Abstract: Recognizing fossil microorganisms is essential to the study of life's origin and evolution and to the ongoing search for life on Mars. Purported fossil microbes in ancient rocks include common assemblages of iron-mineral filaments and tubes. Recently, such assemblages have been interpreted to represent Earth's oldest body fossils, Earth's oldest fossil fungi, and Earth's best analogues for fossils that might form in the basaltic Martian subsurface. Many of these putative fossils exhibit hollow circular cross-sections, lifelike (non-crystallographic, constant-thickness, and bifurcate) branching, anastomosis, nestedness within ‘sheaths’, and other features interpreted as strong evidence for a biological origin, since no abiotic process consistent with the composition of the filaments has been shown to produce these specific lifelike features either in nature or in the laboratory. Here, I show experimentally that abiotic chemical gardening can mimic such purported fossils in both morphology and composition. In particular, chemical gardens meet morphological criteria previously proposed to establish biogenicity, while also producing the precursors to the iron minerals most commonly constitutive of filaments in the rock record. Chemical gardening is likely to occur in nature. Such microstructures should therefore not be assumed to represent fossil microbes without independent corroborating evidence.
3. Discussion
(a) Comparison with previously reported biomorphs
Here, I have shown that the reaction of ferrous sulfate grains with sodium carbonate and sodium silicate solutions in shallow vessels (Petri dishes, limiting vertical extension and introducing an effect of surface tension) allows for the rapid production of large populations of straight and curved filaments with consistently microbe-like sizes and morphologies, including circular cross-sections, non-crystallographic bifurcation during growth, anastomosis, and nestedness. Compositionally, these biomorphs are typical of iron-based chemical gardens previously described in the experimental literature. Most previous experimental studies following the ‘classic’ procedure have used salt granules or pellets several millimetres in diameter immersed beneath several centimetres of solution within test tubes or similar reaction vessels (e.g. [32–35]). This method produces vertically oriented, chimney-like structures several centimetres in length controlled by buoyancy-driven extension, commonly with several sub-vertical branches, which do not closely resemble candidate fossils in their overall morphology. Other studies have used vertically confined spaces to produce smaller, quasi-2D chemical gardens that form meandering filaments with infrequent branching [42,43]. Interestingly, one of these studies [43] describes self-avoidance during filament growth that would seem to preclude anastomosis. In the present study, anastomosis was present but rare, and filaments sometimes met and grew along with each other as if mutually adhesive but unable to converge into a single filament.
The present results do not exhaust the morphospace accessible to chemical gardens, which can also produce pseudoseptate filaments and spherical bulbous terminations resembling fungal sporangia (e.g. figures 38, 39, 48, 55, and 56 in [44]). Serially twisted/helical filaments or ‘stalks’ of iron oxide, which are widely regarded as biosignatures for iron-oxidizing bacteria (e.g. [28]) were not produced in the present study, but classic work suggests that serially twisted forms can also occur (e.g. figure 55a in [44]). Silica-carbonate biomorphs also show helical forms, and further extend the morphospace of abiotic mineral growth structures to encompass fractally branching dendrites, framboid-like masses, rope-like twisted threads and ribbons, and complex shapes resembling urns, corals, and snails, but not closely resembling the iron-mineral filaments addressed by this study [45,46].
(b) Comparison with iron-mineral filaments in the rock record
Iron-mineral filament assemblages previously interpreted as fossilized microbial populations are composed largely of hematite (e.g. [1,2,7,8,11,15]), iron oxyhydroxides such as goethite and ferrihydrite (e.g. [2,5,7,8,10–12,28]), and iron-rich aluminosilicate clay minerals [6–8,10,25]. In line with previous studies of tubular chemical gardens using iron salts [30,32,33], the Raman, EDX, and XRD analyses in the present study suggest that biomorphs produced by reacting ferrous sulfate with either sodium carbonate or sodium silicate solutions were composed largely of iron oxyhydroxides. These minerals very readily transform to hematite during diagenesis or metamorphism, and may also serve as precursors to Fe-rich phyllosilicates in hydrothermal, silica-rich settings [47,48]. Bacterial iron oxidation can likewise produce iron (oxyhydr)oxides (e.g. [49]), but the present results show that the composition of iron-mineral filaments in the rock record is equally consistent with origination through abiotic processes. A recent study also interpreted hollow silica tubes from hydrothermal deposits of the Arctic Mid-Ocean Ridge as possible chemical gardens [28]; such tubes may be obtainable from experiments like those reported here if terminated before iron oxyhydroxides encrust the initial siliceous membranes; cf. the outer layer in figure 1d [32,33].
When produced from seed grains sieved to less than 63 µm in diameter, 188 out of 200 individual chemical garden filaments measured in this study showed external diameters between 2 and 10 µm (median 3.9 µm); no filaments were narrower than 1 µm, and only four were wider than 12 µm (electronic supplementary material, figure S1). This size distribution and range are similar to numerous assemblages of iron-mineral filaments in the rock record (e.g. [1,6,7,50–52]). The chemical gardens in this study (figure 1) also reproduce almost the full range of morphological characteristics (straight and curved trajectories with changes of direction; filled and unfilled (hollow) interiors; circular cross-sections; multiple attachment to knobs; discrete swellings; non-crystallographic, constant-thickness branching; anastomosis; nestedness) previously thought to show that naturally occurring iron-mineral filaments are likely to be microfossils (e.g. [1,2,4,6–15]). It is important to concede that I did not produce true septate filaments with internal, walled compartments, a feature which has been observed in some natural iron-mineral filament assemblages where carbonaceous residues provide additional evidence for biogenicity (e.g. [9,19,20]). Additionally, although filament thickness was usually conserved during growth even during branching and anastomosis (e.g. figure 1e,f,g), this was not always the case; bifurcation could reduce filament thickness while re-convergence could increase it, leading to some dubiously lifelike morphologies, especially in larger filaments. In addition, some filaments tapered gradually in the direction of growth (e.g. figure 1j).
These results are strikingly similar to the assemblage of hematite tubes and non-septate filaments in hydrothermal chert beds of the 4.0 ± 0.3 giga-annum (Ga) Nuvvuagittuq Greenstone Belt, northeast Canada, recently interpreted as Earth's oldest body fossils [1]. The filaments are reportedly 2–14 µm in diameter and up to 500 µm in length, and have been interpreted as the partly permineralized, partly encrusted remains of iron-oxidizing bacteria [1]. Some filaments are attached to knobs 80–120 µm in diameter, and some are nested within tubes (16–30 µm in diameter and 80–400 µm in length), which also occur without filaments; these features were considered incompatible with an abiotic origin, but are replicated abiotically in the present study (figure 1e,i). Buoyancy- or flow-driven growth of chemical gardens from fairly uniform parent crystals or grains would also explain the straight, unbranched, parallel nature of some of the Nuvvuagittuq tubes and their consistent sizes. Hollow tubes could also have originated via dissolution, diffusion, and re-precipitation of filaments during the polymerization of the surrounding silica, with or without leaving residual filaments inside; filaments in some moss agates are surrounded by (commonly multiple) concentric sheath-like tubes likely to have formed similarly [53,54]. Other evidence adduced to support the biogenicity of the Nuvvuagittuq filaments (e.g. the presence near the filaments of graphite, carbonate rosettes with isotopically light carbon, and phosphate) does not settle the biogenicity of the filaments themselves, which are morphologically simple and strictly non-carbonaceous. It is not implausible that alkaline fluids generated by serpentinization of the mafic (sub)seafloor promoted the growth of chemical gardens in this setting.
The results are also reminiscent of numerous candidate microfossils proposed to have formed in subsurface environments, i.e. the deep biosphere (e.g. [5–10,50,51]; see review in [18]). Among these, one assemblage of special scientific importance is the suite of iron-rich chloritic filaments preserved within calcite- and chlorite-filled amygdales (mineralized vesicles) in basalts from the lower part of the 2.4 Ga Ongeluk Formation of South Africa [6]. These filaments were recently interpreted as the oldest fossil eukaryotes, but are similar to the chemical gardens described in the present study in several respects. They are solid, apparently non-septate, about 2–12 µm in diameter, and up to hundreds of µm in length. They are composed of iron-rich chlorite, a common vein- and amygdale-filling phyllosilicate in hydrothermally altered basaltic rocks, where it also forms the filamentous dubiofossil ‘moss’ found in moss agates [55]. The origin of the Ongeluk chlorite is not precisely known; it could derive from the alteration of smectite that replaced organic matter as proposed by Bengtson et al. [6], but smectite can also form via the interaction of hydrothermal silica and iron oxyhydroxides, i.e. the constituents of chemical garden filaments [48]. Independent evidence for an influx of silica-rich hydrothermal fluids exists in the lower part of the Ongeluk Formation in the form of abundant hydrothermal jasper and chert deposits [56].
While the composition of the Ongeluk filaments is seemingly compatible with both biotic and abiotic interpretations, the argument that they are biotic rests largely on their morphological and organizational resemblance to putative fossil fungi from much younger rocks (including some that preserve organic matter). The Ongeluk filaments show curvilinear trajectories, branching, anastomosis, circular cross-sections, and bulbous protrusions. The results of the present study show that all these features are equally consistent with chemical garden growth. Neither the radiating growth of filaments inwards from cavity walls (also seen in moss agates) nor the occurrence of multifurcate, entangled ‘broom’ structures [6] was replicated in my Petri dish experiments, but these features do not seem fundamentally incompatible with chemical garden growth provided with the appropriate distribution of seed material and the correct flow regime and rate. Chemical garden filaments are flexible in the early, gelatinous phase of growth and can become entangled during growth with or without anastomosing. The irregular chlorite lining Ongeluk amygdales, described by Bengtson et al. [6] as a ‘basal film consisting of a jumbled mass', could represent an amalgamation of the membranes formed around seed material in chemical gardens, which become mineralized along with the filaments (figure 1e,g). More naturalistic experimental systems must be used to test these proposals before the hypothesis that the Ongeluk filaments represent chemical gardens can be evaluated fully.
(c) Plausibility of chemical garden growth in nature
Chemical gardens are already thought to occur in geological settings where silica and/or carbonate-laden alkaline fluids react with metalliferous mineral particles or solutions, most notably forming complex structures at marine hydrothermal vents (e.g. [28]; see also [32] for a discussion of chemical gardens in nature). Deep, isolated groundwater tends to become somewhat alkaline (as well as carbonate- and silica-rich) as a consequence of water–rock reactions that consume H+, and in some settings the hydrolysis of olivine and pyroxene in basalts and ultramafic rocks (serpentinization) leads to groundwater pH values as high as 10–12.6 [48,57–61]. Lakes fed by hydrothermal systems in the East African Rift Valley are sufficiently alkaline and silica-rich to be theoretically compatible with biomorph production at the Earth's surface [46], and it has recently been demonstrated experimentally that naturally occurring silica-rich alkaline spring waters are capable of inducing the growth of classical chemical gardens from iron salts, as well as producing silica-carbonate biomorphs [35]. Moreover, the results presented here show that very high pH is not required to form microbe-like filaments, which grew in sodium carbonate solutions acidified to mildly alkaline and even neutral pH (figure 3). Thus, it is reasonable to suppose that groundwater in many of the settings where iron-mineral filament assemblages have been found—silicifying/calcifying marine hydrothermal systems, volcanic rocks near mid-ocean ridges and deeply buried on land, and limestones—could have become sufficiently alkaline to precipitate iron-mineral chemical garden filaments. Further experimental work is, however, needed to test this supposition. Since naturally occurring iron-mineral filaments are widely associated with the common ferrous sulfide mineral, pyrite (e.g. [4,7,62]), I further speculate that the ferrous sulfate minerals or solutions derived from the oxidation of iron sulfide minerals (not necessarily abiotically) may have stimulated the formation of filamentous chemical gardens in some natural settings (a pyrite precursor for some moss agates was also suggested by Hopkinson et al. [27]).
(d) Discriminating between iron-mineralized chemical gardens and fossil microbes
Some natural iron-mineral filament assemblages contain complex organic matter and phosphate, together with iron-mineral growth-textures strongly suggestive of encrustation onto pre-existing organic material, implying that they are more likely to be fossils than not (e.g. [63,64]). Filaments associated with carbonaceous material of indeterminate origin are not necessarily biogenic [31], and most iron-mineral filament assemblages lack such material altogether. Nevertheless, iron-encrusted microbial filaments and abiotic chemical garden filaments and tubes are unlikely to be perfectly indistinguishable in composition, morphology, texture, or organization at all scales, and the possibility remains that diagnostic differences may be discovered [28]. Statistical analyses of morphometric parameters over large populations of biotic and abiotic filaments may be fruitful; preliminary steps have been taken in this direction (e.g. [8,28,45,52]). The controlled experimental iron-mineral encrustation of large numbers of bacterial and fungal filaments will be necessary to provide suitable datasets. As a corollary, experiments to grow chemical gardens in the presence of filamentous microbes may be worthwhile in case this leads to new morphologies. Submicroscopic internal and external textures of biotic and abiotic filaments, not explored in detail by the present work, should be compared. Both smooth-walled and more coarsely crystalline tubes and filaments are found in natural iron-mineral filament assemblages, even together within the same assemblage (e.g. [1]). In the present study, abiotic filaments grown in sodium silicate solution showed smoother exteriors than those produced in sodium carbonate. Smoothness has recently been shown to respond to growth rate, with slow-forming chemical garden filaments tending to show more coarsely textured walls [65]; it has also been shown that chemical gardens grown from ferrous chloride differ microtexturally (and mineralogically) from their ferric equivalents [34]. It was recently pointed out [28] that concurrent precipitation of silica and iron minerals might produce a diagnostically abiotic internal structure in some natural filament assemblages, i.e. a diffuse filament core zone composed of iron-mineral spherules supported by a silica matrix; this was not observed in the present study, but might perhaps occur if more highly polymerized silica media were used.
Posts
