Conjugated enrichments in arsenic and antimony in marine deposits used as paleoenvironmental proxies: preliminary results

Arsenic (As) and antimony (Sb) are two metalloids often mentioned together due to their belonging to group V of the periodic table, which gives them common characteristics (Tab. 1). They are most often mentioned for problems of toxicity and pollution of soils, lakes and drinking water supply; a large literature is available on this subject. On the other hand, relatively little work has been devoted to the conditions of accumulation of these two elements in marine sediments, when they are considered as paleo-environmental markers. Arsenic is much more studied than antimony in this regard. A recent study synthesized the complex geochemistry of sedimentary As (Tribovillard, 2020) and the behavior of this element in marine environments can be briefly summarized as follows. Arsenic is mainly brought to the sediments with (oxyhydr-)oxides of iron and manganese. If reducing conditions develop at or below the water-sediment interface, As can react with sulfide ions to form soluble species that can leave the sediment. On the other hand, As will remain trapped if it can react with iron sulfides (pyrite; see details in Tribovillard, 2020 and references therein). Arsenic and antimony show similarities and differences in the marine environment (Cutter, 1991; Cutter and Cutter, 1995, 2006; Cutter et al., 2001; Chaillou et al., 2008). Both are primarily transferred from the water column to the sediments by Fe and Mn (oxyhydr-)oxides (Qin et al., 2019). However, they can have contrasting behaviors when subjected to variable redox conditions, depending on the amount of reactive iron available in the sediment (Chaillou et al., 2008; Polack et al., 2009; Ye et al., 2020). Arsenic and Sb are readily incorporated into pyrite when it can form (Gregory et al., 2015) but if trapping does not take place, it has been observed that As could be mobile under reducing conditions whereas Sb was preferably mobile under oxic conditions (Ye et al., 2020 and references therein).

The goal of this study is to use the behavioral differences between these two metalloids to progress in paleoenvironmental reconstruction based on geochemical data. In a recent study (Tribovillard, 2020), a set of geological formations of very different age and depositional environment were examined to better understand how the enrichments in As could serve as a paleo-environmental proxy. The formations for which the data of Sb exist will be examined here and the relative enrichments in As and Sb will be compared. The differences and similarities between the distributions of these two elements will allow us to propose paleoenvironmental interpretations that can be transposed to many other marine deposits of the Phanerozoic.

For the present work, the same geological formations as those examined in Tribovillard (2020) will be studied, except for the ones with no Sb data available. The formations are described in Tribovillard (2020) and references therein; the descriptions are summarized in Table 2. The analytical methods are also described in Tribovillard (2020). Some of the results are expressed through the enrichment factors of the elements. Enrichment factors (EF) were calculated as: X-EF = [(X/Al)_sample/(X/Al)_{upper crust}], where X and Al represent the weight % concentrations of element X and Al, respectively. Samples were normalized using the average composition of the upper continental crust of the Earth (McLennan, 2001). The aluminum normalization is used to avoid the effects of variable dilution by carbonate and/or biogenic silica, although certain pitfalls may (seldom) accompany this approach when aluminum content is minimal as may be the case with carbonate rocks (see discussion in Van der Weijden, 2002 and Tribovillard et al., 2006). Any value larger than 1.0 theoretically indicates the enrichment of an element relative to its average crustal abundance but, practically, an enrichment may be considered to be detectable when EF > 3 (Algeo and Tribovillard, 2009). Aluminum normalization and the calculation of enrichment factors have been retained in this paper because it is a convenient way to compare geological formations and sediments deposited in very different deposit contexts. However, this type of standardization is not always a panacea and the examination of elemental concentrations can also provide relevant information, especially when the chemical elements can be carried by particular mineralogical phases. By way of illustration, we can cite the work of Baux et al. (2019) who observed that glauconites carried relatively high concentrations of arsenic in the greenish sands of the Bay of Seine in Normandy.

We will focus here upon the relationships between the respective enrichment factors in As and Sb, formation by formation. All the cross plots are illustrated with the Figure 1S available online (supplementary materials) and the results are summarized with Table 3. Four situations have been identified: situation 1 with low enrichments in both As and Sb, the other three situations with more or less pronounced enrichments in As and/or Sb. Situation 2 is when As is more enriched than Sb, situation 3 is when Sb is (slightly) more enriched than As, and situation 4 is when As and Sb enrichments are marked and proportional. Most of the time, the formations studied correspond to one situation only, but the samples of the Pigmy and Cariaco basins can be grouped in subsets belonging to two or three situations.

For the depositional environments subjected to oxic conditions (Jurassic Vaca Muerta, Argiles de Châtillon, Argiles de Wimereux, Cretaceous La Charce, Cenozoic Pigmy Basin prop parte), low enrichments in As and Sb are observed, with however enrichments in As a little more pronounced than those in Sb. This discrepancy may be related to a normalization bias. In the present work (as well as in Tribovillard, 2020), the enrichment factors are calculated using the upper crust composition of McLennan (2001), i.e., 1.5 ppm for As and 0.2 ppm for Sb. This choice wad guided by the fact that the paper of McLennan (2001) is a widely-used reference in paleo-environmental reconstructions. The value for Sb is in the range of those reported by various authors (Hu and Gao, 2008, and references therein) but the As content is lower than that reported by Sims et al. (1990), Gao et al. (1998), Rudnick and Gao (2003) and Hu and Gao (2008). As already pointed out to by Tribovillard (2020), the low As value of McLennan (2001) may artificially create an automatic > 1 value when calculating enrichment factors, thus erroneously suggesting some enrichment whereas such a bias does not occur using the consensual value of 0.2 ppm for Sb. Consequently, it is concluded here not to take into consideration the differences in the enrichment factors observed in the present work for formations deposited under oxic conditions. We may conclude that no significant enrichment in As and Sb are observed for sediments deposited under oxic conditions.

For reducing environments such as those studied here recording the Frasnian-Fammenian boundary, the respective enrichments in As and Sb are marked and largely proportional, except for the euxinic setting of La Serre (Montagne Noire, France) that yields scattered values. However, the sediments of the highly confined and anoxic Orca Basin (Gulf of Mexico) do not show significant enrichments in As and Sb. We have no explanation for this counter-intuition absence of enrichment despite reducing conditions: is it related to the over-salinity of the basin (Tribovillard et al., 2008) or some basin reservoir effect (Algeo and Lyons, 2006) or the shortage of reactive iron and manganese (Tribovillard, 2020) or some other unsuspected factor? However, the idea to be retained is that sediments deposited under reducing conditions do not all show co-enrichments in As and Sb. This observation suggests that these metalloids are not robust redox proxies.

Still considering anoxic and/or euxinic depositional milieus, the Cariaco basin shows some enrichments in As and Sb. We observe similar distributions for both elements, with (1) gradually lower enrichments in As and Sb with increasing enrichment in molybdenum (Mo), that is, increasingly reducing conditions, and (2) a comparatively slightly higher enrichment in Sb relative to As when both elements are relatively enriched (Fig. 1). The Cariaco Basin yields a rather specific conjunction of factors:

• seasonal upwelling, stimulating a high productivity (Haug et al., 1998);

• highly stratified water column prone to the development of euxinic conditions (Haug et al., 1998; Aycard et al., 2003; Quijada et al., 2015, 2016);

• high sedimentation rates (Algeo and Lyons, 2006);

• active shuttling via Mn and Fe (oxy-hydr)oxides (Algeo and Tribovillard, 2009);

• relatively limited availability of reactive iron (Tribovillard, 2020).

The role of a high sedimentation rate can be discussed to account for the diminished enrichments in As and Sb when the basin was highly confined (e.g., Crombez et al., 2020; Liu and Algeo, 2020). Alternatively highly confined, hence reducing, conditions may have favored the formation of soluble As species, explaining why this element could be impoverished in the sediments (see discussion in Tribovillard, 2020, but this mechanism is less suitable for Sb). Lastly, the relative lack of reactive iron under highly reducing conditions may have limited the efficiency of As and Sb transfer from the water column to the sediment. The important point is that the Cariaco Basin is one of the rare situations where Sb is slightly more enriched than As. This is especially the case for samples showing relatively lower Mo enrichments, that is, regarding Cariaco, the samples for which the Mn/Fe shuttling was the highest.

This influence of the shuttle effect is even more visible in the case of the Pigmy Basin (Gulf of Mexico). Figure 2 shows the Pigmy samples in a diagram opposing the respective enrichments in U and Mo (Algeo and Tribovillard, 2009). The samples plotting in the area typical of the shuttle effect are also those with the highest Sb enrichment compared to As (Fig. 1). Thus, though our dataset regarding settings with a shuttle effect and available As & Sb data is quite limited, our preliminary results strongly suggest that a higher Sb enrichment compared to As could be the mark of the Fe/Mn shuttling.

Qin et al. (2019) report that As and Sb are commonly captured by Fe and Mn (oxyhydr-)oxides but Sb is more reactive than As to the capture by Mn species. In addition, as reported by He and Hering (2009) and He et al. (2019), previous studies have reported the potential of manganese oxides (sensu lato, that, is, hydroxides and oxyhydroxides) in immobilizing the Sb present in water or sediments (Wang et al., 2012; Basu et al., 2014). Moreover, the capacity of Mn oxides for oxidizing and trapping Sb species has been demonstrated to be much higher than that of Fe oxyhydroxides (Thanabalasingam and Pickering, 1990; Belzile et al., 2001; Wang et al., 2012; He et al., 2019). Lastly, As would be less sensitive to such a Mn-mediated capture, as indicated by the light enrichment in As observed on Mn species in Mediterranean sapropels (Robertson et al., 2019; see He and Hering (2009) for a discussion about As solubilization/immobilization in presence of variable proportions of soluble Mn and Fe). Such a Mn-related transfer to the sediment may be hard to decipher after diagenesis because Mn is easily remobilized in sediments undergoing reducing conditions and released back to the water column. Iron may also be remobilized under reducing conditions but it is usually trapped within sediments in the form of insoluble iron sulfide or pyrite. Manganese is not so easily trapped: Mn carbonates and sulfides (rhodocrosite and alabandite, respectively) require specific conditions to be precipitated authigenically (Calvert and Pedersen, 1993; Tribovillard et al., 2006). Therefore, Mn-mediated transfer of Sb and As may be carried out with no discernible Mn enrichment being recorded, but this shuttling could account for Sb being more enriched than As in some occurrences. If confirmed by further studies, this contrasted behavior of As and Sb could be a clue to detect past transfer of metalloids from seawater to sediment through Mn oxides (hydroxides and oxyhydroxides).

Enrichments in As and Sb are occasionally reported for sediments that underwent the influence of hydrothermal-fluid circulation or exhalation (plumes; Zeng et al., 2018 and references therein); however, associations of these two elements have been seldom mentioned for sediments impacted by cold (hydrocarbon) seepage (Tribovillard et al., 2013; Hu et al., 2014; Chen et al., 2016; Wang et al., 2018, 2019a, 2019b; Zwicker et al., 2018). Regarding cold seeps, most often, analyses are performed on associated authigenic minerals such as carbonates and/or sulfides, but only rarely on bulk sediments as is the case in the present study. Such authigenic substances may yield enrichments in Ba and Sr, as well as Ni, Co, Cu, Mo and W (Meyer-Dombard et al., 2012; Wang et al., 2019a) but these enrichments are not systematically observed (Liang et al., 2017). Consequently, identifying the chemical signature of cold seepage is not straightforward for sediments that do not show obvious structures such as authigenic carbonate/sulfide chimneys, concretions or nodules. An additional pitfall is that authigenic carbonates may recrystallize during earliest diagenesis, which expels Mg and Sr and other trace elements from carbonates (Hatem et al., 2014, 2016; Smrzka et al., 2017) thus blurring the original imprint of cold seep fluids. Most often, unobtrusive influences of cold seepage may be deciphered using C, O and S stable isotope composition. Here, our observations show that the As-Sb covariations may be used as a diagnostic tool allowing for typifying cold seep signatures. Considering here the geological formations that undoubtedly underwent cold seepage influences, namely, the Bancs Jumeaux Fm. and the pseudo-bioherms of Beauvoisin, we observe no enrichment in Ba or Sr but the samples show enrichments in both As and Sb, with As-enrichment factors being larger than Sb-enrichment factors. The same is true for the cold seep-associated samples examined by Wang et al. (2019a). Lastly the same is also the case for samples of the VM1 section of the Vaca Muerta Fm., for which cold seepage influences have been suspected (Krim et al., 2019). The geological objects concerned by hydrothermalism discussed here are in fact all linked to cold seeps. The limestone levels of the Boulonnais Tithonian Bancs Jumeaux Fm. and the Beauvoisin Oxfordian bioherms are linked to fluid circulations along synsedimentary faults. These fluids were rich in dissolved organic carbon (probably methane) and were at the origin of the development of particular faunas: oysters of small sizes (cm) almost exclusive in the Boulonnais, when faunas were associated there with cold seeps; Hatem et al. (2016), and dominant lucinids in the case of Beauvoisin bioherms. In the work of Wang et al. (2019a) on the Pliocene of the Chiahsien region (SW Taiwan), the objects of study were calcareous fluid conduits accompanied by lucinids. Finally, in the case of the VM1 section of the vaca Muerta formation, no evidence of fluid circulation was observed in the field (Krim et al., 2017, 2019).

The enrichment in As of seepage-impacted sediments has already been reported and discussed at length by Tribovillard et al. (2013), Hu et al. (2014), Chen et al. (2016), Zwicker et al. (2018), Wang et al. (2018, 2019b). The debated point is to identify the driving force causing As enrichment: As-rich ascending fluids or reactive iron being released at seep sites and inducing a local shuttling, inducing in turn the capture and transfer of As? Here we complement the picture with Sb data and we observe that As is more enriched than Sb in the sediments of such environments (Fig. 3). From another standpoint, Bardelli et al. (2011) observed that bacterially-mediated carbonates can incorporate significant amounts of As (see also Smrzka et al., 2019, 2020). In (past) cold seep settings, most of the authigenic carbonates result from bacterial processes, which could account for As being enriched more than Sb in the seep-related sediments examined here.

However, Table 3 and Figure 1S show that some depositional environments yield As enrichment relative to Sb but no discernible presence of past seep fluids (Weddell Sea, Vaca Muerta Covunco section, Pigmy Basin pro parte). We cannot be sure that no seepage occurred for these deposits but no evidences have ever been observed. Consequently, we cannot state unambiguously that seepage-impacted sediments always show As and Sb enrichments with As being more enriched than Sb. Nevertheless, our results suggest that a shuttle-mediated enrichment would favor Sb over As (especially if the shuttling is mediated by Mn and not Fe alone), whereas the opposite is observed here for the sites of ascertained seepage influences. Therefore, it is suggested that shuttling was not the driving force accounted for As and Sb enrichments at Beauvoisin (pseudo-bioherms) or in the Boulonnais (Bancs Jumeaux Fm.).

A recent synthesis illustrated that As cannot be looked at as a simple-to-use redox proxy (Tribovillard, 2020). The present work further shows that the combination of As and Sb cannot reliably help to reconstruct paleoredox situations, although these metalloids are redox-sensitive elements. However, our results indicate that As and Sb enrichments may be used to discuss two paleoenvironmental situations in the sedimentary record: cold seep-impacted sediments and settings prone to the particulate shuttle process. For these two types of situations, an enrichment is As is observed; however, disentangling the very cause of metal(loid) enrichment in the case of cold seepage is not easy: direct enrichment by ascending fluids or shuttling induced by the seep panache (Tribovillard, 2020 and discussion therein)? To be further assessed, the preliminary results presented here need to be tested in a larger number of cases. Nevertheless, our work suggests that coeval, marked enrichments in As and Sb with Sb being (slightly) more enriched than As, could be a signature of Fe/Mn shuttling. If true, on the basis of the situations studied here, the metalloid enrichments observed at some seep sites would result from direct enrichment by seeping fluids rather than from seepage-released iron and/or manganese, shuttling around seep sites.

Thanks a lot to all those who shared their data with me. I am grateful to Anne Murat (CNAM-Intechmer at Cherbourg) and an anonymous reviewer for their helpful suggestions that opened perspectives. Thanks to the editors of the journal, Cécile Robin (Université Rennes 1) and Laurent Jolivet (Sorbonne Université, Paris).