Evidence for the existence of life during the Archaean segment of Earth history (more than 2500 Myr ago) is summarized. Data are presented for 48 Archaean deposits reported to contain biogenic stromatolites, for 14 such units reported to contain 40 morphotypes of putative microfossils, and for 13 especially ancient, 3200–3500 Myr old geologic units for which available organic geochemical data are also summarized. These compilations support the view that life's existence dates from more than or equal to 3500 Myr ago.
At the request of Thomas Cavalier-Smith, the lead organizer of the Discussion Meeting on which this volume is based, I agreed in December 2002 to address in this contribution the known fossil evidence for the timing of the origin of various prokaryotic lineages. In the interim between then and now, however, questions have been raised regarding the previously widely accepted lines of evidence on which any such discussion would be based—(i) stromatolites; (ii) microfossils; (iii) molecular biomarkers; and (iv) carbon and sulphur isotopic data—for the most part dating from rocks as old as ca 3500 Myr. Recently, for example, Stephen Moorbath, in the pages of Nature and citing Knoll (2003a) as his authority (perhaps in error?), suggested that ‘true consensus for life's existence’ dates only from ‘the bacterial fossils of 1.9-billion-year-old Gunflint Formation of Ontario’ (Moorbath 2005). Evidently, all supposed evidences of earlier life, ‘the many claims of life in the first 2.0–2.5 billion years of Earth's history’, are in doubt (Moorbath 2005). Yet it is precisely during this period of Earth history, prior to 2000 Myr ago (Ma), that most workers have assumed that the major varieties of prokaryotes originated. If the fossil record is to make any contribution to defining the timing of origin of the major prokaryotic lineages, such doubts must be resolved. Clearly, here it would be awkward to address the timing early in Earth history of the origin of various prokaryotic lineages when the evidence supporting such suppositions—and, indeed, of the very existence of life itself—has been called into question. I have therefore elected to recast this contribution to address a more fundamental, first-order question: what evidence exists for life's existence during the Archaean Eon of Earth history, prior to 2500 Ma?
This discussion will not be exhaustive. Elsewhere in this volume, Roger Summons and his colleagues consider the molecular biomarker data available from Archaean deposits, and John M. Hayes and Jacob R. Waldbauer address the relevant evidence from carbon isotopes. Thus, I need to address only two of the four generally accepted evidences of early life: stromatolites and microfossils. But given the amount of data reported from even these two categories of evidence, and the space constraints imposed by this journal, my discussion will necessarily be perfunctory. This article is not the place for a full-blown description of each of the 48 occurrences of Archaean stromatolites listed in table 1, nor of the 40 morphotypes of objects described as microfossils known from Archaean-age rocks (table 2). Rather, my aim is to summarize and illustrate representative examples of such occurrences, so that others may have a better understanding of why it is that most workers worldwide are of the opinion that the ‘true consensus for life's existence’ dates from more than 3500 Myr ago.
2. Preservation of the Archaean rock record
Relatively few workers have considered, even in broad-brush outline, the temporal distribution of the known sedimentary rock record. One of the best such syntheses is provided by Garrels & Mackenzie (1971). According to these authors, the average lifetime of such geologic units is about 200 Myr old and, thus, as a result of geological recycling, about 50% of such sediments that have survived to the present date from the Phanerozoic (less than 550 Myr in age), the remainder from the earlier, ca 4000 Myr long Precambrian (Garrels & Mackenzie 1971, pp. 255–276). As estimated by Garrels & Mackenzie (1971, p. 275), ‘about 90 percent of the Precambrian once deposited is gone’, surviving rocks petering out rapidly with increasing geologic age to yield a severely depleted Archaean rock record. Hence, the oldest surviving igneous rocks, from the Acasta Gneiss Complex of Northwest Territories, Canada, are but ca 4000 Myr in age (Stern & Bleeker 1998), hundreds of millions of years younger than the age of the planet itself. The oldest known sedimentary rocks, amphibolite facies (ca 450–700 °C, ca 3–10 kb; Klein & Hurlbut 1985, p. 505) metasediments of the Isua Supracrustal Group of southwestern Greenland, date from ca 3800 Ma (Nutman et al. 1984; Moorbath 2005). And only two relatively thick ancient sedimentary sequences, those of the Pilbara Craton of Western Australia and the Barberton Greenstone Belt of South Africa and Swaziland, have survived to the present—both spanning the period between ca 3500 and 3000 Ma and both regionally metamorphosed to lower greenschist facies (ca 250–300 °C, ca 2–5 kb; Klein & Hurlbut 1985, p. 505).
Given the markedly depleted Archaean rock record and the fossil-destroying effects of metamorphism typical of such ancient terrains, it is not surprising that ‘in comparison with the fossil record of the Proterozoic (less than 2500 Myr old) Precambrian, that of the Archaean is minuscule’ (Schopf et al. 2005). Nevertheless, it is notable that both of the two relatively thick Archaean sedimentary sequences that have survived to the present contain what has been widely regarded as firm evidence of life, each containing layered megascopic structures interpreted to be microbially deposited stromatolites (table 1; figures 1 and 2), each containing microscopic objects regarded as prokaryotic microfossils (tables 2 and 3; figures 3–5).
3. Archaean stromatolites
As used here, the term ‘stromatolite’ refers to accretionary sedimentary structures, commonly thinly layered, megascopic and calcareous, interpreted to have been produced by the activities of mat-building communities of mucilage-secreting micro-organisms, mainly photoautotrophic prokaryotes. Other definitions have been proposed, some similarly emphasizing the inferred biogenic, organosedimentary nature of such structures (e.g. Walter 1976; Awramik, in Semikhatov et al. 1979; Buick et al. 1981), others focusing solely on the sedimentological morphology of such structures (e.g. Semikhatov et al. 1979, excluding Awramik; Grotzinger & Knoll 1999), and still others searching for a middle ground (Hofmann 1973, pp. 348–350; 2000). Such divergence reflects the difficulties inherent in consistently differentiating unambiguously between assuredly biogenic stromatolites and accretionary abiotic look-alikes (e.g. geyserites, stalagmites and similar cave deposits, tectonically or otherwise deformed sediments, and duricrusts such as calcretes, silcretes and the like). Criteria for such differentiation have been enumerated by Buick et al. (1981, pp. 165–167) and Walter (1983, pp. 189–190), in which firm establishment of biogenicity revolves chiefly around detection within such structures of cellularly preserved microfossils or trace fossils (‘palimpsest microstructures’) of the microscopic organisms responsible for their formation. Unfortunately, even this criterion falls short, since the mere presence of fossilized micro-organisms within an ancient stromatolite-like structure cannot demonstrate that the structure accreted as a direct result of microbial mat-building activities. Moreover, because almost all known ancient stromatolites are or were originally of calcareous composition, presumably composed initially of metastable aragonite or high-Mg calcite (Grotzinger & Knoll 1999), growth of carbonate grains (aggrading neomorphism) during early diagenesis as well as changes occurring during subsequent diagenetic alteration have in all but a relatively few instances served to obliterate morphologically identifiable evidence of the formative mat-building microbes. For this reason, cellularly preserved fossil microbes are known almost without exception only from those stromatolitic deposits in which the initial carbonate matrix was replaced by silica very early during diagenesis, prior to the onset of widespread cellular decay and microbial disintegration and before the development of carbonate neomorphic alteration. Thus, ‘it is probably conservative to estimate that less than 1 percent of all stromatolites ever described have a fossilized microbiota associated with them’ (Grotzinger & Knoll 1999, p. 316).
Given the general absence of microscopic fossils in stromatolitic structures, it clearly is difficult, and is perhaps impossible, to prove beyond question that the vast majority of reported stromatolites, even those of the Proterozoic Precambrian, are assuredly biogenic. Yet in the Proterozoic, stromatolites are so widespread and abundant, and their biological interpretation is so firmly backed by studies of microbial communities cellularly preserved in Proterozoic cherty stromatolites (e.g. Mendelson & Schopf 1992; Schopf 1999; Knoll 2003a; Schopf et al. 2005), that there can be no doubt that nearly all are products of biological activity.
In the Archaean, the problem of proving the biogenicity of such structures presents a greater problem, due chiefly to the paucity of Archaean sediments (for reasons such as those discussed above) and the correspondingly small number of known occurrences of stromatolites and preserved microbial assemblages. Still, it is important to recognize that reports of Archaean stromatolites have increased markedly over the past decade, with such structures now known to be decidedly more abundant and more diverse than was previously appreciated (table 1). Virtually all of the workers who have recorded Archaean stromatolites have also studied in detail stromatolites of the Proterozoic; interpretation of the biogenicity of the Archaean forms and the differentiation of such structures from abiotic accretionary look-alikes are based on the same criteria as those applied to stromatolites of unquestioned biogenicity in the younger Precambrian (including analyses of their laminar microstructure, morphogenesis, mineralogy, diagenetic alteration and so forth; e.g. Buick et al. 1981; Walter 1983). All of the Archaean stromatolites listed in table 1, and the representative examples shown in figure 1, are regarded by those who have reported them as meeting the definition of stromatolite used here, ‘accretionary sedimentary structures, commonly thinly layered, megascopic and calcareous, interpreted to have been produced by the activities of mat-building communities of mucilage-secreting micro-organisms, mainly photoautotrophic prokaryotes’.
Table 1 lists 48 occurrences of Archaean stromatolites reported to date. Occurrences regarded as being of questionable biogenicity and/or Archaean age are not included. Three principal observations are suggested by this compilation.
Despite the scarcity of Archaean geologic units relative to those of the younger Proterozoic, the temporal distribution of such reported occurrences is more or less continuous throughout much of the Archaean, ranging from ca 2500 to 3500 Ma (table 1; figure 2).
An impressive array of morphological diversity has been recorded in numerous Archaean units: sediments of the Transvaal Supergroup (ca 2560 Myr old) and of the Insuzi (ca 2985 Myr old), Steeprock (ca 2800 Myr old) and Fortescue (ca 2723 Myr old) Groups are all reported to contain stratiform (e.g. figure 1c,d,f,g,n), pseudocolumnar (e.g. figure 1a,f), domical (figure 1a,b,m), conical (figure 1e,j–l), columnar (figure 1a) and branching stromatolites (figure 1a,f), whereas those of the Yellowknife Supergroup (ca 2650 Myr old) are reported to contain all of these stromatolite/microbial mat types with the exception of conical forms (Hofmann 2000). Such diversity in stromatolite morphology in a given geologic unit, not uncommonly in the same sedimentary facies, suggests that such forms are not a product of a single set of non-biogenic accretionary processes.
Conical stromatolites have been recorded in 17 of the 48 units listed in table 1 (Hofmann 2000). Present in fully one-third of these deposits—notably including the more than 3300 Myr old Strelley Pool Chert (Hofmann et al. 1999; Allwood et al. 2004) and Kromberg and Panorama Formations (Hofmann 2000)—such ‘conoform stromatolites appear to constitute a special case’, distinctive structures evidently requiring for their formation ‘both highly motile [microbial] mat builders and penecontemporaneous mineral precipitation’ (Grotzinger & Knoll 1999, pp. 342–343). Thus, Archaean conoform stromatolites, ‘especially the conical structures found in 3.45 Gyr rocks of the Warrawoona Group [viz., the Stelley Pool Chert]’, have been regarded as having their accretion most probably ‘facilitated by micro-organisms’ (Knoll 2003b, p. 6).
4. Archaean microfossils
Over the past decades, the rules for accepting Precambrian microfossil-like objects as bona fide have come to be well established; namely, that such objects be demonstrably biogenic, and indigenous to and syngenetic with the formation of rocks of known provenance and well-defined Precambrian age (Schopf & Walter 1983; Schopf 2004). Of these criteria, the most difficult to satisfy has been that of biogenicity (Hofmann & Schopf 1983; Schopf & Walter 1983; Mendelson & Schopf 1992). A nested suite of seven traits for establishment of such biogenicity has been proposed (Buick 1990, pp. 443–446); sets of traits, six for spheroidal microfossils and nine for filamentous forms, that can be used to demonstrate a biological origin of these two particularly common Precambrian morphotypes, have been enumerated (Schopf 2004, pp. 521–523); and the use of this multi-trait strategy to establish the biogenicity of members of Proterozoic microbial communities has been documented (Schopf et al. 2005, pp. 362–365).
Such analyses have shown that a prime indicator of the biological origin of such objects is the micron-scale co-occurrence of identifiable biological morphology and geochemically altered remnants of biological chemistry. Thus, evidence consistent with a biogenic interpretation would be provided were chemical data to show that populations of objects well characterized morphologically as ‘cellular microfossils’ were composed of carbonaceous matter, as would be expected of organically preserved micro-organisms, and would seem especially strong were the data to demonstrate that such carbonaceous matter was unquestionably of biological origin (Schopf et al. 2005). Analytical techniques now available permit a one-to-one correlation, at micron-scale spatial resolution, of cellular morphology and carbonaceous chemistry in objects claimed to be microscopic fossils—for specimens exposed at the surface of samples studied, by use of ion microprobe (House et al. 2000; Ueno et al. 2001a), electron microprobe (Boyce et al. 2001) and Raman spectroscopy (Arouri et al. 2000); and for rock-embedded specimens, by Raman point spectra and by both two-dimensional and three-dimensional Raman imaging (Kudryavtsev et al. 2001; Schopf et al. 2002, 2005; Schopf & Kudryavtsev 2005).
Given that the co-occurrence of biological morphology and carbonaceous chemistry in ancient microfossil-like objects is a presumptive indicator of biogenicity, recent reports of sinuous microscopic tubular structures reported from Archaean pillow lavas (Furnes et al. 2004) are not regarded here as bona fide evidence of ancient life: such tubes exhibit neither distinctively biological morphology nor are they of demonstrable carbonaceous composition. In contrast, each of the many morphotypes of Archaean microfossil-like objects listed in table 2 meets both of these traits, and all of the spheroidal and filamentous morphotypes satisfy the enumerated sets of criteria required for establishment of biogenicity of forms of such morphology (Schopf 2004). Many of the rod-shaped to spheroidal morphotypes are juxtaposed in adpressed pairs (figures 3h–k,n–q and 4c–f,h–j), presumptive evidence of biologic cell division. Similarly, numerous filamentous specimens exhibit uniseriate sequences of discoidal to boxlike chert-filled cavities defined in three dimensions by transverse and lateral carbonaceous walls (figures 3r,s, 4o and 5a–d), presumptive cell lumina and a definitive feature of bona fide cellular filamentous microbes, both modern and Proterozoic (e.g. figure 4k,l, unquestioned microbial filaments from the ca 2000 Myr old Gunflint Formation of Canada; Hofmann 1971). The combined use of Raman imagery (Schopf & Kudryavtsev 2005) and confocal laser scanning microscopy (Schopf et al. 2006) can be expected to more fully demonstrate the three-dimensional cellularity of reported Archaean microbe-like filaments.
Table 2 summarizes salient characteristics of 40 morphotypes of microfossil-like objects reported from 14 Archaean geologic units. Although the biogenicity of such objects in only one of these 14 units has been questioned (those of the Apex chert; Brasier et al. 2002, 2004), in deference to Moorbath (2005) such microfossil-like objects are here referred to as ‘putative microfossils.’ Two principal observations are suggested by the compilation presented in table 2. First, all of the 40 morphotypes are morphologically simple—small rod-shaped bodies, unornamented coccoids, or sinuous tubular or uniseriate filaments—microbe-like morphologies typical of Proterozoic microscopic fossils (e.g. Hofmann & Schopf 1983; Mendelson & Schopf 1992; Schopf 1999; Knoll 2003a) and a simplicity of form, thus, consistent with their interpretation as early-evolved Archaean members of the microbial evolutionary continuum well established in the younger Precambrian. Second, of the six distinctive classes of microbe-like morphotypes reported from these Archaean deposits summarized below, all but one are known from several or many geologic units of markedly differing geologic age, a redundancy of occurrence that is similarly consistent with the interpretation of such putative microfossils as members of the Precambrian microbial evolutionary continuum. The six classes of morphotypes now known are as follows.
Very small solitary, paired or clustered rods (ca 0.75 μm broad, ca 1.5 μm long), inferred to be prokaryotic (bacterial) unicells: one unit (ca 2600 Myr old), one morphotype (figure 3l–q).
Small, solitary, paired or clustered coccoids (average diameter ca 3 μm, range ca 2–5 μm), inferred to be prokaryotic (bacterial, perhaps cyanobacterial) unicells: three units (range 3320–2600 Myr old), three morphotypes (e.g. figures 3f–k and 4a–f).
Large solitary or colonial coccoids (average diameter ca 13 μm, range ca 5–23 μm), inferred to be prokaryotic (bacterial, perhaps cyanobacterial) unicells: three units (range 3388–2560 Myr old), four morphotypes (e.g. figure 4g,m,n).
Narrow unbranched sinuous filaments (average diameter ca 1.25 μm, range ca 0.2–3 μm), with or without discernable septations, inferred to be prokaryotic (bacterial, perhaps cyanobacterial) cellular trichomes and/or trichome-encompassing sheaths: 10 units (range 3496–2560 Myr old), 17 morphotypes (e.g. figures 4k,p and 5e–i).
Broad unbranched septate filaments (average diameter ca 8 μm, range ca 2–19.5 μm), inferred to be prokaryotic (perhaps cyanobacterial) cellular trichomes: four units (range 3496–2723 Myr old), 10 morphotypes (figures 3s, 4o and 5a–d).
Broad unbranched tubular or partially flattened cylinders (average diameter ca 13 μm, range ca 3–28 μm), inferred to be prokaryotic (perhaps cyanobacterial) trichome-encompassing sheaths: five units (range 3496–2516 Myr old), five morphotypes (e.g. figures 3a–e and 4l).
5. Evidence of palaeoarchean life
Table 3 summarizes palaeontologic and organic geochemical data currently available from 13 Palaeoarchean (range 3200–3600 Myr old) geologic units of the two relatively thick ancient sedimentary sequences that have survived to the present, those of the Pilbara Craton of Western Australia and the Barberton Greenstone Belt of South Africa and Swaziland. As is there tabulated, 10 of these particularly ancient units are reported to contain stromatolites/microbial mats; 11 contain putative microfossils; and carbon isotopic data are available from nine of the units, including data from analyses of individual microfossil-like objects in the oldest such unit, the Dresser Formation, as well as Raman spectra and/or two-dimensional Raman images of the carbonaceous components of putative microfossils from five of the units. As shown in figure 6, bulk analyses of the particulate carbonaceous kerogen in these deposits yield average δ13CPDB values that range from −27 to −32‰, whereas such values for carbonate carbon measured in five of the units centre at ca 0‰, values consistent with carbon isotopic fractionation by autotrophic micro-organisms (−25±10‰) and, thus, with a biologic origin of the reduced carbon (e.g. Hayes et al. 2002).
Although most of the 13 units listed in table 3 are interpreted to have been deposited under shallow marine conditions, four units—the Apex chert, Dresser and Dixon Island Formations, and the Sulphur Springs Group—are regarded as representing hydrothermal settings. Chert units of each of these four deposits contain carbonaceous microfossil-like filaments, those from the two oldest deposits (the ca 3496 Myr old Dresser Formation and the ca 3465 Myr old Apex chert), including specimens composed of cell-like segments as broad as ca 20 μm in diameter. Owing to their preservation in ancient hydrothermal springs, some such putative fossils have been suggested to be chemotrophs (e.g. Rasmussen 2000). Based on morphology, the broad filaments of the Dresser Formation cherts have been compared both with chemotrophic beggiatoacean gliding bacteria and with photoautotrophic oscillatoriacean cyanobacteria (Ueno et al. 2001b). In contrast, the Apex filaments were formally described as ‘prokaryotes Incertae Sedis’ (Schopf 1993, p. 643), microbes of uncertain systematic position that because of their ‘undetermined phylogenetic relations’ were specifically ‘not referred to previously described’ fossil microbial taxa.
Nevertheless, the fact that about two-thirds of the named Apex taxa exhibit ‘cyanobacterium-like’ morphology (Schopf 1993, 1999) has presented a puzzle. Cyanobacteria comprise the evolutionarily most advanced lineage of the Bacterial Domain (e.g. Blankenship 1992), all members of the group being capable both of photosynthetic oxygen production and respiratory oxygen consumption. Thus, the presence early in Earth history of these morphologically cyanobacterium-like putative fossils has been widely assumed to suggest that oxygenic photosynthesis and aerobic respiration, both highly derived processes of microbial biochemistry, had already evolved by ca 3500 Myr ago.
Such, however, may not have been the case. In particular, as was specifically noted in the paper in which the Apex fossils were formally described and named (Schopf 1993, p. 644): ‘it is conceivable that the external similarity of the Apex micro-organisms to younger oxygen-producing [cyanobacteria] masks significant differences of internal biochemical machinery; thus, their morphology may provide a weak basis on which to infer paleophysiology’. Consistent with this contention is the present-day occurrence of non-photosynthetic cyanobacterium-like filaments in microbial mats associated with deep-sea thermal vents (Jannasch & Wirsen 1981). Samples collected at a depth of 2550 m from thermal submarine springs at the Galapagos Rift oceanic spreading centre by use of the research submersible ALVIN (Jannasch & Wirsen 1981) contain short tapered uniseriate filamentous microbes that are morphologically comparable to Primaevifilum amoenum (figure 4a–d) of the Apex chert (Schopf 1993). Notably, Jannasch & Wirsen (1981, pp. 528, 537) are impressed by the cyanobacterium-like morphology of these deep-sea microbes, describing them as ‘highly differentiated forms [that appear to be] analogues of certain cyanobacteria…tapered trichomes [that] compare strikingly with members of…[the modern cyanobacterial genus] Homeothrix’.
Thus, though of cyanobacterium-like morphology, the relatively broad carbonaceous filaments of the Dresser Formation and the Apex chert need not be regarded as being physiologically modern cyanobacteria. Rather, like the modern cyanobacterium Oscillatoria and its living chemotrophic apochlorotic analogue Beggiatoa, the fossil forms may represent non-photosynthetic cyanobacterium-like microbial mimics.
Evidence for the existence of life during the Archaean seems strong. Data are presented here from 48 Archaean deposits reported to contain biogenic stromatolites (table 1; figures 1 and 2), from 14 Archaean units reported to contain 40 morphotypes of putative microfossils (tables 2 and 3; figures 3–5), and from 13 especially ancient, 3200–3500 Myr old, Palaeoarchean units, for which available organic geochemical data are also summarized (table 3; figure 6). These compilations support the view of most workers in the field of Precambrian palaeobiology, worldwide, that the ‘true consensus for life's existence’ dates from more than or equal to 3500 Myr ago.
I am pleased to acknowledge the help of H. J. Hofmann, K. Grey and M. R. Walter in amassing the data on Archaean stromatolites presented in table 1, and the suggestions for improvement of an earlier version of this manuscript by A. D. Czaja, A. B. Kudryatvsev, J. Shen-Miller and A. Tripathi. This work was supported by NASA Exobiology grant NAG5-12357 (to J.W.S.) and by CSEOL, the IGPP Center for Study of the Origin and Evolution of Life at UCLA.
One contribution of 14 to a Discussion Meeting Issue ‘Major steps in cell evolution’.
- © 2006 The Royal Society