The desired effect of EW and OAE is to increase the alkalinity in aqueous media so that additional (ideally atmospheric) CO₂ can be absorbed (Figure 2; Box 1). Idealized modeling studies suggest that EW/OAE could increase surface seawater alkalinity by about 100 to >2,000 μmol kg⁻¹ by the year 2100 although the upper estimates are based on extreme, likely unrealistic, application scales (Köhler et al., 2010, 2013; Ilyina et al., 2013; Paquay and Zeebe, 2013; Keller et al., 2014; González and Ilyina, 2016; Taylor et al., 2016; Feng et al., 2017; Lenton et al., 2018). Alkalinity itself does not affect biology directly because it is a chemical concept, not an ion or molecule that could be sensed by organisms. Nevertheless, the increase in alkalinity substantially modifies pH and the concentrations of various ions and molecules of e.g., the carbonate system which can directly affect biological processes.

In EW applications, new alkalinity will enter the oceans mostly via rivers or streams. These watercourses are usually oversaturated with CO₂ relative to the atmosphere and release globally about 1.8 Gt C year⁻¹ mostly from remineralization of terrestrial biomass (Raymond et al., 2013). Hence, the additional alkalinity from EW would likely have exploited its CO₂ uptake capacity before entering the oceans. In the case of OAE, alkalinity is generated in seawater. The perturbation of seawater carbonate chemistry depends on how much alkalinity is added per volume and/or how quickly this volume mixes with surrounding waters. It also critically depends on how fast the perturbed seawater equilibrates with the atmosphere. For example, alkalinity addition would consume CO₂ from the surrounding seawater without immediate replenishment from the atmosphere when mineral dissolution occurs in a water body with slow air-sea gas exchange. In such a scenario carbonate chemistry perturbations are more severe than for fast air-sea gas exchange because high pH (i.e., low H⁺) excursions are not immediately mitigated by in-gassing CO₂ (Figure 3). To generate negative emissions with OAE, it must be guaranteed that the generated CO₂ deficit in seawater is compensated with atmospheric CO₂. Accordingly, carbonate chemistry perturbations will ultimately have to conform with the “air-equilibrated” scenario shown in Figure 3. Nevertheless, reaching this equilibrium may take several years depending on the oceanographic setting (Harvey, 2008; Feng et al., 2017). Thus, more extreme carbonate chemistry perturbations due to “non-equilibrated OAE” at perturbation hotspots may transiently occur (Figure 3) and these hotspots will also have a higher potential to affect marine biota.

The strong shift in the carbonate system from CO₂ to HCO3- /CO32- associated with transient non-equilibrated OAE could potentially drive primary production into CO₂ limitation (Riebesell et al., 1993). Phytoplankton, for example, can acclimate to CO₂ limitation with carbon concentrating mechanisms (CCMs) which utilize HCO3- to support C-fixation (Giordano et al., 2005). However, HCO3- uptake cannot fully compensate for limited CO₂ since HCO3- must ultimately be converted to CO₂ for C-fixation and can leak out of the cells before being incorporated in biomass (Rost et al., 2006). The threshold below which phytoplankton growth rates become notably affected by CO₂ is genotype-specific and depends on growth conditions (e.g., light or temperature) but should generally be roughly below ~100 μatm (Riebesell et al., 1993; Goldman, 1999; Hansen, 2002; Bach et al., 2011; Sett et al., 2014). Thus, phytoplankton growth could be slowed to a variable degree along gradients of mineral dissolution hotspots. However, dilution with un-perturbed water masses as well as CO₂ in-gassing should quickly alleviate the problem so that new production supported by the available nutrients may shift in space and time but rather not with respect to its overall magnitude as it seems unlikely that the nutrients remain unutilized over the course of a seasonal cycle. Nevertheless, the pulses of low CO₂/high pH will affect species differentially as some will be more sensitive than others (Hansen, 2002; Pedersen and Hansen, 2003).

Perhaps the most fundamental outcome of more than two decades of ocean acidification research is that carbonate chemistry perturbations affect calcifying organisms disproportionately (Gattuso et al., 2015). Thus, it is meaningful to pay special attention to calcification when considering impacts of increased alkalinity on marine organisms and ecosystems. Ocean acidification (black arrow in Figure 3) is predicted to have adverse effects on calcifiers (Orr et al., 2005; Kroeker et al., 2010), mainly due to the increased proton (H⁺) concentration and decreased CaCO₃ saturation state (Ω) (Jokiel, 2011; Bach, 2015; Cyronak et al., 2015; Waldbusser et al., 2015). EW and OAE have the opposite effect on carbonate chemistry conditions. Here, H⁺ decreases while Ω increases which could significantly improve the conditions for calcification (Figure 3). The Black Sea is a suitable analog for a highly OAE perturbed marginal sea as it has a surface alkalinity of ~3,300 μmol kg⁻¹ already today (Figure 3). The major planktonic processes in the Black Sea are similar to other oceans (Sorokin, 2002) but there is a remarkable particularity. The Black Sea harbors some of the most extensive blooms of calcifying phytoplankton (coccolithophores) and thick layers of calcareous ooze cover the sediments (Hay, 1988; Kopelevich et al., 2014). These blooms are likely promoted by the favorable carbonate chemistry conditions since coccolithophores are largely absent in the Baltic Sea—a lower alkalinity marginal sea (Müller et al., 2016) with in other respects comparable characteristics to the Black Sea (Tyrrell et al., 2008).

We have argued in the previous section that transient shifts in carbonate chemistry conditions should have little impact on the overall productivity of marine ecosystems but could alter the species composition. The latter may be important because even though the perturbation is transient, it could induce knock on effects altering species composition and trophic interactions during the succession following the initial perturbation. Such a restructuring could have more severe biogeochemical consequences than initially anticipated. For example, the phytoplankton spring bloom is typically dominated by large phytoplankton species (e.g., diatoms) but in the case of acute OAE perturbations the large species could be replaced by smaller ones which are morphologically better suited to deal with low CO₂ concentrations (Wolf-Gladrow and Riebesell, 1997; Flynn et al., 2012; Chrachri et al., 2018). Boyd and Newton (1995) compared organic carbon export to the deep ocean of the 1989 and 1990 North Atlantic spring blooms where a large diatom species was dominant in 1989 and a smaller one in 1990. They found that, while primary production was similar in both years, particulate organic carbon export from the surface to 3,100 m depth was substantially lower in 1990 when the smaller species dominated (Boyd and Newton, 1995). Thus, it will be important to reveal if the transient but possibly pronounced carbonate chemistry perturbations associated with OAE have the potential to alter succession patterns and associated biogeochemical processes such as carbon export on larger scales.

EW/OAE will very likely have a much more sustained impact on calcifiers because the carbonate chemistry changes beneficial for calcification (i.e., increased Ω, HCO3-; lower H⁺) will be permanent (Figure 3). An expected promotion of calcification through EW/OAE will occur at various scales. At first, low H⁺/ high Ω conditions will be sensed at the physiological level and facilitated CaCO₃ precipitation could allow the organisms to divert more energy into other metabolic processes and growth (Jokiel, 2011; Bach, 2015; Monteiro et al., 2016) (b). This may give calcifiers an energetic advantage and could increase their competitiveness relative to non-calcifiers. Indeed, a recent experiment with natural plankton communities demonstrated that differences in seawater pH can cause seemingly small differences in the growth rates of the dominant calcifier (in this case the coccolithophore Emiliania huxleyi) which were sufficient to induce order of magnitude differences in maximum population sizes among treatments (Riebesell et al., 2017). This not only had a considerable influence on the composition of the food web but also strongly altered biogeochemical fluxes such as carbon export or emissions of climate relevant trace gases (Bach et al., 2016; Webb et al., 2016; Riebesell et al., 2017). Accordingly, seemingly small effects of EW/OAE on organism physiology may be amplified in natural environments leading to ecological regime shifts toward calcifying species. Specifying the influence of EW/OAE on pelagic calcification will be important considering the biogeochemical relevance of this process as will be highlighted in the following paragraph.

About 1–2.6 Gt C are precipitated as CaCO₃ by marine organisms every year (Lebrato et al., 2010). The vast majority of the CaCO₃ is formed by coccolithophores, foraminifera, and pteropods (Sarmiento and Gruber, 2006; Lebrato et al., 2010). The formation of one mole of CaCO₃ reduces alkalinity by 2 moles (Zeebe and Wolf-Gladrow, 2001) so that calcification counteracts the desired effect of EW and OAE. For example, net calcification in the surface ocean was increased by 10 % (in mol) through OAE, then 20 % more alkalinity would be consumed and the effectiveness of OAE would be strongly reduced. Furthermore, the EW/OAE-induced increase of Ω could decrease CaCO₃ dissolution in sinking aggregates which are CaCO₃ under-saturated microenvironments in the surface ocean (Milliman et al., 1999). As a consequence, more alkalinity bound into sinking biogenic CaCO₃, would be exported to the deep ocean where its capacity to store atmospheric CO₂ is not exploited. These relatively easily quantifiable negative feedbacks are complicated by the influence of pelagic calcification on the efficiency of the biological pump (Armstrong et al., 2002; Francois et al., 2002; Klaas and Archer, 2002). If EW/OAE stimulates calcification, then more CaCO₃ ballast material becomes available to accelerate sinking velocities of particulate organic carbon (Honjo et al., 2008; Bach et al., 2016). The accelerated sinking could amplify the efficiency of the biological carbon pump because organic carbon that is built with CO₂ originating from the surface ocean would on average sink deeper into the ocean and lock the respired CO₂ in the deep ocean for a longer time (Kwon et al., 2009). This amplified deep ocean CO₂ sequestration of photosynthetically fixed carbon through the higher availability of ballast CaCO₃ could allow CO₂ to be taken up more efficiently by the oceans and therefore constitute a positive feedback on the CDR potential of EW/OAE. Thus, the hypothetical proliferation of calcification under EW/OAE applications could have a positive or negative feedback on EW/OAE efficacy, depending on if the “more ballast feedback mechanism” or the “less alkalinity feedback mechanism” become dominant [see (Riebesell et al., 2009) who discussed the influence of these two feedbacks in the context of ocean acidification]. An assessment of the “real” EW/OAE efficacy therefore needs to account for the response of calcifiers to EW/OAE and the biogeochemical feedbacks they drive.

EW and OAE can be realized with different minerals (Hartmann et al., 2013; Renforth and Henderson, 2017). Olivine-rich silicate rocks like dunite are considered as most effective because they occur naturally, are relatively fast dissolving, and do not require energy-intense chemical processing before their dissolution (Schuiling and Krijgsman, 2006; Renforth and Henderson, 2017). The flipside is that the desired generation of alkalinity coincides with the release of other chemicals such as dissolved silicate (DSi). Dissolving one mole of olivine leads to an increase in total alkalinity by 4 moles and in silicic acid (H₄SiO₄) by one mole (Köhler et al., 2013; Box 1).

Estimating how much DSi enters pelagic ecosystems is difficult for EW but possible for OAE. The difficulty for EW comes from the unknown fraction of DSi that is retained in soils, as well as on land and in freshwater ecosystems. This retention will most likely be significant although a considerable amount will undoubtedly be discharged into in coastal environments (Köhler et al., 2010; Dürr et al., 2011). OAE inputs are easier to assess because OAE only makes sense when all, or at least the vast majority, of the mineral added to the oceans dissolves in water layers which are in contact with the atmosphere within a few years after dissolution. Therefore, it can be assumed that most of the Si contained in silicates such as olivine would dissolve and enrich the DSi pool in the surface ocean. Assessments under more realistic constraints (e.g., accounting for mineral grinding and distribution) suggest that the maximum scale of mineral addition through OAE lies somewhere at around 10 Gt year⁻¹ which is roughly 70 Tmol DSi year⁻¹ if dunite was used (Box 2). Thus, OAE would be a source of DSi to the surface ocean that 5–14 fold higher than all natural sources combined (Box 2).

DSi is an essential nutrient for silicifying organisms and by far the most important silicifiers in the ocean are diatoms (Tréguer and De La Rocha, 2013). These are a group of globally distributed phytoplankton that utilize DSi to form a shell (frustule) made of biogenic silica (opal) (Sarthou et al., 2005). Diatoms were estimated to contribute ~25% to primary production on Earth (Nelson et al., 1995; Field et al., 1998). They usually initiate the seasonal phytoplankton succession and form blooms until they become nutrient-limited and other taxa with alternative nutrient acquisition strategies take over (Sommer et al., 2012).

DSi is often the nutrient that constrains diatom proliferation because it is highly deficient relative to other macronutrients throughout most subsurface water masses that supply nutrients to the euphotic zone through upwelling (Sarmiento et al., 2004). In fact it is the very “success” of diatoms throughout the Cenozoic that is thought to have driven the decrease in global ocean Si concentration (Conley et al., 2017). The deficiency in modern subsurface waters likely has two reasons. First, the opaline frustule dissolves relatively slowly so that DSi is remineralized on average at greater depths than other nutrients (Dugdale et al., 1995). Second, much of the subsurface water is formed by the subduction of Southern Ocean surface water (Sloyan and Rintoul, 2001). Before this water mass is subducted and spreads northwards, diatoms strip out and export DSi more than other nutrients thereby trapping DSi in the Southern Ocean (Sarmiento et al., 2004; Primeau et al., 2013; Holzer et al., 2014). These mechanisms explain why global diatom productivity in the present oceans is lower than it could be without prevailing shortage of DSi (Ragueneau et al., 2000; Sarmiento et al., 2004).

Applications of silicate-based EW and OAE could relieve the present day DSi deficiency (Köhler et al., 2013). The DSi concentrations where diatom DSi uptake rates reach half of the theoretical maximum (K_1/2) are species specific and range between 0.2 and 22 μmol kg⁻¹ (Sarthou et al., 2005). High DSi affinity (i.e., low K_1/2) typically coincides with lower maximum uptake rates (V_max). Accordingly, increased DSi in the surface ocean would favor diatoms with high V_max and K_1/2 and therefore presumably fast-growing species (Sarthou et al., 2005). Furthermore, it could allow diatoms to reduce energy investments for DSi acquisition and enable them to construct thicker frustules (Martin-Jézéquel et al., 2000; Sarthou et al., 2005). Energy savings could accelerate diatom growth rates because available energy can be diverted into the acquisition of other nutrients while reinforced frustules would reduce grazing mortality (Hamm et al., 2003; Wilken et al., 2011; Friedrichs et al., 2013; Liu et al., 2016). Indeed, diatoms typically out-compete other phytoplankton when DSi is plentiful and other essential resources are sufficiently available (Armbrust, 2009). Thus, under silicate-based EW/OAE, diatoms could consume a larger fraction of the other limiting nutrients like N and P and ultimately become more dominant than they are already today. However, it will be important to assess if all diatoms profit equally from DSi enrichment due to the large range of DSi affinities among species (Sarthou et al., 2005). In theory, more heavily silicified diatom species should profit over proportionally from silicate-based EW/OAE as these have inherently higher DSi requirements to satisfy. Thus, it will be important to unravel if silicate-based EW/OAE has the potential to shift diatom communities toward more heavily silicified species.

A particularly interesting aspect in the context of community shifts is the competition between calcifiers and silicifiers. We hypothesized above that calcifiers profit from alkalinity enhancement and would become more dominant under EW/OAE. However, a prerequisite for the success of calcifiers would be that growth of the usually more competitive silicifiers is kept in check by DSi limitation (Egge and Jacobsen, 1997; Tyrrell and Merico, 2004). Under silicate-based EW/OAE, this limitation is relieved so that the fertilization of calcifiers through alkalinity enhancement could be overcompensated by DSi fertilization on silicifiers. Humborg et al. (1997) provide an excellent case study where this effect may actually have been observed. The river Danube contributes ~70 % of the river inputs into the Black Sea but was dammed in the early 1970s. The damming reduced DSi inputs into the Black Sea and promoted coccolithophores at the expense of diatoms (Humborg et al., 1997). The reverse case (i.e., increase of DSi) would be true for silicate-based EW/OAE. Here, calcifiers may not be able to benefit from more favorable carbonate chemistry because silicifiers benefit at the same time. Addressing this is very important because the balance between calcification and silicification has tremendous influence on the efficiency of EW/OAE. For example, if DSi inputs reduce net calcification and the subsequent export of CaCO₃ into the deep ocean by 10 % (in mol), then 20% more alkalinity remains in the surface ocean to absorb atmospheric CO₂. Such a reduction of marine alkalinity sink would enhance the CO₂ uptake capacity of the oceans and therefore increase the effectiveness of EW/OAE. Indeed, paleo-oceanographic studies suggest that increased DSi leakage from the Southern Ocean during glacial periods may have led to a shift from calcification to silicification globally (Brzezinski et al., 2002). This could have reduced the marine alkalinity sink and enhanced the oceanic CO₂ uptake capacity to such an extent that it can explain parts of the glacial-interglacial difference in atmospheric pCO₂ (Matsumoto et al., 2002). It may therefore be an argument to use silicate minerals for EW/OAE as this may reduce the biotic alkalinity sink and thereby indirectly increase the EW/OAE efficacy.

A shift from calcification to silicification under silicate-based EW/OAE would also alter quality and quantity of ballast material in water column. CaCO₃ has a higher density (2.7 g cm⁻³) than BSi (2.1 g cm⁻³) and is therefore considered to be more effectively accelerating the sinking of organic matter into the deep ocean (Klaas and Archer, 2002). However, CaCO₃ ballast is currently considered to be only regionally important in the oceans whereas the influence of BSi ballast is thought to be more widespread (Wilson et al., 2012; Le Moigne et al., 2014; Tréguer et al., 2018). Accordingly, a further increase in diatom biomass and potential shifts toward more heavily silicified diatoms at elevated DSi inputs (see above) may amplify the ballast potential of BSi and therefore carbon export in silicate-based EW/OAE schemes (Köhler et al., 2013). Thus, we speculate that the proliferation of silicifiers, even if this comes at the expanse of less pelagic calcifiers, could lead to a net increase of CO₂ sequestration into the deep ocean and therefore constitute a positive feedback on EW/OAE efficacy. If true, this could be another argument for silicate minerals.

It is expected that the most likely minerals to be used for EW/OAE are Ca or Mg oxides (e.g., CaO) and/or Ca or Mg silicates (e.g., Mg₂SiO₄). Ca and Mg are important ions for all organisms and in particular for calcifiers: Ca as a main ingredient and Mg as an inhibitor of precipitation which is excluded from the mineralisation space (Davis et al., 2000). Due to the high background concentrations of these alkali Earth metals in seawater, the relative change in their concentrations through addition to the surface ocean will be small (Box 2). However, it is likely that there will be considerable local perturbations through EW/OAE, and therefore Ca and Mg, at the input location for these schemes, on a regional scale. Biotic impacts through Ca or Mg may be pronounced at these perturbation hotspots before mixing through ocean currents disperses these chemicals to lower concentrations.

High intracellular calcium is incompatible with life. At high intracellular concentrations, Ca²⁺ causes aggregation of proteins and nucleic acids, affects the integrity of lipid membranes, and can precipitate with cytosolic phosphate removing a major nutrient from bioavailability within the cell (Case et al., 2007). All forms of life, from their origin, need an effective system for Ca homeostasis, which keeps intracellular Ca²⁺ at concentrations low enough to prevent apatite precipitation, at levels of ~0.1 μM, and ~10,000–20,000 times lower than that in seawater (Carafoli, 1987). Although maintenance of such a large Ca²⁺ concentration gradient can be used effectively for Ca signaling as a core component of cellular regulation, it comes at a metabolic cost. Thus, any perturbation of the environmental Ca²⁺ concentration could impose an additional metabolic cost to all forms of life from bacteria to the highest eukaryotes, meaning that marine food webs could be affected by high Ca²⁺ at perturbation hotspots.

Magnesium is the second most abundant cellular cation after potassium. In comparison to Ca, high intracellular concentrations of total and free magnesium ion (Mg²⁺) appear to be essential to regulate numerous cellular functions and enzymes, including ion channels, metabolic cycles, and signaling pathways up with intracellular concentrations ranging from 17-20 mmol kg⁻¹, just 2.5 times lower than ambient seawater (Romani, 2011). Understanding how cells regulate Mg²⁺ homeostasis remains incomplete but again must occur at some, but lesser than Ca, energetic cost to the cell (Romani, 2011).

From a biomineralization perspective, it is possible that calcification will benefit from additional Ca²⁺ in a quicklime OAE scheme (Stanley et al., 2005). The influence of Mg in an olivine scheme is more complicated. Mg is known to be a strong inhibitor of inorganic calcite precipitation through incorporation into the CaCO₃ lattice (Davis et al., 2000). Mg incorporation raises the dissolution rate of the advancing crystal edge, which subsequently increases the mineral solubility, resulting in corresponding reduced net calcification. Aragonite is not inhibited by Mg incorporation into the CaCO₃ lattice in the same way that calcite is. Incorporation of Mg into aragonite's crystal structure has been shown experimentally to have no effect on growth rate of aragonite crystals (Berner, 1975; Gutjahr et al., 1996). Changes in the concentration of Mg in seawater relative to Ca do appear to be reflected in the plankton fossil record. Here, high Mg/Ca ratios through low absolute Ca concentrations, as in the modern ocean, promoted aragonite, and high Mg-calcite mineralizers. Conversely, low Mg-calcite mineralizers such as the coccolithophores, were more prevalent during the Mesozoic, which was characterized by oceans with low Mg/Ca and high absolute Ca²⁺ concentrations (Stanley and Hardie, 1998; Ries, 2010). Therefore, studies in palaeoceanography may be useful to elucidate future changes we could expect from EW/OAE scenarios by acting as suitable analogs.

There is also the possibility that higher Ca²⁺ and Mg²⁺ concentrations may interfere with the uptake of nutrients by phytoplankton (and/or bacteria). For example, Ca and Mg have been shown to influence Fe uptake in the freshwater cyanobacteria Microcystis aeruginosa (Fujii et al., 2015). It is therefore possible that Ca and Mg have similar effects on marine phytoplankton which may have implications from primary production at EW/OAE perturbation hotspots.

Given that foraminifera mineralize first by the engulfing of seawater by a membrane, the low Mg/Ca of their calcite compared to inorganic predictions suggests an active role for pumping of Mg away from the region of mineralization (Segev and Erez, 2006). Two species of benthic foraminifera incorporate more Mg into shells as Mg/Ca increases, suggestive that control on the Mg content of the mineralizing space is not strictly regulated. The greatest increase in precipitation of CaCO₃ was obtained at Mg/Ca of 1, and not at present-day seawater ratio (Mg/Ca = 5). Thus, it is possible that any decrease (increase) in the Mg/Ca ratio of the ocean will both increase (decrease) the degree of calcification of foraminifera, in addition to making the carbonate less (more) soluble by lowering its Mg/Ca ratio.

Pteropods are holoplanktonic gastropod molluscs which are the major producers of aragonite in the surface ocean today (Andersson et al., 2008). The response of pteropods to changing Mg/Ca is not well-constrained due to the difficulty of keeping pteropods alive in culture (Howes et al., 2014). The origination of the pteropods and rise in prevalence during high Mg/Ca in the Cenozoic may suggest that pteropods are favored by the emerging aragonite sea conditions (Burridge et al., 2017). Accordingly, they could be influenced oppositely to coccolithophores and foraminifera by Mg/Ca alterations associated with EW/OAE.

As a first step, it will be essential to constrain the impact of elevated Ca and/or Mg on the order of magnitude predicted in Table 1. To our knowledge, no work has specifically been completed on the response of plankton other than coccolithophores or foraminifera to increased Mg, Ca or altered Mg/Ca. This work may also be important to complete for the major group of silicifiers, the diatoms, due to the competition between silicifiers and calcifiers mentioned before. If CaO were added to the ocean, thereby decreasing Mg/Ca, it is reasonable to hypothesize that coccolithophores as competitors of diatoms would be further strengthened and take a larger share of the available nutrient pool. There is huge opportunity for the utilization of palaeoceanographic studies to help with these predictions, due to the rich fossil and sediment record providing evidence of secular changes in Mg/Ca ratios over geological time showing that silicifiers have risen to open ocean dominance in high Mg/Ca oceans (Falkowski et al., 2004).

All minerals suitable for EW/OAE will contain elements considered as “trace metals” (Hartmann et al., 2013; Schlitzer et al., 2018). Trace metals occur at relatively low concentrations in seawater so that perturbations at the scale anticipated for EW/OAE are likely sufficient to have a fertilizing and/or toxic effect on organisms (Hartmann et al., 2013; Moore et al., 2013; Hauck et al., 2016). The diversity of trace metals added to the oceans via mineral dissolution would be large and highly dependent on the source mineral used for EW/OAE. For example, alkaline industrial residuals such as steel slags are considered to be useful for EW/OAE but they can contain trace metals such as Cr, Mo, Ni, Pb which could negatively influence the environment above critical threshold concentrations (Mayes et al., 2008; Renforth, 2019). Likewise, naturally occurring minerals like olivine or source CaCO₃ for the production of CaO are associated with various elements, depending on their source region. It is beyond the scope of this paper to discuss the potential impacts of all trace elements contained in source material appropriately. We therefore focus on two particularly important ones—iron (Fe) and nickel (Ni). Fe is in focus because it is one of the key nutrients influencing pelagic ecosystems in the surface ocean (Moore et al., 2013; Tagliabue et al., 2017) and is abundant in minerals considered for EW/OAE (Hartmann et al., 2013). Ni is currently not considered as a key nutrient in the surface ocean (Moore et al., 2013) but its abundance in dunite, one of the major candidate rocks for EW/OAE, is very high so that highlighting this element is very reasonable (Meysman and Montserrat, 2017; Beerling et al., 2018).

Fe is rather immobile in soils (Gaillardet et al., 2014) so that Fe enrichment of rivers through EW should be relatively small. However, differences exist among rivers with those that contain large amounts of dissolved organic ligands (“blackwater rivers”) likely transporting more dissolved Fe to the oceans (Krachler et al., 2015; Vieira et al., submitted) which is an important aspect to consider when selecting EW application sites on land. Once riverine Fe reaches the ocean, much of it will be rapidly scavenged, for example by salt-induced flocculation or by settling particles so that Fe perturbations through EW would be restricted mostly to coastal regions (Boyd and Ellwood, 2010) even though physical conditions at some estuaries (e.g., Congo) could transport material further offshore (Vieira et al., submitted). Ni is not as abundant as Fe in olivine but is more mobile in soils and more soluble in seawater (Gaillardet et al., 2014; Montserrat et al., 2017) so that Ni inputs through EW could be higher in coastal systems and potentially further off-shore. OAE short-circuits the terrestrial filter so that the trace metal contamination of the ocean realm would be significantly larger and directly depend on the dissolution rates of mineral powder and the solubilities of the dissolution products (Montserrat et al., 2017).

Fe is an essential micronutrient needed as a co-factor in many enzymes (Tagliabue et al., 2017). The concentration of dissolved Fe (DFe) is mostly <1 nmol L⁻¹ in the surface ocean where productivity is in large parts Fe-limited (Moore et al., 2013; Schlitzer et al., 2018). Therefore, even a relatively small Fe addition of a few tons to Fe-limited surface ocean regions can induce a massive phytoplankton bloom visible from space (Boyd et al., 2007). The ideal olivine mineral for EW/OAE would be pure forsterite (Mg₂SiO₄; Box 1) but in nature such purity rarely exists. Here, olivines typically occur as a mixture of forsterite and fayalite (Fe₂SiO₄) (Taylor et al., 2016). A commonly found forsterite/fayalite mixing ratio is ~9:1 and this rock type, termed “dunite,” exists in sufficiently large quantities to sequester maybe some hundreds of Gt CO₂ from the atmosphere (Taylor et al., 2016). Fe enrichment scenarios tested with an Earth System model suggested that this iron fertilization of ocean primary production would increase OAE efficacy roughly by one third (Hauck et al., 2016). However, more recent thermodynamic modeling indicated that fayalite dissolution reduces alkalinity through secondary reactions (Griffioen, 2017), although this could not be confirmed in a dissolution experiment (Montserrat et al., 2017). Thus, the overall influence of Fe perturbations on the CDR potential of EW/OAE is very difficult to determine with our current level of understanding. Fe enrichment could, to a smaller extent, also occur under applications of artificially processed minerals. Quicklime for example is made by calcining natural CaCO₃ rich source rocks with purities of <98 % typically achieved at industrial scale (Renforth and Kruger, 2013; Renforth et al., 2013). The majority of the remaining mass is composed of silicate and iron oxides. Therefore, Fe or Si addition when using calcined limestone will certainly be lower than from silicates such as dunite (Box 2).

Iron is often not the most limiting nutrient for phytoplankton in coastal environments as these usually receive DFe supply from sediment fluxes or terrestrial sources (Boyd and Ellwood, 2010). Increased DFe river discharge through EW may therefore have a limited effect on primary production near coasts. In the case of OAE the situation could be different. Counterintuitively, the addition of mineral particles could scavenge bioavailable DFe thereby reducing primary production. This may apply for mineral powder additions in DFe replete regions where DFe scavenging (adsorption) by volcanic ash particles has been shown to outweigh the Fe release through dissolution thereby leading to a net removal of DFe (Rogan et al., 2016). However, when particles are added to a DFe deficient system the dissolution term outweighs the scavenging term leading to a net increase in DFe (Rogan et al., 2016). Thus, OAE applied in DFe-limited areas should enhance primary production unless the ecosystem is acutely co-limited by other nutrients such as nitrate. Indeed, DFe and N co-limitation is widespread, especially on the edges of sub-tropical gyres (Browning et al., 2017). OAE would strongly relieve DFe but not N limitation thereby favoring N₂ fixing cyanobacteria (e.g., Trichodesmium) which are largely controlled by DFe availability in oligotrophic systems (Mills et al., 2004). It will therefore be essential to test if OAE could stimulate N₂ fixation rates of cyanobacteria in these regions that may ultimately increase primary production and the CDR efficacy of OAE.

Altogether, the DFe related effects on primary production would most likely coincide with pronounced shifts in the composition of the plankton community. For example, a specific stimulation of cyanobacteria may initiate a reorganization of the energy flow along trophic pathways, ultimately affecting the marine food web in its entirety. Such indirect effects of EW/OAE may even be stronger than the first order chemical perturbation and therefore need to be taken into account (Strauss, 1991; Walsh, 2013). Ultimately, all the positive and negative side-effects that have been anticipated for ocean fertilization with iron (e.g., increased ocean productivity, deep-ocean hypoxia; Oschlies et al., 2010) would also have to be considered in the context of EW/OAE when minerals with considerable Fe content are used.

Ni is rare in the Earth's crust but over-proportionally abundant in forsterite-rich olivine (Simkin and Smith, 1970). Ni is more mobile in soils than Fe especially at low pH and dissolves relatively quickly so that it is currently considered as the most important contaminant under silicate-based EW/OAE (Nieminen et al., 2007; Hartmann et al., 2013; Meysman and Montserrat, 2017; Montserrat et al., 2017; Beerling et al., 2018). It is an essential co-factor in variety of enzymes but Ni nutrient requirements and sensitivities to excess concentrations vary greatly among organisms (Figure 4). Eukaryotic microalgae have lower Ni requirements (mostly <2 nmol kg⁻¹) and are more sensitive to high concentrations than prokaryotes such as cyanobacteria (Figure 4). Ni has a higher solubility than Fe and occurs in the surface ocean mostly as Ni²⁺ in concentrations of about 2–12 nmol kg⁻¹ so that bulk primary production is usually not limited by it (Konhauser et al., 2009; Gall et al., 2013; Moore et al., 2013). This also explains the general absence of a nutrient-like vertical concentration profile of Ni with the exception of the North Pacific (Glass and Dupont, 2017) where occasional Ni limitation of cyanobacteria communities have been observed (Dupont et al., 2010).

The influence of dissolved Ni (DNi) on pelagic systems has not been a particularly strong focus in Biological Oceanography so far but it may become important in the context of EW/OAE. Here, DNi could either function as nutrient or toxin with pronounced impacts on primary production and plankton community composition.

Recent dissolution experiments with olivine have shown that an increase of alkalinity of about 100 μmol L⁻¹ coincided with an increase of ~3 μmol L⁻¹ in dissolved Ni (DNi) due to non-stoichiometric dissolution (Montserrat et al., 2017). These DNi concentrations are ~3 orders of magnitude higher than ambient concentrations and would likely be harmful for many organisms (Figure 4). Thus, olivine-based EW/OAE could strongly affect marine plankton communities near perturbation hotspots. As for Fe, impacts of DNi will likely differ regionally and depend on the constitution and evolutionary history of plankton communities. For example, communities near estuaries may be adapted to higher DNi concentrations, especially when discharging rivers come from basaltic, olivine rich catchment areas (Nieminen et al., 2007). However, this should be the exception rather than the rule so that we would expect rather detrimental effects on biodiversity and productivity when too much DNi is released locally.

DNi is generally not considered to be an element that is bio-accumulated strongly by organisms (bio-accumulation = enrichment within organism tissues) (Nieminen et al., 2007). Without bio-accumulation, there is also little potential for Ni bio-magnification along the food chain (i.e., Ni enrichment with increasing trophic level). However, as mentioned earlier, Ni is likely one of the most important trace metal contaminants under olivine-based EW/OAE but certainly not the only one. Other potential trace metals (e.g., Cr, Cu, or Cd) (Simkin and Smith, 1970; Beerling et al., 2018) may become bio-accumulated and bio-magnified, particularly in phytoplankton evolutionarily associated to the “red lineage” (Wilson et al., 2019). Furthermore, a suite of anthropogenic materials (e.g., slag from the production of steel, or mining residuals) could be used for EW (Renforth, 2019). These can contain a range of elemental compositions, and a material by material assessment is required for wider-scale impact. Thus, it may be possible that certain trace metals introduced into the environment via EW/OAE could bio-accumulate and bio-magnify which may cause health problems, particularly for end-members in the food chain like humans (Järup, 2003). It is therefore important to identify minerals associated with such risks and not use them for EW/OAE. Indeed, trace metal toxicity may become a key decisive factor for the acceptability of EW/OAE applications and strongly influence the choice of rocks that could ultimately be used.

The toxicity-related hypothesis outlined above refers to EW/OAE scenarios where very high DNi concentrations occur. However, after dilution of the perturbation signal in the oceans through space and time DNi may dilute from harmful to fertilizing concentrations (Figure 4). Ni is an essential co-factor for enzymes involved in urea cycling and utilization (Nieminen et al., 2007; Glass and Dupont, 2017) and has been shown to limit growth of cyanobacteria- and picoeukaryota-dominated phytoplankton communities (Dupont et al., 2010). Therefore, EW/OAE-related DNi inputs could stimulate primary production in certain oligotrophic settings. Indeed, the growth rate of the predominant N₂-fixer Trichodesmium has been shown to profit strongly from DNi amendments (Ho, 2013) such that ecological regime shifts toward more cyanobacteria in subtropical gyres would be a plausible consequence of olivine-based EW/OAE. If true, moderate increases of DNi may under certain circumstances locally enhance marine productivity and therefore EW/OAE efficacy through a stimulation of N₂-fixation.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.