Proc. Natl. Acad. Sci. USAVol. 73, No. 9, pp. 2973-2976, September 1976Chemistry

Fossil microorganisms from the approximately 2800 to 2500 million-year-old Bulawayan stromatolite: Application of ultramicrochemicalanalyses

(Precambrian microfossils/electron microprobe analyses/mass spectrometry/scanning electron microscopy/evolution of life)

Lois ANNE NAGY AND JOHN E. ZUMBERGELaboratory of Organic Geochemistry, Department of Geosciences, The University of Arizona, Tucson, Ariz. 85721

Communicated by Harold C. Urey, July 9,1976

ABSTRACT Microfossils, probably representing membersof Precambrian photosynthetic communities of bacteria andblue-green algae, have been found in the approximately2800-2500 million-year-old Bulawayan stromatolites fromRhodesia. Several populations of coccoid and elongate micro-fossils have been observed in the dark, carbon-rich stromatolitelaminae. Some of these elongate forms are morphologicallysimilar to modern bacterial spores. These microfossils werestudied in petrographic thin sections and identified by com-bined scanning electron microscopy-electron microprob andby analyses of energy dispersive spectra of individual micro-fossils. The microfossils contain 1-20% organic carbon; somemorphotypes contain traces of sulfur and one other, traces ofphosphorus. The polymeric nature of the organic carbon wasestablished by analyzing aggegates of microfossils at elevatedtemperatures in the solid inlet system of an organic mass spec-trometer. The coccoid microfossils range in size from 1.2 to 4.3Am, the elongate microfossils are from 2.4 to 9.8 ;sm. They aremineralized with dolomite, embedded in a calcite matrix, andare shown to be both indigenous and syngenous with the rock.Identical microfossils also containing organic carbon but min-eralized with quartz have been observed in the stromatolitesfrom Belingwe which are part of the Bulawayan Group fromRhodesia. Caution must be used in the interpretation of whatthese forms are because of their great age and relatively simplemorphologies. However, based on morphology and chemicalanalyses, they represent fossilized bacteria, blue-green algae,or, most likely, both.

The earlier the development stage of Precambrian microfossils,the simpler their morphology appears; the unambiguousdiagnoses of these morphologically simple Early Precambrianmicrofossils are essential for evolutionary studies. Conventionalmethods of micropaleontological diagnoses should be aug-mented by ultramicrochemical analyses on the microfossilsthemselves; this approach was used for studying the microfossilsin the Bulawayan stromatolite. As an ancillary part of this ap-proach, the porosity and permeability of a younger stromatolite[approximately 2300 million-year (m.y.) old] from theTransvaal Sequence, which contains unambiguous indigenousmicrofossils (1), have been determined to evaluate the potentialof Recent algal contaminations.-

MATERIALS AND METHODSThe samples of Bulawayan carbonate stromatolite came fromthe Huntsman Quarry near Bulawayo, Rhodesia; this geologicalsetting has been extensively investigated (2, 3). The age of thesestromatolites (and the corresponding Belingwe stromatolitesthat were recently discovered, also from the Bulawayan Groupin Rhodesia) was determined to be approximately 2800-2500m.y. by the Rb-Sr method (4). It has been reported that minuteparticles, marginally resembling bacteria, appear to be present

Abbreviation: m-y., million years.

in the Bulawayan samples as determined by transmitted lightmicroscopy (5). A younger, laterally linked stromatolite fromthe Transvaal Sequence in South Africa with an age of ap-proximately 2300 m.y. has been studied. Its age has been es-tablished from radiometric dates of igneous events at the topand at the base of this Sequence (6, 7). The Transvaal carbonatestromatolites have been extensively studied. (8-11) and blue-green algal microfossils, some of which contain differentiatedcells, have been described (1, 12) from these rocks.The Bulawayan samples were studied in nine petrographic

thin sections by transmitted light microscopy at high magni-fications with oil immersion lenses. In addition, the polishedand etched rock surfaces were studied with a combined scan-ning electron microscope-electron microprobe using an ARLscanning electron microprobe quantometer. This instrumentenables one to obtain scanning electron micrographs of highresolution as well as elemental analyses by electron probe onthe very same micron-sized particles. The polished rock surfaceswere etched with either 0.1 or 0.2 M HCl for 3 or 1 min, re-spectively, then washed with H20, and then by 48% HF for 30sec, then washed again with H20, and finally etched with 0.1or 0.2M HCO for 3 or 2 min, respectively. A control experimentusing the same concentrations of HCl, HF, and HCI for thesame periods of time on the smooth Bulawayan carbonate awayfrom the laminae did not result in the synthesis of any fluoridesor other inorganic artifacts resembling the morphology or sizerange of the microfossils. Some polished surfaces were alsoetched with only 0.2M HCl for 3 min. In the first set of etchingexperiments, the rock samples were coated with aluminium,and in the second set with carbon in order to make them con-ductive. The size distribution of 60 particles, which morpho-logically resembled microfossils, were measured from calibratedmicrographs made with the scanning electron microscope.Electron microprobe analyses (optimized and calibrated forspecific elements from known standards, particularly for car-bon, fluorine, calcium, and iron) were performed both by ele-mental mapping and by point counts on individual microfossils.Energy dispersive spectra were also obtained from electronmicroprobe analyses to ensure that elements >Z = 8 have beendetected. For additional chemical confirmation, aggregates ofparticles from the Bulawayan stromatolite laminae werescraped into the solid inlet system of a Hitachi RMU-6E massspectrometer and then heated to 380° under approximately10-6 torr (0.06 mPa) vacuum. For a blank, mineral matter awayfrom the stromatolite laminae and containing no visible mi-crofossils was analyzed by mass spectrometry in an identicalmanner. Prior to inserting the particles and the blank mineralmatter into the mass spectrometer, the rock was extracted withbenzene-methanol, methanol, and then with water to removepotential soluble chemical contaminations. All glassware in the

2973

2974 Chemistry: Nagy and Zumberge

lk..

X9 0 ^m

A -

FIG. 1. Petrographic microscopy, s electron microscopy, and electron microprobe analyses. (A) Photomicrograph of a petrographicthin section showing a Bulawayan stromatolite lamina; horizontal black line represents 100 .m. (B-E) Photomicrographs taken at higher mag-nification of the same petrographic thin section showing lens-shaped/elongated and (E) rod-shaped particles in or adjacent to the lamina embeddedin the rock; black lines represent 10 lAm. (F, G, K, and L) Scag electron micrographs showing lens-shaped (F, G), rod-shaped (K, note arrows)and coccoid (L) particles. These samples were etched with HCl, HF, and then HCI; qualitative electron microprobe analyses showed C, Mg,Ca, F, and traces of Fe in the particles. Black line represents 10 ,m. (H-J) Particles quantitatively analyzed with the electron microprobe, whichwas optimized for C analysis. These particles were found in the same sample, but in this experiment the sample was etched only with HCL Theseparticles contain 2 ± 0.5% organic C and are mineralized with dolomite; (particle H contains a trace of P). The black lines represent 1 Mum.

experiments were cleaned with a mixture of 85% hot concen-trated H2SO4 and 15% hot concentrated HNO3; all solventswere freshly distilled.The porosity as well as air and water permeabilities of a

Transvaal Sequence stromatolite have been determined. Inaddition, the sedimentary petrology of one of the fossiliferouspetrographic thin sections from the Transvaal stromatolite hasbeen studied [R. L. Folk (1976) personal communication],particularly for any "microcracks" which could have intro-duced Recent algal contaminations. This was advisable sincethe Transvaal stromatolite contains fossil blue-green algaewhich may have modern analogs; therefore, these studieseliminate the potential danger of Recent algal contamina-tions.

RESULTSA transmitted light photomicrograph of a lamina from theBulawayan stromatolite is shown in Fig. IA. The laminae

generally contain dolomite while the rock matrix consists ofMg-free calcite. Microfossils embedded inside the rock in pet-rographic thin sections which resemble those seen in scanningelectron micrographs are shown in Figs. IB-E. Scanning elec-tron micrographs of coccoid, rod-, and lens-shaped microfossilsare shown in Figs. 1F-L; particles shown in Figs. iF, G, K, andL were found in HC1, HF, and HCl-etched samples. The mi-crofossils shown in Figs. 1H-J were obtained by scanning mi-croscopy from rock surfaces which were only HCI etched.These particles were individually analyzed by the electronmicroprobe and contained 2 + 0.5% organic C and are miner-alized with dolomite. (The microfossil shown in Fig. 1H alsocontains a trace of P.)The diameter of the coccoid forms varied between 1.2 and

4.3 Am with a mean diameter of 2.3 Am and standard deviation= 0.7,m. The lengths of the elongate morphotypes variedbetween 2.4 and 9.8 Am with a mean length of 5.9 gm andstandard deviation = 1.9 ,tm; the rod-shaped particles showedmore variation.

Proc. Natl. Acad. Sci. USA 73 (1976)

1% . ,

Proc. Natl. Acad. Sci. USA 73 (1976) 2975

Fourteen individual lens- or rod-shaped particles were an-alyzed specifically for C, P, S, F, Cl, Ca, Mg, Fe, Si, Al, K, andchecked for any other element >Z = 8. All 'particI ih

morphologically resembled microfossils contained organiccarbon (mean: 9%; range about 1-20%), they all containedcalcium (25-37%), significant but varying amounts of Mg, andsmall amounts of Fe (0.05-0.10%). With the energy dispersivespectral analyses covering the elements >Z = 8 in the periodictable, it is not possible to stoichiometrically account for all ofthe carbon found by any possible combination of permissiblemineral compositions. Thus, this excess carbon does appear tobe organic carbon. The microfossils were fossilized with dolo-mite or calcite. The rod-shaped particles contained smallquantities of sulfur in addition to organic C and dolomite. Thelens-shaped particles contained no S; however, one of themcontained traces of phosphorus. The particles treated with HFcontained the same abundance of fluorine as did the smoothrock surface away from the laminae. Samples treated with HClonly contained no F in the microfossils nor in the smooth sur-

faces. Cl was absent in all analyses. Al, Si, and K were occa-

sionally present in what appeared to be clay mineral particles.It is noteworthy that the morphology and C content of F con-

taining particles and those that lack F were identical. Conse-quently, the rod- and lens-shaped microfossils observed in theHCI, HF, and HCl-etched surfaces were not fluoride artifactsnor some iron or sulfur artifacts which have been described as

potential problems in scanning electron microscopy (13). Thefact that organic carbon is present in these microfossils was

confirmed by the mass spectrometric analyses which showedthe polymer breakdown products C3H4+, C3H5+, CSH7+,C4H7+, C4H8+, C4H9+, and C5H9+ but no ions due to benzeneor methanol. The background calcite away from the laminaewas a virtual blank. In another study, the Belingwe lime-stone-chert stromatolite (which is of the same age and is stra-tigraphically equivalent to the Bulawayan) was found to containmicrofossils morphologically identical to those in this Bula-wayan stromatolite. It should be mentioned that these Belingwemicrofossils again contain organic carbon but are mineralizedwith quartz rather than dolomite.A younger stromatolite from the Transvaal Sequence, which

has yielded complex microfossils, was further studied as a

control for contamination. In studies of this kind, Recent con-

taminations are always a threat; this is complicated by the trendof increasing morphological complexity with younger age. ThisTransvaal stromatolite proved to be an unusually massivelimestone. Its porosity is 0.14% (an average massive limestoneis about 8%). Its air permeability is 2.3 X io-3 millidarcies (theaverage massive limestone has 1-2 millidarcies), and its waterpermeability is 2 X 10-6 millidarcies (the average massivelimestone has 0.1 or higher millidarcies water permeability)(14). In addition, the examination of a thin section of fossili-ferous Transvaal stromatolite [R. L. Folk (1976) personalcommunication] showed no open microcracks; however, a closenetwork of micron-sized veins, filled by calcite, penetrated therock. These microtectonic veins were formed at depth duringdiagenesis or incipient mild metamorphism and became ce--

mented with calcite under this same environment. Algal con-

taminations could not have been introduced during such bio-logically severe environments at depth during mild metamor-phism. Thus, the algal microfossils described from the Transvaalstromatolite are both indigenous and syngenous.

DISCUSSIONSince these Bulawayan microfossils were found in a stromatolite(a sedimentary rock structure generally accepted to be built by

blue-green algae and/or bacteria) one would expect that theseforms are microfossils and not prebiological particles. One mayaSk' if th-0e structures are mineral grains coated and/or im-pregnated with polymeric organic matter. However, this seemsnot to be the case since they fall in a narrow size range and havespecific and repetitive morphologies resembling bacteriaand/or blue-green algae. Furthermore, these coccoid, rod-, andlens-shaped particles containing organic carbon are embeddedin mineral grains which contain no organic carbon and showdefinite but random grain outlines. From the chemical analysesit can be concluded that these Bulawayan forms are microfossilsand not sample preparation artifacts or mineral grains. Neithercould they be Recent contaminations because they are miner-alized with dolomite. Although aided by the scanning electronmicroscope, even with chemical information the identificationof primitive microfossils is difficult.To illustrate some of the difficulties in morphological iden-

tification, a scanning electron micrograph of spores of Bacilluspolymyxa seen in cross section shows 8-fold symmetry (15)which is reminiscent of mineral matter. (B. polymyxa is re-ferred to for morphological comparison only, and it is in no wayinferred that there were aerobic bacteria approximately 2800m.y. ago.) Such spores are morphologically similar to one of theBulawayan morphotypes; Bulawayan morphotypes may rep-resent Clostridia-type analogs. Rod-like aragonite needles seenby scanning electron microscopy (16) also morphologicallyresemble simple organisms. However, it should be noted thataragonite is not stable in the geological environment for longperiods of time (17). There are many more morphologicalsimilarities between biological and mineral material, but thesetwo examples will suffice. It is not possible at this point to dif-ferentiate between what may be bacterial spores or blue-greenalgae in the Bulawayan samples because both may have simplemorphologies and their size ranges overlap; bacterial spores canbe >7 jm long (18), and cyanophytae cells can be as small as0.8 1um (19).

This study points out that extreme caution must be used whendetermining the biogenicity of Early Precambrian microfossilsand that the chemical composition of individual structures mustbe known. These Bulawayan forms are the oldest known diverseassemblage of microfossils which have been chemically con-firmed.

We thank Dr. K. A. Kvenvolden for collecting and providing thesamples and Drs. J. W. Anthony, P. E. Cloud, F. Drouet, R. W. Hoshaw,R. J. Janssen, W. S. Jeter, L. M. Kelley, G. 0. W. Kremp, L. Margulis,B. Nagy, J. W. Schopf, J. F. Schreiber, N. A. Sinclair, S. R. Titley, andI. Yall for constructive comments during various phases of this research.This work was supported by National Science Foundation GrantDES75-14859.

1. Nagy, L. A. (1974) "Transvaal stromatolite: first evidence for thediversification of cells about 2.2 X 109 years ago," Science 183,514-516.

2. Macgregor, A. M. (1940) "A pre-Cambrian algal limestone inSouthern Rhodesia," Trans. Geol. Soc. S. Afr. 43, 9-16.

3. Macgregor, A. M. (1947) "An outline of the geological history ofSouthern Rhodesia," South. Rhod. Geol. Surv. Bull. 38, 73 p.

4. Bickle, M. J., Martin, A. & Nisbet, E. G. (1975) "Basaltic andperidotitic komatiites and stromatolites above a basal unconfor-mity in the Belingwe greenstone belt, Rhodesia," Earth Planet.Sci. Lett. 27,155-162.

5. Schopf, J. W., Oehler, D. Z., Horodyski, R. J. & Kvenvolden, K.A. (1971) "Biogenicity and significance of the oldest knownstromatolites," J. Paleontol. 45, 477-485.

6. Nicolaysen, L. 0. (1962) "Stratigraphic interpretations of agemeasurements in Southern Africa," in Petrologic Studies: a

Chemistry: Nagy and Zumberge

2976 Chemistry: Nagy and Zumberge

Volume in Honor of A. F. Buddington, eds. Engel, A. E. J.,James, H. L. & Leonard, B. F. (Geological Society of America,New York), pp. 569-598.

7. Van Niekerk, C. B. & Burger, A. J. (1964) "The age of the Ven-tersdorp System," Ann. Geol. Surv. S. Afr. 3, 75-86.

8. Young, R. B. (1933) "Conditions of deposition of the Dolomite

Series," Trans. Geol. Soc. S. Afr. 36,121-135.9. Button, A. (1973) "Algal stromatolites of the early Proterozoic

Wolkberg Group, Transvaal Sequence," J. Sediment. Petrol. 43,160-167.

10. Truswell, J. F. & Eriksson, K. A. (1972) "The morphology ofstromatolites from the Transvaal dolomite North-West ofJohannesburg, South Africa," Trans. Geol. Soc. S. Aft. 75, 99-110.

11. Zumberge, J. E. & Nagy, B. (1975) "Alkyl substituted cyclic ethersin 2,300 Myr old Transvaal stromatolite," Nature 255, 695-696.

12. MacGregor, J. M., Truswell, J. F. & Eriksson, K. A. (1974) "Fil-amentous algae from the 2300 m.y. old Transvaal dolomite,"

Proc. Nati. Acad. Sci. USA 73 (1976)

Nature 247,538-539.13. Peat, C. J. & Lloyd, B. J. (1974) "Direct observation of rock ma-

cerates," Nature 251, 294-295.14. Core Laboratory, Dallas, Porosity and Permeability Determi-

nations for The University of Arizona, Report SCAL-76154(1976).

15. Holt, S. C. & Leadbetter, E. R. (1969) "Comparative ultrastruc-ture of selected aerobic spore-forming bacteria: a freeze-etchedstudy," Bacteriol. Rev. 2, 346-378.

16. Bathurst, R. G. C. (1974) "Marine diagenesis of shallow watercalcium carbonate sediments," Annu. Rev. Earth Planet. Sd. 2,257-274.

17. Krauskopf, K. B. (1967) in Introduction to Geochemistry(McGraw-Hill Book Co., New York), pp. 66-68.

18. Bauchop, T., Clarke, T. J. & Mewhook, J. C. (1975) "Scanningelectron microscope study of bacteria associated with rumenepithelium of sheep," Appl. Microbiol. 30, 668-675.

19. Desikachary, T. V. (1959) in Cyanophyta (Indian Council ofAgricultural Res., New Delhi), p. 686.