Mg>Na>Sr. The concentration of Na ranged, on average, from 2.0 ± 0.3 mg g -1 in M. trossulus to 3.2 ± 0.2 mg g -260 The trace elements (Mn, Ba, Cu, Pb, V, Y, U, Cd) exhibit a general trend of highest concentrations in A.
The results of the Kruskal-Wallis nonparametric ANOVA test, which was used to compare the element concentrations between the four size classes in each species, revealed the lack of variability within M. arenaria. U (H = 18.202, p < 0.001) into shells decreased in larger mussels (Fig. 4). 285 Detailed analyses of the differences in the studied elements between the size classes based on post-hoc Dunn's tests for multiple independent groups indicated that the significant variations were not linear (Fig. 4). In L. balthica, Na concentration decreased in larger shells, showing differences between the size classes I and III and I and IV. In the shells of C. glaucum, Sr concentration increased gradually, reaching a peak in the size class IV, while statistically significant differences were observed between the size classes I and IV and the III and IV. An 290 inverse pattern was observed for Na, yet its concentration differed only between the smallest and largest clams.
Shells of C. glaucum were also characterized by a common trend of the Mn and Cd, which decreased from I size class and later reached a plateau. In shells of A. improvisus, the concentrations of Mg, V, Cu and Pb decreased in larger individuals. The levels of Mg in shells statically differed between the size classes III and IV; likewise, shells from the size class III had the highest concentrations. The elements V and Pb occurred at the highest concentrations 295 in shells of the smallest individuals, and later, they oscillated around a similar mean value. Cu decreased in growing shells of barnacles, reaching the minimum in the size class III. There were no statistically significant differences between the classes III and IV, and the oldest half of the group maintained the downward trend. However, it is worth noting that Cu concentrations in many of the oldest shells are several times higher. Shells of M. trossulus were characterized by the highest variability of trace element concentrations between size classes. The trend of 300 decreasing concentrations was clearly marked for V, Cu, Cd and U. The concentration of Y decreased in larger shells, but the oldest shells showed an upward trend; therefore, significant differences were found between the size classes I and III (Fig. 5).
The comparison of elemental concentrations in shells of mussels and barnacles from different regions, based on literature data is presented in Table 4.

CaCO3 polymorph type and elemental concentrations
The average Ca concentration of all species collected in the Baltic Sea was found to be 343 mg g -1 (Table 3), which corresponds to c. 86% weight per weight (w/w) of pure CaCO3. For comparison, same species or congeners from different regions contain between 310 and 420 mg g -1 of Ca in shell (Table 4), the results above 400 mg g -1 310 probably being caused by analytical bias.
Among shell impurities, the most concentrated elements were Na, Sr, Mg in all studied CaCO3 polymorphs (Fig. 3). These compositions are considered to be typical in calcareous skeletons of marine invertebrates, as those elements are energetically favoured in CaCO3 crystal lattice, substituting for Ca (Allison et al., 2001;Iglikowska et al., 2016;Reeder, 1983;Sugawara and Kato, 2000;Wang and Xu, 2001, see also Table  315 4). The main elemental constituents of the Baltic shells showed generally uniform distribution in collected samples, and within the ranges typical for a given species. Clams, that precipitate fully aragonitic shells and bimineralic mussels were characterized by the following order of accumulated concentrations: Na>Sr>Mg, while in barnacles this order was different: Mg>Na>Sr (Fig. 3). The most evident observation was that the aragonitic shells of C.
glaucum, M. arenaria and L. balthica contained over 15 times more Sr than Mg. In the calcitic A. improvisus, Mg 320 was 1.8 times more concentrated than Sr, while the bimineralic shells of M. trossulus containing layers of calcite and aragonite, were distinguished by equalized concentrations of Mg and Sr in shells (Table 3, Fig. 3). Such pattern of Mg and Sr is observed in number of calcareous species including bivalves and barnacles (Table 4; Dalbeck, 2008;Iglikowska et al., 2016;Wang and Xu, 2001;Zhao et al., 2017). However, by comparing species, it is evident that the calcite of barnacles A. improvisus deviated from this general trend. Out of the species studied in this work, 325 both Mg and Sr concentrations were higher in calcitic barnacles than in aragonitic clams (Table 3, Fig. 3). Kinetic and biological effects influence the partitioning of Sr between the shell and seawater (Urey et al., 1951), and Sr in shells is known to significantly exceed its concentration expected at the thermodynamic equilibrium (Schöne et al., 2010). Furthermore, previous studies revealed that barnacles have significantly higher Sr shell concentration (up to 2.3 mg g -1 ) than other marine invertebrates (Carpenter and Kyger, 1992;Ullmann et al., 2018). The 330 concentration of Sr in shells of A. improvisus from this study (2.3 ± 0.2 mg g -1 ) supports these findings, suggesting that barnacle precipitation process differs from other marine organisms. Carpenter and Kyger (1992) concluded that Sr is a sensitive indicator of precipitation rate this may explain the elevated Sr concentrations in barnacles.
The shell increment in barnacles in rapid (15 -30 mm per year, Milliman, 1974)  properties, yet in a strongly species-specific way (Skinner and Elderfield, 2005). 340 The concentrations of Mn and Ba are several orders of magnitude higher in the calcitic shells of barnacle A. improvisus than in other species (Table 3, Fig. 3). Based on the ionic radii of Mn and Ba, it is expected that aragonite shells would incorporate Ba more intensively than would calcitic shells (Findlater et al., 2014;Gillikin et al., 2006), yet this trend is not observed in this study, indicating Ba as not clearly related to the crystal lattice orientation. Furthermore, barnacles appear to have a stronger capacity for Mn incorporation into shells than other 345 benthic calcifiers (Pilkey and Harris, 1966), even A. improvisus collected at the same location (GN, Fig. 1), possibly due to the species-specific biological factors (Bourget, 1974). Yet, as both species with highest concentration of Mn produced calcite (Table 3, Fig. 3), we should not rule out that the polymorph type of CaCO3 to some degree regulates the level of shell Mn in low salinity environment.
It was observed that an increased incorporation of Mg and Sr into shell can contribute to distortion of 350 CaCO3 crystal lattice and results in parallel increased incorporation of trace elements into the shells (Harriss, 1965;Davis et al., 2000;Finch and Allison, 2007;Dalbeck, 2008;). In our study, shell lattices of mussels and barnacles could be changed through the concentration of Mg and Sr higher than in clams. Therefore, the concentration of Cu, V, Cd, Y and U seems to be driven to some degree by crystal lattice properties having the highest values in species containing calcite in their shells (Fig. 3). It is also important to note that the smaller ionic radii of V, Cd, 355 Y, U and Cu are energetically favoured in calcite, while larger Pb radius is favored in aragonite structure (Morse et al., 1997;Reeder, 1983;Wang and Xu, 2001). However, trace elements Cu, V, Cd, Y were present at the highest levels in bimineralic M. trossulus, while only U, as well as Pb, in calcitic A. improvisus (Table 3), which suggests that additional factors, other than the crystal lattice effects, determine those concentrations. Furthermore, this study likewise revealed inconsistent variability of trace element concentrations between aragonitic clams. Out of the 360 three species of clams, M. arenaria was characterized by the lowest concentration of all trace elements (Table 3).
This also indicates that factors other than mineral properties co-regulate trace elements accumulation in shells.
This study considers element concentrations in shells that were not subjected to chemical removal of organic matter prior to the dissolution of the carbonate matrix. Chemical cleaning of carbonate skeletons prior to chemical analysis and both improvement of the data quality and potential artefacts associated with this are widely 365 discussed in the literature (Barker et al., 2003;Holcomb et al., 2015;Loxton et al., 2017), but the plausible pretreatment method for the removal of organics still needs to be found (Inoue et al., 2004). Mannella et al. (2020) showed that the suitability of chemical pre-treatments for organic matter removal from carbonate matrices should be evaluated on a case-by-case basis and, in case of relatively low organic content, should be avoided. In addition to CaCO3 crystal lattice, shells of barnacles and bivalves usually contain up to 5.0 wt% of organic matter (Bourget, 370 1987;Wolowicz and Goulletquer, 1999;Marin and Luquet, 2004;Rueda and Smaal, 2004). However, the specific features and composition of given organic matrix might be a cause of the inter-species and inter-individual variability of elemental concentration in shells (Fig. 3, Table 3; Takesue and van Geen, 2004). The organic fraction was found to be generally not associated with significant level of trace elements (Lingard et al., 1992;Takesue et al., 2008), which are strongly incorporated into the crystal phase. Yet, many authors found correlations, especially 375 for Mg and Mn, associated with the shell organic matrix (Bourget, 1974;Walls et al., 1977;Lorens et al., 1980;Rosenberg et al., 2001;Takesue and van Geen, 2004). Despite replacing with Ca in shell lattice, these elements are biologically essential (Bellotto and Miekeley, 2007) and inter-species variability of Mg and Mn between barnacles, mussels and clams (Fig. 3) could be enhanced by specific properties of organic phases. Furthermore, trace elements in shells may be present in microscopic aqueous fluid inclusions that were trapped within crystals 380 during their growth from solution (Gaffery, 1998). Lécuyer and O'Neil (1994) found that such inclusion waters constitute up to 2% of the shell and probably represent the remnants of metabolic fluids produced by the mantle epithelium. Therefore, their composition most likely results from the specific biological features of an organism rather than from the structural properties of calcium carbonate.

Size classes and potential biological impact on elemental concentrations
The recorded concentrations of trace elements in all populations exhibited marked variability among individuals (Table 3, Fig. 3), a feature previously recorded by several authors (Gillikin et al., 2005a;Vander Putten et al., 2000). In this study, individuals were collected over a wide range of sizes (Table 2), representing different ages and life spans. Bivalves are long-living organisms, and these of the Gulf of Gdańsk have life expectancy of 4 -12 390 years (Gofas, 2004;Żmudziński, 1990), contrary to barnacles with the relatively short lifespan of approximately one year (Bornhold and Milliman, 1973). The southern Baltic Sea is driven by cyclical environmental dynamics, which evoke physiological stress, determine the food base and drive its biogeochemical cycles (Elder and Collins, 1991). Shells of A. improvisus experienced one-year variability of environmental factors, while bivalves represent long-term variability. Thus, the life span to some extent, may explain the lowest variability of trace elements in 395 barnacles (Fig. 3). However, this relation is not noticeable in mussels, for which the inter-individual variability of trace elements in the youngest individuals (size class I) was not lower than in the oldest ones (size class IV, Fig.   4). Thébault et al. (2009) revealed low inter-individual elemental variability in bivalves and on this basis indicated the environment as a factor controlling their incorporation within shells. Therefore, the heterogeneous elemental concentrations between individuals from the brackish Gulf of Gdańsk may to some extent be caused by biological 400 factors, that could lead to a deviation from what is expected with purely environmental control. The biological influence on the shell chemistry in the southern Baltic Sea could be reinforced by unfavourable conditions for calcification. The low salinity (~7) and alkalinity, which is typical for the studied area of the Gulf of Gdańsk, cause a reduced CaCO3 saturation state (Beldowski et al., 2010;Cai et al., 2010;Findlay et al., 2008). Ions of Ca 2+ and CO3 2are essential components for the crystal formation, and when their concentrations in seawater are low, 405 calcifying organisms exert selective Ca 2+ channels to enable an active ion capture from solution (Sather and Mccleskey, 2003). The required higher contribution of Ca 2+ active pumping results in greater degree of biological control over the calcification process (Sather and Mccleskey, 2003;Waldbusser et al., 2016) and shells are not produced in equilibrium with environmental conditions when it comes to an elemental concentration.
Within species, organisms from juveniles to adults experience morphological and functional changes 410 related to sex, metabolic rate or reproductive stage, which complicate the biomineralization process (Carré et al., 2006;Freitas et al., 2006;Gillikin et al., 2005b;Schöne et al., 2010Schöne et al., , 2011Warter et al., 2018). The size-related elemental patterns in shells of A. improvisus, C. glaucum, M. arenaria, L. balthica and M. trossulus from the Gulf of Gdańsk indicate that if a significant variability exists, it is specifically expressed in trace element concentrations.
The studied mussel M. trossulus and barnacle A. improvisus showed the greatest variability between size classes, 415 while the size class effects were less pronounced in clams (Fig. 4). However, the variability of trace elements was not uniform for M. trossulus and A. improvisus, even though the organisms came from one location ( Table 1). The size-related trend was observed for V, Cu, Y, Cd and U in molluscs, and Mg, V, Cu and Pb in barnacles (Fig. 3).
Among clams, we found the lack of size-dependent changes within M. arenaria. In L. balthica only the concentration of Na decreased with shell growth, while C. glaucum showed variability of Sr, Na, Mn and Cd (Fig.  420   4). Different patterns of elemental accumulation for species in the same habitat were observed before (Rainbow, 2002(Rainbow, , 1995. Rainbow et al. (2000) tested A. improvisus and M. trossulus from the Gulf of Gdańsk as environmental biomonitors by measuring the concentrations of Co, Zn, Fe, Cd, Pb, Mn and Ni in soft tissues. They found that mussels and barnacles occurring at the same location did not show the same variation in elemental bioavailabilities, probably because barnacles were particularly strong accumulators of trace elements (Rainbow, 12 2002(Rainbow, 12 , 1998. This shows that biological differences between species, such as growth rate, feeding rate, assimilation efficiency (Luoma and Rainbow, 2005), route and degrees of element uptake (Rainbow and Wang, 2001) are significant factors determining the elemental accumulation in shells.
In this study, it was generally observed that, when statistical differences between size classes were recorded, the concentrations of trace elements decreased with the shell size. The reverse was found only for Sr in 430 the shells of C. glaucum, in which the concentration of Sr increased with size (Figs. 4 and 5). Large mussels pump less water per unit of the body weight, and their uptake of trace elements is lower than that in smaller individuals.
When the concentrations of trace elements decrease with increasing shell size (Fig. 4), the incorporation might depend on the growth rate. The younger specimens could have a greater growth rate and shell precipitation rate, resulting in a greater uptake of trace elements (Dalbeck, 2008;Szefer et al., 2002). Rosenberg and Hughes (1991) 435 suggested that areas of higher shell curvature, such as the umbo, require greater metabolic expenditure, resulting in an increase of element uptake. When the metabolic activity of an organism decreases, the ionic flux likewise decreases, increasing the tendency of Ca 2+ to block other ions fluxes (Carré et al., 2006;Friel and Tsien, 1989). Therefore, in a low salinity environment of the Gulf of Gdańsk, metabolic fluctuations of organisms can have an exceptionally strong effect on elemental variability, which was high among studied individuals (Table 3, Fig. 3). 440 The surface-to-volume ratio decreases with size and affects the contribution of the adsorbed element content to the bulk concentration (Azizi et al., 2018). Therefore, the negative correlation between the bulk elemental concentration and shell size (Fig. 4), noted in some previous studies (Martincic et al., 1992;Piwoni-Piórewicz et al., 2017;Ritz et al., 1982), could have been caused by a greater potential of surface adsorption in smaller individuals. This is most pronounced for a few trace elements, the concentrations of which decreased across the 445 four size classes in shells of A. improvisus (V, Cu, Pb), M. trossulus (V, Cu, Y, Cd, U) and C. glaucum (Mn,Cd,Figs. 4 and 5). Catsiki et al. (1994) suggested that, apart from metabolic processes, an active detoxification mechanism in tissues is responsible for this trend, and its efficiency is higher in older and larger individuals.

Environmental factors and elemental concentrations 460
Organisms derive trace elements in dissolved and particulate forms primarily from surrounding water, sediments and their food base (Freitas et al., 2006;Gillikin and Dehairs, 2005;Poulain et al., 2015). The concentration of basic macro elements Na, Mg and Sr in surrounding seawater is generally proportional to salinity (Beldowski et al., 2010;Cai et al., 2010;Findlater et al., 2014;Wit et al., 2013), therefore the environmental level of these elements should be rather homogenous in the study area (salinity range 6.9 -7.3, Table 1). And indeed, our analysis 465 has shown low variability of the concentrations of Na (2.0 ± 0.3 mg g -1 -3.2 ± 0.2 mg g -1 ) and Sr (1.2 ± 0.1 mg g -1 -2.3 ± 0.2 mg g -1 , Table 3) in the shells of studied species, which might reflect environmental stability.
However, the concentration of Mg in shells as this element was highly variable (80 ± 30 µg g -1 -3.9 ± 0.3 mg g -1 , Table 3, Fig. 3). Although salinity was indicated as the factor controlling the concentration of Na, Sr and Mg in environment and skeletons built there (Elderfield and Ganssen, 2000;Rosenheim et al., 2004), the range of surface 470 salinity in the study region was unlikely to explain the observed range of Mg concentrations. Mg is the most energetically preferred skeletal element (Allison et al., 2001;Wang and Xu, 2001), is associated with the shell organic matrix (Bourget, 1974) and its concentrations has been found to depend on seawater temperature (Freitas et al., 2005;Dalbeck, 2008;Schöne et al., 2011). Vander Putten at al. (2000 observed that the skeletal Mg/Ca variations in M. edulis could not be interpreted solely based on variations in the seawater Mg, while Dodd and 475 Crisp (1982) showed that the Mg/Ca ratios of most estuarine waters only differ significantly from the open-ocean ratios at salinities below 10. Therefore, the variations in the shell Mg concentration must be the combination of many biological and environmental factors.
Among trace elements (Mn, Ba, Cu, Pb, V, Y, U and Cd), Mn and Ba were definitely the most concentrated in shells (Fig. 3), as previously reported for bivalves (Lazareth et al., 2003;Vander Putten et al., 480 2000). Such high concentrations of Mn and Ba are usually associated with freshwater inputs to estuarine systems, which likewise causes phytoplankton blooms (Gillikin et al., 2006;Vander Putten et al., 2000;Thébault et al., 2009). Consequently, Ba and Mn may be taken up by calcifiers through ingestion of phytoplankton (Stecher et al., 1996) or decaying algal flocs (Brannon and Rao, 1979;Stecher et al., 1996). The elemental incorporation into shell has been shown to correlate at species level with changes in primary production and phytoplankton blooms 485 (Freitas et al., 2006;Lazareth et al., 2003;Vander Putten et al., 2000). Therefore, the food base and elemental transport in the trophic chain might cause the spatial and species-specific elemental variability in shells (Fig. 3).
This pattern is generally confirmed in different regions by the contents of Mn and Ba relative to other trace elements (Fig. 4).
Seawater trace element concentrations in the study area depend on seasonality and human activity, but 490 elemental concentrations in sediments represent long-term chemical background. In coastal regions, trace elements discharged from various sources, such as atmosphere, rivers, and plankton blooms, can be rapidly transported from the water column to the bottom sediments (Bendell-Young et al., 2002;Szefer et al., 1995). In this study, samples were collected in two regions of the Gulf of Gdańsk: the outer part of the Puck Bay (stations MA, M2 and MW) and central part of the Gulf of Gdańsk (station GN, Fig. 1). The shells of clams collected in the outer Puck Bay, 495 with sandy sediment (Szefer et al., 1998;Szefer and Grembecka, 2009;Uścinowicz, 2011), were depleted in trace elements. On the other hand, mussels and barnacles from the central Gulf of Gdańsk, with sandy mud, were characterized by higher elemental concentrations than clams (Fig. 3). On this basis we can assume that the concentrations of trace elements Mn, Ba, Cu, Pb, V, Y, U and Cd in shells might be related to sediment granulometry and increase from sandy to silty type (Góral et al., 2009;Uścinowicz, 2011). The maximum 500 concentration of trace elements within a given region is commonly associated with the finest sediment fraction (<2 µm) as compared to the sandy fraction (Kim et al., 2004;Szefer et al., 1998). In addition, the formation of Mn-Fe oxyhydroxides in the surface sediments (Glasby and Szefer, 1998;Szefer et al., 2002) has a particular role in absorbing trace elements in the fine sediments (Pruysers et al., 1991;Szefer et al., 1995;Tessier et al., 1979). It has been previously shown that for mussel M. trossulus, clam L. balthica and barnacle A. improvisus from the 505 central part of the Gulf of Gdańsk (near the GN station), elemental concentrations in shells (Szefer and Szefer, 1985) and tissues (Rainbow et al., 2000(Rainbow et al., , 2004Sokolowski et al., 2001) were similar to that in the Vistula River plume and higher than trace elements concentration in macrozoobenthos from the outer Puck Bay.
However, the studied trace elements exhibited variations in accumulated concentrations between species, both within the Puck Bay (M2: C. glaucum, MA: M. arenaria, MW: L. balthica) and within the central Gulf of 510 Gdańsk (GN station: M. trossulus. and A. improvisus, Fig. 3). The highest concentration of trace elements in C. glaucum (M2: 10 m) and L. balthica (MW: 31 m) within clams from the sandy Puck Bay could be driven by elevated amounts of trace elements in oxygenated zones where Fe-Mn oxyhydroxides accumulate. At the Mn(II)/Mn(IV) redox interface manganese oxides may predominantly precipitate on the periostracum of molluscs comparing to inorganic surfaces (Strekopytov et al., 2005), which may, in turn, influence the incorporation of trace 515 elements into shells. M. arenaria likewise was collected in a shallow zone (MA: 10 m), but had, nevertheless, the lowest concentrations of all studied trace elements (Fig. 3). This bivalve mainly spends life buried 20 -30 cm in a sediment (Żmudziński, 1990). Szefer et al. (1998) reported that in the Gulf of Gdańsk the enrichment factors for Cu, Zn, Ag, Cd and Pb are highest in the <2 µm fraction and decrease with increasing both fraction size and depth of the sediment. This may by a reason why C. glaucum and L. balthica have similar patterns of elemental 520 accumulation, while M. arenaria was characterized by the lowest values (Fig. 3). Such dependence indicates that the sediment properties belong to factors controlling concentration of trace elements in the shells.
Mussels M. trossulus and barnacles A. improvisus from the same location (Fig. 1, Table 1) showed greater chemical differentiation than clams. This relationship is most evident for Mn and Ba. Much lower concentrations of Mn and Ba were found in the shells of molluscs (Mn: 54.0 ± 15.0 µg g -1 , Ba: 17.0 ± 6.70 µg g -1 ) than in 525 barnacles (Mn: 625 ± 160 µg g -1 , Ba: 73.0 ± 20.1 µg g -1 , Table 3). A similar relationship was found in the soft tissues of M. trossulus and A. improvisus collected from different locations in the Gulf of Gdańsk in May 1998 (Rainbow et al., 2000). The range of Mn in the soft tissues of M. trossulus varied from 19.0 to 41.0 µg g -1 , while that in A. improvisus ranged from 187 to 307 µg g -1 and was interpreted as species-specific accumulation efficiency. The concentration of Mn has been repeatedly reported to be influenced by phytoplankton blooms 530 (Vander Putten et al., 2000;Gillikin et al., 2006;Zhao et al., 2017), thus the diet seems to affect the shell Mn concentration. Due to size of the organism, the diet of A. improvisus depends more on Mn-rich phytoplankton than larger M. trossulus that easily picks zooplankton. The lower proportion of zooplankton in the barnacle diet could have hereby add to a higher Mn concentration in their shells (Fig. 3). This suggests that the Mn concentration in shell is highly species dependent. 535 Some studies have investigated trace element concentrations in shells (but many more have concentrated on the soft tissues) of marine invertebrates as a tool to assess trace elements contamination of the aquatic environment. The concentrations of trace elements Mn, Ba, Cu, Pb, Cd and U in the shells of a studied organism in Gulf of Gdańsk and in other regions was found to be highly variable, even within a single taxon (Table 4). Therefore, the environmental conditions prevailing during biomineralization are largely reflected in the trace 540 element concentrations of the shells; nevertheless, their interpretation requires consideration of biological factors specific to the species.

Conclusions
The shells of calcitic Amphibalanus improvisus, aragonitic Cerastoderma glaucum, Limecola balthica, Mya arenaria and bimineralic Mytilus trossulus from the Gulf of Gdańsk are accumulators of a wide spectrum of trace 545 elements from the surrounding environment. The elemental concentration levels in studied species are not only determined by their bioavailability in the environment. Many biotic and abiotic factors are acting on the shell incorporation mechanism and their effect is likely to be species-specific.
By determining Ca, Na, Sr, Mg, Mn, Ba, Cu, Pb, V, Y, U and Cd in the shells of given species we found some patterns of elemental accumulation. At a local scale of the Gulf of Gdańsk, the main elements Na, Sr and 550 Mg are mostly dependent on crystal lattice properties of calcite and aragonite. Clams that precipitate fully aragonitic shells have a clear preference for accumulating Sr over Mg in shells, contrary to dominant Mg content over Sr in barnacle shell calcite. It is energetically more favourable for larger cations such as Na and Sr to enter the aragonite lattice with smaller cations (e.g. Mg) favouring calcite. However, this relationship breaks down when comparing shells of different species or genera. For example, the barnacle calcite contains higher Sr concentration 555 than the bivalve aragonite. The level of main elements, especially Sr and Mg, seems to be determined by specific biological factors, such as growth rate.
In case of trace elements Mn, Ba, Cu, Pb, V, Y, U and Cd, factors other than given crystal lattice presence, seem to determine their concentrations. The elemental variability between size-grouped shells indicates that trace elements were more variable than Na, Sr, Mg, but this varies between species. Moreover, there is a trend of the 560 elemental concentrations being lower in larger than in smaller shells. Biological differences between and within species, such as feeding including its rate and assimilation efficiency related to age of organisms (size of the shell), Żmudziński, L.: Świat zwierzęcy Bałtyku -atlas makrofauny, WSiP, Warszawa., 1990.   (Heinemann et al., 2008), g Baltic Sea, Gulf of Gdańsk (Szefer and Szefer, 1985), h Bay of Bengal, India (Raman and Kumar, 2011), i Black Sea (Mititelu et al., 2014), j Canada (Klein et al., 1996, k Cultured (Freitas et al., 2009), l Cultured (Heinemann  Figure C shows the sampling stations as black circles (see Table 1 for station details). The grey lines indicate 20 m isobaths.   . Pairwise comparisons of elements in the shells of studied species with statistically significant differences (p < 0.05) between size classes (for size class details see Table 2).