Enhanced terrestrial Fe(II) mobilization identified

source:nature

Abstract

Most cave formation requires mass separation from a host rock in a process that operates outward from permeable pathways to create the cave void. Given the poor solubility of Fe(III) phases, such processes are insufficient to account for the significant iron formation caves (IFCs) seen in Brazilian banded iron formations (BIF) and associated rock. In this study we demonstrate that microbially-mediated reductive Fe(III) dissolution is solubilizing the poorly soluble Fe(III) phases to soluble Fe(II) in the anoxic zone behind cave walls. The resultant Fe(III)-depleted material (termed sub muros) is unable to maintain the structural integrity of the walls and repeated rounds of wall collapse lead to formation of the cave void in an active, measurable process. This mechanism may move significant quantities of Fe(II) into ground water and may help to explain the mechanism of BIF dissolution and REE enrichment in the generation of canga. The role of Fe(III) reducing microorganism and mass separation behind the walls (outward-in, rather than inward-out) is not only a novel mechanism of speleogenesis, but it also may identify a previously overlooked source of continental Fe that may have contributed to Archaean BIF formation.

Introduction

The tropical regions of Brazil, including Carajás, Iron Quadrangle (IQ), and Southern Espinhaço Range, contain some of the most extensive landscapes of Proterozoic iron deposits in the world1, 2. The upper sequence is comprised of a laminated quartz-hematite banded iron formation, known as itabirite or jaspilite, which hosts some of the largest iron ores deposits in the world3, 4. The formation of these ore bodies, which represent a heterogenous mix of hematite, and goethite, and other Fe-hydroxides, is not clearly understood, but required silica removal followed by Fe replacement/deposition, with a total Fe content up to 67 wt%2, 3, 5, 6. These high-grade ores, which are generally low in P, Al and Si, are among the most economically important Fe deposits in the world, representing > 20% of global iron reserves2, 7, 8.

The iron-rich landscapes and subsurface features in Brazil are covered in a ferruginous duricrust known as canga (derived from the indigenous word itapanhoacanga9), which protects the more friable BIF and ore deposits from weathering1, 2, 9,10,11,12,13. This duricrust, which can range in thickness from a few centimeters to 30 m (average 3 m) is composed of detrital fragments of BIF cemented primarily by Fe-oxides, including hematite, goethite, and relatively poorly-crystalline Fe(III) (hydr)oxides1, 9, 10, 14, 15. This cementation generates a well indurated material surface that is extremely resistant to weathering, with rates of 0.17–0.54 m Myr−1 reported10, 12, 16. Similar to the Fe ores, the mechanism of canga formation, which includes enrichment of P oxides and rare earth elements (REEs) compared to itabirite, is poorly understood9, 17.

Canga does not exist as a separate bedded layer, but rather lays over the BIF landscape like a thick blanket, which led Dorr (1964) to suggest that there must be continuous turnover, otherwise it would have been removed thorough denudation long ago, exposing the friable, underlaying BIF. While canga is still much younger than BIF, it has been dated to ~ 65 Mya, making it one of the oldest exposed landscapes in Brazil9, 10, 18. Nonetheless, the age of canga is not homogenous, and dating has revealed that younger canga is found at depth in the same sampling cores9, 10, 18. In recent years, there has been increasing evidence that microbial Fe-cycling may be responsible for the maintenance of canga, with Fe(III) reduction releasing Fe(II) that is then oxidized as it moves toward the surface9, 12, 13, 19, 20. The abundance of consolidated Fe(III) (hydr)oxide cements at the surface make it resistant to weathering and limits water infiltration; however, discontinuities in the surface of the canga, such as tension joints, fractures and penetration by plant roots allow water to enter the subsurface, where the crust-like surface give ways to a high-porosity matrix in the canga, with an internal porosity up to 29%1, 11, 21,22,23. The routes for water into the subsurface and relatively high internal porosity of canga results in the formation of regionally significant aquifers, and water flow can reach 2.80 × 10–4 m s−1, comparable to highly fractured rocks and even karst aquifers, with primary porosity occurring at the canga-BIF interface1, 11, 24. Despite this porosity, the weathering-resistant nature of canga would suggest that karstification is limited; however, these iron landscapes represent some of the most cave-dense regions of Brazil, containing over 3,000 documented iron formation caves (IFCs), representing ~ 20% of all the known caves in Brazil1, 21, 22, 25, 26.

Brazilian IFCs were first described in 1818, and remained a relative curiosity until 1988, when the new Brazilian constitution included caves as a natural resource that required a preservation zone1, 2, 25, 27. This occurred during a significant increase in mining activities in Brazil, which have grown from ~ 1.5 million tons annually in the 1950s to approximately 20% of world production today (~ 400 million tons annually)28. Ore extraction has primarily occurred through opencast mining, and due to the preservation of identified IFCs, has necessitated a meandering pattern across the landscape, impacting ecosystems through a combination of habitat loss and the impact of mining waste effluent27,28,29,30.

Most IFCs are short, averaging 30 m in length with an average 2 m diameter1. A mechanism of formation of these IFCs was put forward by Simmons25, who postulated that IFCs formed due to dissolution by Fe(III) reducing microorganisms (FeRM) or through solubilization of dolomitic cements within the Itabirite. In support of this hypothesis, Parker et al.31 cultured FeRM from IFC sediments and demonstrated their ability to reduce Fe(III) phases within BIF and canga31, 32. Microbial Fe(III) reduction by cave-asssociated microorganisms was driven by fermentative organisms, which demonstrated pitting of Fe(III) (hydr)oxide surfaces31. While fermentation has not previously been associated with large-scale Fe-reduction, Parker et al.31, 32 demonstrated that it dramatically accelerated Fe-reduction when compared to respiratory Fe-reduction. Nonetheless, the reduction experiments of Parker et al.31, 32 were carried out in batch incubations, where passivation of Fe-oxides by Fe(II) and the closed system could reduce Fe-reduction rates and influence the drivers of FeRM metabolism31, 33. To better reflect the conditions experienced in canga, we demonstrated that under flow conditions bacterial fermentation Fe-reduction led to an accelerated rate of dissolution, which enhanced permeability33.

Together these data suggested that cave formation processes are driven by water bringing organic carbon from surficial primary productivity into the canga via the surface recharge zone through surface unconformities. Then FeRM activities reduce insoluble Fe(III) (hydr)oxides to relatively soluble Fe(II), which can then be transported via the developing aquifer1, 26, 31, 33. Such activity consolidates into flow paths that coalesce around cave conduits, with a mechanism of speleogenesis similar to the mass separation and transport seen in other karst systems, albeit driven by FeRM1, 31, 34; however, there are limitations in this model, as Fe-reduction cannot occur in the presence of oxygen (making cave formation difficult to reconcile with a conduit model) and a conduit model does not match the morphology of the observed IFCs (intercalating rooms, carved floors, no relationship with lithology)1. In this work, we reconcile these observations with FeRM activity using a combination of techniques in geology, materials chemistry, and geomicrobiology, to demonstrate that Fe-reduction is occurring behind the walls of the cave, sequestered from atmospheric O2. We demonstrate that this FeRM activity leads to extensive Fe(III)-reduction in situ, promoting passage collapse and enlargement, in an active and ongoing process that matches the observed morphology1. The low pH and anoxic conditions, along with the presence of apparent electron shuttles, may explain the enrichment of P oxides and REEs in canga9. These data demonstrate not only a novel method of cave formation, but suggest a more significant mechanism of subsurface Fe mobilization and weathering than has previously been considered35,36,37.

Results

In our past work we characterized the microbial activity of the floor sediments of the IFCs, as this would match the traditional gravity flow path for water in epigenic caves34. But if dissolution was primarily occurring at floor level in these caves, this would result in a down-cutting passage morphology, which is not observed34. We have since quantified that the majority of IFCs (> 95%) have irregular, elliptical-shaped passageways and flat floors that align with the dip of the canga-BIF contact1. Given the location of the IFCs at the tops of ridges, it is untenable to invoke phreatic conditions being established to create the elliptical passage shape, and the bulbous plan of these caves (larger chambers intersected by small passageways as shown in Fig. 1) is more suggestive of wall retreat1. Nonetheless, if FeRM activity were driving dissolution on wall surfaces within the IFCs, we would expect there to be evidence of weathering and Fe-oxides on these surfaces; however, IFC walls are hard and smooth, with no obvious weathering. In an attempt to identify weathering zones on the cave walls we used a Schmidt hammer, which measures compressive strength of a material38. The data (Fig. 2) suggest that the strength of the canga walls within the caves is significantly reduced (Piety Cave, Rh 9.0 ± 2.4; VL-02 Cave, Rh 13.4 ± 4.9) compared to unaltered canga (Rh 38.0 ± 6.6), the canga surface directly above the cave (Rh 39.1 ± 9.1) and BIF (Rh 58.8 ± 1.0); the larger range of Rh values for canga are likely due to the heterogeneity of these samples manifested by the presence of hematite clasts .

Given that Fe(III) reduction generally requires anoxic conditions, we hypothesized that this loss of strength could be caused by Fe(III) reduction occurring behind the walls of IFCs, enhancing host rock dissolution and weakening the walls. To test this, we drilled multiple cores in a variety of locations into the walls of two separate IFCs. In all cases, the drill penetrated a hard crust, up to 3 cm thick, before entering an unconsolidated material that ranged from dry-granular to paste-like in consistency (Supplemental Fig. S1). Probing the material with a sterile spatula revealed that this unconsolidated material, which we term sub muros (from Latin sub behind + muros wall), extends into the canga up to 25 cm behind the cave wall surface, in all directions. A computed tomography (CT) scan of a hand sample (Fig. 3) reveals that the cave side of the wall contains cemented clasts (the hard crust we drilled through), with unconsolidated material beneath this crust. Canga is composed predominantly of goethite and hematite, while BIF contains quartz with goethite and hematite (Fig. 4). An XRD analysis of sub muros demonstrated that it is depleted in Fe(III) phases in comparison to the canga and BIF material (Fig. 4). Concentrated HNO3/HCl-extractable Fe (designated Fetot in Fig. 4) of BIF is low in comparison to other phases, and this is likely attributable to the quartz content and extreme resistance of the hematite to acid attack. Indeed, no Fe(III) could be extracted using hydroxylamine-HCl, which is used to reductively dissolve poorly crystalline Fe(III) phases (Fig. 4). Besides being relatively depleted in Fe, the sub muros was also enriched in Fe(II) in comparison to the other phases, indicating that Fe(III) reduction is occurring in this part of the IFCs (Fig. 4). Taken together, these results confirm that reductive dissolution of host rock Fe(III) occurs behind the cave walls, rather than at the cave wall-void interface.

Characterization of cave-associated microorganisms

We have demonstrated enhanced porosity by FeRM activities in flow-through experiments of columns packed with crushed canga33, which matches the increased porosity demonstrated in the CT scan (Fig. 3). Such increased porosity would increase the capacity for meteoric water flow behind the walls and facilitate the mass export necessary for cave formation34. We have previously measured high levels of organic carbon in water dripping into the caves from cracks in the canga (up to 57 mg L−1 compared to 0.5 mg L−1 in carbonate caves31, 39), providing the carbon and energy necessary to support microbial growth and drive FeRM within the sub muros31, 37. When we modeled this chemistry with a synthetic pore water in vitro, we demonstrated significant Fe-reduction that developed a higher activity rate with repeated rounds of induced flow, presumably through FeRM selection33. Direct cell counting revealed ~ 3.23 × 107 cells g-1 in the sub muros across the sampled caves (Supplemental Table S1), which is several orders of magnitude higher than has been seen in limestone caves (~ 104–106 cells g−1) and is just below the microbial abundance observed in soils40, 41.

To determine the metabolic potential of the microorganisms found within the sub muros, we extracted DNA from a sample and carried out a metagenomic analysis. Illumina sequencing of sub muric material from Triangle Cave yielded ~ 16.78 Gbp of data for analysis. The generated dendrogram using normalized gene copy data from 103 identified KEGG orthologs genes (Fig. 5) demonstrated that the sub muric metabolic profile did not cluster with other microbial communities examined in other known Fe-rich systems identified in the JGI IMG database (e.g., acid mine drainage)42, 43. Nonetheless, the sub muros did share a metabolic profile with Fe-rich Brazilian rupestrian soils, similar to that expected in soils above the IFCs (Fig. 5). We observed enrichment of genes involved in amino acid and organic carbon metabolism compared to AMD and other cave environments, particularly in the breakdown of organic polymers and aromatic compounds, as would be anticipated in soil detritus (Fig. 5 and Supplemental Table S2). No photosynthetic genes were observed, confirming that the metabolic activity within the sub muros was isolated from photosynthetic activity, as would be expected in the aphotic environment within the cave (Fig. 5). Sequences associated with CO2 fixation and methane oxidation were infrequently encountered, suggesting a microbial ecosystem supported by the catabolism and assimilation of surface-derived organic carbon rather than autotrophic growth, although transcriptomics will be necessary to more clearly understand the active metabolic processes under these conditions .

Interestingly, the metagenomic data did not demonstrate an enrichment in genes known to be associated with iron metabolism, including mtrABCF and omcABES37, 44. Nonetheless, the microbial community within the sub muros contained members of the Chloroflexi, Acidobacteria, WPS-2 and GAL15, which have limited cultured representatives and for which iron metabolism has not been observed or is poorly described (Fig. 6A)37, 44. Given the importance of c-type cytochromes in Fe(III) reduction, we examined the metagenomic data for the presence of the heme binding Cys-X-X-Cys-His amino acid motif. Three MAGs from the community affiliated with the Betaproteobacteria, the most complete of which (MAG2_5) encoded ~ 85 Cys-X-X-Cys-His motif-containing proteins, while the most complete Chloroflexi (MAG2_5) contained ~ 50 and the GAL15 MAG contained ~ 35 proteins with this motif (Fig. 6A). These data compare with 42 predicted c-type cytochromes in Shewanella oneidensis MR1 and 111 in Geobacter sulfurreducens that were identified using the same method (Fig. 6B). These findings suggest that extracellular electron transfer and dissimilatory Fe(III) reduction pathways may be present in sub muros communities, but involve as-yet poorly characterized genes in the microorganisms identified45. We also observed a relatively high abundance of genes involved in secondary metabolite production (e.g. phenazines; Supplemental Fig. S4) in the metagenomics data, which can act as electron shuttles to Fe(III) phases46, 47, thus providing an additional mechanism of Fe(III) reduction.

To determine if microbial dissolution of Fe(III) (hydr)oxides was occurring, several coupons of hematite were embedded in the sub muros material within the caves for approximately one year (Fig. 7). Hematite coupons were chosen because of the predominance of that phase in materials associated with the caves (Fig. 4 and Ref.32) and the smooth surface of the available hematite coupons which allowed us to visualize alterations of the coupon surfaces. Several attempts were made to identify microorganisms associated with the hematite chips using a number of preservation techniques (including paraformaldehyde or alcohol preservation), but none were found attached, suggesting that these organisms do not grow directly on the hematite surfaces. Instead, we observed pyramidal dissolution pits (Fig. 7B), which were identical to those seen in laboratory incubations of hematite coupons and an IFC-derived FeRM enrichment culture (Fig. 7C)31. Optical profilometry of the hematite coupon surface (Fig. 7D) allowed us to determine the average pit volume to be 19.7 μm3. This generated a total volume (v) loss of 5.25 × 104 μ3 cm−2 of hematite surface area year−1. Taking into account a conservative porosity for canga (16.98%), we normalized to reactive surface area (S¯¯¯) of 148.1 cm2 cm−348, allowing us to scale the in situ dissolution rate (dS) in μm2 dissolution per cm3 canga as:

dS=vcm2×S¯¯¯

Discussion

Our goal has been to understand the processes that lead to the formation of IFCs in the relatively impermeable iron landscapes of Brazil. Caves form in many types of rock, either through erosion or dissolution34. The most common type of caves on Earth are epigenic caves, which form when meteoric water reacts with CO2 to form a weak H2CO3 solution, followed by dissolution of rock outward from the forming cave conduit and mass removal by ground water outward from the forming cave conduit49, 50. Microorganisms may either accelerate or play the primary role in the formation of the second most common cave type, hypogene caves51, 52. For example, hypogene speleogenesis in some sulfidic caves is driven by microbial metabolic activity within groundwater52, where aerobic oxidation of sulfide to sulfuric acid at or near the cave void-wall interface drives dissolution and cave formation53. In all cases, enlargement of the void occurs via dissolution outwards from the developing conduit34. In this “classical” model of cave formation, passage enlargement would be limited by intrusion of atmospheric O2 into the cave, which would limit Fe(III) reduction.

To address this paradox, we suggest a new mechanism of speleogenesis based on our results, where FeRM reductively dissolve the Fe(III)-rich matrix at the BIF/canga interface. Our previous work indicates that pulsed water flow through canga inoculated with sub muros enhances canga-Fe(III) reduction, at least partially due to removal of Fe(II) passivates, which would otherwise limit further Fe(III) reduction31, 33. This pulsed water delivery, which is consistent with rainfall patterns in the rainy season, increases FeRM activity and porosity, accelerating dissolution and the formation of Fe-depleted sub muros (Fig. 8)33. Over time, this weakening of the rock matrix causes a collapse of the Fe(III)-depleted sub muros material into the cave void (Fig. 8). After the collapse, oxygen within the cave atmosphere auto-oxidizes the newly exposed Fe(II)-rich fluids within the still-consolidated wall structure, cementing this matrix at the cave/wall interface (Figs. 3 and 8) and restoring the anoxic interior of the cave wall, wherein the process repeats. The mass removal of Fe(II) by water behind the wall and resultant collapse over time creates the cave void and observed cave morphology (Fig. 1)1. The role of Fe(III) reduction and movement of water behind the walls (outside-inwards), rather than through the cave conduit (inside-outward), is a previously undescribed mechanism of speleogenesis34. We term this newly recognized process exothenic (from the Latin sub behind + muros wall) biospeleogenesis.

Canga is enriched in P oxides, REEs, and has a positive Eu/Eu* ratio (1.8) compared to itabirite9, 17. It has been proposed that the enriched REEs in canga may be due to scavenging by secondary ferrihydrite oxides; however, this hypothesis does not take into account a role for microorganisms in this process9, 33. Given the recent disruption in the global REE supply chain, there has been a renewed interest in microbial sequestration of REEs54. This renewed interest has helped identify a variety of mechanisms that microorganisms utilize to immobilize REEs, including biosorption and bioaccumulation, and includes member of the Chloroflexi, Proteobacteria, Acidobacteria and Actinobacteria, that are enriched in sub muros (Fig. 3)31, 33, 54, 55. When we attempted to culture FeRM species from sub muros, we recovered fermentative Fe-reducing Clostridia spp.33. Similarly, these Clostridia appeared to use organic molecules to transfer electrons to Fe(III) with glucose as an electron donor with these lab-based FeRM cultures generating similar surface pitting profiles as those seen on hematite surfaces in situ (Fig. 4)31, 33, 56. Under the redox and pH conditions of sub muros, Eu would exist as Eu(III) and similar electron shuttles have been demonstrated to play a role in its reduction, which Clostridia bioaccumulate as intracellular polyphosphate56, 57. Microbial activity could therefore explain the formation of canga and relative enrichment of Eu and Nd (1.8 and 4.2 compared to itabirite, respectively)9.

In addition to identifying a new process of cave formation, these caves are an indication of the extensive potential for microbially mediated weathering of the rocks in the Carajás, Iron Quadrangle (IQ), and Southern Espinhaço Range, which are typically considered quite resistant to dissolutional weathering1, 9. Recent work12, 13 has indicated extensive Fe cycling in canga despite the seeming permanence of the canga. The canga is continuously reworked through alternating Fe(III) reduction and then reoxidation of the biogenic Fe(II)12, 13. In addition to the dynamic stability of the canga duricrust, the presence of caves and their microbial origins indicate that the itabirite phases are susceptible to substantial weathering and export of material. Here, reductively dissolved Fe may be transported through extensive subsurface aquifers, which may contribute as much as four times the freshwater discharge of rivers and streams58, 59. Many of these aquifers drain directly into marine environments below the surface, such as the Mediterranean Sea, which receives up to 75% its freshwater from groundwater springs that drain without detrital material that is often associated with continental runoff60. With a calculated dissolution value within sub muros of 7.78 cm3 m−3, using a density for hematite as 5.26 g cm−3, this would mobilize 40.9 g Fe-oxides per m3 canga year−161. If this measured reductive dissolution of Fe occurred uniformly on a regional scale, assuming the canga layer is 3 m thick, this is equivalent to ~ 122 tons of Fe annually moving into the subsurface per km2. Given that the Iron Quadrangle alone constitutes 7200 km2, even if the efficiency of this system were 1% of the in situ observed values, this is equivalent to ~ 9 million tons of subsurface Fe in this region annually. As such, itabirite weathering could be an underappreciated contributor to marine Fe budgets62, 63. Indeed, continental Fe is proposed to be a substantial contributor to the high dissolved Fe in Archaean oceans from which BIF formed65.