Paleoecology of South African Australopithecines: Taung Revisited'

CURRENT ANTHROPOLOGY Vol. 15, No. 4, December 1974 ) 1974 by The Wenner-GrenFoundation for AnthropologicalResearch

RECENT THINKING ON HUMAN EVOLUTION

Paleoecology of South African Australopithecines:Taung Revisited'

byKarl W. Butzer

INTRODUCTION to a gracilelineage of earlyhominids. Almost a half-century later, the continuingspate of hominid discoveriesin East The Taung2 skull was the first australopithecine fossil Africa and the problems they have generated lend ever discovered and, as Australopithecusafricanus, gave its name greater pertinence to the South African type fossils and type sites.3 Yet the key Taung specimen has not been 'The field and laboratorystudies reflectedin this paper were published in full, and the geological context has been made possible by support from the Wenner-Gren Foundation establishedin a broad way only. The stereotypedgeneral- (grant-in-aid2344), the National Science Foundation (grant GS- izations about Taung found in the secondary literature 3013 to R. G. Klein and K. W. Butzer), and the Anthropology have remained unchanged for several decades. In fact, and Geography Departments of the Universityof Chicago. D. M. Helgren (Toronto) assisted with the laboratoryanalyses and however, the stratigraphicage commonlyassigned to the provided substantialnew informationand useful criticismon the Taung site must be challenged, thus raising new questions basis of his ongoing field research. Dr. and Mrs. G. J. Fock about the positionof theTaung childin hominidphylogeny. (McGregor Memorial Museum, Kimberley)and Mr. and Mrs. L. Ultimatelyperhaps of greater importance than relative Matter (Ulco) generouslyprovided the hospitality,assistance, and encouragement that made the fieldworkpossible. Maps and dia- age and basic taxonomyare the ecological implicationsof grams were drawn by C. Mueller-Wille (Chicago). The X-ray the Taung site, located at the highest latitude (27?32'S) diffractogramswere run and interpreted by Romano Rinaldi of any australopithecinesite and on the very edge of the (Chicago), while B. E. Bowen and Thure Cerling (Iowa State) Kalahari Desert. Here, too, recent work indicates a need examined four of the sand samples by electron microscope. The Grootkloofsnails were identifiedand commented on by A. Zilch for fundamental revision of prevailing concepts. Such a (Forschungs-InstitutSenckenberg), while C 14 determinationswere revisionbecomes particularlyurgent at a time when paleo- promptlyprovided by Robert Stuckenrath(Smithsonian Institu- biological and paleobehavioral theorists have begun to tion). P. V. Tobias and M. E. Marker (Witwatersrand)offered explore a number of ecological hypothesesto extensive and detailed suggestions on an interim draft of the help explain diverse aspects of early hominid differentiationor the specificevolution of the hominine line. Both dietary and locomotory adaptations play a central role in several of KARL W. BUTZER is Professorof Anthropologyand Geography these hypotheses. at the Universityof Chicago (Chicago, Ill. 60637, U.S.A.). Born in 1934, he was educated at McGill University(B.Sc., 1954; Whereas the fossilsthemselves are at long last proving M.Sc., 1955) and at RheinischeFriedrich-Wilhelms-Universitait, capable of yieldingdirect information on masticatoryactivi- Bonn (Dr.rer.nat., 1957). He was Assistant and Associate ties and locomotorymechanisms, environmental informa- Professorof Geography at the Universityof Wisconsin from 1959 to 1966. He has done extensivefieldwork in Egypt,Spain, Ethiopia, and South Africa. His research interestsare Pleisto- manuscript.I appreciate discussions with C. K. Brain (Transvaal cene environmentsand stratigraphy,ecological adaptations of Museum), R. Liversidge (McGregor Memorial Museum, Kimber- early hominids, and prehistoricman-land rel.Itionships.He is ley), R. G. Klein (Universityof Washington),and D. C. Johanson author or co-author of Die NaturlandschaftAgyptens (Mainz, (Case Western), as well as the unforgettableintroduction to the 1959); QuaternaryStratigraphy and Climatein theNear East (Bonn, Taung juvenile given me by Phillip Tobias in 1969. 1968); Environmentand Archeology(Chicago: Aldine, 1964; 2Taung is Tswana for 'place of the lion,' composed of tau 'lion' Chicago: Aldine-Atherton/London:Methuen, 1971); Desertand and ng 'place of,' as I am informed by D. T. Cole and P. V. Riverin Nubia (Madison: Universityof WisconsinPress, 1968); Tobias. It also means 'place of Tau,' a legendary 18th-century and RecentHistory of an EthiopianDelta (Universityof Chicago chief (Breutz 1968:15). Since the totem of the local Tlhaping Department of Geography, 1971). He is co-editor of Afterthe tribe is the kudu (Breutz 1968:99), the latter interpretationis Australopithecines(The Hague: Mouton, in press). probably correct. The common spelling Ta'ung is therefore as The present paper, submittedin final form 15 iII 73, was incorrectas the obsolete toponymTaungs. sent for comment,along withthe papers of Tuttle, Todd and 3 As became particularlyapparent at the Wenner-GrenFounda- Blumenberg, and Porshnev that follow, to 50 scholars. The tion workshop "Fossil Hominids fromEast Rudolf, Kenya," orga- responses to all the papers are printedbelow and are followed nized by G. LI. Isaac and R. E. Leakey, New York, February 9, by renlies fromthe authors 1973. See also Tobias (1972).

Vol. 15 No. 4 December1974 367

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions tion has traditionallybeen based on associated faunal Norlim.5This specimen was to be published by Dart (1926) assemblages and sedimentinterpretation. Realistic ecologi- as Australopithecusafricanus. The Buxton-Norlimlime pre- cal models will ultimatelyneed to be the product of an cipitates-"calc-tufas" or "travertines"6-from which the interdisciplinarysynthesis, employing the input of critically fossil was derived had first been briefly described by evaluated data processed by the diverse methodologies of Wybergh (1920). In view of the interestdue to the aus- the broad spectrumof scientistswith paleoanthropological tralopithecinediscovery, Young subsequentlyexamined a focus or avocation. Thus far there has probably been a number of tufas at and near Buxton-Norlim,describing littletoo much assertionof apparent interrelationshipsthat occurrences, recognizing but a single generation of tufa are based on an incompleteappreciation of the complexity precipitates,discussing their mode of origin (in part by and conditionaltestimony of data fromunfamiliar subdisci- analogy to the Brazilian study of Branner [1911]), and plines. Despite the continuing compartmentalizationof inferringa single cycle of accentuated fluvialactivity (with researchinstitutions and fundingagencies, individuals must deposition of tufas interbedded with coarse detritus),fol- striveto communicateto the degree thatthey can appreciate lowed by uninterrupted tufa accretion, and ultimately that no data of any category is whollyunequivocal. Thus fluvialerosion. Young also provided a pertinentdescription a late Tertiaryfaunal assemblage cannot be simplyequated of the cave fillin which the australopithecinewas found. witha specificvegetation and ecozone, nor can a sediment A detailed study of the Buxton-Norlimarea was finally necessarilyyield a reliable prognosis on either macro- or undertakenin 1948-49 by Peabody (1954), who recognized microdepositionalenvironments. at least four major generationsof tufas or travertinesand No paleoecological generalizations-as opposed to limited establisheda complex geomorphologic sequence. Peabody site-specificinferences-relevant to Tertiary and Pleisto- demonstrated facies patterning within the older tufas, cene hominids are established beyond reasonable doubt. mapped the occurrences at approximately 1:10,000, and No matter how quantitativea contributingsubdiscipline reconstructedthe appearance of the area priorto quarrying may aspire to be, the compendium of paleoanthropological by means of a panorama. A varietyof fossiliferousassem- sciences is necessarily an inexact one by virtue of the blages and archeological occurrences were collected, al- multivariatecharacter of interactingorganic and inorganic thoughwith exception of the australopithecinecave proper phenomena in nature. As scientistswe remain obligated (Cooke 1963; also Wells 1969 and Freedman and Stenhouse to continue to question the assumptions underlying our 1972), the faunas have been incompletelypublished, and methodologies,and search for new ways to achieve fuller only one archeological site (Witkrans; see Peabody 1954) and more accurate resolution.The techniquesof the behav- has been systematicallyanalyzed (Clark 1971). Peabody ioral, biological, or geological subdisciplines relevant to suggesteda climaticinterpretation for Buxton-Norlimand anthropologyare never infallibleand must thereforere- other nearby tufas and then offered climato-stratigraphic main flexible. Similarly,the body of general information correlationswith the controversialVaal River sequence. must never be elevated to the status of dogma, whether Despite the valuable and fundamental work of Young it be from our own or a cognate subdiscipline with data and Peabody, both the stratigraphicage and paleoclimatic we are less able to evaluate critically. implicationsof the older Buxton-Norlimdeposits remain The potentialcontributions of the earth sciences to both uncertain (see also Cooke 1967:176). It requires little data input and synthesisin paleoanthropologyare consid- emphasisthat this state of affairsis unfortunate,particularly erable, if for no other reason than that a paleo-physical at a time when the relative ages of the South African geography is indispensable in both paleoprimatologyand australopithecinebreccias-with respect to each other and prehistory.But interpretationin geomorphology, sedi- to the East African deposits-are of as crucial interestas mentology,and geological stratigraphyis no less difficult the possible but as yet unproven ecological differentiation than is the isolation of specificvariables in any ecological of the key australopithecinelineages (see Butzer 1971 a: system.Sophistication in data analysis and synthesisdoes chap. 25). improve with time, even if not necessarily in a linear In the course of fieldworkin the Vaal and Harts valleys progression. Older studies, no matterhow good for their during 1970 and 1971, I made several trips to different time or how venerable or beloved their authors, must parts of the Union Lime Co. concession at Ulco. I also periodicallybe reconsidered.Our appreciationof the South visitedBuxton-Norlim and the nearbyquarries at Thoming, Africanaustralopithecines can be no exception to this law togetherwith C. K*.Brain and R. Liversidge.Fresh perspec- of disciplinarysurvival. tives were provided by an aircraftreconnaissance in 1971 Prior to the discoveryof australopithecinesin Tanzania, and furtherfieldwork in 1973. Finally, renewed studies Ethiopia, and Kenya, the detailed geological studies of Peabody (1954) and Brain (1958) in South Africasuggested 5The older name of Buxton has been changed to Norlim, a strongsupport for an explicit ecological differentiationof contractionof NorthernLime Company. To simplifyidentification the gracile and robust lineages. In an earlier paper, I and to avoid confusion with a specific local tufa designated as suggested that the patternsinferred were less than proven Norlim, the double name Buxton-Norlimis probably preferable. (Butzer 1971 b). The present paper attemptsto reconsider 6There is no generallyaccepted terminologyto describe sheets of lime precipitates."Travertine" best refersto dense, horizontal, the case of the Taung australopithecine.The results are or undulating bands of calcite, with successive increments of equally far from conclusive, but they may help clarifya macrocrystalline,columnar structurealternating with the more numberof pointsand suggestavenues forfurther research. common laminated, nonporous to semiporous, cryptocrystalline calcite. Many authors prefer to limit this term further,to cave flowstonesand dripstones. The term "tufa," on the other hand, has often been reserved for calcite precipitatedon and between GENERAL BACKGROUND growing or decaying plants; the primary accretions commonly forma spongy, cellular, and somewhat brittlerock, while porous In November, 1924, Young (1926) recognized a primate calcitesands tend to fillthe voids betweenthe verticalor horizontal organic structures(Butzer and Hansen 1968:355). Such organic skul14 recentlyuncovered by quarry operations at Buxton- tufasare, nonetheless,more often than notembedded withinsheets of banded precipitates,and a numberof basic studies(e.g., Branner 4The first partial skulls of small fossil baboons had already 1911, Gardner 1932) have used "tufa"to designate mixed, external been discoveredin 1920 and described in an unpublished address facies of this kind. I consequently suggest that dripstone and to the Royal Society of South Africa by S. H. Haughton in the flowstonebe used to describe carbonate precipitatesin caves, while same year(Tobias 1973a). Relationshipsof the fauna to the hominid tufa be employed for external deposits, including both organic are discussed furtherbelow. tufas and travertinessensu strictu as facies.

368 C U R R E N T A N T H R O P O L O G Y

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions by D. M. Helgren are under way. Butzer:SOUTH AFRICAN AUSTRALOPITHECINES Sediment samples collected in both areas were primarily analyzed in the of the Paleo-Ecology Laboratory University GROOTKLOOF of Chicago. The results of this work underscore that correlations of the various escarpment tufas with other RecentAlluvial presumed sequences of geomorphologicevents in southern Wash 117m Africa were premature. However, the tufa generations at widelyseparated points of the escarpmentare remarkably comparable and do suggest that a detailed and systematic studycould establisha firmregional stratigraphy that might possiblybe linkeddirectly to erosional surfacesand calcrete deposits found to the west of the Vaal and Harts rivers. Furthermore,the sediment analyses provide new insights TufaIll Tufa 11 as to paleoclimatic interpretation,in particular for the matrixof the Taung australopithecine.The present paper attemptsto illustratethe mechanismsof tufa accretionand destructionon the basis of the small-scale but exemplary Breccia development of a single locality, the Grootkloof. These Terraces resultsare then applied to a revisionof the classic Buxton- GrootkloofCave Norlim sequence. BreachedDoline

Old GAAP urn FossilSites Alluvial ESCARPMENT H~~art$ Depositss THE GAAP ESCARPMENT euxiun Thuming Tufa I The Buxton-Norlimand other key tufas of the northern Cape Province are found along the Campbell Rand (from Afrikaans rand 'ridge, rim') or, as it is now more usually GrootloUoI 1066m termed, the Gaap (=Kaap) Escarpment. This prominent ealoni s Klootf. scarp, averaging approximately100 m in reliefand 1,150- KIMBERLEY 1,200 m in elevation, extends some km from 280 27?7'S 0 250 near Vryburg to 29?7'S near Douglas, trending from northeast to southwest. The cliffed escarpment (40-70?) m Gaap Plateau Gaap m and the generally flat plateau (less than behind it 1200 Escarpmtnt 1200 0.5?) Karoo are formed of gently warped, Precambrian dolomites of P 1100 the Transvaal System.Small facetson the scarp face reflect minor lithologicalcontrasts in the gray dolomites and the dark gray shaley interbeds (see also DuToit 1907, Young 1926) thatare locally importantas impermeable aquiludes, controllingspring lines and cavern development.The scarp rim is angular, the footslopemore oftensmoothly concave, averaging25? as itdescends onto a rock-cutplain descending to the valley margins of the Vaal and Harts with average r noff.Simil sar mulinticycl-icxandso foinmrpooyhv gradientsof less than 0.5?. surfaces(below).ulticyclic sands ofeolianmorphologyahav The Gaap Escarpment is an exhumed, early Paleozoic erosional feature,subsequently buried under Karroo sediments during the late Paleozoic and reexposed by late Tertiaryplanation in the Vaal Basin (DuToit 1910; Stratten found their way into most of the tufas, both at the time 1968; D. M. Helgren, personal communication).Although of primaryaccretion and during later periods of corrosion and this relativelyrecent erosion has cut across outcrops of redeposition.(b) Karsticerosion of the dolomiteupland underlyingPrecambrian dolomites, quartzites, and slates is relativelyinsignificant, being limited to restricteddevel- (see DuToit 1907-8), Karroo sedimentscommonly extend opment of underground caverns, with associated solution into proximityof the escarpment,so that there can have shafts(ponors) and superficialmicrofluting (lapies). Dolines been littlesignificant scarp retreatin Cenozoic times (see are rare, at least in scarp proximity.Nonetheless, surface fig. 1). Consequently, although the major drainage lines drainage occurs only after protracted,heavy rains, when have continued to incise short,V-shaped gorges along the all caverns and fissuresare periodicallyfilled with vadose edge of the escarpment,tufas could have formed at any water. At that point streams "rise in flood, descending timesince the Karroo coverwas strippedoff in Mio-Pliocene in raging torrents from the plateau" (Young 1926:55). (and later) times (see King 1968:272, fig. 119). Activechannel erosion is confinedto such sporadic floods, Drainage is primarilycontrolled by surficialsediments, although spring sapping continues through much of the subsurfacelithology, and climate.(a) Water-workedquartz year (see also Young 1926). Most of the effectiverainfall sands are locally prominent in the shallow, intermittent is, in fact,converted into spring discharge, primarilysea- watercourses above the escarpment. They appear to be sonal in nature, both above the scarp rim and along the derived by wind and water action from the thick and aquiludes exposed in the escarpmentitself. (c) Precipitation extensive Kalahari cover sands (see Grove 1969) further in scarp proximitytoday appears to increase from 300 west,7and they absorb a high percentage of any initial

appear to be rare, of limited extent, and irregularlydistributed 7 However, on a recent reconnaissance of the Gaap Plateau to on the plateau, except forthe area immediatelysouth of beyond Kuruman. Kuruman and Griquatown, D. M. Helgren (personal Thus the derivation of eolian components in the older Gaap communication)failed to find widespread interveningmantles of deposits remains problematical.For modern prevailingand effec- "Kalahari Sand." Free-standingbodies of sand or sandy soil mantles tive wind directions,see Climateof SouthAfrica (1960). Vol. 15 * No. 4 * December1974 369

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions mm at the latitude of Douglas to 375 mm near Ulco and In addition to a sample fromthe matrixadhering directly 450 mm near Buxton-Norlim.8By the Koeppen classifica- to the Taung skull, 32 sediment samples were collected tion this yields a winter-dry,cool steppe climate (BSkw), from Ulco Grootkloof,nearby fossiliferousexposures, the by the Thornthwaitesystem a semiarid, mesothermalcli- Ulco Quarries, Buxton-Norlim,and alluvialfills of the Harts mate withno water surplus at any season (DB3 d) (Schulze and Dry Harts rivers at the Delportshope and Buxton 1958) and, theoretically,no runoffexcept of a local and bridges respectively.These samples cannot be considered sporadic type. This regime is reflected by the discharge representativeof the range of geological units,and modern statisticsof the Dry Harts at Taung, a large catchment counterpartswere barely sampled. However, the selection representativeof the northernmostescarpment with over is adequate to provide a more objective description of 450 mm precipitation;average annual flow 1923-45 was several depositional suites and to allow some informative 88 days, with a double maximum, in March (26%) and comparisons.After preliminary processing, 23 samples were November (19%), and a minimumin July(0.1%) (Welling- analyzed in some detail (for selected results,see table 1).13 ton 1955:374-76). Runoff is restrictedfar more by me- Two problems of assessing the significanceof eolian chanical percolationand evaporationthan by the incomplete components warrantfurther discussion here: and degraded vegetativemat. The "thornveld"of the Gaap 1. Almost every sample has a textural maximum in the Escarpment and its forezones is an acacia scrub-parkland finesand grade (60-200 ,u) witha secondaryclay maximum with a subcontinuous grass cover.9 Where protected, as (under 2 ,u). This bimodal distributionunderscores the throughthe effortsof L. Mattersouth of Ulco, a dry thorn mixed derivationor transportof the sedimentand renders woodland maydevelop, but torrentialrunoff and sheetwash difficultstatistical manipulation by available techniques (see remain characteristic. Folk 1966). An additional complication is the nature of The escarpment is well developed south of the Ulco the clay component, including clay minerals as well as Quarries, where reliefattains 90-120 M.'" Structuralridges silica-ranging the fullspectrum from crystalline to mobile, (hogbacks and cuestas) and outliers,amid local planation amorphous forms. Titration and filtrationafter boiling in surfaces above the rim, create an additional local relief acid removed only an arbitrarypart of this silica, and of 20-50 m, yielding an undulating topography marked rigorous statisticaltechniques cannot be applied to this by smooth inflectionsof slope. At the Ulco Quarries the mixed suspended and colloidallydissolved component. As scarp is completelyobscured by a complex tufa fan some a result, the sieve components above 37 ,u were treated 4 km wide and 1.3 km deep, while furthernorth, smaller independently,allowing discrimination in termsof median tufaaprons continueto maskthe cliffs.The Ulco Grootkloof grades and sortingand skewnesscoefficients. In the present (Afrikaans for 'great cleft') lies 3.3 km southwest of the context,Trask sortingvalues under 1.40 are "good," 1.40- quarry fan. Another major fan complex, Malony's Kloof," 1.90 "moderate," and greater than 1.90 "poor." Good projects from the scarp another 2.8 km farther south. sortingcan be expected with eolian and constant-velocity Drainage above the escarpment is generally shallow, with fluvialtransport; poor sortingis typicalof colluvial sheet- repeated divergence and convergence amid various struc- wash or the torrentialdeposits of intermittentstreams. turallineaments until the scarp is reached. Here Grootkloof Optimal sortingwill occur witheolian derivativesdominant is the most conspicuous erosional feature,although similar in fluvialdeposits because of a persistentmoderate-velocity deep valleys are presumably found under the tufa fans current.Trask skewnessvalues above 1.00 and below 0.60 at Ulco Quarries and Malony's Kloof. Below the scarp, reflect asymmetrical grain-size distributions and imply drainage once again dissipates and diverges into multiple mixed sand origins. In particular, higher values have a rills that run perpendicular to the Vaal-Harts system. "tail" of coarser sands, while lower values have a "tail" of silt,both typesof skewnesscommonly reflecting colluvial redistribution. LABORATORY TECHNIQUES EMPLOYED 2. An established procedure that has also been applied to South Africa (see Bond 1954, Brain 1958) claims to The Gaap Escarpment now has two major agencies of sedimenttransfer and deposition: water-both surfaceand spring-and wind. The '3 The analysisconsisted of the followingsteps: (a) Decalcification primarypurpose of applying sedi- in 15% hydrochloricacid measured calcium carbonate content ment analysesis to attemptto differentiatebetween fluvial, for bulk samples as well as provided residues suitable for further spring, and eolian components in the geological record. study.Since the rare dolomite sand grains do not appear to have The analytic techniques offer no conclusive answers, but been affected by cold acid, the resulting sand fraction remains instead providecriteria for representative.(b) Wet-sievingof the residues remaining after differentiatingbetween deposits decalcificationand removal of the mobilized amorphous silica and allow comparisons with modern "standards," where provided texturalinformation on sand-sized materials. Standard- available, collected from specific depositional environ- mesh sieves (37, 63, 210, and 595 ,u, 2.0 and 6.35 mm) were ments.12 employed. Textures quoted refer to the noncarbonate, detrital residue (i.e., after removal of the mobile silica) and follow the textural classificationof Link (1966). Table 1 gives the textural 8 componentsgreater than 37 i,uafter exclusion of the finergrades Generalized from over 50 stations with records in excess of in order to apply Trask's coefficients of size sorting, So 25 years (Climateof SouthAfrica 1954). These values differappre- = V'Q, / Q3, and skewness,Sk = Q, Q3 /(Md),2 whereQ, is the ciablyfrom the 250-450 mm range estimatedby Clark (1971:1223) firstquartile, Md the median, and Q3 the third quartile of the from precipitation maps, or the patentlyerroneous value of 150 cumulativetextural curve, with > Q3 . (c) Hydrometer mm long quoted for Buxton-Norlim. QI analyses, using a 5% solution of sodium pyrophosphateas peptizing agent, 9 For a detailed and accessible outline of modern vegetation were applied to the Harts and Dry Harts suites and to the communities,based on the work of J. P. H. Acocks, see Clark Australopithecusmatrix (eight samples) to differentiatethe clay (1971:1226). and silt components into four classes: under 2 ,u, 2-6 ,u, 6-20 0 See the South African 1:50,000 topographic series,map 2824 ,u, and 20-60 ,u. X-ray diffractogramanalyses were also carried AC ("Ulco"), as well as the airphotos. out on the matrixsample. (d) pH values and electricalresistance " This sectoris perhaps betterknown as Gorrokop or Gerrakop. (mV) were determined for the ten friable samples. The former 12 General discussionsof sedimentaryprocesses and analyses are were generally alkaline and ranged from 7.0 to 8.9, the latter availableelsewhere (e.g., Butzer 1971 a), and mostof the techniques varied but little,from +80 to + 150. (e) Sand-sized residues were and nomenclature employed here are either outlined or docu- microscopicallyscanned for minerals other than quartz, with mented in Butzer and Hansen (1968:13-16, 461-63), Pettijohn semiquantitativeestimates of quartz-grainrounding and frosting. (1957:30-38), and Soil Survey Staff(1960:30-33 with references, In addition, four sand samples were examined under the scanning appendix). electron microscope.

370 C U R R E N T A N T H R O P O L O G Y

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions differentiatequartz sands on the basis of shape (degree Butzer: SOUTH AFRICAN AUSTRALOPITHECINES of rounding) and surface texture (frostingversus polish). In effect,eolian sand grains are supposed to be rounded electron microscope elucidates the historyof sand grains, and frosted(the frostingbeing due to microscopic impact not of the sediment in which they come to rest. There scars),while river-transported grains have a luminous polish are, then, no simple criteria for evaluating the eolian and are less symmetricallyrounded (see Cailleux and Tricart "appearance" of a sand grain. This is particularlytrue in 1963:53-102; compare Shepard and Young 1961). the case of the deep sands of the Kalahari Basin. Initially However, Kuenen and Perdok (1962) have shown thatsuch probablyof fluvialor marine origin,these sands have been interpretations,based on low-magnificationanalyses, are repeatedly reworked by various processes, primarilyby facile and of dubious quantitativeapplication.'4 Far more wind. Polycyclicsands of thistype together with more local, conclusiveare the resultsobtained fromthe electronmicro- residual types have been successivelyintroduced into the scope at veryhigh magnificationsthat allow differentiation Harts and Vaal drainage by fluvialas well as eolian forces of various distinctivesurface textures, many of them pecu- (see Van Rooyen 1971: chap. 5). In general,these suspected liar to certain environments (Krinsley and Doornkamp "eolian" components of "Kalahari" type are subrounded 1973). Unfortunately,this complex and expensive tech- to rounded, eitherfrosted or polished,and partlydiscolored nique is also not foolproof.Of four samples analyzed from bymicroscopic skins of ferricoxide. The degree of rounding the Gaap Escarpment, one (Ulco Quarry) showed quartz slowlyincreases with transport, while frosting may be added

TABLE 1

SELECTED SEDIMENT DATA FROM THE GAAP ESCARPMENT AND HARTS DRAINAGE

RESIDUAL TEXTURAL COMPONENTS ABOVE 37,u

FIRST THIRD CaCO3 QUARTILE MEDIAN QUARTILE TEXTURAL UNIT (%) (,U) (,U) (,U) SORTING SKEWNESS CLASSIFICATION

Harts River(Delportshope Bridge) Flood Silts 12.1 125 89 54 1.52 0.85 Clayey-sandsilt Younger Sandy Marl 64.4 145 101 80 1.34 1.03 Sandy-claysilt Basal Silts 20.4 115 87 54 1.49 0.82 Clayey-sandsilt Dry Harts River(Buxton Bridge) 5.0 m Terrace 0.4 120 68 46 1.62 1.19 Sandy-claysilt 8.5 m Terrace (2) 0.9 340 180 107 1.79 1.12 Siltysand 8.5 m Terrace (1) 1.1 170 113 80 1.49 1.07 Silty-claysand Younger Sandy Marl 54.6 186 124 92 1.42 0.87 Silty-claysand Ulco Grootkloof Cave Floor 46.0 145 105 71 1.43 0.85 Clayey-siltsand Tufa III 99.6 145 110 88 1.28 1.05 Sandy-siltclay Breccia II 87.9 190 118 71 1.63 0.97 Silty-claysand Tufa II 86.3 145 110 85 1.31 1.02 Clayey sand Breccia I 86.3 1020 180 105 3.12 3.31 Clayey sand Fossil Beds (2) 60.5 - - - - - Fossil Beds (1) 81.9 145 105 85 1.31 1.02 Sand Tufa I 97.9 125 80 51 1.56 1.00 Silty-sandclay Ulco Quarry Fissure Fill 42.9 162 112 89 1.35 1.15 Sand Buxton-Norlim Channel Alluvium 50.0 162 115 88 1.36 1.08 Siltysand Equus Cave Fill 37.1 145 102 70 1.44 0.89 Silty-claysand Oxland Apron 96.4 145 106 70 1.44 0.83 Silty-sandclay Norlim Carapace 98.4 145 95 59 1.57 0.95 Clayey-siltsand Wet-Phase Cave Filla 88.4 118 82 53 1.50 0.93 Clayey silt Dry-Phase Cave Fill 81.3 145 115 84 1.32 0.92 Sand Thabaseek Carapace 99.1 115 66 48 1.55 1.27 Clayey-siltsand aTaung fossilmatrix sands of glacial modification (Permocarboniferous Ice or eliminated; the oxide coatings appear to be commonly Age!), another(Buxton Equuscave fill)was multicycleeolian, removed in highlyalkaline lakes or afterappreciable river a third (Buxton dry-phase cave fill) fluvial with eolian transport.Altogether this allows no more than a subjective overprinting,and the fourth (Buxton wet-phase cave fill) evaluationof theexotic sand grains,although the authigenic a mixture of all these agencies. Yet all four samples quartz sands of the local dolomite are fortunatelyquite represented colluvial cave or fissure infillings.Thus the distinctive:angular, glossy,and fine-grained.

14 Experimental work indicated that "eolian" frosting was impossibleto reproduce bymechanical abrasion, but must be largely TUFAS OF GROOTKLOOF attributedto chemical micropitting,while in the case of fluvial sands chemical attackproduces eithertotal frostingor total polish, depending on the chemicalproperties of the particularwater body. DESCRIPrION These conclusions were confirmed by a comprehensive study of over 14,000 sand grains fromdifferent past depositional environ- Tufa I, the oldest preserved, is a steeply dipping body ments in Egypt (Butzer and Gladfelter1968). thatextends over 80 m fromthe scarp rim to the footslope Vol. 15 No. 4 Decemnber1974 371

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions (fig.2). Contactswith the dolomiteare irregular,corroded, ? 620 B.P. (SI-1301) above, indicatinga mid-Upper Pleisto- and marked by subrounded to rounded, cobble conglom- cene age.'5 erates bedded into former fissures. Beds are primarily A phase of erosion followed, removing part of Tufas foresetat 45-65? and constitutea "carapace" (in the termi- I and II, while leading to some scarp retreatof the dolomite nology of Peabody 1954). All these deposits are well walls of the kloof. indurated,with extensive recrystallization and typicalwavy Breccia Terrace I, 2-3 m thick, is a crudely stratified "banding" of variable crystallinityand discoloration.Mate- rubble of grit, pebbles, cobbles, and boulders, generally rials range from mesocrystalline,white (10 YR), clean angular to subangular, and consistingprimarily of uncor- travertinesto cryptocrystalline,pink (5-7.5 YR), oxide-rich roded dolomite, with some blocks derived from Tufas I precipitates,interbedded with manganese-enriched laminae and II. The matrixis a very pale brown (10 YR), poorly of lightgray (10 YR) color. (All colors cited were obtained sorted and heavily skewed clayey sand, subsequently cal- fromthe MunsellColor Charts for dry samples.j Amorphous creted (Tufa III phase) and even later dissected to form silica amounts to 17% of the noncarbonate residue, while the 6-7 m terraceof today. The sand component includes the detrital residue is a silty-sandclay, with the above-37 both abundant shale/slate and traces of black chert from ,i component moderately sorted. Except for a trace of the dolomite, while the quartz sands range from angular shale/slate,the sands are angular to subrounded, polished to rounded, withsome typicalfrosted, oxide-coated grains quartz. Althoughundoubtedly badly weatheredtoday, Tufa of eolian type.Some diffuseas well as macroscopicorganics I includes primaryinterbeds of clayey,reddish soil deriva- are present,but littleamorphous silica (2%). tives.The sands include a high proportionwith ferric oxide Tufa III is another restrictedcarapace withinthe gorge, skinsotherwise characteristic of eolian "Kalahari" sands. fallingsome 50 m withbedding dips of 40-70?. The material

m GROOTKLOOF

-100 TufaIII Tufa 11 140-70?dips) 140-40? dips)

0 T 6merraceI

Fl,,, Stream no horizontalscale

FIG. 2. Cross-sectionof the Ulco Grootkloof.

The Gaap Escarpment has suffered next to no retreat is a wavy,banded to laminated,cryptocrystalline precipitate, since deposition of Tufa I, and a valley-likeincision at varyingin color fromvery pale brown (10 YR), light gray the kloof preceded initial accumulation. Possibly con- (10 YR), to pinkish gray (7.5 YR). Although the tufa is temporary is a hanging valley at 70-75 m immediately cemented, recrystallizationis uncommon, and diffuse or- northof the kloof(fig. 1). During a long,subsequent interval ganic matter is evident. The noncarbonate residue is a of erosion, the kloof was deeply incised and elongated well-sortedbut skewed sandy-siltclay, with an angular to to approximatelyits present configuration.Several stages subrounded sand component of quartz, predominantly of dissectionare indicated by conspicuous steps at 20 and frosted,and a trace of shale/slate. The cumulativetextural 35 m above the modern klooffloor (fig. 2). Related deposits curve shows small but distinct"tails" of coarse sand and and any intermediatetufa generations,such as are repre- of silt. Hygrophyticsalts are indicated, but little or no sented on the Gaap Escarpment opposite Douglas, appear amorphous silica. The C 14 date of 20,825 ? 230 B.P. to have been completelyremoved. (SI-1640) suggestsa late Upper Pleistocene age. Tufa IIis relativelyminor, falling some 35 m, with40-70? Another period of erosion followed,with 2-3 m down- dips (fig. 2). The material is an impure, cryptocrystalline, wavy-bedded,light gray to verypale brown (10 YR) precipitate,cemented and withevidence of some secondaryrecrys- '5Samples were washed in 0.5 NHCI before CO2 evolution in H 3 PO 4 (Robert Stuckenrath,personal communication).Although tallizationand 18% amorphous silica in the noncarbonate "dead" carbonates are indisputablyavailable it is equally pertinent residue. The detrital residue is a well-sortedbut skewed that fresh atmosphericcarbon dioxide would have been incorpo- clayeysand, withhalf of the exclusivelyquartz sand compo- rated during the breakdown of the dolomite into a bicarbonate, nent frosted,subrounded to rounded, and oxide-coated. and again at the time these bicarbonates remained in solution; Samples of dense, similarly,younger carbonates have been repeatedlyor continually cryptocrystallinecalcite gave consistent added through postdepositionalre-solution. It is thereforeproba- C 1 ages of 30,760 ? 1,035 B.P. (SI-1639) below and 26,130 ble that the apparent C 14 age gives a true order of magnitude.

372 C U R R E N T A N T H R O P O L O G Y

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions cutting into Breccia I, destruction of much of Tufas II Butzer:SOUTH AFRICAN AUSTRALOPITHECINES and III, and some concomitantscarp retreat.This material accumulated at the bottomof the kloof as Breccia II. Hansen (1968: chap. 7). The Grootklooftufas are primarily Breccia TerraceII is a 2-m-thickmass of poorly stratified, foreset and well bedded, with few identifiable organic angular dolomite grit and rubble, with a noncarbonate impressions or structures,even though algal laminations residual matrix of light gray (10 YR), moderately sorted are probably an integralpart of the accretion process. In silty-sandclay as well as almost 25% amorphous silica. In contrast,the Kharga and Kurkur tufas are primarilyhori- addition to a moderate concentrationof diffuse organic zontal, occasionallycurrent-bedded, with crude basal detri- matter,there are numerous microimpressionsof rootlets tus generally followed by massive organic tufas, related or grass stems,as well as terrestrialsnails (Xerocerastussp.). to multiple spring vents, spring-fedpools, and vegetation The sand is rich in shale/slate, with the quartz predomi- thickets,and late'ral facies including pond marls rich in nantly frosted and angular to rounded. The matrix is plant castsand impressions.On the otherhand, the Catinga primarilyderived frombreakdown of local materials,with limestonesof northeasternBrazil include great valley-floor littleevidence of freshexternal sediment. lime sheets aggrading by sheetwash and stream discharge Cementationof Breccia II was completed during accre- (Branner 1911) in areas with 650-700 mm precipitation; tion of minor kloof-walltufa cascades (Tufa IV), rich in theirsimilarities with the Grootklooftufas are remarkable. organic matter.One such deposit has a C " age of 9,310 The breccia terracesare a differentmatter. They record ? 105 B.P. (SI-1641), indicatingan early Holocene age. thedebris produced during phases of erosion thatdestroyed A final phase of erosion was responsible for dissection great masses of the Grootklooftufas. Thus formationof of Breccia II which,in conjunctionwith a corroded bedrock breccias is antitheticalto tufa accretion. Similarly,the fine channel, now formsa 4 m terrace. residues of the breccia terracesare abnormallyrich in very Spring activityduring the dry season is today limited coarse sands and otherwise differentfrom those of the to a trickleof water over bedrock cascades into a plunge tufas (see table 1). Consequently, although the tufas are pool below. Contemporarysedimentation involves a grayish nowhere preserved directly overlying the breccias at brown (2.5 Y) sand (58% CaCO3), covered withabundant Grootkloof,they are temporallydistinct, although theyare algae and some mosses. The snails,identified as Xerocerastus not separated by appreciable time intervals. cf. psammophilus(O. Boettger), are widespread in the The breccia record of kloofcutting is highlyinformative, Kalahari Basin.'6 since the fresh,angular dolomite detritustestifies to either violent corrasion or accelerated mechanical weathering DISCUSSION ratherthan protractedcorrosion. The issue of mechanical abrasion versus weathering is resolved by the sediment The three generationsof tufaat Grootkloofreflect different record of the Grootkloofcave (see fig. 2), where the lower depositional trends: part is filled with angular dolomite blocks in a matrix I. A long period, initiated by crude basal detritus,but identical to that of the breccia terraces and overlain by mainly involvingslow accretion of pure, verticallycrystal- flowstonecascades (with Xerocerastussp.) that can best be lized travertinesand clayey, laminated tufa. The limited linked to Tufa III. In other words, the cave breccia is quantityof polished sands presumablyincludes both fluvial contemporarywith Breccia Terrace II. Yet the cave breccia and eolian components. Any initial organic impressions is not the result of torrentialfluvial activity,but is best and snail faunas have probably been destroyed through explained by cumulative roof collapse below the ponor repeated recrystallization. at the apex of thecave. Thus mechanicalweathering, almost II. A brief of phase accretionof sandy and clayey tufa, certainly frost-shatteringprompted by cold Pleistocene in unusuallylow carbonates and witha prominentcompo- climates,was primarilyresponsible. nent of derived or primaryeolian-type sands. If frost-weatheringis indeed invoked to explain breccia III. Another brief phase involvingslow accretion of a formationhere, a wintertemperature depression of at very laminated tufa,rich in siltand clay,with a trace of organic least 2.50 C must be inferred, since the mean monthly matter and salts. Some eolian-typesands are in evidence. minimumfor Julyat Kimberleyis -2.4? C (Climateof South Thus the three generations have points in common as Africa 1965) and only the modern 56-year record lows well as significantdifferences. All imply low to moderate of -5 to -7? C would&be adequate to produce the deep discharge velocities,but protractedperiods of spring and frost-wedgingimplied by block-by-blockdisintegration in stream flow. All indicate that infiltrationwaters-issuing highlyresistant, massive dolomite. from an underground, dolomitic honeycomb reservoir- Fissures or caverns filled with reddish sands are not were the primarycontributors of carbonates. Everything developed in the Grootklooftufas. The floor sediment of points to sustained rains capable of maintaininga high the Grootkloofcave is shallow, dark brown (7.5-10 YR), phreaticwatertable in the dolomiteand abundant quantities calcareous, organic, and loamy. It is identicalto the much of vadose waterin general. Similarly,the upland vegetative thickerfloor fillof dark grayishbrown (10 YR), clayey-silt mat must have been relativelycomplete to ensure adequate sand (table 1) in a cave that opens on the breached doline percolationand effectivelyinhibit surface runoff. Nonethe- just south of Grootkloof.Here, hyrax and some bat dung, less, the ratio of eolian-type sands changes with each as well as baboon generation, as do the coarser detrital components (clay feces, contributea high proportion of the sediment. However, the detrital residue of the cave sand -- silt) and the nature of (perennial -- sporadic fill is well seasonal?) fluvialactivity. sorted, with a grade maximum near 100 ,u. Despite superficial resemblances, the Ulco Grootkloof The almost exclusivelyquartz sand is angular to rounded, with a tufasare differentfrom the escarpmenttufas of the Libyan moderatelyhigh proportion of oxide skins. These Desert Plateaus originallydescribed fromthe Kharga Oasis obvious eolian components, which constitute some 20% by Gardner (1932; also in Caton-Thompson 1952:1-14) by total weight, must be obtained primarilyfrom sandy and studied in greater detail at Kurkur by Butzer and upland soils that are washed in through fissures,with a small part introduced organically. The cave entrance is sheltered from blowing sand, and silt-sizedcomponents are almost absent. "6Littleis knownabout the ecological preferencesof thisarchaic Decomposition of the organic matter genus, which has been recovered from a number of tufa occur- could ultimatelyproduce a reddish sandy sediment similar rences (A. Zilch, personal communication). to those found in older fissuresand caverns at the Ulco

Vol. 15 No. 4 December1974 373

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions Quarries and at Buxton-Norlim.Yet there are no primary carbonate cycles are initiated repeatedly by minor shifts eolian deposits above the klooftoday. Consequendy, eolian of the hydrological balance. Nonetheless,-primaryaccu- components in the geological record are presumably mulation of the key tufa masses would have been inimical derived in similarways by (a) fluvialreworking of mixed to the development of any deep fissures. eolian and residual materials into colluvial or alluvial The various points of interpretationraised here will be deposits-ultimately transported to the escarpment edge applied to a new consideration of the Buxton-Norlim by running water-and (b) direct but slow contribution sequence and then synthesizedin the concluding section. of eolian materialsto developing soils. The stratigraphicframework is collated in table 2. Three fissurefills were analyzed fromthe major quarry Except for Bushman paintings in the Grootkloofcave, tufa at Ulco, a sediment body lithologicallyidentical to no archeological or fossiliferousoccurrences are indicated Tufa I at Grootkloof.7 Each fissuredeposit was a colluvial in the gorge itself. However, below the breached doline wash with a well-sorted but skewed detrital residue of immediatelysouth, there is a veneer of younger tufa (II medium-grade sand, its quartz component subangular to or III) on the escarpment face, and a small core and a rounded, polished, and partly oxide-coated. Carbonate Levallois flake of Middle Stone Age aspect were found content varied from 8 to 60%, with a high pH (8.8) and embedded in this tufa. colors ranging from reddish brown and yellowishred (5 More interestingare the stronglyeroded formerfissure YR) to pink (7.5 YR). One example was sufficiendyfresh fillsexposed on the footslope some 200 m south of Groot- to preserve traces of carbonized macro-organics.Excellent kloof. Both of these contiguous"sites" contain fragmented section faces in the Ulco Quarries show that the larger and disarticulatedbone of large and small mammals. In fills are found in geometricallyor verticallyas well as one instance the matrix is a stratifiedaccumulation of horizontallyarranged solution hollows. This leaves little reddish yellow (5 YR), well-sortedsilty sand interbedded doubt that most of the responsible corrosive activitywas witha 10-15-cm travertinehorizon. In the otherthe matrix postdepositional. "Constructional"hollows are limited to is a laminated,pink (5-7.5 YR), calcretedand recrystallized pockets with maximum dimensions of 15 cm, generally sedimentwith a detritalresidue of well-sorted(So = 1.31) in association with rapidly aggraded organic tufas (D. M. sand, interbeddedwith clean, white (10 YR), cryptocrystal- Helgren, personal communication).Consequendy, most or line tufa similarto that of Tufa I. It is debatable whether all of the larger examples of red sandy pockets,inclusions, these fills antedate or postdate Tufa I, although I provi- and fillswithin the Ulco Quarry tufa must be considered sionally suggest that they form the base to the Tufa I as intrusiverather than primarydeposits, with a full range aggradation. These modest finds,made by L. Matter and of sedimentage fromthe presentback to the initialphases recentlyexamined by J. Kieser and R. Hockman for the of postdepositionalkarstic erosion. This does not preclude Universityof the WitwatersrandAnatomy Department, are some tufaaccretion after karstic erosion began, since minor symptomaticof numerous incidentaldiscoveries or collec- tions made along the scarp over the past several decades.

TABLE 2 TUFAS OF BUXTON-NORLIM GEOMORPHOLOGIC EVOLUTION OF ULCO GROOTKLOOF The tufa sequence of Buxton-Norlim(Peabody 1954) ac- 9. Bedrockdissection and corrosion(4 m downcutting).Mini- cumulated with respect to two successive bedrock valleys malspring activity. of the Thabaseek River, which now occupies a canyon 8. TufaIV. Minorkloof-wall cascades; cementation of Breccia measuring as much as 200 m from shoulder to shoulder TerraceII. and 50 m in depth. The Thabaseek River begins some 7. Acceleratedmechanical weathering of cliffs, with accumulationof BrecciaTerrace II and, throughponor enlargement, 30 km to the west, amid a poorly defined watershed on of brecciain cave. the Gaap Plateau. This large surface catchment and 6. Bedrockdissection and corrosion(2-3 m downcutting). groundwaterbasin presumablyexplains the dimensions of 5. TufaIII. Accumulationof carapace,with some eolian-type both the erosional and depositional forms at Buxton- sands,derived silts, and tracesof salts.Seasonal (?), mainly Norlim. spring-generatedfluvial activity. Cementation of Breccia The ThabaseekTufa, the oldest recognized by Peabody TerraceI. (1954), formerlyextended some 35 m down the escarpment 4. Acceleratedmechanical weathering of cliffs,with accumula- face (see fig. 3). The material is conspicuously white (10 tionof crude,angular rubble in Breccia TerraceI. YR 3. TufaII. Relativelyrapid accumulation of carapace with high 10/2) and correspondinglypure (table 1); the limited proportionof eolian-typesands. Sporadic torrential fluvial detritalresidue is a clayey-siltsand, moderatelysorted and activitywith local spring flow. ?Local embedding of Middle skewed,with a trace of hygrophyticsalts, and subrounded, StoneAge artifactson escarpment;cascades and dripstones polished quartz sands. Cavitation and macrosolution are in cave. correspondinglywell developed, and pink to reddish yellow 2. Protractedand complexinterval of bedrockdissection and (5 YR) soil pocketsare commonplace. Two pointsof major corrosion.Local karstdevelopment (dolines, ponors, caves interest are (a) the australopithecine breccia, discussed in Tufa I and escarpmentdolomite). furtherbelow, and (b) the basal breccia ("apron"). The 1. Tufa I. Slow accumulationof carapace facies,with some latter (ThabaseekBreccia) is described as "a thick surface basal, cobble conglomerate;contains intermixed, older, reddishclayey soils as well as some eolian-typesands. rubble of dolomite and shale from pebble to boulder size" Perennial(?), mainly spring-generated fluvial activity. Apron cemented into "a coarse compact breccia" (Peabody and carapacepreserved in contemporarytufa fan at Ulco 1954:677 and fig. 6). These field similaritiesto the breccia Quarries;local fossiliferouscave fillsnear Grootkloofmay terraces of Grootkloofare impossible to overlook, so the representanother basal facies. basal Thabaseek detritus may well be a frost-weathered footslope rubble. Significantly,too, it lacks derived, older travertines(see also Peabody 1954). The NorlimTufa is representedby a carapace extending 7M. E. Marker and a student spent time between the Ulco 65 m down the escarpmentface. The macroscopic Quarries and Malony's Kloof in 1972 and collected C 14 samples proper- fromthe main quarry tufa. These are all older than the effective ties are identicalto those of GrootkloofTufa I. The detrital dating range of C 14 (J. C. Vogel, personal communication). residue is a moderatelysorted clayey-siltsand. The sands

374 C U R R E N T A N T H R O P O L O G Y

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions BUXTON Butzer:SOUTH AFRICAN AUSTRALOPITHECINES today and the others apparentlyrestricted to moisterparts B ue of the Transvaal.) The Norlim Tufa was followed by a long period of bedrock dissection,during which a 35-m-deep, now aban- doned gorge was cut some 900 m back into the escarpment. The terminal period of cutting left a record of crude, x an 111111I footslope talus "of sand and angular boulders of bedrock and of the oldest breccia" (Peabody 1954:678). As visible EquusCave today, these beds (the Oxland Breccia) offerclose genetic #NorlimCara Pace parallelsto the Grootkloofbreccia terracesand demonstrate unequivocallythat bedrock dissectioninvolved rapid back- ChannelAlluvium wearing that included accelerated mechanical weathering. The Oxland Tufa developed withinthe uppermost part tNorlim of a gorge eventuallychoked by the 30 m of totalaccumula- - hnnel tion. The apron facies is exposed to a thicknessof 10-12 m, with a basal conglomerate followed by 4-5 m of well- stratifiedor current-bedded,semicemented, reddish yellow (5 YR) siltysand withdolomite grit and lenses of rounded, HrdlickaCave coarse to cobble-grade dolomite gravel. The upper half Thabaseek of the apron consists of wavy banded or laminated, very Caaae AustralopithecineCave pale brown (10 YR), porous and cryptocrystallineprecipi- tates including bulbous, spring-ventinterbeds. Abundant diffusehumus is present,as wellas tracesof macro-organics, witha detritalresidue of moderatelysorted silty-sandclay. The finer sands are mainly subangular and polished, the 0 250 coarser ones rounded and frosted; all are of quartz, and m ferricoxide skins are absent. The carapace limestonesare steeplyforeset at 40-70? but otherwiselithologically almost identical. After accumulation of the Oxland Tufa, a widespread patina of pyrolusitewas formed on most surfaces in the FIG. 3. Tufasat Buxton-Norlim(modified after Peabody 1954:fig. Buxton-Norlim area, an event of evident stratigraphic 2). import (see Peabody 1954) although there is no ready paleoclimaticinterpretation (see Engel and Sharp 1958). The Channel Alluvium is a 4-5-m fill that dips 20 are exclusively quartz, subangular to subrounded and downvalley within the defunct pre-Oxland valley. This overwhelminglypolished. Of interestare local pockets of material grades up from (a) a stratified,calcreted, light cemented breccia, consistingof crude, angular dolomite reddish brown (5 YR) silty sand with angular dolomite rubble in a sandy, reddish-yellow(5 YR) matrix. grit and lenses of subrounded, coarse to cobble-grade This generation of tufa is currentlybeing exploited in gravel,to (b) an unconsolidated,brown (7.5 YR) siltysand a large quarry at Thoming, 8 km south of Buxton-Norlim with columnar structure,grit, and reworked concretions (see fig. 1). Here Peabody (1954) identified one of two and finally(c) a calcified, reddish yellow (7.5 YR) subsoil examples of his typicalfacies sequence 18 witha subhorizon- withslightly humified top soil. The materialis well sorted tal tufa "apron"~em bedding sand and rubble,disconforma- but skewed, and the sands include quartz withsome oxide bly overlain by inclined, foreset beds of an encroaching skins as well as grains of shale/slate. The coarser quartz tufa fan or "carapace." An identical vertical progression is rounded and frosted,the finer subangular and partly can now be observed in the lower of the two Ulco quarries, polished. The Channel Alluvium must be considered as with a relativelypure apron limestone dipping at 10 or a valley fillderived fromcolluvial deposits, from a source less, the superposed carapace inclined at 5-1Oo or more. other than the Oxland Tufa. It includes both residual Both the Ulco and Thoming quarries hav&yielded valuable dolomite soil and reworked or primaryeolian sands from organic inclusionsfrom local interbedsor pockets in what above the escarpment. It has no attributesof a swampy, are, byall lithologicand geomorphologiccriteria, analogous waterholedeposit (contraPeabody 1954:697). The topogra- and coeval tufa bodies. These were collected by the respec- phic and facies parallels of the Oxland Breccia, Oxland tive general works managers of the two companies, Mr. Tufa, and Channel Alluviumwith the Tufa II/III complex Matter and Mr. Duffy. They include molluscs, a freshwater and related breccias at Ulco Grootkloofare unmistakable. crab now common to both the Thabaseek and Vaal rivers, The final stages of geomorphic evolution at Buxton- and impressions of mosses, reed stems, and broad-leaved Norlim were marked by bedrock dissectionof the present plantsprovisionally identified by Richard Liversidgeas OIea Thabaseek canyon,followed by accretionand partial block- capensis,Zizyphus sp., and another unidentified broad- ing of the channel by some 30 m or so of the Blue Pool leaved tree now absent fromthe escarpment.(The 13 plant Brecciaand Tufa. This carapace is macroscopicallyidentical species identifiedfor Peabody [(1954:694] by Robert Rodin to the Oxland. The two C'4 dates of 15,980 ? 230 B.P. are not listed; 11 of them are reported to be found there (SI-1642) and 14,010 ? 170 B.P. (SI-1643) show that the top of the Blue Pool Tufa is temporallyequivalent to Tufa III at Ulco Grootkloof. Furthermore,it implies that the "8Peabody(1954) correlates the northernend of the Thoming Blue Pool Breccia is equivalent to Breccia Terrace I. It exposures with the (younger) Oxiand Tufa at Buxton. However, the current exposures in the southern half of the quarry are is currentlybeing corroded and dissected (Peabody 1954). lithologicallytypical of the Norlim Tufa. The possibilityof two Peabody's (1954) Channel Tufa is probably an equivalent generationsat Thoming remains open. of t.heBlue Poo] Tufa in the old Thabaseek valley.

Vol. 15 No. 4 December1974 375

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions THE AUSTRALOPITHECINE DEPOSITS AND 8/4), inhomogeneous sediment with dispersed grit and OTHER CAVE FILLS small pebbles of corroded, reworked travertine;it has a detrital residue of well-sorted,medium-grade sand (see The Taung australopithecinewas recovered from a cave table 1) and colloidal silica (8% of the total 19% noncarbon- fill intrusivewithin the Thabaseek Tufa. The exact site ate residue) and is marked by pyrolusitestreaks. Except had already been destroyed by quarrying at the time of for a trace of chert, the sand is quartz, overwhelmingly Young's visit in 1924 (Young 1926), but Peabody (1954) subrounded to rounded and frosted,with some oxide skins. has meticulouslyattempted to reconstructthe destroyed The electronmicroscope reveals fluvialmarking with strong cavern topography and its successive fills on the basis of eolian overprinting.In other words, this is a wash derived available reports, test excavations, local testimony,matrix in major part from eolian-type sands, with a little local adheringto thevarious fossils,and the exposures remaining rubble. It clearly represents the sediment type described in 1947-48. Peabody (1954:681-85) identifiedfour deposits by Peabody (1954) as his "dry" phase, sandy limestone, from bottom to top (fig. 4): (1) a massive, reddish, sandy and it is similar to other, cemented reddish fillingsthat "limestone,"fossiliferous, representing a "dry phase"; (2) carry fossils, primarilybaboons, from various points on pure lime intercalatedwith lenses of reddish, sandy lime- the Gaap Escarpment.

BUXTON Sterile ReddishSand ~~~~~~~~~Black Earth

3 Australopithecinelevel) (ThabasekiTufas

L no horizontalscale

FIG. 4. Section of the Hrdli'ckaCave, part of the australopithecine cave complex (modified after Peabody 1954:fig. 4). stone,fossiliferous, marking a "wet phase"; (3) semicement- A sample of the matrixadhering directlyto the Taung ed, coarse reddish sand, sterile; (4) unconsolidated "black skull was kindlyprovided by P. V. Tobias. This is a pink earth" with Middle Stone Age artifacts.Whereas deposits (5 YR 7/4), fine-grainedsediment with contorted ("mar- 1 and 2 are conformable,3 and 4 fillsuccessive generations ble") veins of white, cryptocrystallinecalcite and small of solutionfissures. Peabody marshallssound evidence that hollows with mesocrystallinecalcite. The residue is a mod- the australopithecineskull was recovered from the "wet- erately sorted clayey silt with a primary clay maximum phase" deposits. Unfortunately,the petrographicanalyses under 2 ,u(41.5% compared withonly 3% in the "dry-phase" he obtained were highly generalized and allowed no dif- deposits) and a secondary silt maximum between 6 and ferentiationof deposits 1, 2, and 3.19 I have analyzed samples 20 ,u (23%). The primarycomponent above 37 ,uis quartz, of "dry-" and "wet-phase" deposits, and they are indeed mainly polished, and increasing with size from subangu- quite distinctive. lar/angular to subrounded/rounded. Many grains have A sample of well-cemented,sandy breccia was selected "ferric"skins which, according to the X-raydiffractogram, from one of several occurrences in solution hollows of include 9-IOA clay minerals. Eight diffractograms-with the Thabaseek Tufa, froma point veryclose to the former various preparations and several fractions-show a strong australopithecinecave (see fig. 3). This is a pink (5 YR dominance of montmorilloniticclay, the other minerals, in decreasing order of importance,being illite,muscovite, '9These sedimentswere collectivelydescribed as quartz grains kaolinite, potassium feldspar(s), quartz, calcite/dolomite, in a calcite (totaling over 55%) matrix with less than 1% chert goethite, and rutile (?). Petrographic examination shows and 2% feldspar; the only heavy minerals noted were traces of zircon; average grain diameter is near 100 ,u,with quartz grains the presence of some graydolomite, ferric-sandy aggregates occasionallycoated with oxide skins and generally hyperangular and opaque iron (either ilmenite or magnetite). Study of to subrounded (Ada Swineford,cited by Peabody 1954:683). Young the 63-250 p. sand fractionwith the electron microscope (1926:62) also noted the ferricoxide coatings, but described the indicatedthe presence of some characteristiceolian compo- quartz grainsas angular to rounded, furtherrecovering magnetite, rutile,and tourmaline in the heavy-mineralcategory. Finally, G. nents, but in general the sample is of extremely mixed Bond (cited by Oakley 1954:n. 5) found the quartz grains to be provenance. This and the cumulativetextural curve argue of eolian tVDe. Dresumablvrounded and frosted. stronglyfor a derived soil, incorporating an incidental

376 C U RRE N T A N T H RO POLOG Y

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions amount of secondary,eolian sand. Butzer:SOUTH AFRICAN AUSTRALOPITHECINES Consequently,there is a fundamentaldifference between this"wet-phase" sediment and the "dry-phase"sandy lime- Equus cave fill (fig. 3). The deposits" occupy a solution stone: the Taung matrixis neithera reworkedeolian sand, cavernin theOxland Tufa apron, below an erosional surface as are most other secondarytufa fills,nor a highlycalcified that grades down below the Channel Alluvium. The cave cave flowstone.It is a cemented soil sedimentwhose calcite fill is a stratified,partly calcified, reddish brown (5 YR), has been recrystallizedseveral times and whose coarser moderatelysorted silty-claysand, withdispersed, reworked componentscompare most closelywith those of the Norlim tufa as well as angular dolomite debris.23As also verified Tufa (see table 1). by electronmicroscope, the sands are in good part of eolian Accordingly,one can suggest the followingevolution for type-subrounded to rounded, mainlyfrosted quartz, with the australopithecinecave and its successive fills: some oxide skins-with a few grains of gray chert. This 1. Since the australopithecinecave complex was found Equus cave fill includes eolian components,soil wash, and at variable depths of 5-15 m below the surface of the evidence of mechanical weathering,all indicativeof a new Thabaseek Tufa, dipping steeplyin places as it cut across aggradation hemicycle.These deposits were originallysep- the bedding planes (Young 1926; Peabody 1954:682-85 arated from the Channel Alluvium by some intermediate and pl. 3, fig.1), it is unquestionablyof erosional origin. conglomerates,but are probably only a little older (see It is unlikelythat active karsticsolution would accompany table 3). tufa accumulation, and the cavern itselfmust have been excavated shortlyafter the Thabaseek aggradation had TABLE 3 ceased, whilevadose waterwas stillcirculating in abundance. 2. After solution became negligible, eolian sands were GEOMORPHOLOGIC EVOLUTION OF BUXTON-NORLIM blown across the tufa land surface or washed down across (in partbased on Peabody1954) the escarpment. Ultimatelysands of this type found their way into deep fissures,together with a little tufa debris. 12. Limiteddissection and corrosion,with karstic fissures in The excellentsorting, the micromorphology,and the dura- severaltufas, including Later Stone Age soilwash. ble class of heavy minerals suggest fairly direct eolian 11. Blue Pool Tufa.Relatively rapid accumulationof carapace transportfrom mantles of "Kalahari"-typesands, but the (? coevalwith Channel Tufa in old Thabaseekvalley) and bedding of the "dry-phase"deposits speaks for waterlaid local fissurefills (? "blackearth" of australopithecinecave fissure/cavern fill, incorporating large proportions of and Witkransarcheological level) with Middle Stone Age reworked eolian sand and subsequently impregnated by artifacts. 10. Initialaccumulation of soil wash,eolian derivatives,and carbonates.20 angularrubble in fissures(Equus cave and basal Witkrans 3. Eventually rainwash began to erode and transport deposits),with Middle Stone Age artifacts,followed by residual soils, while spring activityonce more increased aggradationof ChannelAlluvium and thick,basal Blue Pool and lime-charged vadose waters began circulatingagain. Breccia. The resulting"wet-phase" deposits correspondingly include 9. Protractedbedrock dissection and corrosion,with limited interbeddedsoil wash, clayeyflowstones, and clean traver- karstdevelopment in all older tufas.Manganese discolor- tines.21 These deposits can be correlated with the initial ationson exposedsurfaces. accretion of the Norlim Tufa and date the Taung fossil 8. OxlandTufa. Accretionof apron tufasand, ultimately, accordingly. carapacedeposits with soil and eolianderivatives. Seasonal 4. Followingthe tufa-accretionphase, a numberof fresh, (?), increasinglyspring-generated fluvial activity. 7. Acceleratedmechanical weathering of cliffs, with accumula- primaryfissures were created by freshkarstic solution. tion of crude,angular talus in OxlandBreccia. Followed 5. Accumulation of sterile reddish sands in these new by accumulationof basal fluvialconglomerates and finer fissures,or in empty pockets of the old caverns, probably alluvium,contemporary with reworked eolian derivatives preceded the Oxland Tufa. or angulardetritus in numerousNorlim and Thabaseek 6. Further fresh fissures were formed after renewed fissures,including australopithecine cave "sterile red sands." karsticsolution, at the same time that other caverns were 6. Protractedbedrock dissection and corrosion,with karst dissolved in the Oxland Tufa. developmentin Thabaseek and NorlimTufas, including 7. Finally,a dark soil-either vertisolicor highlyorgan- thesolution fissure penetrating the australopithecine depos- ic-with Middle Stone Age artifacts filled a variety of its. 5. NorlimTufa. Slow accumulation of carapace facies, overlying fissures,including one in the australopithecinecave (fig. apronfacies at Thoming.Contains soil derivatives and some 4). Travertineassociated withthe lower part of the Middle eolian-typesands. Perennial(?), mainlyspring-generated Stone Age breccia at nearby WitkransCave yielded a C 14 fluvialactivity. Initial aggradation recorded by reddish clayey age of 33,150 + 2,500 B.P. (Clark 1971). Furthermore, soil wash and by flowstoneinterbeds of the upper,"wet" sound C'4 dates on the South African Middle Stone australopithecinedeposits. Hominid level. Age now run ca. 20,000 to greater than 55,000 B.P. (Klein 4. Colluvialaccumulation of eolianderivatives into valley and 1970, Beaumont and Vogel 1972), withcoastal occupation fissurefills, including the lower,"dry" australopithecine extendingback well into the Last Interglacialtransgression depositsand localbreccias. Faunal level. 3. Dissectionand localkarst development in Thabaseek (Butzer 1973a). Consequently,an Tufa, early Upper Pleistocene creatingaustralopithecine cave complex. age for the post-Oxland solution phase seems plausible. 2. ThabaseekTufa. Slow accumulation of carapacefacies with The australopithecinecave stratigraphyoutlined here minimalimpurities of mixed derivation.Perennial (?), is readily reconciled with the tufa sequence (table 3). It spring-generatedfluvial activity. is furthercorroborated by the evidence fromseveral minor 1. Acceleratedmechanical weathering of cliffs, with accumula- caves described by Peabody (1954), particularlythe massive tionof cruderubble in ThabaseekBreccia.

22Examined by C. K. Brain, Richard Liversidge, and me in 20These are comparable to Brain's (1967b) "Phase II" deposits 1971. of the Transvaal Caves, i.e., derived from the surface outside 23Peabody (1954) noted fossil fragments,including the tooth of the cave. of a small Equus. P. Verhoef (Universityof South Africa)recovered 2"Comparable to Brain's (1967b) "Phase I" deposits, i.e., mixed a furthercollection of bones in 1971; these are being studied sediment fromboth outside and inside and inside of the cavern. by C. K. Brain.

Vol. 15 No. 4 December1974 377

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions AN ENVIRONMENTAL MODEL FOR THE GAAP for considerable latitude of variation to the dry side of TUFAS modern averages. Phase2. Protractedrains capable of outpacing infiltration The sequences outlined forGrootkloof and Buxton-Norlim begin to increase,leading to activesheetwash or sheetflood- show that a cyclicform of erosion and deposition has been ing and intermittenttorrential discharge in streams.Diffuse typicalof the Gaap Escarpmentduring the recentgeological surficial deposits of diverse origin are washed together past. Table 4 suggests a possible explanatory model for in concentratedcolluvial spreads, as valley-floorfills, and thesecycles and allows opportunityfor more comprehensive intocave fissures.Limited, lime-charged cavern waters favor discussion of the related environmental changes. This fossilizationof bone embedded in sandy cave fills ("dry- model has four phases, as follows: phase" baboon deposits). Parallel with accelerated fluvial and colluvialactivity, accelerated frost-weatheringproduces TABLE 4 unusual quantities of crude slope rubble, forming talus below.the cliffsand markedlyincreasing the rate of head- AN EXPLANATORYMODEL FOR GEOMORPHOLOGIC CYCLES ON THE GAAP ward stream-cuttinginto bedrock.24Major canyon forma- ESCARPMENT tion and extension are made possible both by stream dynamismand preparatoryfrost-wedging. PHASE DOMINANTGEOMORPHIC VEGETATIONAND This morphodynamicmodel presupposes a cooler and PROCESS CLIMATE increasinglywet climate. Since the severityof winterfrosts has only an indirectrelationship with the evaporativelosses of surfacedischarge fromheavy summerrains, an absolute 4 Declining spring and fluvial Grassveld and increase of precipitationto somewhat above the 450-mm activity,leading to erosion; groundcover mark is suggested. Vegetation typesare difficultto assess, karsticsolution with fissure deteriorating; and cave formation. increasinglydry and sincecolder climateswould affectmany species of the woody warm. thornveld. Groundcover was incomplete, however, and 3 Maximum spring discharge Grassveld providing infiltrationpoor. This precludes a closed grassy mat of and protracted fluvial acti- optimal groundcover and sweetveldtype and suggestsa transitionaltype of vegetation, vity, leading to large-scale infiltration;humid and perhaps including elements of both the thornveldand the tufa accumulation-initially probablycool. more cold-tolerantkarroo shrub. of subhorizontal aprons, Phase3. Eventuallythe environmentalchange responsible later of steep carapaces; for Phase 2 crosses a criticalthreshold, and a new dynamic and clayey beds flowstones between the variables in cave fissuresof previous equilibriumis established interacting tufa sheets ("wet-phase" vegetation,weathering, erosion, and deposition. This equi- australopithecinelevel), with librium is not established simultaneouslyeverywhere, be- cementation of older brec- ginningearlier where hydrogeologicalconditions favor an cias and cave fills. early crudescence of spring discharge as groundwater 2 Accelerated fluvial and Incomplete groundcover reservoirsincrease. The relativeincrease of spring versus colluvialactivity, washing to- with poor infiltration; surface waters in the key sedimentaryareas is gradual. gether surficial sediments increasinglywet and cool. At firstsurface discharge transports and deposits detritus- intovalleys and cave fissures gravels,sands, silts-in the wake of heavy rains; but spring (including ba- "dry-phase" the season and a little boon deposits); accelerated discharge persiststhroughout rainy frost-weatheringleading to thereafter,impregnating these deposits with lime and talus accumulation. ultimatelyaccumulating in thin sheets of chemical precipi- 1 Accelerated eolian activity, Thornveld with limited tates. Clayey soil wash alternateswith primaryprecipitates with limited colluvial and groundcoverand in caves and fissures("wet-phase" australopithecine bed). fluvial processes, leading to infiltration;dry and Vegetation cover in the catchmentultimately increases thinbut extensivemantles of warm, in part drier than to the point where next to no soil products are exposed eolian sedimentsand wash. today. to erosion and the velocityof surface runoffis inhibited, while infiltrationaccounts for a higher proportion of the rainfall. True fluv'ialactivity is diminished in favor of strongerand increasinglysustained spring flow, most of Phase 1. During long periods of time there is little it vadose waterwith only limitedunderground trajectories. significantgeomorphic change along the escarpment.Most Sheets of relativelypure lime now begin to accumulate rainfall evaporates or percolates into surficial sediments. along the valley floorsand as broad fans ("aprons") below Any excess rains lead to local sheetwash that, in view of the embouchuresof the canyonsand kloofs.Each rainstorm the incompletevegetation mat, serves to redistributeloose produces sheetfloodsthat sweep across these aprons, carry- sediment and soil. Surficial materials include humified ing a littleclay and sand or an occasional pebble, but the residual products, stream alluvium, colluvial wash, and major impact is temporary,surface corrosion that is more veneers of eolian sands blown in from the west or north than compensated for afterresumption of spring-feddis- by the dominant storm winds. Caves are dry, with slow charge. In this way the characteristictufa banding and accumulation, much of this organic. The limited laminationis produced. groundwatersfeed a small number of insignificantsprings, Eventually,significant spring discharge at higher eleva- mainly seasonal. Most of these springs are sufficiently tions along the scarp walls is favored by a rise of the lime-chargedto deposit carbonateprecipitates, algal traver- watertablein thescarp dolomites.Spring waters now cascade tines,and plant tufason a smallscale. However,occurrences do.wn the walls of kloofs or plunge into canyons over are too fewand restrictedto be of geomorphic significance. lime-encrustedwaterfalls. These deposits build forwardat The inferred vegetation is comparable to that of the early 19th century-a thorn scrub-woodland and grassy parkland with limited groundcover and infiltration.Since 24At least two phases of substantial Pleistocene cooling are recorded in the Drakensberg (Ellenberger 1960, Harper 1969, rainfallranges from300 to over 450 mm along the escarp- Hastenrathand Wilkinson1972) and at least three in the southern ment today, this "warm-dry"morphostatic model allows Cape Province (Butzer and Helgren 1972; Butzer 1973a, c).

378 CURRENT ANTHROPOLOGY

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions steep angles as "carapaces" that locally replace the aprons. Butzer:SOUTH AFRICAN AUSTRALOPITHECINES The angles of these carapace sheets decline from almost 900 in waterfallsto less than 400 at the scarp front and currentlyin progress between Douglas and Ulco by D. 50 or so in low cones that locally build out onto the flat M. Helgren. plains below the escarpment. The surfeitof lime-charged More difficultis the task of external correlation.Direct waters ultimatelyleads to thorough cementation of any links with the alluvial terraces of the Harts and Vaal may permeable cave and fissurefills. ultimatelybe provided by the local sheets of increasingly The inferredvegetation for Phase 3 is a closed mat of dolomitic travertinesthat extend from the escarpment to grass withabundant aquatic formsand fringingtrees near the valley margins,as tested by the Union Lime Co. (L. springs,seeps, and streams. Analogous plant associations Matter,personal communication). Whether or not sufficient may have been found early in the 19th century in the bore data can be obtained in an area of minimalexposures higher grassveld,which now enjoys 500-1,000 mm rainfall remains a moot point. Pending furtherstudy by D. M. and cooler temperatures.25Probably the WiurmPleniglacial Helgren of the general planation surface rising from the vegetationof Florisbad-grassveld withlocal aquatic plants Vaal and Harts valley-marginto the footof the escarpment, (Van Zinderen Bakker 1957; also Coetzee 1967)-and the the best possibilitiesof external correlation available are climateof Kimberley-over 800 mmprecipitation and much the facies alternationsof the Harts terrace deposits (see cooler (Butzer et al. 1973)-are most applicable.26 By table 1). These allow the tentativestratigraphic suggestions implication,Phases 2 and 3 saw a rainfall increase from of table 5. The nature of the Harts sequence is basic to 450 mm to as much as 800 mm along the Gaap Escarpment. our argumentand can be summed up as follows: Phase 4. The water supply begins to fait. As spring 1. The lowermost Harts is followed by a subrecent dischargewanes, surface dischargebecomes dominant and floodplain deposit at 2.5 m above low water. This flood tufaattrition outweighs accretion-except near springvents terrace continues below the Harts-Vaal confluence and and seeps. Carapaces and aprons cease to grow and begin along the Vaal River provides a reference datum at the to show evidence of superficialcorrosion. The watertable type site of the late-Pleistocene-to-Holocene Riverton falls, too, and incipient underground caverns are rapidly Formation (Butzer et al. 1973). These Harts flood silts enlarged as verticalfissure networks begin to channel water consistof well-stratified,light brownish gray (2.5 Y) clayey- throughthe nonfunctionaltufa masses, both old and new. sand siltsthat are moderatelysorted (table 1). Quartz sands In thisway new fissuresor caves are created in both newer are mainlysubrounded and polished. and older tufa bodies liable to circulationof vadose waters. 2. The Harts also has nonfunctional+3 m (5.5-6.0 m Increasingly sporadic stream discharge eventuallycarves above low water)and +6-8 m terraces.These can be readily out new channels throughthe tufaswhile conditions return related to interfingeringVaal terraces that are identical to those of Phase 1. with Members V and IV of the RivertonFormation. Near Groundcover is initiallystill complete enough to inhibit the Vaal confluence, these two alluvia consist of brown soil stripping.But protractedrains are decreasingin impor- flood siltsidentical to those along the Vaal, which appears tance, so that as the grassveld deteriorates,the frequency to have ponded intothe lowermost Harts channel. However, of sheetflooding is already limited and geomorphologic upstream,along the Dry Harts at the Buxton Bridge, the adjustments remain incomplete. As the climate becomes two sedimentarybodies are both distinctand informative warmer and drier, a thornveld vegetation returns and (see table 1). The lower is 5 m above low waterand consists essentiallymorphostatic conditions prevail once again. of well-stratified,reddish brown (5 YR)_sandy-clay silt, The australopithecine cave complex was first eroded moderatelysorted but skewed,with angular, polished quartz during Phase 4, whereas the younger, corrosion fissures dominantin thefiner sands and subrounded, frostedquartz representidentical stages within successively younger cycles. with oxide skins in the coarser grades. The coarse sand "tail"and dispositionsuggest a mixtureof eolian-typesands with other fluvial deposits, laid down on the floor of an MORPHO-STRATIGRAPHIC CONTEXT OF THE intermittentstream. The higher terrace, at 8.5-9.5 m, is GAAP TUFAS a massive bedded, verticallystructured, brown (10 YR) sediment grading up from silty-claysand to a siltysand. The preceding model is based on the three and four cycles Sorting decreases and ,skewness and coarseness increase representedat Grootkloofand Buxton-Norlimrespectively. upwards. An internalcorrelation of these cyclesis suggestedby table The implicationis that of channel sediments deposited 5. Peabody (1954), whose study encompassed the major in a stream with increasing contrastsin current velocity tufasof the northernmostthird of the escarpment,suggest- and briefer,more violentflood crests.There is a significant ed the presence of pre-Thabaseek deposits at Boetsap; proportion of oxide-coated, subrounded quartz sands furtherstratigraphic details can be expected from work throughout.As in otherparts of thenorthern Cape Province and Orange Free State, downcuttingand alluviation were 25In terms of geomorphologic processes, both the Brazilian in response to broad regional shiftsof the geomorphologic analog of Branner (1911) and the Mediterranean counterparts balance (see Butzer 1971 b), and at Riverton itself there (e.g., Yugoslavia, Mallorca) presuppose a precipitationof at least is indirect evidence that Members IV and V are of mid- 600-650 mm. Yet such conditions are not replicated today in and late-Holocene age respectively(Butzer et al. 1973). the even wetter dolomite country of the eastern Transvaal (see Markerand Brook 1970). M. E. Marker (personal communication, For the Harts drainage, thistentatively suggests that eolian- November 16, 1972) concurs that "indicationsare that tufa forma- typecomponents have been available throughoutthe Holo- tion required adequate seasonal rainfalland a denser vegetation cene, although a more favorable hydrologicalbalance ob- cover than now obtains,even in areas of ? 500-600 mm rainfall tained in mid-Holocene times; discharge has since been in the northeasternTransvaal." 26Contraryto the conclusions of Young (1926) and Peabody sporadic. (1954), I believe that the major tufas record optimal moisture. 3. The youngestcalcified deposits are found immediately Young and Peabody interpretthe dramatic record of fluvialerosion west of the Vaal, Harts, and Dry Harts, where they form in my Phase 2 as the period of maximum rainfall; I, instead, the"Younger Sandy Marl" terrace(+ 7.5 m at Delportshope, attributeit to ruptureof equilibriumand to the torrentialdischarge typicalof semiarid environmentswithout adequate groundcover. but eroded and partlysuballuvial upvalley). The material See also the analogous interpretationof freshwaterlimestones is white to light gray (2.5-5 Y), calcreted by over 50% of the KimberleyDistrict (Butzer 1973b). CaCO 3, and has bimodal residues quite similar to the

Vol. 15 No. 4 December1974 379

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions younger, Riverton Member IV alluvium (see table 1). Middle Pleistocene fauna (Cooke 1963, Wells 1964) and Lithologicallythis terrace allows a correlationwith Riverton Acheulian industries.27 Member III, which has terminal C'4 dates of ? 17,000 This Harts sequence, although based on fragmentary B.P. Correlations with the youngest tufas of the Gaap exposures, is almost as completeas the post-"Older Gravel" Escarpment are reasonable both on geomorphologic alluvial sequence of the Vaal at eitherRiverton or Windsor- grounds and on the basis of the Tufa II/III dates from ton. It thereforepresents a stratigraphicyardstick of some Grootkloof.Consequently, Tufa II/III and Breccia I/II value for the younger part of the Gaap sequence. at Grootkloof,as well as both the Channel Alluvium and If the preceding argument is accepted and if major Blue Pool Tufa at Buxton-Norlim, can be confidently geomorphologic cycles of the Gaap and Vaal are placed assigned to the Upper Pleistocene. Rolled Acheulian and side by side, the Thabaseek and Norlim cycles can be no freshMiddle Stone Age artifactsrecovered by Jones (1920) older than the last two Vaal terracesof a fourfold Older

TABLE 5

STRATIGRAPHIC RELATIONS BETWEEN BUXTON-NORLIM AND ULCO GROOTKLOOF, WITH POSSIBLE HARTS-VAAL CORRELATIONS

SUGGESTED BUXTON-NORLIM GROOTKLOOF HARTS (AND VAAL) RIVER GEOLOGICAL AGE

(Dissection,solution) (Dissection) Floodplain Terrace Holocene Tufa IV (9,300 B.P.) +6-8 m Fill (RivertonMember V)

Blue Pool Tufa and Breccia; ?Channel Tufa II/III and Younger Sandy Marl (Riverton Upper Pleistocene Tufa (2 16,000-< 14,000 B.P.) BrecciaI/II complex Member III with Middle Stone C (: 31,000- Age artifacts,2 17,000 B.P.) ChannelAlluvium; cave fillswith Middle -' 20,000 B.P.) Basal Silt (Riverton Stone Age artifacts Member I or II) (Dissection,solution) (Dissection) (Dissection) Oxland Tufa Older Sandy Marl (Vaal Middle Pleistocene Younger Gravels "C"; Acheulian, with Middle Pleistocene fauna) Oxland Breccia (?Vaal Younger Gravels "A") (Dissection,solution) (Dissection) (Dissection) NorlimTufa; cave clays and flowstones Tufa I (Vaal Older Gravels, stage 4: Lower Pleistocene withAustralopithecus africanus "70-ft.";alternatively, Younger Gravels "A") Colluvia, partlyeolian, sandy cave fills and local breccias (baboon deposits) (Dissection,solution of (Dissection) australopithecinecave) ThabaseekTufa (Vaal Older Gravels, stage 3: "100-ft.";alternatively, stage 4) ThabaseekBreccia ?Older tufas at Boetsap (Vaal Older Gravels, stage 2: Upper Pliocene Holpan Platform) (Vaal Older Gravels, stage 1: Nooitgedacht Platform)

fromHarts Riverlateral gravelsnear Taung Mission clearly Gravels complex (see Butzer et al. 1973). Specifically,this come fromthis Younger Sandy Marl. would place the australopithecinelevel of early Phase 3 4. At the Delportshope Bridge, an older body of truncat- of the Norlim cycle as equivalent to the ruptureof equilib- ed flood silts is found under the Younger Sandy Marl rium suggested by the final ("70-ft.") unit of the Older terrace. This "Basal Silt" is a prismatic-structured,gleyed, Gravels.Alternatively, the Norlimcycle could be correlated light gray (5 YR) sediment otherwisequite similar to the withYounger Gravels "A," the Oxland cyclewith Younger modern flood silts (see table 1). Its significancelies in Gravels "C." Although eithercorrelation is possible, I feel lithologicaland stratigraphicanalogs to RivertonMembers that this is the best approximation possible on available I and II at the type site, providing strong confirmation evidence and certainlystronger than any existinggeological for the terracecorrelations offered here. arguments.28I furthersuggest that the Norlim cycle is 5. Finally, there are two "Older Sandy Marl" terraces at + 18 m and + 26-27 m, just below the Vaal-Harts 27The artifactualassociations of the Buxton-Norlimarea are compatiblewith this, since all post-Oxlandcave fillshave either confluence. The older of these terraces comprises gravels Middle or Later Stone Age materials. interdigitatedwith calcretes that slope up towardsthe Gaap 28The only otherline of geologicalargumentation has been Escarpment,where theycan alternativelybe correlatedwith based on incorrectclimatic interpretation for the Taung aus- the Oxland Tufa or withGrootkloof Tufa I and the Norlim tralopithecine correlated with unacceptable climatic inferences Tufa. This also happens to be the west bank Vaal facies from the Vaal terrace sequence (by Peabody [1954]) and from the climatic cycles purportedly reflected in the Transvaal aus- of the "Younger Gravels" substage "A" (D. M. Helgren, tralopithecinebreccias (see Brain 1958, and Cooke 1963, contra personal communication; see Butzer et al. 1973) with its Butzer 1971cand Iq71a:427-31).

380 C U RRE N T AN T H RO PO LO G Y

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions of late Lower Pleistocene or even early Middle Pleistocene Butzer:SOUTH AFRICAN AUSTRALOPITHECINES age, although concepts such as Lower and Middle Pleisto- cene cannot yet be generally defined by rigorous criteria the more mesic, transitionalvegetation posited above for in southern Africa, even if we accept the Matuyama- phase 2. Omitting the sand-burrowingrodents, it would Brunhes geomagnetic boundary (700,000 B.P.) as the base also be a reasonable fauna to expect with the grassland of the Middle Pleistocene. and fringingforest suggested for phase 3. Traditionally,the Taung skull has been accepted as the oldest australopithecineknown fromsouthern Africa. This IMPLICATIONS OF THE TAUNG FAUNA case has been most comprehensivelyargued on faunistic grounds by De Graaf (1960), Cooke (1963, 1967), and Ewer The fauna derived fromthe australopithecinecave complex and Cooke (1964), who placed the Taung fauna in the in the 1920s raises problemsof provenienceand interpreta- earliest segment of the Sterkfontein-Swartkrans"faunal tion. span." A strong reiterationof this view was recentlygiven The great mass of baboon deposits,including those from by Sampson (1971). However, Wells (1967:105-6; 1969) Hrdlicka Cave, clearly came from the "dry-phase" sands has repeatedlyquestioned this assignment,and R. F. Ewer (see in particularPeabody 1954). In fact,most and perhaps (in Wells 1967:105) has admitted that the supporting all of the mammalian fauna, other than the hominid, arguments are slim indeed. Instead, the new short-faced appears to come fromthese lower sands. A piece of turtle baboon (Papio wellsi)from Taung seems to be very close carapace was recently found in the hominid matrix (P. to one from Swartkrans(Wells 1967:105). Similarly,the V. Tobias, personal communication), so the turtle and cercopithecoidsfrom Taung tend to align that site with freshwatercrab probablycome fromthe upper, "wet-phase" Swartkransand Kromdraai rather than with Sterkfontein clayey silt. The basic Taung faunal assemblage therefore and Makapansgat (Freedman and Stenhouse 1972). It is, pertains to phase 2 and has only limited applicabilityto in fact,quite strikingthat Parapapio broomi,the commonest the hominid of phase 3. cercopithecoidat Sterkfonteinand Makapansgat, is totally Over half of the Taung fauna consists of murid and lacking fromTaung. bathyergidrodents (15 genera or species) or insectivores Althoughthe revised Taung geology,as presented here, (3 species or genera) (De Graaf 1960, Cooke 1963), while does not allow direct correlation with the Transvaal aus- half of the remainder are forms specificallyadapted to tralopithecinebreccias, it demonstrates that the Taung rockyoutcrops or caves: 2 species of hyrax,a klipspringer fossilis significantlyyounger than previouslysupposed and (Oreotragus),a bat, a small owl, and-if recent baboon consequentlyprovides implicitsupport that Taung is in- occupation serves as a guide-5 species of baboon (Papio deed no older than Swartkransor Kromdraai. Indirectly, and Parapapio). Thus at least 28 of 33 mammalian species too, the Taung matrixis basically similarto the Transvaal were adapted to cave, rocky,or edaphic microhabitats,and breccias, since all are derived by colluvial processes from 30 of 33 (91%) are nonmigratory(in the sense of Clark residual soils of mixed originand subsequentlyaccumulated 1971:1228). Pending furtherwork on the microfauna of in sizeable caverns. A large suite of X-ray diffractograms the Transvaal cave breccias, this renders difficultany currentlyunder studyshows that the Swartkransand upper stratigraphicand ecological comparisonswith the Transvaal Sterkfonteindeposits are often dominated by illite and faunas,in whichlarge, mobile mammals are well represent- montmorilloniticclays, whereas hillwash exposed in the ed. Sterkfonteinvalley today is characterizedby illite,possibly Cartmill (1967) has attempted to analyze the rodent in combinationwith kaolinite. Consequently the weathering paleoecology,indicating that 6 out of 8 species withecologi- environmentsresponsible for the pedogenesis on dolomite cal preferences point to a rainfall in excess of 300 mm, thatpreceded soil erosion in the twoareas were fundamen- compared with9 of 10 at Sterkfontein,8 of 8 at Kromdraai, tallysimilar. and 13 of 15 at Makapansgat. This essentiallyargues that the Taung "dry-phase" context was almost as moist as that of the Transvaal cave breccias, a point that is not CONCLUSIONS unreasonable. Significantly,too, the closest modern counterparts of the two most commonly represented forms, Since the original publication of the Taung discoveryby whichinclude several species each of Mystromysand Otomys, Dart (1926) it has been believed that the type specimen are overwhelminglyfound in moist, well-vegetatedareas. of Australopithecusafricanus lived on the treeless margin Of equal interest is that at least several of the genera of the Kalahair Desert, with a way of life differentfrom represented may have preferreddry, sandy soil for their that of extant forest-livingapes (see also Tobias 1973a). burrows,probably not a fortuitouscoincidence when their This concept ultimatelyled to the theory of ecological remains come from sandy sediments. Of the remaining differentiationbetween gracile and robust australopithe- mammals,the silver-backedjackal (Canis cf. mesomelas)and cines. A belief in the great antiquityof the Taung fossil Africanhare (Lepus cf. capensis)range fromopen grassland was maintained while furtheraustralopithecines were dis- to deciduous forest;further, an extinctkudu (Tragelaphus) covered, and it is still widely asserted today that Taung and primitivebushpig (Potamochoerides)argue for a savan- is as old as any early hominid in southern Africa. The na-woodland mosaic, while a small duiker (Cephalophuscf. evidence marshalledhere argues thatneither of these claims caerulea; see Wells 1967) and Cercopithecoidea required is acceptable: at least some forest. 1. The Taung australopithecine was penecontempora- As an assemblage, the "dry-phase"fauna would be most neous not withthe Thabaseek Tufa (as Peabody suggested), compatible with the complex of microhabitatsoffered by but with the younger Norlim Tufa. Even the solution a rocky escarpment flanked by plains and dominated by responsiblefor creating the cavern postdates accumulation grassy vegetation,yet broken by sizeable stands of thick, of the Thabaseek Tufa. fringingwoodland near water.29This accords well with 2. The fossil is now conclusively associated with the "wet-phase"deposits of the australopithecinecave, whereas 29As mentioned earlier, the plant impressions from Thoming most or all of the faunal assemblage appears to have come include two species that argue for moister conditions (Peabody fromthe older "dry-phase"deposits. 1954). 2 rl It 441- __,- _ - .el.. ts-1 . .- __ ? l-1__ -1 -__ *

Vol. 15 - No. 4 December1974 381

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions contemporarywith the onset of the Norlim sedimentary such as that of D. M. Helgren and myselfalong the Gaap hemicycle;they are sands witha strongeolian component. Escarpment and in the Vaal Valley, will be necessary. The "wet-phase" deposits date from the early phases of Equally evidentis the urgentneed for palynologicaltesting Norlim tufa accretion; they are clayey silts, interbedded of the Gaap tufas,systematic study of the plant impressions withflowstones and overwhelminglyderived from residual and fossil invertebrates,and radiometricassays by urani- dolomite soils. um-seriesdating on mammalian bone fromboth the Gaap 4. A new environmentalmodel for the geomorphologic sites and the Vaal Valley. Substantial improvements in cyclesof the Gaap Escarpmentargues stronglyfor a wetter paleoenvironmentalinterpretation and stratigraphicreso- and possiblycooler climate contemporarywith the Taung lution can, therefore, be expected. As a corollary, the hominid. A precipitationmean of 600-800 mm (compared presentconclusions may be strengthenedor found in need with450 mm today)is suggested,implying that this hominid of revision. No less urgent, in my opinion, is attention of the gracile lineage lived in a subhumid or humid and to two other perspectives of the overall problem: (1) a not a semidesert environment. Unfortunately,the new fullpublication of the mammalianfaunas long ago collected geologicalevidence offersno clues as to whetherthe Taung from many differentGaap sites, togetherwith a compre- remains were primaryor derived, i.e., occupant, prey, or hensive analysisof the rodents recovered fromall relevant intrusiveaccident. South African sites, and (2) an up-to-date monographic 5. Since the Sterkfonteinvalley only receives some 750 study of the type specimen of Australopithecusafricanus.32 mm rainfall today,30the Taung context speaks for an Fresh evaluationsin these directionsmay be no less surpris- environment similar to that coeval with the Transvaal ing than my own. australopithecines. This is compatible with the overall geological evidence fromboth areas,3' although the Trans- vaal breccias span a much longer period of time that was admittedlynot homogeneous in terms of geomorphologic Abstract trends. 6. Recent reinvestigationof the Harts and Vaal valleys The environmentalcontext and stratigraphicposition of shows that the youngest tufa generation (Blue Pool) can the first australopithecine discovery, at Taung, South be correlated with Upper Pleistocene alluvia (Riverton Africa,need to be reconsidered,since the Taung juvenile, Formation Member III), while the penultimate tufa and as Australopithecusafricanus, is the type specimen of the itsbasal breccia (Oxland) were generallycontemporary with gracile lineage of Plio-Pleistocenehominids. Recent study the Vaal Younger Gravels. These have a Middle Pleistocene of the site and its regional settingshows that two general- fauna and evolved Acheulian industries.By extrapolation, izations widelyheld today are either incorrector dubious. the Norlim Tufa would appear to be no older than the In particular, the Taung hominid is found in deposits last ("70-ft.")stage of the Vaal Older Gravels, presumably that clearly indicate a subhumid or humid environment of late Lower Pleistocene age. and not a semidesert,thus questioning the original basis 7. Reconsiderationof the Taung "dry-phase"fauna (by for hypothesesconcerning an ecological differentiationof L. H. Wells, L. Freedman, and others) shows an increasing gracile and robust australopithecines. Furthermore, the number of similaritieswith those from Swartkrans and Taung specimenrelates to a youngergeomorphologic cycle Kromdraai. than postulated by F. E. Peabody, and this cycle appears 8. The implication is that the Taung hominid clearly to immediatelypredate the Middle Pleistocene "Younger postdates the gracile australopithecinesfrom Sterkfontein Gravels" of the Vaal River. This and other, faunal criteria and Makapansgatand would thereforealso seem to postdate suggestthat Taung is contemporarywith or even younger the penetrationof true Homo into southern Africa. If the than Swartkransand Kromdraai ratherthan broadly coeval Taung specimenis indeed no older thanthe youngest robust with Sterkfonteinand Makapansgat. By implication, the australopithecinesof the Transvaal, then such a late, local Taung hominid may postdate the arrival of true Homo survival of the gracile lineage would seem to pose new in southern Africa,opening up a new range of problems evolutionaryand ecological problems. concerning the phylogenyand ecological adaptations of It requireslittle emphasis thatfurther stratigraphic work, the australopithecines.

30The long-termrainfall records at Kromdraai, Krugersdorp, and Roodepoort range from 716 to 787 mm (Climate of South 32In fairnessto R. A. Dart, it should be mentioned that Dart Africa1965). did originallywrite a lengthymonograph on the Taung specimen, 3'As based on unpublished, detailed study of a representative but left it unpublished after its appearance in Sir Arthur Keith's suiteof samples fromthe keyTransvaal australopithecinebreccias. writingsin 1931 (P. V. Tobias, personal communication).

382 CURRENT AN THROPOLOGY

This content downloaded from 128.62.59.252 on Fri, 13 Mar 2015 20:58:15 UTC All use subject to JSTOR Terms and Conditions