Environmental influence on calcification of the bivalve Chamelea gallina along a latitudinal gradient in the Adriatic Sea

Environmental factors are encoded in shells of marine bivalves in the form of geochemical properties, shell microstructure and shell growth rate. Few studies have investigated how shell growth is affected by habitat conditions in natural populations of the commercial clam Chamelea gallina. Here, skeletal parameters (micro-density and apparent porosity) and growth parameters (bulk density, linear extension and net calcification rates) were investigated in relation to shell sizes and environmental parameters along a latitudinal gradient in the Adriatic Sea (400 km). Net calcification rates increased with increasing solar radiation, sea surface temperature and salinity and decreasing Chlorophyll concentration in immature and mature shells. In immature shells, which are generally more porous than mature shells, enhanced calcification was due to an increase in bulk density, while in mature shells was due to an increase in linear extension rates. The presence of the Po river in the Northern Adriatic Sea was likely the main driver of the fluctuations observed in environmental parameters, especially salinity and Chlorophyll concentration, and seemed to negatively affect the growth of C. gallina.


Results
SR, SST and SSS and Chl varied among sites along the latitudinal gradient (Kruskal-Wallis test, df = 5, p < 0.001; Table 1), between August 2013 and April 2015. SR, SST, SSS showed to have significant negative relationships with latitude, while Chl showed positive relationship (p < 0.001; Supplementary Fig. S1). In particular, SSS displayed two sites, Chioggia and Goro, detached from the other sites and Chl presented the same detached trend in the sites of Goro and Cesenatico, diverging both SSS and Chl from a linear relationship with latitude ( Supplementary  Fig. S1a). The sites of Chioggia, Goro and Cesenatico are the sites located near the Po delta ( Fig. 1), in which the high variability in SSS and Chl values is due to the freshwater inflows during seasons ( Supplementary Fig. S1a). All the environmental parameters were significantly correlated with each other, in particular SST with SR and SSS (p < 0.001; Supplementary Fig. S1b).
Shell lengths were homogeneous among sites (Kruskal-Wallis test, df = 5 and p > 0.05; Supplementary Table S1). Also age did not differ among sites both for external rings and internal bands (Kruskal-Wallis test, df = 5 and p > 0.05; Supplementary Table S1).
There were no significant differences among the growth curves obtained from the external and internal rings within each site (Supplementary Table S2), therefore, a generalised VBG curve was obtained for each site (Fig. 3). The δ 18 O values along the shell growth direction exhibited a roughly sinusoidal sequence of lower (summer) and  www.nature.com/scientificreports www.nature.com/scientificreports/ higher (winter) values, and the number of observed seasons allowed estimation of age (Fig. 2c). Age from δ 18 O values validated the data from the other two ageing methods fitting the VBG curves ( Fig. 3).
At each site, shell skeletal and growth parameters were significantly correlated with shell length (Fig. 4). At all sites bigger shells exhibited higher micro-density and bulk density and lower apparent porosity compared to smaller ones (Fig. 4). Moreover, linear extension rates and net calcification rates decreased with shell length at all sites (Fig. 4).
Variation of shell skeletal and growth parameters was then analysed in relation to environmental variables along the latitudinal gradient. Correlations were performed in the whole dataset of 84 shells for each site ( Table 2, Fig. 5a). Apparent porosity and bulk density were not correlated with environmental parameters, while linear extension rate and net calcification showed significant positive correlations with SR, SST, SSS and negative correlations with Chl (Fig. 5a). Correlations with the environment were also performed in the immature and mature shells separately and in a subgroup of the mature shells including only shells of commercial size (>22 mm). In all groups, linear extension and net calcification were positively correlated with SR, SST, SSS and were negatively correlated with Chl, except extension rate which did not correlate with SST in immature shells (Fig. 5b, Table S3).

Discussion
In this study, we successfully used shell external and internal growth rings to estimate the age of C. gallina and build growth curves for this species at six sites along a wide latitudinal gradient. In some samples, the annual growth rings were hardly to determine because the shell external surface was smooth or damaged and because the blue-staining pattern in the internal section was not very defined, highlighting ambiguous internal growth lines. Both methods were validated by δ 18 O profiles along shell growth direction, suggesting the methods were appropriate and fairly accurate for the age estimation of C. gallina specimens. This work was the first attempt to determine the age from the internal shell section of C. gallina using the Mutvei's solution. We can conclude that counting of internal growth rings, after shell sectioning and Mutvei's treatment is a time-consuming method with no evident growth pattern at all times, suggesting to better adopt external growth rings and oxygen isotope analyses in future age studies of C. gallina. Our values are in conformity with the estimated maximum shell length and growth constant of C. gallina from previous studies in the Adriatic Sea 35 , in the Western Mediterranean Sea 19 and in the Algarve coast 17 . The differences reported in maximum asymptotic length (L inf ) and von Bertalanffy growth constants (K) among sites (Supplementary Table S2) are probably due to local environmental conditions, www.nature.com/scientificreports www.nature.com/scientificreports/ as already suggested by previous studies on the growth of C. gallina conducted in different areas 17 . In previous studies, the eastern populations of this species, from the Marmara and Adriatic Sea, showed greater longevities than western populations along the Spanish coast and the Algarve coast 17,32 .
As previously observed for molluscs and for other organisms 36,37 , C. gallina extension rate decreased with increasing length. The reduction in net calcification rates with size was determined by decreasing linear extension rates, partly countered by increasing bulk density. A higher apparent porosity was observed in shells of small size and it sharply decreased to less than 20% approaching the length at sexual maturity (about 18 mm) 38 . High porosity influenced bulk density which was conversely lower in small size shells. This suggests that during the first year of life, C. gallina seems to promote porosity, enabling it to keep higher linear extension rates in order to reach the size at sexual maturity. Although juveniles are more vulnerable than adults to most predators, a denser skeleton could limit the rate of body growth, increasing the time spent at smaller, non reproductive sizes, while more porous ones could lead to an increase in shell's linear extension rate allowing C. gallina to reach size at sexual maturity faster 39 . C. gallina is a gonochoric species with external spawning thus having a larger shell could mean more space available for gonads. From about 20 mm in length, C. gallina seems to change its biomineralization behavior, showing small variations in apparent porosity and bulk density and a continuous decrease in linear extension rate and net calcification. Moreover, bigger and older individuals fully allocated net calcification in making denser shells likely to be less vulnerable to predators, by depressing linear extension rate 39 .
A previous study conducted along the same latitudinal gradient in the Adriatic Sea showed that solar radiation and sea surface temperature directly affected shell skeletal properties in specimens of C. gallina of commercial size, showing more porous and less dense shells in the most irradiated and warm populations 26 , but these trends  www.nature.com/scientificreports www.nature.com/scientificreports/ were not analysed in specimens less than 25 mm long. This is the first study investigating shell skeletal and growth parameters during the lifespan of the clam C. gallina and in relation to solar radiation, temperature, salinity and chlorophyll concentration in both immature and mature shells. Environmental parameters seemed to have a greater influence on large shells over 22 mm long. Large clams with lower growth rates tend to have higher standard metabolic rates (SMR) as confirmed by the "principle of allocation" theory 40 , such as the potential tradeoff in the allocation of energy between growth and maintenance metabolism that result to be negatively correlated. Earlier studies in marine bivalves showed that individuals with higher SMR are less resistant to environmental stress and this is in agreement with our findings that mature shells were more influenced by environmental variables than immature shells 40,41 . Individuals with higher SMR, like mature clams, must depend more on their reserves to sustain vital functions and support physiological responses to stress; instead individuals with lower energetic requirements, like immature clams, have a surplus of energy with which to withstand stressful conditions 41 .
Net calcification rates increased towards southern populations, with increasing SR, SST and SSS and lower Chl, in immature and mature shells. The immature shells allocated increased calcification rates on bulk density. A possible explanation could be that denser clams could make them less vulnerable to predation, which often affects early life stages 42 . This hypothesis is in agree with the general trend of reducing vulnerability with incrementing prey size reported for example for molluscs preyed by decapods 43 . Predation likely relies on the ability of the predators to crack or perforate the shells and changes in shell characteristics to enhance strength, such as increase in bulk density, could lead to higher survival of small individuals, particularly where predation pressure is high 42 . Very little is known on predatory species of C. gallina in the Adriatic Sea; some of these are the starfish, Astropecten spp., the gastropod Neverita josephinia and the fish, Gobius niger and Lithognathus mormyrus 44 . However, further studies on the population density of these species along the latitudinal gradient in the Adriatic Sea are needed to confirm this hypothesis. The differences in shell density that we observed along the gradient could depend on different mineralization rates driven by environmental parameters, especially by temperature. Towards South, immature clams could raise their shell density due to a decrease in the energetic costs of shell www.nature.com/scientificreports www.nature.com/scientificreports/ formation with increasing aragonite saturation state in warmer waters 45 . In contrast to immature clams, in mature ones, the increase in calcification rates towards Southern populations was invested on increasing linear extension rates. A possible explanation could be that bigger shells could make them more fertile, by suppling more space for gonads, but the hypothesized biological significance for this calcification pattern has still to be investigated.
Since C. gallina is an infaunal bivalve, the effect of SR along the latitudinal gradient is probably related to other abiotic and/or biotic parameters, such as temperature and phytoplankton density. The increase in net calcification rates with increasing SST could be due to a decrease in the energetic costs of shell formation with increasing aragonite saturation state in warmer waters 45 . Studies on molluscs highlight that calcification increases with aragonite saturation state 46 . In this study, SST does not exceed 19 °C of mean annual sea surface temperature in the southernmost site and producing carbonate shells could be less expensive than in colder sites with mean SST of 16 °C. There are still no aragonite saturation data available along the latitudinal gradient to deeply understand relationships between shell calcification and seawater chemistry in the Adriatic Sea.
Chlorophyll a concentration is a good food proxy for clams, based on the assumption that phytoplankton is the main component of suspension feeding bivalves diet 47 . The Adriatic Sea is characterized by a lower phytoplankton concentration in the South compared to the North 48 . Previous results indicate that the clam Venus verrucosa grew faster in areas with high Chlorophyll a concentrations in the Eastern Adriatic Sea 49 and that food availability played an important role in determining the growth rates of population of bivalves 50 . Despite growth increases as a function of food concentration 51 , in this study shell linear extension and net calcification rates resulted to be lower with high Chl in all size shells, suggesting there could be other environmental parameters that synergistically affected the growth of C. gallina. Indeed, along the latitudinal gradient in the Adriatic Sea the enhanced growth of C. gallina could be due to higher temperature and salinity that mitigate the contingent lower food availability.
Here we show that in the northern sites under Po delta influence, in particular in the site of Goro, linear extension and calcification rates of C. gallina were reduced compared to Southern sites. One possible explanation could be non-optimal salinity conditions (SSS < 30). The Po river flow reduces salinity, exposing C. gallina to strong seasonal variations in salinity values with intense reductions in autumn and rises in summer 52 , especially in the site of Goro, while the other sites show increasing salinity moving from Goro towards North and South. While C. gallina is an euryhaline species with a high ability to acclimatize to extremely brackish conditions 53 , suboptimal salinity may constitute a stressor leading to modify biochemical mechanisms, such as incrementing their antioxidant defences to face the higher oxidative stress 54 and to vary physiological responses, such as valve closure, reducing feeding activity and slower growth rates 54,55 . Moreover, lower salinity water also leads to lower aragonite saturation state 56 , with increasing cost of calcification 57 . Another possible explanation could depend on eutrophication. In the Adriatic Sea, the marine environment is strongly affecting by the input of Po river nutrients, that influence sea water transparency and lead to anoxic events 58 . Slow growth rates associated with eutrophicated habitats have been previously recorded for the bivalve Cerastoderma edule 59 and for the bivalve Austrovenus stutchburyi 60 and these studies seem to confirm the results found for C. gallina that showed higher extension and calcification rates in oligotrophic conditions. Moreover, silt and clay that characterise the bottom of the Po delta area could interfere with the feeding mechanism, leading to low extension and net calcification rates and growth rates were found to be higher in sand than mud 60 . Thus, the observed reduction in extension and calcification rates of C. gallina in those sites that are most influenced by the Po delta could depend on a number of factors acting synergistically. However, further studies are needed to test these hypotheses.
The geographic information of growth reported in this paper, with an increase in calcification and linear extension rates towards Southern sites, could suppose that fisheries in the Southern sites could be characterised by bigger clams with a significant raising of inside edible mass compared to the clams of the Northern sites. But further analysis on the clam flesh is essential to investigate this hypothesis.
A complex pattern of interactions between the organism and several habitat conditions (e.g. temperature, salinity, food availability) shapes the growth of the species over time and space. Investigating these relationships and modelling the environmental control of marine species growth is essential in sustainable management of coastal ecosystems. Valuable insights for developing ecosystem-based management tools of aquaculture activities derives from studies providing relevant information on environmental conditions affecting mollusc growth 61 . In addition, knowledge of the growth rates allows a proper management of bivalve fisheries and the observations highlight in this study can be used in predictive models to explore the evolution of C. gallina resource exploited during time under extended changes in their habitat conditions, for example seawater temperature, salinity or food. The present paper, together with continue monitoring of clam stocks, is aimed at providing in-depth insights on age and growth, considered the prerequisite to generate the information on recruitment, longevity, mortality, and their application in developing sustainable and efficient fisheries management policies imposed by the Ministry of Agricultural, Food and Forestry Policies in Italy; (e.g. DM 27/12/2016 on clams discard in the Italian stocks).

Conclusions
Differences found in shell skeletal parameters, especially apparent porosity and bulk density, with length could be due to different biomineralization patterns between immature and mature shells of C. gallina. Before reaching sexual maturity, C. gallina seemed to promote porosity enabling it to keep higher linear extension rates in order to reach the size at sexual maturity faster, while after sexual maturity shells seemed to depress linear extension rate and make denser shells.
Moving far from the Po delta towards South, warmer seawater, low fluctuations in salinity and oligotrophic conditions suggested that these environmental conditions may be most favourable for the clam C. gallina, leading to higher net calcification rates. Net calcification rates were significantly reduced in sites around the Po delta, possibly as a result of lower temperature and reduced salinity that increase the energetic costs of shell formation www.nature.com/scientificreports www.nature.com/scientificreports/ with decreasing aragonite saturation state. Net calcification could also be reduced as a result of increased eutrophication and silt and clay of the bottom driven by the river discharges that could interfere with the feeding mechanisms. The present study therefore points out the importance of considering multiple environmental parameters to investigate bivalve growth. In addition, knowledge of the growth rates allows a proper management of bivalve fisheries. Given the great socio-economic relevance of C. gallina in all the Italian Adriatic coasts, studies like this one are crucial to guarantee a knowledge-based management of this important resource.

Materials and Methods
Between August 2013 and April 2015, specimens of C. gallina were collected from six sites along a latitudinal gradient in the Adriatic Sea from 45°42′N to 41°55′N (Fig. 1). Clams were sampled at each site using hydraulic dredges at 3-5 m depth. 84 shells of different size from each site were used for the analyses. Shells were divided in three groups: immature (up to 18 mm), mature shells 62 (over 18 mm) and commercial shells (over 22 mm, new experimental commercial size adopted from January 2017).
Skeletal apparent porosity (percentage of the pore volume connected to the external surface; %) and micro-density (mass per unit volume of the material which composes the shell, excluding the volume of pores; g cm −3 ) were measured by buoyant weight analysis, using a density determination kit Ohaus Explorer Pro balance (±0.1 mg; Ohaus Corp., Pine Brook, NJ, USA) 26 . Measurements required for calculating apparent porosity and micro-density were 26 : ρ density of the fluid (double distilled water: 0.998 g cm −3 at 20 °C and 1 atm) DW dry mass of the shell SW saturated mass of the shell = mass of the shell plus mass of the water enclosed in its pores BW buoyant mass of the shell = mass of the shell fully saturated with water minus mass of the water displaced by it. In addition, clam shell length (anterior-posterior maximum distance) was obtained with ImageJ software after data capture of each shell shape with a scanner and dry shell weight was measured using an analytical balance (±0.1 mg) 26 .
Bulk density (shell mass/volume ratio, including the volume of pores) was also measured by buoyant weight analysis. Shell linear extension rates were obtained with the length/age ratio (cm y −1 ), while the net calcification rate (mass of CaCO 3 deposited per year per unit area g y −1 cm 2 ) was calculated for each shell by the formula: net calcification (g cm −2 y −1 ) = bulk density (g cm −3 ) x shell extension (cm y −1 ) 63 .
Age was measured in a subsample of 30 shells of different size in each site, by using three methods: shell surface growth rings (Fig. 2a), shell internal bands (shell cross-sections and Mutvei's solution; Fig. 2b) and stable δ 18 O composition ( Fig. 2c; see ESM for the ageing methods details). By counting the total number of visible external and internal rings in each shell, the age-length keys were obtained for the two methods and fitted with the von Bertalanffy growth (VBG) functions, using a non-linear model that provides estimates of the parameters in the VBG equation through bootstrap method: where L t is individual length at age t, L inf is asymptotic length (maximum expected length in the population), K is a growth constant, t is the age of the individual. Two growth curves for each sites were produced and a chi-square test of maximum likelihood ratios was used to examine the significance of differences in growth functions between the two ageing method. Kimura's method allows the testing of several hypotheses to compare the two curves by analysing one or more growth parameters simultaneously. For these purposes, the FSA (Simple Fisheries Stock Assessment Methods) and the Fishmethods packages in R studio were used. If no differences were revealed between VBG curves from shell surface growth rings and internal bands, generalised growth curves for each site were constructed by merging age-length keys from both methods and the resulted generalised VBG function for each site were taken into account for extrapolating age in all the 84 shells. Age were calculated from the inverse of generalised VBG function of each site: www.nature.com/scientificreports www.nature.com/scientificreports/ To validate the data from the two counting rings methods and the generalised VBG curves, oxygen isotopic measurements (δ 18 O) were carried out at the Godwin Laboratory for Palaeoclimate Research, Department of Earth Sciences, University of Cambridge. δ 18 O measurements were carried out on "spot" samples collected from the prismatic layer and from the cross-lamellar layer and drilled in sequence along the shell growth direction by means of a 0.5 mm dental diamond drill 15 . Dried homogenized powdered samples were treated with helium then added an acidified solution consisting of 104% orthophosphoric acid and left to react for 1 hour at 70 °C. Each sample was then analysed using a Thermo Gasbench preparation system attached to a Thermo Delta V Advantage mass spectrometer in continuous flow mode. Age resulted from counting lighter δ 18 O (summer) and heavier δ 18 O (winter) peaks were then plotted with the age-length key from the two ageing methods.

Data Availability
The dataset generated and analysed during the current study is available from the corresponding author on reasonable request.