The results of this study demonstrated that the chemical composition of Sparus aurata otoliths have potential to differentiate between nursery areas, and between Y-O-Y specimens and young adults in the eastern Adriatic. Concentrations of five analysed elements (Mg, Ba, Sr, Zn and Pb) were above the detection and quantification limits and varied between sites. The most discriminant elements were Sr and Mg, while Zn and Pb spatial variations were visible but were not statistically significant for Pb. The concentration of Ba did not reveal any specific patterns.
Comparisons of nursery areas. Testing for natal homing is very challenging due to the difficulty in tracking all life stages44. Generally, otolith chemistry is affected by water chemistry35,49-51, and it has proved useful in examining movements among life stages in regions where there is detectable spatial variation37. Considering Sparus aurata, previous studies showed that juveniles occupy seasonally contiguous lagoons10,25,27,32,34, while adults inhabited open sea waters27,28,32. In this study, the most distinctive elemental signature was related to low salinity sites near the coast where Y-O-Y specimens were sampled. As expected, the chemical signatures of these three sites differed significantly, which can possibly be attributed to their specific relief features (fluvial or karstic relief), freshwater inflow (continuous or sporadic), water quality and/or surroundings (urban or sparsely populated area). However, the majority of sampling sites for young adults did not differ significantly, implying that the middle Adriatic coast generally provides uniform living conditions for S. aurata recruits. Additionally, sites situated far offshore exhibited certain specificities that resulted in significant differences from nursery sites, likely due to the higher and more stable hydrographic conditions and high salinity (>37 ppt) at these offshore sites.
In this study, Mg:Ca and Sr:Ca concentrations were significantly different between the analysed sites. Otolith Mg concentrations had elevated values in the coastal lagoon (S5), while the lowest values were recorded at an open-sea site (S12). Interesting, the opposite was found for Sr:Ca, where the lowest values were recorded at S5 and the maximum at S12. However, certain caution is needed when relating these values to environmental conditions. Higher values of Mg uptake into otoliths may more likely be related to endogenous processes52-54, and as such Mg is not a reliable environmental indicator55. Previously, Vrdoljak et al.45 detected elevated but individually specific concentrations of the Mg:Ca ratio for juveniles of S. aurata from the Pantan lagoon (S5), suggesting that Mg incorporation is likely affected by an individual physiological mechanism. Two other elements analysed here, Sr and Ba, were previously successful in reconstructing environmental and coastal-estuary migrations56,57. Their ratios in otoliths are usually linked to concentrations of Sr:Ca and Ba:Ca in ambient water and salinity, with a higher Sr:Ca found in marine, and higher Ba:Ca found in freshwater58,59. Gillanders and Kingsford60 reported higher Sr concentrations in otoliths of recruits from coastal reefs in comparison with those from estuarine seagrass habitats. However, no specific patterns of Ba incorporation were found in relation to salinity. Additionally, Zn:Ca and Pb:Ca concentrations were highest at the coastal lagoon Mala Neretva (S1) and a proximal site (S4), while the most reduced values of these elements were recorded at open-sea sites (S7). Besides known fact about natural origin of Zn61 and its individual physiological regulation62, the obtained elevated values of Zn can be also associated with anthropogenic pollution due to intensive marine traffic63,64. In addition, the highest Pb concentrations were found near known contamination sources45,60,65,66, i.e., at the Mala Neretva coastal lagoon (S1), though without prominent incorporation in specimen otoliths that would demonstrate a clear relationship with environmental contamination levels45.
Re-allocation to nursery habitats. The accurate discrimination by multivariate analyses indicated that S. aurata specimens originating from specific geographic areas in the Adriatic Sea have distinct elemental signatures that potentially enable their re-allocation to specific nursery habitats along the coast. Multivariate analyses have become a desirable tool for such purposes, since they are capable of separating different perturbations from natural spatio-temporal variability displayed by most populations67 with an analytical procedure52,49,58. However, the lower rate of success (35–45%) achieved in differentiating between the analysed nursery areas may be driven by two potential mechanisms. The first by the intrinsic, physiological mechanisms of each individual that determine the supply of trace elements to the inner-ear and their subsequent incorporation into the otolith during its growth, and the second by differences in environmental exposure68. Using mixture models to estimate each nursery habitat contribution, Niklitschek and Darnaude69 found that these results reflect the large interannual variability in nursery habitats12, which may be common to the shallowest water and estuarine nursery areas70, thereby making it difficult to find a clear pattern of overlap between different localities or ontogenetic groups. Since this is experimental study, no doubt the rate of success will increase with larger sample size.
The difference in element uptake by individual otoliths or intrinsic, physiological mechanisms, was most evident for Zn, which varied greatly in value from individual to individual but was clearly separated between the 0+ and 2+ aged specimens, with reduced values in young adults. That fact was previously proposed by Avigliano et al.71 for the differentiation of cohorts, providing temporally constrained information on habitat, behaviour or nutrient supply for Y-O-Y specimens and young adults. Additionally, since cohorts occupy specific space-time frames, it is necessary to avoid mixing different cohorts in multivariate analyses due to the reliability of obtained results. On the other hand, the importance of environmental exposure is evident in the incorporation of Ba:Ca in S. aurata otoliths. This ratio gave the lowest discriminating power, which may suggest that the impact of freshwater inflow either from rivers or continuous or sporadic freshwater grounds along the eastern Adriatic coast is so strong that it prevents a clear salinity boundary between estuarine and marine waters. Therefore, all specimens caught in the sampling area exhibit Ba concentrations without any regular pattern, regardless of whether they were sampled in coastal lagoons, shallow coves or offshore. Therefore, the lowest percentage of Ba as a discriminant can also be considered to reduce the overall success of re-allocation, further highlighting that much of the marine waters along the middle Adriatic coast have lower than expected salinity. As mentioned above, there are a number of shallow, semi-protected bays with continuous submarine freshwater springs in the Adriatic Sea. These are karst phenomena created by geomorphological processes under the influence of recent sea-level changes72. Like coastal lagoons, these bays also comprise a diverse mosaic of environments with different physiochemical characteristics2,9.
The obtained results revealed that the most specimens were correctly re-allocated to nursery area N3, which lacks specific coastal lagoons as potential nursery sites for S. aurata. This is a broad coastal area in the middle Adriatic with a number of coves that could be potential nursery sites, followed by two analysed coastal lagoons previously identified as nursery grounds for this species16,45,73. Namely, Mala Neretva and Pantan are small, shallow, brackish coastal lagoons that represent sites with obvious, optimal conditions required by this species. The specimens sampled within N3 were most correctly re-allocated (45.6%) based on their Mg:Ca, Sr:Ca and Pb:Ca ratios. Moreover, these elements were allocated with 100% accuracy. Contrary to what we expected, Ba had the lowest, but expressed percentage of correct allocation (75%). The shallow cove Prosika, although not a typical coastal lagoon, thus proved to have a role equivalent to a nursery. This is a shallow, semi-closed cove with an annual influence of freshwater springs, which is highest in the winter-spring period after snow melt and intensive rain74-76. These coves were previously identified as having higher fish species richness16 than coastal lagoons, due to the wide connection with the sea that favours the entrance of fish eggs, larvae and young of-the-year by tides and currents from spawning areas70,77-79. Meanwhile, the entrance of marine fish eggs and larvae in coastal lagoons is limited because of the narrow sea connection, and therefore lagoons may suffer from severe conditions depending on precipitation and evaporation80,81. We do not claim that all young adults of S. aurata from nursery area N3 originated from a unique cove as a nursery habitat (e.g., S10). Moreover, we suggest that S10 is representative of a number of similar localities along the 120 km stretch of Adriatic coastline82 sheltered by numerous islands that can serve as suitable nursery habitats for S. aurata. Morais et al.32 suggested that S. aurata larvae displayed a cautionary behavioural strategy, relying on feeding plasticity and batch-spawning to compensate for the lack of apparent behavioural plasticity, to ensure metapopulation stability and resilience. This is reflected in the alterations in nursery selection from coastal lagoons, with reduced quality as habitats for juvenile fish due changed environmental conditions34,83,84 and significant increase in eutrophication and anoxic crises85,86 to other coastal habitats that could ensure the conservation of S. aurata local populations. The lowest correctness of re-allocation was associated with N1, due to the lowest correctness of all element allocation in young adults in comparison with specimens from N2 and N3. A possible explanation is that in the highly variable environment of the Mala Neretva coastal lagoon, individual physiological mechanisms prevail over environmental mechanisms due different diurnal movements, environmental preferences/tolerances, and prey-choice/availability across the habitat mosaic87,88.
Generally, the loss of biodiversity, ecosystem functions and vegetation in estuarine and coastal ecosystems highlight the need to better understand the environmental drivers in coastal habitats1,89. In this sense, conservation measures to maintain sustainable stock levels must be aimed not only at protecting coastal lagoons but at all essential habitats or even a part of the coast26. Mediterranean coasts have been significantly affected by anthropogenic destruction over the past 50 years. Matić-Skoko et al. 90 suggested that significantly modified juvenile fish communities in nurseries may be due more to constant human embankment and marine infrastructure construction along the coast in recent decades, than to climate change or fishing pressures, as generally considered. This indicates the necessity for accurate knowledge of fish lifetime movements and critical habitats 91,92, and proper estimations of the contribution of each habitat to adult populations 3.