Authigenic iron oxide proxies for marine zinc over geological time and 1 implications for eukaryotic metallome evolution

27 Here we explore enrichments in paleomarine Zn as recorded by authigenic iron oxides 28 including Precambrian iron formations, ironstones and Phanerozoic hydrothermal 29 exhalites. This compilation of new and literature-based iron formation analyses track 30 dissolved Zn abundances and constrain the magnitude of the marine reservoir over 31 geological time. Overall, the iron formation record is characterized by a fairly static range 32 in Zn/Fe ratios throughout the Precambrian, consistent with the shale record (Scott et al., 33 2013, Nature Geoscience, 6, 125-128). When hypothetical partitioning scenarios are 34 applied to this record, paleomarine Zn concentrations within about an order of magnitude 35 of modern are indicated. We couple this examination with new chemical speciation 36 models used to interpret the iron formation record. We present two scenarios: first, under 37 all but the most sulfidic conditions and with Zn binding organic ligand concentrations 38 similar to modern oceans, the amount of bioavailable Zn remained relatively unchanged 39 through time. Late proliferation of Zn in eukaryotic metallomes has previously been 40 linked to marine Zn biolimitation, but under this scenario, the expansion in eukaryotic Zn 41 metallomes may be better linked to biologically intrinsic evolutionary factors. In this case 42 zinc’s geochemical and biological evolution may be decoupled, and viewed as a function 43 of increasing need for genome regulation and diversification of Zn-binding transcription 44 factors. In the second scenario, we consider Archean organic ligand complexation in such 45 excess that it may render Zn bioavailability low. However, this is dependent on Zn 46 organic ligand complexes not being bioavailable, which remains unclear. In this case, 47 although bioavailability may be low, sphalerite precipitation is prevented, thereby 48 maintaining a constant Zn inventory throughout both ferruginous and euxinic conditions. 49 These results provide new perspectives and constraints on potential couplings between 50 the trajectory of biological and marine geochemical coevolution. 51 52 Introduction 53 Zinc is the most common inorganic co-factor in eukaryotic metalloenzymes (Berg and 54 Shi, 1996; Maret, 2001; Dupont et al., 2010), and as a consequence, it has become the 55 basis for a hypothesis that the biological use of Zn may have evolved in the late 56 Precambrian when it became available in seawater (Williams and da Silva, 1996; Dupont 57 et al., 2006; 2010; Saito et al., 2003). Modern marine phytoplankton differ significantly 58 in their ability to grow at low Zn concentrations; modern surface seawater has 59 concentrations that range from ~0.04-0.5 nM (e.g. Bruland, 1989; Lohan et al., 2002). 60 Studies of marine cyanobacteria have found little to no measurable Zn requirement under 61 the conditions tested thus far in the globally abundant Prochlorococcus and 62 Synechococcus (Sunda and Huntsman 1995; Saito et al., 2002). In contrast, eukaryotic 63 phytoplankton have been observed to be quite sensitive to low zinc conditions. Some, 64 notably neritic diatoms (those that inhabit shallow marine waters from the littoral zone to 65 the edge of the continental shelf), are more sensitive and experience growth rates that are 66 significantly reduced at free Zn concentrations below 10 M, while others show only 67 minor reductions in growth rates at 10 M (Brand et al., 1983). The centric diatom 68 Thalassiosira sp. displays dramatically reduced growth rates in coastal species at Zn 69 concentrations of 10 M (Tortell and Price, 1996); however, species from offshore 70 oligotrophic environments, such as Thalassiosira oceanica, are more tolerant of low Zn 71 conditions (Sunda and Huntsman, 1995). The unicellular algae, Emiliania huxleyi (a 72 coccolithopharad), shows decreased levels of alkaline phosphatase activity as Zn 73 approaches picomolar concentrations (Shaked et al., 2006). Concentrations are low 74 enough in modern environments that Zn stimulation or co-limitations of marine 75 phytoplankton communities have been observed in some high nutrient, low chlorophyll 76 environments (Franck et al., 2003; Jakuba et al., 2012), but not other environments 77 (Crawford et al., 2003; Coale et al., 2005). While physiological studies of phytoplankton 78 Zn requirements are limited and it is apparent that no single dissolved Zn concentration 79 can be pinpointed as universally limiting, it is also apparent that severely suppressed 80 marine Zn concentrations would have significant consequences for the activity and 81 abundance of modern eukaryotic phytoplankton. 82 Such may have been the case in deep geological time. A survey of physiological 83 experiments found that marine prokaryotic microbes, particularly the cyanobacteria, show 84 metal nutritional requirements consistent with hypothesized Precambrian seawater 85 compositions (Saito et al., 2003). From this it was suggested that during much of the 86 Precambrian, the bioavailable marine Zn reservoir, as well as those of Cu and Cd, may 87 have been much lower than in the modern oxygenated ocean due to the formation of 88 strong aqueous complexes between Zn and sulfide that are likely not bioavailable (Luther 89 et al., 1996; Edgcomb et al., 2004). Proteomic-based phylogenetic analyses also indicate 90 a relatively late origin for most Zn-binding domains in eukaryotic metalloenzymes, 91 leading to the suggestion that in addition to depressed oxygen availability, marine Zn 92 biolimitation stemming from higher Precambrian sulfide concentrations and expanded 93 euxinia during the mid-Proterozoic (1800 to 800 Ma) may have impeded eukaryotic 94 diversification (Dupont et al., 2006; 2010). Accordingly, seemingly rapid eukaryotic 95 diversification in the Neoproterozoic (1000 to 542 Ma) may, in part, be tied to an 96 enhanced bioavailable marine Zn reservoir accompanying oxygenation of the oceans. 97 This model provides a simple link between the enigmatic and protracted diversification of 98 eukaryotes and the shifting availability of bio-essential metals in a manner akin to a ‘bio99 inorganic bridge’ (Anbar and Knoll, 2002). 100 However, until recently this possibility has yet to be evaluated in light of 101 sedimentary proxies for the evolution of the paleomarine Zn reservoir. A recent 102 examination of the shale record (Scott et al., 2013) indicates that Zn may have been near 103 modern abundances and was likely bioavailable to eukaryotes throughout the 104 Precambrian, casting doubt on the coupled geochemical and eukaryotic evolutions with 105 respect to Zn utilization. Here, guided by new chemical speciation models, we explore 106 eukaryotic evolution as revealed by a ~3.8 billion year record of marine authigenic iron 107 oxide deposition, in the form of Precambrian iron formations, Phanerozoic ironstones and 108 Fe-rich exhalites, herein collectively referred to as iron formations (IF). We use the rock 109 record to shed light on the poorly understood relationships between marine trace metal 110 availability, metalloenzyme proliferation, and biological innovation. 111 Zn is predominantly bound by organic ligands in modern seawater (e.g., Crawford 112 et al., 2003), but it also occurs in aqueous form, i.e., Zn, Zn(OH), Zn(OH)2, ZnCO3, 113 ZnSO4, and ZnCl2, suspended solids (e.g., ZnS), or adsorbed onto particle (e.g., Zirino 114 and Yamamoto, 1972). Furthermore, Zn may be strongly complexed by aqueous sulfide 115 (Luther et al., 1996) such that in anoxic environments, where HS is present, inorganic 116 bisulfide and potentially polysulfide Zn complexes may play key roles in dominating the 117 speciation of dissolved Zn (e.g., Gardner, 1974). Luther et al. (1999) provide an example 118 of when polysulfides may become dominant, which occurs when 10 μM Zn is titrated 119 with sulfide in excess of 5 μM. In some conditions where a strong redoxcline exists, such 120 as Jellyfish Lake, Palau (Landing et al., 1991), total dissolved Zn concentrations may 121 actually increase at depth due to the formation of aqueous sulfide complexes. The 122 proportion of Zn that is bioavailable is controlled by either sulfide complexation (Luther 123 et al., 1996; Edgcomb et al., 2004) or by organic ligand complexation. However, 124 complicating the issue of bioavailability is recent evidence that suggests organic 125 complexation of Zn may in fact increase the potential for uptake (Aristilde et al., 2012). 126 In surface layers of the open ocean, horizontal and vertical mixing, atmospheric 127 fallout, biological uptake, and particulate removal are the main controls on Zn abundance 128 (Bruland, 1980). Accordingly, total dissolved Zn follows a nutrient profile in the modern 129 oceans, where Zn is between ~0.04-0.5 nM in the surface layers, increasing below the 130 photic zone to ~8-10 nM, where it remains relatively constant down to the seafloor 131 (Bruland et al., 1994; Lohan et al., 2002; Nolting and de Baar, 1994). Despite this 132 variability, for the purposes of discussion we assume a concentration of 10 nM as a 133 “modern” value effectively describing the majority of the water column. While uptake in 134 the photic zone could have lead to surface-deplete nutrient-type Zn depth profiles in the 135 deep past, especially if concentrations were limiting, the Zn concentration of deep waters 136 obviously plays an important role in the upward diffusive resupply of Zn (John, 2007) 137 and thus upper water column Zn concentrations. 138 In terms of the modern zinc budget, inputs from mid-ocean ridge hydrothermal 139 systems (~4.4x10 mol/yr) dominate riverine fluxes (~3.4x10 mol/yr) by a factor of 13, 140 while diffuse off-axis venting contributes little marine Zn (~1x10 mol/yr)(Wheat et al., 141 2002). However, the efficiency of scavenging of hydrothermal zinc is poorly constrained. 142 Modern aeolian Zn deposition is significant (~0.7 – 3.5x10 mol/yr), although roughly 143 75% is anthropogenic (Duce et al., 1991). Modern sinks are poorly constrained but likely 144 include organic matter, metal hydroxides, and sulfide burial fluxes. Modern hydrothermal 145 fluids, the primary natural Zn input, are enriched by 16,000 88,000 times the seawater 146 Zn value at their source (Doe, 1994), but these values drop significantly with distance 147 from the vent due to seawater dilution and Zn incorporation into sulfide and metal 148 hydroxide phases (Trocine and Trefry, 1988; German et al., 1991). Given higher mantle 149 recycling rates (e.g., Sleep and Windley, 1982), we consider a hydrothermal Zn 150 component to be more relevant during the Precambrian. Under anoxic and ferruginous 151 (Fe-rich) seawater conditions, with Fe(II)>S(-II) as required for iron(III) oxyhydroxide 152 formation, it is likely that hydrothermal Zn would have dispersed over wider areas of the 153 deep ocean for a lack of an effective sink, with a spatial distribution and areal extent 154 similar to Fe in the case of Precambrian iron formations (e.g., 10 km in the Hamersley 155 basin; Morris, 1993). 156 Based on the low solubility of Zn sulfide minerals and the formation of strong 157 aqueous Zn-S complexes, expanded euxinia during the late Paleoproterozoic and 158 Mesoproterozoic has been proposed to have limited the bioavailability of Zn and other 159 sulfide-reactive trace metals (e.g., Cu, Cd), and thereby influenced metallome evolution 160 (Williams and Da Silva, 1996; Sunda and Huntsman, 1995; Saito et al., 2003). However, 161 recent work suggests that Proterozoic oceans were almost certainly laterally 162 heterogeneous in their geochemical characteristics (e.g., Planavsky et al., 2011), such that 163 sulfidic conditions may have been limited to shallow or coastal areas (e.g. Poulton et al., 164 2010). Such considerations therefore warrant a re-examination of trace metal evolution in 165 the context of a dominantly ferruginous Proterozoic ocean, especially with regards to 166 elements vital for eukaryotic evolution such as Zn. Precambrian authigenic iron oxides, 167 comprising laterally extensive IF that are highly Fe-rich and S-poor, necessitate Fe-rich 168 and sulfide-poor conditions at the time of their deposition (Klein, 2005; Bekker et al., 169 2010). In this regard, the Precambrian IF record may be considered to represent large 170 areas with conditions where Fe(II)>>S(-II). Such chemical deposits thus record ancient 171 seawater where no strong euxinic metal sink was locally present; this makes the IF record 172 an ideal target for exploring paleomarine concentrations of Zn. 173 174 Methods 175 Geochemical equilibrium calculations (Fig. 1) were performed using Visual MINTEQ 3.0 176 (Gustafsson, 2011) and the primary thermodynamic database provided (thermo.vdb) was 177 modified to account for multiple aqueous Zn sulfide complexes as well as Zn 178 complexation by organic ligands (SI Table 1). Modeling conditions included seawater179 like salinity (0.56 M NaCl), standard temperature (25oC), the exclusion of molecular O2 180 and a pCO2 of 10 times present atmospheric levels (PAL). Calcium, pH and Fe were 181 determined by equilibrium reactions with excess calcite and siderite. Redox 182 considerations were omitted such that all Fe and S species are +II and –II, respectively. 183 Supersaturated minerals were permitted to precipitate and activity corrections were made 184 using the Davies equation. Figure 1 presents chemical equilibrium models of speciation 185 in the Fe(II)-S(-II)-Zn-organic-ligand system in terms of molar concentrations and 186 mineral saturation indices (IAP/Ksp) and as a function of increasing total system sulfide 187 concentration (sulfide in both dissolved form and bound in minerals). In this model, 188 Fe(II) is available at concentrations in equilibrium with siderite (as per Holland, 1984), 189 pH is determined by equilibrium with siderite and calcite at pCO2 = 10X PAL (present 190 atmospheric level). We consider a pCO2 of 10 times PAL as an intermediate value 191 between high-end estimates for pre-1.8 Ga (>100 PAL, Ohmoto et al., 2004) and the 192 modern. Total zinc is fixed at an approximately modern value of 10 M (consistent with 193 Zn concentrations derived from the IF record presented below, as well as those used by 194 Saito et al., 2003), and the system is effectively titrated with increasing quantities of 195 sulfide (total S-II added). 196 Our dataset of Zn in authigenic iron oxides includes new analyses and a 197 comprehensive literature compilation (SI Table 2). We assign iron formations to one of 198 four broad categories – Algoma IF, Superior IF, ironstones, and Phanerozoic 199 hydrothermal deposits. Algoma IF are characterized by limited areal extent and close 200 association with submarine volcanic sources. Superior IF are laterally extensive and 201 typically formed on continental shelves. Ironstones encompass Precambrian granular and 202 oolitic iron formations, as well as more modern iron oolite-pisolite occurrences that 203 formed in shallow, nearshore environments. Phanerozoic hydrothermal deposits represent 204 modern, oxic seawater deposits where Fe(III) deposition occurred near hydrothermal 205 sources (Bekker et al., 2010). 206 Zn was analyzed in drill core and fresh hand samples (i.e., samples collected in 207 the field from outcrop), which were sub-sampled, then powdered and subjected to trace 208 element analysis. Importantly, samples showing evidence of weathering, alteration or 209 signs of severe metamorphic or diagenetic overprinting were excluded. Exclusion criteria 210 include association with a lateritic profile, Fe concentrations greater than 60%, extensive 211 veining, for all but Eoarchean samples recrystallization of chert to macrocrystalline 212 quartz, intense folding, and above-greenschist facies metamorphism. Analyses were 213 performed on powder digests or by in situ laser ablation (New Wave Research UP-213) 214 using a Perkin Elmer Elan6000 Quadrupole – Inductively Coupled Plasma Mass 215 Spectrometer (Q-ICP-MS) at the University of Alberta (U of A). Precision was monitored 216 by repeated analyses of well-constrained international standards (BE-N Basalt, CRPG 217 Nancy for digests and NIST SRM 610 or 612 for laser ablation). Sample analyses at 218 Woods Hole Oceanographic Institute (WHOI) were conducted on a ThermoElectron Inc. 219 Element 2 high-resolution sector field ICP-MS and precision and accuracy assessed by 220 analysis of USGS geostandard BHVO-1. Sample selection and analytical methods are 221 identical to our previous work (Konhauser et al., 2009, 2011) and are described in detail 222 therein. At the U of A repeated analyses (n = 3) of BE-N produced a value for Zn of 223 128±19 ppm at the two standard deviation level, compared to a recommended value of 224 120±13 ppm. Repeat analyses of laser ablation standards NIST SRM 610 and 612 at the 225 U of A yielded average values of 474±66 ppm (n = 46) and 38.7±5.0 ppm (n=31) at the 226 single standard deviation level. These are compared to mean literature values for laser 227 ablation of NIST STM 610 and 612 of 469±34 ppm and 40±2 ppm, respectively (Jochum 228 et al., 2011). At WHOI repeated analysis of BHVO-1 produced a value for Zn of 91±19 229 ppm at the two standard deviation level, compared to a recommended value of 230 105±5ppm. 231 From a database of over 3800 new and literature IF analyses, 1660 have available 232 Zn data, and of those, 590 samples passed filters for detrital contamination (<1% Al2O3 233 and <0.1% TiO2; Fig. 2) and compatible mineralogy; the unfiltered and filtered records 234 are presented in Fig 3A and B, respectively. Compatible mineralogies were restricted to 235 Fe and Si-rich chemical sediments, thereby excluding volcanics, sulfides, and carbonates. 236 For authigenic iron oxide sediments with low detrital contamination, molar Zn and Fe 237 data were compared to hypothetical partitioning scenarios to constrain potential 238 paleomarine Zn concentrations. 239 The simple partitioning models presented herein (lines in Figure 4) constitute an 240 effort towards developing trace element proxies in IF that are better informed by the rock 241 record itself and are independent of experimentally-determined partition coefficients. The 242 slope in Zn-Fe space (a Zn/Fe ratio) is calculated by assuming quantitative precipitation 243 of both Zn and Fe from a given volume of seawater, such that hypothetical seawater Zn 244 and Fe concentration scenarios may be compared directly with rock record data. In 245 reality, only a fraction of total dissolved Zn will be removed at any given time, but as this 246 is also the case with Fe, and as Zn adsorption depends on available Fe(III) oxyhydroxide 247 surface sites, partial co-removal of Zn and Fe approaches the scenario of quantitative 248 removal proposed by the simple, hypothetical partitioning scenarios. 249 The hypothetical partitioning scenarios presented in Figure 4 are dependent on 250 several important assumptions: (1) that adsorption occurred to Fe(III) oxyhydroxides, 251 such that any particular trace element should scale with Fe (but not Si), (2) that maximum 252 dissolved Fe concentrations may be constrained (as per Holland, 1984) by either mineral 253 solubility (e.g., ~1-10 ppm for siderite) or sedimentation rate (e.g., 20 mg/cm per year 254 under a water column of at least 100 m, thus 2 ppm), and (3) that Zn and Fe precipitated 255 quantitatively. Assumption (1) is supported by Fig. 4, and while assumption (3) is 256 unlikely, it is conservative in that a maximum estimate of partitioning efficiency and thus 257 a minimal potential seawater concentration is achieved. 258 259 260 Results 261 Results of the geochemical models are presented in Figure 1 and described in detail 262 herein. Mineral saturation indices (upper dashed lines) indicate that saturation with 263 respect to sphalerite is achieved at total sulfide concentrations over 10 M and limits total 264 dissolved Zn (combination of Zn and ZnS) thereafter. Total sulfide concentrations 265 above 5 x 10 M (saturation with respect to mackinawite) are effectively excluded by the 266 S-poor mineralogy of IF samples. Three models are considered: (1) in the absence of 267 organic ligands (Fig. 1A), (2) with 1 nM of an uncharacterized organic ligand binding 268 Zn with a conditional log K of 11, as described for Central North Pacific seawater by 269 Bruland (1989)(Fig. 1B), and (3) with 100 nM of the same organic ligand, as a sensitivity 270 test for historical differences in the availability of organic ligands (Fig. 1C). In all 271 models, total Zn concentration is effectively limited by the sulfide-dependent solubility of 272 sphalerite. In the absence of organic complexation (Fig. 1A), it can be seen that the 273 hydrated metal aquo complexes of Zn and Fe dominate under all conditions, with the 274 exception of the highest permitted sulfide concentrations, where Zn and ZnS(aq) become 275 approximately equimolar (see discussion). When organic complexation of zinc is 276 considered, at modern concentrations of strong Zn binding ligands (~1-3 nM, Bruland, 277 1989; Jakuba et al., 2012), organic zinc complexes quickly become the dominant form of 278 dissolved Zn. When total S(-II) is further increased, sphalerite precipitation draws down 279 the total dissolved reservoir to parity with the strong Zn binding organic ligand (at ~10 280 M total S(-II) added). In the case of Zn-binding organic ligand concentrations 100X that 281 of modern (Fig. 1C), regardless of the ambient sulfide concentration, the total dissolved 282 Zn pool is effectively dominated by organic complexes, free Zn is suppressed even 283 under sulfide-poor regimes, and the total Zn inventory is buffered against sphalerite 284 precipitation losses. While these geochemical models reaffirm a strong role for organic 285 complexation in determining the bioavailable Zn pool, bioavailable Zn does not 286 descend significantly below concentrations considered limiting for all organisms 287 investigated (10 M), unless upper water column depletion of Zn is also considered (see 288 discussion). 289 Compositional data for modern and ancient authigenic iron oxides are presented 290 in Figures 2 through 4. Zn concentrations in detritally-filtered samples average 130 ppm 291 (nearly twice the upper crustal value of 67 ppm; Rudnick and Gao, 2003), with average 292 molar Zn/Fe ratios of 0.00228 and a standard deviation of 0.0238. For samples with Al or 293 Ti values above detrital filter cutoffs, Zn concentrations tend towards upper crustal 294 values, suggesting an increased Zn contribution from siliciclastic sources (Fig. 2); 295 samples below detrital filter cutoffs lack correlation of Zn with Al and Ti but show Zn 296 concentrations that scale with Fe (Fig. 4). This indicates authigenic Zn enrichment 297 conforming to distribution coefficient behavior in these samples. It is likely that Zn was 298 acquired during initial ferric oxyhydroxide precipitation by adsorption processes (e.g., 299 Benjamin and Leckie, 1981; Planavsky et al., 2010), which during the Archean and early 300 Proterozoic, most likely occurred in the marine photic zone (Konhauser et al., 2002; 301 Kappler et al., 2005; Planavsky et al., 2010). 302 Figure 3A-B display all available data and those passing detrital filters, 303 respectively, as a time series of molar Zn/Fe ratios. While significant variability exists at 304 any given time, the overall trend is a relatively static range in Zn/Fe over geological time, 305 except for the most modern samples (see discussion). Figure 4A puts these ratios in 306 perspective by presenting simple models for quantitative Zn removal at marine Fe 307 concentrations of 179 μM (10 ppm) and near-modern Zn concentrations of 10 nM (0.65 308 ppb); nearly all data fall within the range predicted by our models and crucially, indicate 309 paleomarine Zn concentrations within an order of magnitude of modern oceans. An iron 310 concentration of 179 μM was applied as it represents the upper limit of conservative 311 estimates based on the work of Holland (1984) and would subsequently correspond to the 312 lowest estimate of paleomarine Zn (i.e., decreasing Fe from 179 to 17.9 μM results in 313 increasing estimates for paleomarine Zn). 314 At assumed Fe concentrations of 179 μM, a minimum Zn concentration of 0.1 315 nM is indicated by Zn/Fe ratios preserved in ancient iron oxides, yet the majority of 316 samples are well represented by a concentration within 10-fold of modern Zn values (Fig. 317 4A). Estimates are considered conservative as our models assume 100% adsorption of Zn 318 onto the primary ferric oxyhydroxide; this quantitative scavenging relationship represents 319 maximum possible partitioning efficiency and thereby returns a minimum possible 320 concentration; partial Zn adsorption would lead to increased estimates for paleomarine Zn 321 concentrations. Conversely, increased dissolved Fe concentrations would result in a 322 decreased estimate of paleomarine Zn concentrations. However, even at a high-end 323 estimate of 1790 μM Fe, realistic only for essentially undiluted hydrothermal fluids 324 (Edmond et al., 1982) and ~30X higher than limits imposed by siderite solubility 325 (Holland, 1984), near modern Zn levels are still indicated by the authigenic iron mineral 326 record (Fig. 4B). There exists a high level of agreement between the filtered and 327 unfiltered records in this regard (Fig. 3; SI Fig. 1). 328 329 Discussion 330 Zn/Fe ratios in IF through time appear relatively constant (Fig. 3) despite dramatic 331 changes in ocean chemistry from the Archean to today. The spread in Zn enrichments 332 may be related to (1) vertical/lateral paleomarine spatial variability, (2) local or short333 term fluctuations in the marine Zn reservoir, or (3) diagenetic effects. Firstly, it is 334 anticipated that vertical/lateral spatial variability might be similar to that of modern 335 oceans where Zn concentrations are heterogeneous between, and within, ocean basins. In 336 terms of vertical distribution in the Precambrian, we expect that similar to today, enriched 337 Zn fluids would resupply a depleted photic zone via diffusion and advection from deep 338 waters. Future work examining Zn isotope compositions may reveal whether ancient 339 upper water columns were depleted due to biological activity (c.f. Kunzmann et al., 34