The analyzed organisms included zooplankton, mollusks [bivalves—Crassostrea gigas (n = 9 individuals), Pecten maximus (n = 4)—and cephalopods—Loligo vulgaris (n = 3), Onykia ingens (n = 3), Sepioteuthis lessoniana (n = 3)], crustaceans [Malacostraca—Maja brachydactyla (n = 3), Litopenaeus stylirostris (n = 3)], and fish [Chondrichthyes and Actinopterygii—Lophius piscatorius (n = 2), Merluccius merluccius (n = 2), Sardina pilchardus (n = 2), Scyliorhinus canicula (n = 2), Solea solea (n = 2), Spondyliosoma cantharus (n = 2), Zeus faber (n = 2), Plectropomus leopardus (n = 2)]. These were divided into trophic groups to assess the influence of food and feeding ecology characteristics on Li bioaccumulation and Li isotopy. We defined the trophic groups according to feeding mode (filter-feeders, grazers/scavengers, predators) and the preferred predator diet (i.e., invertebrates, small fish, or both) (Metian et al., 2013; Cipro et al., 2018). For precision on the ecology of the studied species, see Supplementary Table S1 and previous work (Bustamante et al., 2003; Bustamante and Miramand, 2005; Chouvelon et al., 2009; Chouvelon et al., 2012; Metian et al., 2013; Cipro et al., 2018; Noger-Huet et al., 2022).

Fifty-eight of the 78 samples were previously characterized in Li concentration (Thibon and Weppe et al., 2021c) and selected from distinct biogeographic areas: the Bay of Villefranche-sur-Mer (North Mediterranean sea—2020), the Bay of Biscay (northeast Atlantic Ocean—2018), New Caledonia (western tropical Pacific Ocean—2007), and the Kerguelen Islands (southern Indian Ocean—1998). The remaining 30 samples consist of six zooplankton samples and 14 fish organs sampled from 2 John Dory. These two fish were collected in the Bay of Biscay in 2020. Zooplankton was collected on 12 March 2020 (9:30–10:30a.m.) from the outer part of the Villefranche-sur-Mer Bay using the CNRS Sagitta-III boat. To sample all the zooplanktonic community, we used a WP2 net of 200 µm size along a bottom-surface (75–0 m) transect. We then sub-sampled the planktonic community >500 µm with appropriate sieves. In these coastal Mediterranean waters and at this period of the year, zooplankton from 200 to 500 µm are mainly constituted by filter-feeder copepods (Clausocalanids), whereas zooplankton >500 µm are mainly carnivorous copepods (Candacia spp.) and jellyfishes (gelatinous zooplankton) (Seguin, 1981; Romagnan et al., 2015; Nival et al., 2021). Concerning bivalves, cephalopods, and fish, some individuals were dissected in a clean laboratory to sample specific organs (dorsal muscle, liver, kidneys, gills, and brain for fish; adductor muscle, gonads, digestive gland, and gills for bivalves; mantle muscle and digestive gland for cephalopods). Tissue samples were stored in clean individual plastic bags, fresh weighed (wet weight—ww), and put at −20°C until being freeze-dried for 48 h. After drying, samples were again weighed (dry weight—dw), ground to a fine powder with a porcelain mortar and pestle, and stored in clean individual plastic vials. For more details on bivalves, crustaceans, cephalopods and fish sampling and sample preparation, see Thibon and Weppe et al. (2021c).

All analytical preparations were performed in a pressurized clean laboratory under a fume-hood to minimize procedural blanks using ultra pure reagents (HCl, HNO₃, and Milli-Q water) and pre-cleaned vessels. The samples were first dissolved with a mixture of concentrated HNO₃ (64%—3 mL) and H₂O₂ (32%—.4 mL). After 48 h, concentrated HCl (37%—1.5 mL) was added to pursue the digestion in inverse aqua regia, and heated for 24 h before being evaporated. A second step of inverse aqua regia ensured total digestion.

Prior to Li isotopic analyses, the Li in the samples must be isolated. Lithium extraction and purification were performed following the procedures of Vigier et al. (2008, 2009). Sample aliquots of ∼30 mg were dried out and taken up in .5 mL of 1 N HCl and loaded on AG 50-X12 (200–400 mesh) cationic resin in 8.5 cm-high Teflon columns. Lithium was eluted and purified with 1 N HCl. This fraction was evaporated to dryness before the analysis. Five total procedural blanks, five replicates of seawater standard solution (SW-1G, Thibon et al., 2021b) were included, as well as several biological materials, allowing us to evaluate the precision and accuracy of the overall procedure (Thibon et al., 2021b).

Lithium has two naturally occurring isotopes, one heavy (⁷Li, 92.41% abundance) and one light (⁶Li, 7.59% abundance). The isotope compositions of geological and biological materials are reported as per-mil variations relative to an international isotopic standard (NIST RM 8545 reference material, known as LSVEC) as δ⁷Li_sample [‰] = [((R_sample/R_LSVEC)−1) × 1000], where R is the corresponding ratio (⁷Li/⁶Li) in the sample and in the international standard. Li isotopic analyses were performed using a multi-collector inductively coupled plasma mass spectrometer (Thermo-Fisher Neptune Plus at the CNRS-INSU Facilities, École Normale Supérieure de Lyon, France) following the procedure developed for low Li concentrations in biological materials (Balter and Vigier, 2014).

Samples and standards were analyzed under dry plasma conditions in low-resolution mode. The standard bracketing method (using the LSVEC reference material) was used to performed instrumental mass bias correction. The five total Li procedural blanks, ranged between 7 and 22 pg, with an average of 13 pg (n = 5). For 30 mg of material with a Li mass fraction of 1 μg/g, the maximum blank contribution is negligible (<.07%). On average, the internal reproducibility (within-run precision) was ±.04‰. Between-run-precision (external reproducibility) was estimated from two to five replicate measurements of five different aliquots, and are about ±.3‰, except for DORM-2 and TORT-2 references, which display Li concentrations lower than .2 mg/g (Thibon et al., 2021b). The accuracy of this method has been confirmed by a standard addition method applied to fish muscle reference material (Thibon et al., 2021b). Seawater standard solution (SW-1G: EDMM-1-G from Analytical Environnemental Laboratory), previously published at 31.3‰ ± .8‰ (n = 4, 2 s) (Thibon et al., 2021b), has been measured here at 30.8‰ ± .5‰ (n = 5, 2 s), which confirmed the absence of measurements bias.

Lithium concentrations were previously published (Thibon and Weppe et al., 2021c), and determined at the LIENSs laboratory using an Inductively Coupled Plasma Mass Spectrometer (Thermo Fisher X-Series II). For samples collected in the Bay of Biscay, concentrations of several trace metals (Ag, As, Cd, Co, Cr, Cu, Fe, Hg, Mn, Ni, Pb, Se, V, and Zn) were also performed at LIENSs using the same instrument or a Varian Pro Inductively Coupled Plasma Optical Emission Spectrometer and, only for the Hg, a Altec AMA254 Atomic Absorption spectroscopy (Noger-Huet et al., 2022).

Since normality and homoscedasticity of the Li compositional and isotopic data were not validated, we adopted non-parametric statistical tests to compare population of more than two samples. The Mann-Whitney-Wilcoxon and Kruskal-Wallis tests were used to compare median values between two or more observational series, respectively. Tests were performed using R software (R Core Team, 2017) using the functions wilcox.test or kruskal.test. The level of significance for statistical analyses was set at p-value < .05.

A multifactor ANOVA has been performed using R software (R Core Team, 2017) on the whole dataset to test whether the δ⁷Li (bcPower-transformed to approach normality and homoscedasticity) is influenced by any of the environmental and ecological factors investigated or by an interaction of the factors.

In addition, the relationship between quantitative variables was investigated using principal component analysis (PCA, see Section 4.2), a statistical tool that reduces the number of original quantitative variables to fewer dimensions that may explain the majority of the observed variability. Here, we performed a PCA on two groups (fish and bivalves from the Bay of Biscay), considering the following initial variables: As, Co, Hg, Li, and Zn concentrations, which include nutrients, non-nutrients, moderately to highly toxic elements, and δ⁷Li.

Marine organisms display a wide range of δ⁷Li values, from 4.6‰ (digestive gland of bivalves from the Bay of Biscay) to 32.0‰ (zooplankton from the Bay of Villefranche-sur-Mer; Table 1, Supplementary Table S1 and Figure 1). There is no visible simple correlation with a single environmental or ecological parameter.

For zooplankton and bivalves, whole organisms were analyzed. Zooplankton from 200 to 500 µm in size display δ⁷Li values close to the seawater one (31.2‰), ranging from 31.3‰ to 32.0‰. Zooplankton larger than 500 µm in size display slightly lower values compared to seawater, from 28.1‰ to 31.1‰, although not significantly different than the smaller size one on average (Wilcoxon test: p-val. > .05). Bivalves (without shells) display similar δ⁷Li average values in the Bay of Biscay and in New Caledonia (26.1‰ ± 5.1‰ and 27.5‰ ± 5.9‰ respectively). In the bivalves from Bay of Biscay, muscles, gonads, and gills display similar average δ⁷Li values (muscles: 25.4‰ ± 2.9‰, n = 2; gonads: 24.6‰ ± 8.8‰, n = 3; gills: 23.0‰ ± 1.7‰, n = 3, Table 1). In contrast, digestive glands display the most fractionated δ⁷Li value (6.0‰ ± 3.9‰, n = 2). For cephalopods, muscles and digestive gland were analyzed separately. Muscle δ⁷Li values are 25.0‰ ± .7‰ (n = 3), and 25.6‰ ± 1.9‰ (n = 3) in the Bay of Biscay and in the Kerguelen Islands, respectively (Wilcoxon test: p-val. > .05). In New Caledonia, cephalopod muscles are slightly lower than those from the other studied regions (21.2‰ ± 1.9‰, n = 3), although not significantly (Wilcoxon test: p-val. > .05). The digestive gland of cephalopods from New Caledonia (22.4‰ ± .2‰, n = 3) is not isotopically fractionated compared to the muscles (Wilcoxon test: p-val. > .05), in contrast to bivalves. In the Kerguelen Island, the digestive gland average δ⁷Li value (19.9‰ ± 4.5‰, n = 3) is lower than that in muscles (Table 1), the same feature as observed in bivalves from the Bay of Biscay. In crustaceans, δ⁷Li values were measured in muscles only, from both the Bay of Biscay and New Caledonia individuals, and they show non-significant differences (25.9‰ ± .9‰, n = 3 vs. 20.9‰ ± 4.1‰, n = 3, respectively, Wilcoxon test: p-val. > .05).

For fish, we analyzed Li isotope composition of various organs from samples collected in the Bay of Biscay (muscles, kidneys, gills, brain, heart, and gonads). Liver and muscles from fish collected in New Caledonia were also analyzed for comparison. In the Bay of Biscay, muscles depict an average δ⁷Li value of 23.8‰ ± 4.1‰ (n = 11), similar to the value determined in a New Caledonia fish, although only one data has been collected (n = 1). Bay of Biscay and New Caledonia fish livers have lower δ⁷Li values than corresponding muscles (15.2‰ ± 10.3‰, n = 4, and 19.0‰ ± 4.3‰, n = 2, respectively, Table 1). Kidneys, gills, brain, heart, and gonads also display lower δ⁷Li values when compared to muscles, i.e., 19.7‰ ± 6.4‰ (n = 4), 19.5‰ ± 12.8‰ (n = 3), 18.1‰ ± 1.1‰ (n = 3), 17.9‰ ± 2.1‰ (n = 2), and 16.1‰ ± 1.4‰ (n = 2), respectively, the lowest value being measured in gills (12.1‰).

All taxonomic groups combined (bivalves, crustaceans, cephalopods and fish), muscles from the Bay of Biscay and Kerguelen species present higher δ⁷Li than muscles from New Caledonia species (24.5‰ ± 3.6‰, n = 19, and 25.6‰ ± 1.9‰, n = 2, vs. 21.5‰ ± 3.3‰, n = 7, respectively) (Table 1). In contrast, whole bivalves (without shells) do not depict any difference between individuals from the Bay of Biscay and New Caledonia (26.1‰ ± 5.1‰, n = 6, vs. 27.5‰ ± 5.9‰, n = 3, respectively, Wilcoxon test: p-val. > .05, Table 1; Figure 1).

Within a single location, muscles from the different taxonomic groups display similar average δ⁷Li values (Table 1): 25.4‰ ± 2.9‰ (n = 2), 25.9‰ ± .9‰ (n = 3), 25.0‰ ± .7‰ (n = 3), 23.8‰ ± 4.1‰ (n = 11) for bivalves, crustaceans, cephalopods and fish from the Bay of Biscay, and 20.9‰ ± 4.1‰ (n = 3), 21.2‰ ± 1.9‰ (n = 3), 23.8‰ (n = 1), for crustaceans, cephalopods and fish from New Caledonia, (see also Table 1; Figure 1). This feature is not observed for gills, which display distinct isotope compositions in bivalves and fish from the Bay of Biscay (23.0‰ ± 1.7‰, n = 3, vs. 19.5‰ ± 12.8‰, n = 3), but with a high variability associated to a low number of samples.

To help understand Li behaviour in marine biological systems, the organisms Li concentrations and Li isotopic distribution can be compared to those of other metals measured in the same tissues—using a principal component analysis—PCA applied to the Bay of Biscay. The PCA is performed on two groups (fish and bivalves, all tissues), considering the following initial variables: As, Co, Hg, Li, and Zn concentration and δ⁷Li (Thibon and Weppe et al., 2021c; Noger-Huet et al., 2022). Variables and individuals are plotted on the space constituted by the first two components—PC1 and PC2—in Figure 2. The two first components together represent 87% and 67% of the dataset variability for bivalves and fish, respectively. This implies that PC1 and PC2 are sufficient to explain the majority of the dataset variability. According to their position on PC1-PC2 space, variables can be split into two groups for bivalves (As, Hg, and Zn vs. Co, Li, and δ⁷Li), whereas the separation is not straightforward for fish. The first two components are adequate to fully discriminate the different tissues sampled. In bivalves, Co correlates with Li (r = .75) and anti-correlates with δ⁷Li (r = −.66), whereas it is not the case for fish (Supplementary Figure S1).

To evaluate the percentage of δ⁷Li variability that environmental and ecological factors can explain, we performed a linear model on δ⁷Li, bcPower-transformed to approach normality and homoscedasticity. This model reports an adjusted R-squared around 60%, indicating that the studied factors cannot explain 40% of δ⁷Li variability. Further studies are therefore needed to consider additional extrinsic parameters (environmental ones such as pH, salinity, climate conditions, presence of contaminants, season), and intrinsic parameters (e.g., organisms’ sex, age, reproductive status).

Nevertheless, since 60% of δ⁷Li variability can be explained, a multifactor ANOVA has been performed on the whole dataset, using R software (R Core Team, 2017), to test whether the δ⁷Li is influenced by either any of the factors investigated or by an interaction of these latter. The ANOVA results show that the variability of δ⁷Li is statistically impacted by both the type of sampled tissue and the taxonomic group (p-val. < .001) and at a smaller level by the habitat (pelagic—demersal—benthic) (p-val. < .01) and the trophic group (p-val. < .05). On top of that, there is a significant interaction between the tissue and the location (p-val. < .001). Therefore, their combined effect is stronger than the effect of these latter factors considered separately. We further discuss the impact of tissue and habitat on Li isotopes fractionation in the following sections.

As described in the introduction, the studied samples evidence that Li concentrations in marine organisms follow a biodiminution effect along the food webs, from filter-feeders to predatory fish, i.e., Li concentrations decrease from species of low trophic level to the ones of high trophic level (Thibon and Weppe et al., 2021c). Our study adds to this observation the first Li concentrations of zooplankton samples and these data confirm the biodiminution effect previously highlighted. Indeed, zooplankton displays Li concentrations from 5.9 to 22.3 µg/g, i.e., at least one order of magnitude higher compared to those of bivalves and all other groups analyzed so far (Li = .01–1.20 µg/g, Thibon and Weppe et al., 2021c). The fact that filter-feeders display higher Li concentrations than predators has been interpreted as a major role of filtration on Li uptake and bioaccumulation pathway in this group.

Concerning Li isotopes, we find that δ⁷Li values measured in zooplankton (28.1‰–32.0‰) are higher and statistically different from the other taxonomic groups (Wilcoxon test: p-val. < .01). They are also not different from the seawater value that is homogenous at the world ocean scale (31.0‰ ± .3‰, Millot et al., 2004; Carignan et al., 2007). For the other groups, their average δ⁷Li values are lower than seawater, meaning that they are enriched in the light ⁶Li isotope compared to their environment (Table 1). Although not always significant, their δ⁷Li values seem to decrease from the trophic groups representing low trophic levels—i.e., plankton—to trophic groups representing high trophic levels—i.e., PSM (Table 1; Figure 3). It can also be noticed that the maximum δ⁷Li value of each trophic group decreases from zooplankton to bivalves, and to crustaceans, cephalopods, and fish, suggesting a potential role of the trophic level on Li isotopes that would deserve to be further investigated. The Li isotopic variability among each trophic groups is however high and might be explained by either additional physiological effects or environmental factors.

Our dataset shows that Li isotopes are not homogeneously distributed among organs and tissues. Indeed, the various organs and tissues display a large range of δ⁷Li values (Figure 1), implying a crucial role of physiology in Li isotope fractionation. However, the Li physiology itself in the marine environment is little known (Thibon and Weppe et al., 2021c). Lithium intake, accumulation, and elimination may fractionate Li isotopes, as it has been indicated by cultured bivalves (Thibon et al., 2021a). These processes may vary differently according to the species, tissue physiological function, and environmental parameters. For instance, clams and mussels are able to bioaccumulate Li (Figueroa et al., 2013; Thibon et al., 2021a) and this is accompanied by an enrichment in ⁷Li. Li has long been considered as a non-essential element for the organisms, in contrast to micronutrients such as Zn and Cu. Nevertheless, Li essentiality and functions are currently debated even for human beings for whom the number of studies is consequent, in relation with its role in the brain and the toxic risks for treated bipolar patients (e.g., Schrauzer, 2002; Roux and Dosseto, 2017; Shahzad et al., 2017). It is known that active transporters present in all living organisms play a key role in the Li transfer to cells and organs (Counillon et al., 2016). Kinetic biological Li isotope fractionation have recently been highlighted (Thibon et al., 2021a; Poet et al., 2022) during active Li transfer, along with the intracellular pH and Na regulation, which influences the Li levels in cells and in specific organs (e.g., kidneys, brain).

When considering only bivalves and fish from the Bay of Biscay, the combination of Li concentrations and Li isotopes allows to fully discriminate tissues (Figure 4), which strongly suggests that Li is involved in distinct paths and functions among these internal components. The high variability observed within fish tissues reflects a species effect. Indeed, the four kidneys’ samples depict two groups corresponding to two different species (monk fish L. piscatorius and John Dory Z. faber), although these species have distinct trophic ecology (in the Bay of Biscay, Figure 4).

We selected two individuals from the same fish species (John Dory—Z. faber), from which all organs were analysed (Figure 5; Supplementary Table S1). At first, results confirm significant δ⁷Li heterogeneity between tissues since the values range from 7.7‰ to 23.4‰, similar as it has been reported for one terrestrial mammal, sheep, although over a smaller range (Balter and Vigier, 2014). With the exception of the liver, the two studied individuals highlight a good similarity in their Li concentrations and δ⁷Li values (Figure 5A). One way to comprehend these tissue specific Li signatures is to understand how Li behaves within and between these tissues (e.g., exchange, storage, transfer, remobilisation, …). However, we do not know yet if any specific tissue governs the Li distribution in marine organisms [such as the liver for copper in humans; Kim et al. (2008)]. Liver is particularly depleted in Li, and there is not a unique tissue that strongly bioaccumulates Li within the John Dory organism. At the cellular level, Li homeostasis and regulation are controlled by fluxes through ionic cellular transporters such as Na/H exchangers (NHE, Counillon et al., 2016). Thus, brain, kidneys, heart, and gills—highly implicated with Na exchanges—can actively exchange or bioaccumulate Li in fish. Gills show the highest δ⁷Li values, with a Li concentration close to that of brain, gonads, and heart. Gills are in direct contact with seawater, which may explain its isotopic difference with other tissues and its high δ⁷Li value. On average, brain, heart, and kidneys depict similar δ⁷Li values, suggesting a negligible isotope fractionation of the Li internal transport from the intestine to the excretion system (Figures 5A, B), in contrast to what was observed in sheep (Balter and Vigier, 2014). Another difference is that in sheep, liver and muscles display the heaviest isotope compositions while in the John Dory, gills display higher δ⁷Li values than muscles. This difference of Li behavior between terrestrial mammals and fish may partly come from the primary Li intake pathways (filtration and ingestion for fish and ingestion for sheep), which may fractionate Li isotopes differently.

A noticeable feature is that none of the tissues display a δ⁷Li value heavier than that of seawater or gills, while a Li mass balance implies that if some tissues display lower δ⁷Li than the Li intake, other tissues must be enriched in ⁷Li. This low δ⁷Li values may suggest that the Li intake is not dominated by a seawater-like value and therefore is not due primarily to the gills filtering system. Likely, a non-negligible part of the incorporated Li may come from prey with lower δ⁷Li values than seawater.

It is possible to calculate and compare the whole organisms δ⁷Li for the two John Dory using the following equation: δ 7 L i w h o l e = ∑ i   L i i L i w h o l e × M i M w h o l e δ 7 L i i , (1) with i the various organs, and M the mass. These calculations result in a similar whole body δ⁷Li value: 20.4‰ and 21.5‰, respectively for the two individuals (Figure 5C). These two John Dory having different total body weight (.9 vs. 1.3 kg), this homogeneity of Li isotope composition of a single species supports a species-specific Li regulation, and a weak effect of metabolism, even if more individuals would be needed to confirm this observation. Both John Dory δ⁷Li values are fractionated by −10‰ compared to seawater, in favor of the light ⁶Li. This fractionation is stronger than most Li isotopic data displayed by crustaceans and mollusks muscles (Figure 3), the classical fish prey. Perhaps other prey with lower δ⁷Li values are an important source of Li for John Dory. Another possibility is that gills transfer Li to the fish with a strong Li isotope fractionation in favor of the ⁶Li, but with no effect on the δ⁷Li value of the gill itself. This last hypothesis would a priori be consistent with previous conclusions based on Li concentrations, but would need to be tested in laboratory-controlled conditions.

For bivalves, the different organs and tissues were analyzed in the king scallop P. maximus (Figure 1). Contrary to fish, scallop’s gills δ⁷Li values were similar to those of the muscles and gonads (Figure 1). This may be explained by the different volume of seawater filtered by their gills or by the different roles that gills can play between these groups. Both bivalves and fish use gills for respiration purposes, but bivalves use also gills to feed, and they filter a much larger volume of seawater. This may result in a distinct activity of the NHE transporters, explaining the observed weaker fractionation in bivalves, where gills are used for both respiration and feeding purposes. Another possible explanation is that for invertebrates, blood composition is influenced by fewer processes in contrast to vertebrate fish with bones (Li being able to incorporate bone mineralized tissues; Thibon et al., 2022, in press).

Among bivalve tissues, digestive gland displays the highest Li concentrations, with a distinct δ⁷Li value compared to muscle, gonad, and gills. This likely results from a higher bioaccumulation in this tissues, such as already observed for other trace metals (Bustamante and Miramand, 2004; Metian et al., 2008a; Metian et al., 2008b; Saavedra et al., 2008). However, we cannot exclude some contamination due to the potential presence of small Li-rich/low δ⁷Li aluminosilicate particles within the cells of the digestive gland (Chassard-Bouchaud et al., 1984; Thibon and Weppe et al., 2021c).

Finally, the results of our PCA, which includes Li and other trace metals (see Section 3.3) measured in bivalves and fish from the Bay of Biscay, highlight an intimate link between Co and Li, and a major role of digestive gland on their homeostasis (Figure 2; Section 3.3). As shown in Supplementary Figure S1, digestive gland of scallops is distinct from the other bivalve tissues, with a higher Li concentration and a particularly lower δ⁷Li value, suggesting a major role of this organ in both Co and Li metabolism. This also suggests a role of digestive glands in Li and Co detoxification, by storage or excretion (Bustamante et al., 2000). In contrast, in fish, Li and kidneys are close in the PCA, suggesting a key role of this organ in the Li regulation. However, in Figures 1, 4B, Li concentrations are low in all fish tissues, and kidneys do not display a distinct Li concentration nor Li isotope composition compared to other fish tissues. Overall, these results support distinct Li regulation pathways between those two taxonomic groups, and the Li isotope variability among tissues is higher in the invertebrate while their Li concentrations are higher.

It is possible—as a first approximation—to consider muscles only, from all taxonomic groups combined, in order to examine the effect of environmental factors on δ⁷Li (Supplementary Figure S2). Looking at the role of habitat, δ⁷Li of muscles from the Bay of Biscay organisms does not appear to be different between demersal, pelagic, and benthic species (Supplementary Figure S2b). This suggests a negligible role of the habitat on Li isotopes, which is consistent to what has been observed for Li concentrations (Thibon and Weppe et al., 2021c). It also implies that Li isotopes behave similarly in muscles, independently of the habitat of their host. Nevertheless, the number of individuals is limited for each habitat (n = 2–10) and more data would be needed to verify more precisely the role of this parameter.

In contrast, differences appear at a broad geographic scale. Thus, muscles of the organisms from New Caledonia display a δ⁷Li statistically lower than those from the two other locations (Table 1; Supplementary Figure S2). Such a geographical effect on Li isotopes is mainly triggered by cephalopods and crustaceans (Figure 1). New Caledonia is known to account for a large proportion of Co and Ni worldwide production (3% and 10% in 2016, respectively; Plaza-Toledo, 2016). However, Co and Ni mining releases other trace metals in the environment, such as Cr and Fe (Pasquet et al., 2016). Moreover, New Caledonia Co bearing minerals can also be Li-rich, such as lithiophorite (.3 wt% LiO₂ and 7.9 wt% Co₂O₃) or asbolan-lithiophorite intermediate (.01 wt% LiO₂ and 13.9 wt% Co₂O₃) (Llorca and Monchoux, 1991). Although Li release from Co or Ni mining has not been quantified precisely yet, mining processes may release some Li compounds, similarly to other trace metals. Therefore, the δ⁷Li shift of New Caledonia cephalopods and crustaceans might be related to Li-rich prey carrying local terrestrial Li isotopic signatures, generally enriched in ⁶Li (with low δ⁷Li). Minerals and rocks from the continental crust, and more particularly from mafic rocks as in New Caledonia, have a low δ⁷Li value, generally lower than 4‰. It is interesting to observe that this ⁶Li-rich signature is not recorded in fish (see Figure 1). Focusing on δ⁷Li in low trophic level organisms such as coastal filter-feeders or grazer/scavenger may therefore help to identify anthropogenic or geogenic contamination processes. Indeed, a recent laboratory study highlighted the capacity of Li isotopes in mussel soft tissue to track Li levels in the environment (Thibon et al., 2021a). Moreover, Li isotopes measured in an Asian river revealed their potential to identify the source of Li contamination (Choi et al., 2019).