Past aquatic environments in the Levant inferred from stable isotope compositions of carbonate and phosphate in fish teeth

Here we explore the carbon and oxygen isotope compositions of the co-existing carbonate and phosphate fractions of fish tooth enameloid as a tool to reconstruct past aquatic fish environments and harvesting grounds. The enameloid oxygen isotope compositions of the phosphate fraction (δ18OPO4) vary by as much as ~4‰ for migratory marine fish such as gilthead seabream (Sparus aurata), predominantly reflecting the different saline habitats it occupies during its life cycle. The offset in enameloid Δ18OCO3-PO4 values of modern marine Sparidae and freshwater Cyprinidae from the Southeast Mediterranean region vary between 8.1 and 11.0‰, similar to values reported for modern sharks. The mean δ13C of modern adult S. aurata and Cyprinus carpio teeth of 0.1±0.4‰ and -6.1±0.7‰, respectively, mainly reflect the difference in δ13C of dissolved inorganic carbon (DIC) of the ambient water and dietary carbon sources. The enameloid Δ18OCO3-PO4 and δ13C values of ancient S. aurata (Holocene) and fossil Luciobarbus sp. (Cyprinidae; mid Pleistocene) teeth agree well with those of modern specimens, implying little diagenetic alteration of these tooth samples. Paired δ18OPO4-δ13C data from ancient S. aurata teeth indicate that hypersaline water bodies formed in the Levant region during the Late Holocene from typical Mediterranean coastal water with high evaporation rates and limited carbon input from terrestrial sources. Sparid tooth stable isotopes further suggest that coastal lagoons in the Eastern Mediterranean had already formed by the Early Holocene and were influenced by terrestrial carbon sources. Overall, combined enameloid oxygen and carbon isotope analysis of fish teeth is a powerful tool to infer the hydrologic evolution of aquatic environments and assess past fishing grounds of human populations in antiquity.

Large, but systematic variations in δ 13 C values have also been observed between the dentine and the enameloid of modern shark teeth [24]. The positive δ 13 C values of enameloid (1.6 to 4.8‰) compared to significantly lower values in dentine (-6.4 to -2.3‰) suggest that enameloid approaches isotopic equilibrium with ambient dissolved inorganic carbon (DIC), likely through diffusional exchange with seawater DIC [24]. The δ 13 C values of structural carbonate of enameloid in fossil shark teeth agree well with those documented in modern species, suggesting that in vivo carbon isotope compositions are preserved [19][20]25]. Thus, in cases where the original stable isotope compositions of fish teeth are preserved, the δ 13 C values of the enameloid may serve as a proxy for changes in the isotopic composition of DIC of the ambient water [19][20][24][25]. Yet, the range of enameloid δ 13 C values in modern bony fish, which is essential to assess fossil fish δ 13 C values, is presently limited.
In this study, we use carbon and oxygen isotopic compositions of co-existing carbonate and phosphate fractions of fish tooth enameloid to infer the aquatic environments in which (pre-) historic humans caught their fish. For this purpose, different fish species from the modern Mediterranean (Sparidae) and freshwater fish from Lake Kinneret (Cyprinidae) were analyzed for their δ 18 O CO3 , δ 18 O PO4 and δ 13 C CO3 values. These results are then compared to values obtained from fish teeth (belonging to the same species) recovered from different archaeological sites and periods in the southern Levant. To examine the possible effect of diagenesis on enameloid carbon and oxygen isotopic compositions, we also analysed some fossil pharyngeal teeth (1 st molariform tooth) recovered in deposits of lake 'Ubediya (Lower palaeolithic, ca. 1.4 Ma; Jordan Rift Valley). Combined δ 18 O PO4 -δ 13 C and δ 18 O CO3 -δ 13 C data of fish tooth enameloid is explored as a tool to reconstruct the hydrological history of past fish habitats.

Materials
Teeth of five extant fish species from three different aquatic habitats in the Levant were analysed for their enameloid carbon and oxygen isotope composition: Sparus aurata (n = 3, 9 teeth) and Pagrus caeruleostictus (n = 2, 6 teeth) (Sparidae, Perciformes, seabream) from the Mediterranean Sea (Haifa Bay, Israel), Scarus sp. (Scaridae; parrotfish, n = 1, 3 teeth) from the Red Sea, and the domesticated Cyprinus carpio (Cyprinidae; common carp, n = 2, 6 teeth) from the freshwater of Lake Kinneret in northern Israel ( Table 1). Each of these species has a distinct diet: S. aurata feeds mainly on shellfish, mussels and oysters; P. caeruleostictus feeds on bivalves, crustaceans and fish; Scarus sp. feeds on corals and sponges; C. carpio feeds on plants, insects, crustaceans and benthic worms. Enameloid δ 18 O CO3 , δ 18 O PO4 and δ 13 C CO3 values of these fish are used as reference data for comparison with those of ancient fish teeth of the same species.
The archaeological samples used in this study are the first molariform teeth of Sparus aurata (n = 45) of the family Sparidae (Perciformes, seabream). These fish teeth were excavated from 12 well-dated archaeological sites in Israel [22], covering key periods of the Holocene: (1) the Pre-Pottery Neolithic sub-periods (A, B and C) dated to the Early Holocene (11,700-8,500 yr BP); (2) the Chalcolithic, (7,500-5, The fossil fish were identified to species level using reference collections housed at the University of Haifa (Israel), the National Natural History Collections at the Hebrew University and The Zooarchaeological collections at the Autonomous University of Madrid (Spain).

Oxygen and carbon isotope analyses of the carbonate fraction in bioapatite
The enameloid part (~0.2-0.4 mm layer) of each individual tooth crown was separated from the dentine using a diamond-head micro-dental drill, washed three times with distilled water and dried overnight at 50˚C. Each sample was crushed and ground to powder using an agate mortar and pestle. Organic matter was removed from the sample powder by soaking in 30% H 2 O 2 over night as described in Gehler et al. [32]. This pretreatment was tested on shark's teeth (dentine + enameloid) and seemed to have minor effect on the δ 13 C values, which were comparable to those obtained using different pretreatments, such as NaOCl + acetic acid, as well as untreated samples. [24] The pretreatment effect on the δ 18 O CO3 values, observed in shark's teeth [24] was minimised [26] by removing the dentine entirely from the enameloid. Due to low average CO 3 2− content (2-3 wt %), approximately 2 mg of enameloid powder was reacted with 100% H 3 PO 4 at 70˚C in a Thermo Scientific KIEL IV automated carbonate device. The δ 18 O CO3 and δ 13 C values of the samples were then determined on the CO 2 using a Finnigan Delta Plus gas source mass spectrometer in dual inlet mode at the stable isotope laboratory of the Geoscience Center at the University of Göttingen, Germany. Measured isotope ratios were normalised to an in-house carbonate standard (Solnhofen) which has been calibrated against NBS-19. The analytical precision of this method is better than ±0.1‰ for both δ 18 O CO3 and δ 13 C. Measured isotope ratios are reported in δ-notation, i.e., as the deviation in per mil (‰) from the international measurement standards Vienna Standard Mean Ocean Water (VSMOW; δ 18 O CO3 ) and Vienna Pee Dee Belemnite (VPDB; δ 13 C): where R represents the 18 O/ 16 O or 13 C/ 12 C ratio.

Oxygen isotope analyses of the phosphate fraction in bioapatite
The phosphate fraction of the samples was separated using a method modified after Dettmann et al. [34] and described in detail by Tütken et al. [35] and Sisma-Ventura et al. [22]. Five milligrams of pretreated enameloid powder were placed on vibrating table and digested in 0.8 ml HF (2 M) for 12 h. The samples were centrifuged, and the remaining supernatant solution with the dissolved phosphate was separated from the CaF 2 precipitates. The HF solution transferred into a new vial and was neutralised with 25% NH 4 OH using bromothymol blue as pH indicator, and Ag 3 PO 4 precipitation was completed by adding 0.8 ml of 2 M AgNO 3 . The settled Ag 3 PO 4 crystals were centrifuged, and the supernatant solution containing excess AgNO 3 was removed. The Ag 3 PO 4 precipitate was then rinsed five times with Milli-Q water and dried overnight in an oven at 50˚C. Ag 3 PO 4 aliquots of 0.5 mg were placed into silver capsules and analysed in triplicate by means of high temperature reduction using a Finnigan TC-EA coupled via a Conflo III to a Micromass 100 GC-IRMS at the University of Mainz, or by a Finnigan Delta Plus XL GC-IRMS at the Universities of Tübingen and Lausanne, according to the method of Vennemann et al. [36]. The raw δ 18 O PO4 values were normalised to an Ag 3 PO 4 standard with a certified value of 21.7‰ (silver phosphate P/N IVA33802207, batch no. 180097, distributed by IVA Analysentechnik, Germany). The analytical precision for this standard was better than ±0.3‰ (1σ). For untreated NIST SRM 120c Florida phosphate rock standard reference material, we obtained a δ 18 O PO4 value of 21.9±0.3‰ (n = 15). This value compares well with the value of around 21.7‰ initially proposed by Lécuyer et al. [12] and measured in most other laboratories as compiled in Chenery et al. [37].

Oxygen and carbon isotope compositions of carbonate and phosphate in modern and ancient fish tooth enameloid
Individual δ 18 O CO3 , δ 18 O PO4 and δ 13 C data from the analyses of bulk enameloid powder from five modern Sparidae and two modern Cyprinidae are summarised in Table 1 Table 2).
The modern samples of Sparidae from the southeast Mediterranean littoral yield a relatively narrow δ 18 O PO4 range of about 1.0‰ between 22.5 and 23.4‰ (n = 15). In contrast, δ 18 O PO4 values between 21.1 and 26.4‰ were measured for the ancient S. aurata teeth (n = 45).
A narrow range between -0.5 and 0.6‰ is also observed for the δ 13 C values of modern adult S. aurata specimens ( Table 1). A lower δ 13 C value, averaging -2.3±0.6‰ was found for teeth of a single juvenile sample of S. aurata (<0.5 kg). For adult P. caeruleostictus the δ 13 C values range between -0.7 and 0.6‰. No significant differences are present between the average δ 13 C values of the adult S. aurata and P. caeruleostictus samples. The ancient Sparidae teeth ( Table 2) yield variable δ 13 C values, covering a large range from -5.8 to 6.1‰.

The theoretical δ 18 O PO4 range of Sparidae bioapatite
The calculation of the equilibrium δ 18 O PO4 range of bioapatite forming in the southeast Mediterranean littoral and in hypersaline lagoons, evolving from typical seawater of this region, is based on the temperature-dependent relation for the oxygen isotope fractionation during biomineralisation of apatite by Lécuyer et al. [4]:  [30,40]) and can even reach a variation of up to 2.2‰ within the same tooth ( [30], and refs. therein). Similarly high variability was also observed for Δ 18 O CO3 -PO4 in modern shark teeth (enameloid and dentine), which were found to vary between 6.9 and 11.8‰ [24]. The intra jaw inter-tooth Δ 18 O CO3 -PO4 variation in a single modern shark can also vary by more than 2‰ [24]. The Δ 18 O CO3 -PO4 values of modern Sparidae and Cyprinidae enameloid samples vary between 8.1 and 11.0‰, yielding an average of 9.5±0.7‰, (n = 8 modern fish; Table 1). These values agree well with the average Δ 18 O CO3 -PO4 of 9.1±1.5‰ (n = 12) reported for modern shark teeth [24]. Moreover, similar to sharks, high unsystematic variability (~3‰) was observed in the δ 18 O CO3 values of modern Sparidae teeth, while δ 18 O PO4 values vary less by <1‰ in the same teeth (Fig 2A). The high correlation between the δ 18 O CO3 and the δ 18 O PO4 in modern teeth (r 2 = 0.73), indicates that both fractions precipitate simultaneously from the same body water pool [25]. However, the more variable δ 18 O CO3 values in modern bony fish and shark teeth in comparison to that of the δ 18 O PO4 (Fig 2A) suggests that the two fractions could be subjected to different exchange rates, thus integrating different ranges of ambient conditions. Further support for this distribution can be seen in the data obtained from ancient S. aurata teeth, particularly from the LB-Ir age. The larger scatter of the data obtained from the carbonate fraction of the ancient S. aurata teeth may, therefore, still reflect the in vivo distribution of δ 18 O CO3 values [19][20]24], rather than the effect of diagenetic alteration.
We calculated the offset between the predicted and measured Δ 18 O CO3 -PO4 values using the δ 18 O CO3 to δ 18 O PO4 relationship by Pellegrini et al. [30] which is based on all available modern mammal data. The calculated offset for ancient S. aurata (1.0±0.7‰) and fossil Luciobarbus sp. (Cyprinidae) (0.6±0.36‰) from the Δ 18 O CO3 -PO4 regression falls within the predicted interval of uncertainty for the δ 18 O CO3 -δ 18 O PO4 regression for mammals (Fig 2B). This suggests that the majority of enameloid δ 18 O CO3 (and by implication δ 13 C) values of ancient S. aurata teeth and fossil Cyprinidae teeth are well preserved and likely were not affected by any significant diagenetic alteration.
Enameloid δ 13 C of Sparidae teeth δ 13 C values in animals vary as a function of environment and biochemical pathways. The largest difference in δ 13 C occurs between organic (e.g., lipids, carbohydrates, proteins, and collagen) and inorganic (e.g., shell carbonate or carbonate within bioapatite) carbon components (e.g., [41][42][43]). The δ 13 C values of structurally bound carbonate in the bioapatite of the bones and teeth of herbivorous mammals are~6-15‰ higher than the dietary sources of these animals (e.g., [43][44][45]). With similar carbon isotope fractionation between tooth enamel bioapatite, breath CO 2 , and diet, the inter-species differences in the δ 13 C values of structurally bound carbonate in the bioapatite of mammals result primarily from differences in digestive physiology [44][45]. While the δ 13 C of dentine in shark teeth derives predominantly from a dietary carbon source, the enameloid of the same teeth is typically characterised by 6-8‰ higher δ 13 C values [24]. Secondary, in vivo diffusional processes and isotopic exchange with ambient water DIC during tooth formation have been suggested as possible explanations for the elevated δ 13 C values of shark tooth enameloid [24].
The carbon intake of adult Sparidae is mainly derived from their diet; food items include the soft tissues of bivalves, gastropods and crustaceans [46]. Changing dietary carbon sources at different ontogenetic stages of the fish life cycle may explain the somewhat lower mean δ 13 C value of -2.3‰ found in a single juvenile fish (< 0.5 kg). Indeed, wild S. aurata have been observed to forage for shellfish and zooplankton [47] in a manner that varies with the size of the fish [48]. For instance, soft-bodied animals such as polychaetes and small crustaceans predominantly occur in the stomach contents of small S. aurata (5-9 cm in length), while larger, hard-shelled prey such as barnacles, bivalves and other teleost fish are consumed by larger S. aurata (10 to > 25 cm in length) [48].
Typical fractionation between carbonate and dissolved carbon in the form of bicarbonate ions is 2.5‰ at 20˚C [49]. Hence, for typical Mediterranean seawater DIC δ 13 C values of 1.0 ±0.3‰ [38][39] carbonate δ 13 C values of up to 3.5‰ would approach isotopic equilibrium with DIC. Such positive δ 13 C values between 1.6 and 4.8‰ have been recorded in the carbonate fraction of enameloid bioapatite of modern and ancient shark teeth [19][20]24]. A similar value was also measured in this study in a single sample of Scarus sp. (Scaridea, δ 13 C = 3.8‰) from the Red Sea (raised in a seawater aquarium-open system). Structurally bound carbonate δ 13 C values in enameloid bioapatite of modern Sparidae samples did not reach such high values. Thus, their body water bicarbonate pool seems still to be influenced to some degree by their dietary carbon sources, while approaching the isotopic composition of DIC in their habitat.
We estimate the percentage of metabolic carbon (f food ) in the carbonate δ 13 C values of modern Sparidae and Cyprinidae teeth (Fig 3) by using a simple mixing calculation: where δ 13 C of possible food sources for Sparidae range between -8.5 and -23‰ [50][51] and for Cyprinidae between -15 and -34‰ [52]. The δ 13 C of Mediterranean DIC is 1.0±0.3‰ [38][39] and between -4.0 and -7.0‰ for Lake Kinneret [53]. The f DIC and the f Food are the percent contribution of DIC and food, respectively, to the enameloid δ 13 C value (δ enameloid ), and α is the fractionation factor between the dissolved carbon sources and the δ enameloid value (assuming a 2.5‰ fractionation). The f food in the δ 13 C of juvenile Sparidae range between 20% and 50%, while for adult fish (both Sparidae and Cyprinidae) it ranges between 10% and 30% (Fig  3). Assuming that the structural carbonate in the enameloid of bony fish has a similar carbon isotope fractionation as calcite (assuming 1.0‰ fractionation factor), and not aragonite (as assumed for sharks [24]), then the metabolic effect of both Sparidae and Cyprinidae would be much lower than estimated and their δ 13 C values would further approach isotopic equilibrium with the water DIC. The Scaridea δ 13 C value seems to approach isotopic equilibrium with the Red Sea DIC, but may also relate to its durophageous diet feeding on corals and sponges.
The δ 13 C of the freshwater Cyprinidae enameloid samples range between -6.6 and -5.2‰ while the lake water δ 13 C DIC values vary between -7.0 and -4.0‰ (mean δ 13 C DIC of around -5.5‰), with the Jordan River input ranging from -7.3 to -6.5‰ [53]. Thus, carbonate δ 13 C values of up to -2.0‰ would appear to approach isotopic equilibrium with the lake water DIC. Similar to Sparidae, Cyprinidae samples did not reach such high values, supporting the hypothesis that bioapatite δ 13 C values represent a mixture of ambient water δ 13 C DIC and δ 13 C diet . For the δ 13 C diet values of the food sources see text for details. Theoretical equilibrium δ 13 C CO3 of 3.5‰ and -2‰ for carbonate in fish enameloid for the South East Mediterranean (SEM) and Lake Kinneret, respectively, were calculated for an equilibrium fractionation of 2.5‰ (20˚C), based on the δ 13 C DIC of 1‰ and -5‰, respectively, for the two different water bodies [33][34]48]. https://doi.org/10.1371/journal.pone.0220390.g003 Past aquatic environments inferred from paired δ18O-δ13C of fish tooth enameloid The main control on the δ 13 C DIC in coastal lagoons is the mixing of fresh and marine waters entering the lagoon; meaning it is also a useful palaeosalinity indicator [54]. This is reflected in the co-variation of δ 18 O PO4 -δ 13 C and δ 18 O CO3 -δ 13 C values in tooth enameloid of S. aurata, which migrate between water bodies of different salinities [46][47][48]. Thus, variations in the δ 18   Cyprinidae tooth enameloid. The modern fish δ 13 C data is corrected for the Suess effect of -1.2‰ [33,50] to preindustrial values to enable comparison with the ancient Sparidae and Cyprenidae of the Southeast-Mediterranean region. Note that increasing δ 18 O values often also coincide with higher δ 13 C values recording evaporative changes in salinity and carbon source of the lagoonal water bodies these fish were caught from. A freshwater habitat (Lake Kinneret) of modern Cyprinidae is reflected primarily by the low δ 13 C, but also by δ 18 O values lower than those of marine Sparidae. Note that already in PPN some lagoons were mostly brackish-marine as indicated by low δ 13 C and δ 18 O values of the Sparus teeth but a few high values also imply the existence of hypersaline lagoons. modern fish from the southeast Mediterranean coastal water and from Lake Kinneret, respectively. For comparison with pre-industrial values, modern fish tooth δ 13 C values were adjusted to the east Mediterranean Suess effect of~1.2‰ on seawater δ 13 C DIC (the depletion trend in oceanic δ 13 C DIC , due to the emission of CO 2 from burning fossil fuels) obtained from the vermetid reef carbonate δ 13 C composite record [38,55]. The ancient S. aurata teeth that bear the δ 18 O PO4 signature typical for bioapatite formed in Mediterranean coastal water [21][22] also fit the pre-industrial δ 13 C range for modern fish from this habitat. These values are mainly observed in the Early Holocene specimens, suggesting that the fish were caught in typical coastal water [22]. However, teeth with typical hypersaline δ 18 O PO4 values (> 23.7‰, [21][22] similar to those of extant fish from the Bardawil lagoon [2], show typical marine δ 13 C values (Fig 4B).
The highest δ 13 C values in the ancient S. aurata teeth are observed during the Mid-Late Holocene (Bronze Age and Iron Age) and are consistent with high δ 18 O PO4 values typical for tooth formation in hypersaline lagoons [2,[21][22]. The increase in the δ 13 C DIC of closed, evaporative and warm water bodies, like the Bardawil lagoon, may be related to degassing of CO 2 and seasonal biological effects, resulting in the high correlation between δ 18 O PO4 and δ 13 C in modern fish teeth (Fig 4B). Both, photosynthesis and degassing remove 12 C-rich CO 2 from the DIC pool of seawater leading to higher δ 13 C DIC values ( [56] and refs. therein). These effects are amplified when they occur in a limited isotopic reservoir (i.e. a restricted and shallow water bodies), such as a long-term lakes or lagoons [57][58]. Thus, similar to the coupling of δ 18 O CO3 and δ 13 C in carbonates of closed water bodies [57], the positive relationship between δ 18 O PO4 -δ 13 C (r 2 = 0.42; p<0.001) and δ 18 O CO3 -δ 13 C (r 2 = 0.41; p<0.001) values in the bioapatite of ancient and modern Sparidae teeth can be used as proxies for the formation of coastal lagoons.
The δ 18 O PO4 and δ 13 C values of the ancient fish teeth indicate that the hypersaline habitats of S. aurata in the Levant region formed during the Late Holocene from typical southeast Mediterranean coastal water under high evaporation rates and a limited carbon reservoir, similar to the present day Bardawil lagoon in Northern Sinai. This lagoon represents the main hypersaline habitat of Sparidae in the Levantine basin today [59] as well as in the past millennia [21][22]. These evaporatively-influenced lagoonal habitats evolved during the Mid-to-Late Holocene transition (~3.5 ka ago), when sea level stabilised close to its present-day level [60] approximately 3,620±160 years BP [61]. Sea-level stabilisation resulted in the formation of the perennial shallow hypersaline Bardawil lagoon along the northern Sinai coast, due to the establishment of long-shore currents that transported Nile sands which built up blocking sandbars [58,62].
During evaporation, the lighter 16 O isotope is preferentially fractionated into the gas phase, this results in the gas phase having~9‰ (at 25˚C, the mean annual surface temperature of water bodies in the study area) lower δ 18 O values then the water mass from which it evaporated [63]. This, for the most part, is an inreversable reaction. As such, the residual water mass becomes more fractionated in a way that could be described through Rayleigh distillation equations [64]. As such, if we know the initial δ 18 O value of the water mass, and how fractionated it became, we can calculate the loss of water to the vapor phase. As the salt does not evaporate, the residual water mass will retain the original amount of salt. Dividing the initial amount ot salt by the final mass of water will allow us to calculate the salinity. Taking the observed 4 to 5‰ difference in δ 18 O values between the modern (representing the non-evaporated water mass) and Late Holocene S. aurata tooth assemblages, it is thus possible to calculate the fraction of water lost (Fig 5A). Within a reasonable ambient temperature range for the SE Mediterranean region (15 to 30˚C), this translates to a 33% to 46% loss of the initial water massindicating significant evaporation and restricted water exchange between lagoons and the open Mediterranean Sea during the Late Holocene. Assuming an initial seawater salinity of 37‰, this loss would place the salinity in these basins to between 54‰ and 68‰ (Fig 5B), representing the upper range of habitable conditions for S. aurata [2,59]. This is a maximum estimate, given that about 2.5‰ (of the 5‰ range) in Sparus teeth δ 18 O PO4 of the Bardawil lagoon can potentially be attributed to seasonal changes in water temperatures [22].
Our stable isotope results of S. aurata tooth enameloid indicate that an early phase of lagoon formation in the Levant was already taking place during the Pre-Pottery Neolithic period (Fig 4) between 11,000 and 6,000 yr BC (Early-mid Holocene, Greenlandian to mid Northgrippian), despite the lack of sedimentological evidence for such lagoons. However, δ 13 C values suggest that these lagoons were likely also fed by terrestrial sources. In fact, Sparus can tolerate a wide salinity range, including brackish water [46]. While most of the δ 18 O PO4 values of the PPN (Early Holocene) fish fall within the expected range of bioapatite formed in isotopic equilibrium with Mediterranean temperature and seawater δ 18 O values, some of the PPN samples have relatively low δ 13 C and δ 18 O PO4 values. These low δ 13 C and δ 18 O PO4 values fall along the regression between modern fish from the Mediterranean and Lake Kinneret (Fig 4B). This implies a freshwater influx (with low δ 13 C DIC values) into past brackish habitats of Sparus during the PPN, Early Holocene. This provides evidence that S. aurata were caught in different lagoonal settings between the Early and Late Holocene.
Overall, the pairing of δ 18 O PO4 -δ 13 C and δ 18 O CO3 -δ 13 C in fish tooth enameloid thus represent a potential tool for investigating the hydrological history of marine to freshwater lagoonal settings and assessing the environments of past fish habitats. This isotopic approach can provide valuable information about past human fish exploitation patterns and enable us to distinguish marine from lagoonal fishing grounds.

Conclusions
The environmental history of fish ecosystems in the southeast Mediterranean during the Early to Late Holocene was studied by investigating the stable isotope composition of Sparidae tooth enameloid. The provenance of Sparidae from marine versus hypersaline lagoonal water was recorded in the δ 18 O PO4 , δ 18 O CO3 and δ 13 C values of modern and ancient fish teeth, which reflects the wide range of saline habitats encountered during their migratory life cycle. The enameloid δ 18 O CO3 as well as the Δδ 18 O CO3-PO4 offset and δ 13 C values of Holocene Sparidae teeth from the Levant are consistent with those of modern fish, indicating only minor diagenetic alteration. The δ 13 C of Sparidae tooth enameloid reflects the water DIC of the fish habitat and its diet. Thus, carbon cycling in past fish environments can be characterised using the δ 13 C of fish teeth.
Paired δ 18 O PO4 -δ 13 C and δ 18 O CO3 -δ 13 C values of ancient Sparidae tooth enameloid were used to infer the environmental history of fish from coastal and lagoonal settings in the Levant.
Our results indicate that Late Holocene hypersaline lagoons evolved from typical southeast Mediterranean coastal water under high rates of evaporation and limited carbon input from terrestrial sources. This lagoon formation occurred in association with the postglacial stabilisation of the sea level about 3,500 years ago, however, first lagoons had already formed during the Early Holocene and were likely influenced by freshwater input. Overall, the pairing of δ 18 O PO4 -δ 13 C and δ 18 O CO3 -δ 13 C of fish tooth enameloid is therefore, a potential tool to infer past hydrological balance and environmental shifts of fish ecosystems. This enables us to gain new insights into past fishing grounds that were exploited by (pre-)historic humans.