Taphonomic perspectives on hominid site use and foraging strategies during Bed II times at Olduvai Gorge, Tanzania

The faunal assemblages excavated by Mary Leakey in Bed II of Olduvai Gorge, Tanzania, have, like the more well-known Bed I assemblages, traditionally been interpreted as the result of hominid butchering activities in the lake margin and riverine settings of the paleo-Olduvai Basin. A reexamination of all of Leakey's Bed I sites has shown that hominids played little or no role in the formation of all but one of those faunal assemblages, a finding that prompted the reanalysis of the Bed II sites presented here. We expand upon a previous taphonomic study that provided systematic data for HWK East Levels 1–2, MNK Main, and BK. In addition to these assemblages, we provide data on HWK East Levels 3–5, FC West, TK, and SHK. Our data contradict previous interpretations of MNK Main as a hominid accumulation but uphold the contention that BK represents a primarily hominid accumulation reflecting early access to carcasses. The small and poorly preserved assemblages from FC West and TK are difficult to link unambiguously to either hominids or carnivores. Site MNK Main and HWK East Levels 3–5 appear to be death arenas where carcasses accumulated via natural deaths and/or serial predation. Site SHK is severely biased by selective retention and therefore little can be said of its formational history. Nevertheless, no hominid modifications were documented in this assemblage. Comparisons with other Olduvai sites indicate a more conspicuous hyena taphonomic signal during Bed II times than Bed I times, which appears to mirror the changing configuration of the large carnivore guild. These findings also beg the question of what activities were being carried out by hominids with the stone tools discarded at these sites. Although it seems clear that hominids were utilizing stone tools to carry out subsistence activities unrelated to carcass butchery, more excavation and techniques such as phytolith analysis should be employed to explore alternative explanations.     

Hominid carnivory and foraging strategies, and the socio-economic function of early archaeological sites.

New evidence for the tissue types exploited by early hominids from carcasses possibly acquired through scavenging is derived from the larger mammal bone assemblages from FLK I, level 22 (Zinjanthropus floor), and FLKN levels 1 and 2 from Bed I, Olduvai Gorge, Tanzania. Published skeletal part profiles from the two archaeological sites are evaluated using (i) modern observations on the sequence by which carnivores consume carcass parts in order to assess the timing of hominid access to carcasses, and (ii) measurements of flesh and marrow yields to assess the tissue types sought and acquired. These results suggest that the maximization of marrow (fat) yields, not flesh (protein) yields, was the criterion shaping decisions about carcass processing. Because of evidence for density-dependent destruction of some flesh-bearing parts by scavengers of the hominid-butchered assemblages, however, it is uncertain whether carcass parts were transported and acquired by hominids in a largely defleshed condition. The results on tissue types acquired are combined with a discussion of predation risk, feeding competition and the equipment needs of carcass processing in an attempt to identify archaeological test implications of competing hypotheses for the socio-economic function of the earliest archaeological sites.     

New estimates of tooth mark and percussion mark frequencies at the FLK Zinj site: the carnivore-hominid-carnivore hypothesis falsified

Traditional interpretations of hominid carcass acquisition strategies revolve around the debate over whether early hominids hunted or scavenged. A popular version of the scavenging scenario is the carnivore-hominid-carnivore hypothesis, which argues that hominids acquired animal resources primarily through passive opportunistic scavenging from felid-defleshed carcasses. Its main empirical support comes from the analysis of tooth mark frequency and distribution at the FLK Zinj site reported by Blumenschine (Blumenschine, 1995, J. Hum. Evol. 29, 21-51), in which it was shown that long bone mid-shafts exhibited a high frequency of tooth marks, only explainable if felids had preceded hominids in carcass defleshing. The present work shows that previous estimates of tooth marks on the FLK Zinj assemblage were artificially high, since natural biochemical marks were mistaken for tooth marks. Revised estimates are similar to those obtained in experiments in which hyenas intervene after humans in bone modification. Furthermore, analyses of percussion marks, notches, and breakage patterns provide data which are best interpreted as the results of hominid activity (hammerstone percussion and marrow extraction), based on experimentally-derived referential frameworks. These multiple lines of evidence support previous analyses of cut marks and their anatomical distribution; all indicate that hominids had early access to fleshed carcasses that were transported, processed, and accumulated at the FLK Zinj site.     

Crocodylian and mammalian carnivore feeding traces on hominid fossils from FLK 22 and FLK NN 3, Plio-Pleistocene, Olduvai Gorge, Tanzania

Taphonomic analysis of the Olduvai Hominid (OH) 8 left foot from FLK NN Level 3 and the OH 35 left leg from FLK Level 22 (Zinjanthropus level) in Middle Bed I, Olduvai Gorge, indicates that both were fed upon by crocodiles. Both bear extensive tooth marking, including bisected tooth marks diagnostic of crocodylian feeding. The location of the bisected tooth marks on the distal tibia and the talus indicates disarticulation of the foot by crocodiles. The broken proximal ends of the tibia and fibula are more typical of feeding by a leopard-like carnivore, as is damage to the OH 7 mandible and parietals that are associated with and may derive from the same individual as OH 8. Previous work showing a close articulation of the foot and the leg has been used to suggest that the two specimens belong to the same individual despite deriving from sites separated by 200 m and slightly different stratigraphic levels according to previous work. The location and agent of tooth marking and the nature of gross damage do not refute this hypothesis, but the punctures on the talus and distal tibia differ in size and sharpness. Recent work shows that the stratigraphic discrepancy between OH 8 and OH 35 is greater than previously thought, refuting the single-individual hypothesis. Although seemingly unlikely, this denotes that two hominids represented by rarely found leg and foot elements both lost their left foot to crocodiles at nearby sites within a 6,000 year interval. We cannot determine if the hominids were preyed upon by crocodiles or mammalian carnivores. However, the carnivore damage to them and associated faunal remains suggests that high predation risk constrained hominid activities involving discard of the stone artifacts found at these sites. This finding is inconsistent with the interpretation of the sites as home bases or living floors.     

Beyond leopards: tooth marks and the contribution of multiple carnivore taxa to the accumulation of the Swartkrans Member 3 fossil assemblage

The ca. 1.0 myr old fauna from Swartkrans Member 3 (South Africa) preserves abundant indication of carnivore activity in the form of tooth marks (including pits) on many bone surfaces. This direct paleontological evidence is used to test a recent suggestion that leopards, regardless of prey body size, may have been almost solely responsible for the accumulation of the majority of bones in multiple deposits (including Swartkrans Member 3) from various Sterkfontein Valley cave sites. Our results falsify that hypothesis and corroborate an earlier hypothesis that, while the carcasses of smaller animals may have been deposited in Swartkrans by leopards, other kinds of carnivores (and hominids) were mostly responsible for the deposition of large animal remains. These results demonstrate the importance of choosing appropriate classes of actualistic data for constructing taphonomic inferences of assemblage formation. In addition, they stress that an all-encompassing model of assemblage formation for the hominid-bearing deposits of the Sterkfontein Valley is inadequate and that each must be evaluated individually using not just analogical reasoning but also incorporating empirical data generated in the preserved fossil samples.     

Validation of bone surface modification models for inferring fossil hominin and carnivore feeding interactions, with reapplication to FLK 22, Olduvai Gorge, Tanzania

Resolving the issue of how Early Stone Age hominins acquired large mammal carcasses requires information on their feeding interactions with large carnivores. This ecological information and its behavioral and evolutionary implications are revealed most directly from the tooth, cut, and percussion marks on bone surfaces generated by hominin and carnivore feeding activities. This paper employs a bootstrap method, a form of random resampling with replacement, to refine published neotaphonomic models that use the assemblage-wide proportions of long bones bearing feeding traces to infer the sequences in which Plio-Pleistocene hominins and carnivores accessed flesh, marrow, and/or grease from carcasses. Results validate the sensitivity of the models for inferring hominin feeding ecology, which have been questioned on grounds shown here to be unfounded. The bootstrapped feeding trace models are applied to the late Pliocene larger mammal fossil assemblage from FLK 22 (Zinjanthropus site), Olduvai Gorge, Tanzania. High frequencies of tooth and percussion marking on long bone midshaft fragments from FLK 22 are most consistent with those feeding trace models that simulate hominin scavenging from carcasses defleshed by carnivores, while cut mark data indicate that hominins more often had access to upper forelimb flesh than upper hind limb flesh. Together, the bone surface modification data indicate that hominins typically gained secondary access to partially defleshed carnivore kills, but they also allow for the possibility of some carcasses being processed only by carnivores and only by hominins.     

Scavenging or Hunting in Early Hominids: Theoretical Framework and Tests

Evidence from Bed I, Olduvai, supports the hypothesis that scavenging, not hunting, was the major meat-procurement strategy of hominids between 2 and 1.7 million years ago. Data used to evaluate the hunting and scavenging hypotheses are derived from studying cut marks on Bed I bovids, comparing adaptations necessary for scavenging with those of early hominids, and a pa-leoecological reconstruction of Bed I carcass biotnass, carnivore guild, and hominidforaging area.     

Intrinsic qualities of primate bones as predictors of skeletal element representation in modern and fossil carnivore feeding assemblages

Plio-Pleistocene faunal assemblages from Swartkrans Cave (South Africa) preserve large numbers of primate remains. Brain, C.K., 1981. The Hunters or the Hunted? An Introduction to African Cave Taphonomy. University of Chicago Press, Chicago suggested that these primate subassemblages might have resulted from a focus by carnivores on primate predation and bone accumulation. Brain's hypothesis prompted us to investigate, in a previous study, this taphonomic issue as it relates to density-mediated destruction of primate bones (J. Archaeol. Sci. 29, 2002, 883). Here we extend our investigation of Brain's hypothesis by examining additional intrinsic qualities of baboon bones and their role as mediators of skeletal element representation in carnivore-created assemblages. Using three modern adult baboon skeletons, we collected data on four intrinsic bone qualities (bulk bone mineral density, maximum length, volume, and cross-sectional area) for approximately 81 bones per baboon skeleton. We investigated the relationship between these intrinsic bone qualities and a measure of skeletal part representation (the percentage minimum animal unit) for baboon bones in carnivore refuse and scat assemblages. Refuse assemblages consist of baboon bones not ingested during ten separate experimental feeding episodes in which individual baboon carcasses were fed to individual captive leopards and a spotted hyena. Scat assemblages consist of those baboon bones recovered in carnivore regurgitations and feces resulting from the feeding episodes. In refuse assemblages, volume (i.e., size) was consistently the best predictor of element representation, while cross-sectional area was the poorest predictor in the leopard refuse assemblage and bulk bone mineral density (i.e., a measure of the proportion of cortical to trabecular bone) was the poorest predictor in the hyena refuse assemblage. In light of previous documentation of carnivore-induced density-mediated destruction to bone assemblages, we interpret the current findings as suggestive of the secondary importance of bulk bone mineral density to other intrinsic qualities of skeletal elements (e.g., size, maximum dimension, and average cross-sectional area). It is only when skeletal elements are too large for consumption (e.g., many long bones) that they are fragmented following intra-element patterns of density-mediated carnivore destruction. There appears to be a size threshold beneath which bulk bone mineral density contributes little to mediating carnivore destruction of carcasses. Thus, depending on body size of the predator, body size of the prey, and specific size of the element, bulk bone mineral density may play little or no role of primary importance in mediating the destruction of skeletal elements. We compare patterns in modern comparative assemblages to patterns in primate fossil assemblages from Swartkrans. One of the fossil assemblages, Swartkrans Member 1, Hanging Remnant, most closely approximates a hyena (possibly refuse) assemblage pattern, while the Swartkrans Member 2 assemblage most closely approximates a leopard (possibly scat) assemblage pattern. The Swartkrans Member 1, Lower Bank, assemblage does not closely approximate any of our modern comparative assemblage patterns.     

Examining time trends in the Oldowan technology at Beds I and II,Olduvai Gorge

The lithic analysis of the Bed I and II assemblages from Olduvai Gorge reveals both static and dynamic time trends in early hominids' technology from 1.8 to 1.2 m.y.a. The Bed I Oldowan (1.87-1.75 m.y.a.) is characterized by the least effort strategy in terms of raw material exploitation and tool production. The inclusion of new raw material, chert, for toolmaking in the following Developed Oldowan A (DOA, 1.65-1.53 m.y.a.) facilitated more distinctive and variable flaking strategies depending on the kind of raw materials. The unique characters of DOA are explainable by this raw material factor, rather than technological development of hominids. The disappearance of chert in the subsequent Developed Oldowan B and Acheulian (1.53-1.2 m.y.a.) necessitated a shift in tool production strategy more similar to that of Bed I Oldowan than DOA. However, the evidence suggests that Bed II hominids might have been more skillful toolmakers, intensive tool-users, and engaged in more active transport of stone tools than the Bed I predecessors. Koobi Fora hominids maintained a more static tool-using behavior than their Olduvai counterparts due mainly to a stable supply of raw materials. They differed from Olduvai hominids in terms of less battering of cores, consistent transport behavior, and few productions of side-struck flakes, indicating a regional variation of toolmaking and using practice. However, they shared with Olduvai hominids a temporal trend toward the production of larger flakes from larger cores after 1.6 m.y.a. Increased intake of animal resources and the expansion of ranging area of Homo ergaster would have led to the development of technological organization. Technological changes in the Oldowan industry are attested at Olduvai Gorge, Koobi Fora, and Sterkfontein, suggesting that it was a pan-African synchronous phenomenon, beginning at 1.5 m.y.a.     

Research on late Pliocene Oldowan sites at Kanjera South, Kenya

The late Pliocene is notable for the appearance of two new hominid genera as well as the first archaeological sites, generally attributed to the Oldowan Industrial Complex. However, the behavioral ecology of Oldowan hominids has been little explored, particularly at sites older than 2.0 Ma. Moreover, debates on Oldowan hominid foraging ecology and behavior have centered on data from only two regions, and often from single site levels. Here we describe the preliminary results of our investigation of Oldowan occurrences at Kanjera South. These occurrences preserve the oldest known traces of hominid activity in southwestern Kenya, and unlike most of the Oldowan sites in the 2.0-2.5 Ma time interval, artefacts are found in spatial association with a well-preserved fauna. In 1996 and 1997, this project initiated the first excavation program for Kanjera South. Magneto- and biostratigraphy indicate that deposition began approximately 2.2 Ma, substantially earlier than previously thought. At Excavation 1, artefacts were found in spatial association with a taxonomically diverse faunal assemblage in Beds KS-1 and KS-2. Excavation 2 yielded a partial hippopotamus axial skeleton with artefacts in KS-3. Cores from both sites were incidentally flaked and represent a Mode I lithic technology indistinguishable from the Oldowan. Approximately 15% of the artefacts were manufactured from non-local raw materials, indicating a flow of resources into the area. Stable isotopic analysis of KS-1 and KS-2 pedogenic carbonates suggests that the Excavation 1 assemblages formed in a relatively open (>75% C4 grass) habitat. The Excavation 1 and 2 faunas contain a high proportion of equids relative to Oldowan accumulations from Bed I Olduvai Gorge, Tanzania. Beds KS-1 and KS-2 thus preserve traces of Oldowan hominid activities in a more open setting than has been previously documented.     

Archaeological predictions for hominid land use in the paleo-Olduvai Basin, Tanzania, during lowermost Bed II times

We present a preliminary predictive model of Oldowan stone artefact and scavenged larger mammal bone assemblages for 11 landscape facets modeled earlier to occur across a large portion (>300 km²) of the paleo-Olduvai Basin during lowermost Bed II times. This second phase of model-building is based on our earlier characterizations of the basin's landscape ecostructure and the inter-facet distribution of key resources and hazards probably encountered by Late Pliocene hominids (Peters & Blumenschine, 1995, 1996). Our current extension of the model of hominid–landscape interactions specifies additional theoretical components, including: (1) the assumed capabilities of Oldowan hominids (presumablyHomo habilis, primarily); (2) the landscape-facet-specific tasks they carried out; (3) the immediate stone and bone task residues they produced; and (4) the predicted composition, condition, density, and clustering of stone artefact and butchered and unbutchered bone assemblages for each facet. We develop ecological linkages between these new and formerly reported modeling components, the most fundamental of which is the facet-specific degree of tree/shrub cover abundance, and the correlated degree of competition among larger carnivores and hominids for scavengeable larger mammal carcasses. These factors condition variability among landscape facets in scavenging opportunities encountered by hominids, which in our model is the major predictor of bone and stone artefact assemblage composition. The predictive value of scavenging reflects the bias of paleoanthropological traces toward technology and butchery in their landscape context, but the model is surprisingly insensitive to what are usually thought to be critical social components of hominid land use. The predictions for the traces of hominid–landscape interactions modeled herein can be tested in the future against the landscape archaeological sample being excavated from lowermost Bed II by the Olduvai Landscape Paleoanthropology Project.     

A revised stratigraphic framework for Olduvai Gorge Bed I based on tuff geochemistry

Olduvai Gorge, Tanzania has rich records of Plio-Pleistocene fauna and flora, hominin fossils, and stone artifacts preserved between well-dated tephra layers (tuffs). Accurate correlation between sites in the two million year section is complicated by faulting, erosion, change in physical appearance of the tuffs, and quality of preservation. Traditional tuff geochemical techniques using glass cannot be applied because of poor glass preservation, and previous physical mapping has led to miscorrelations in Bed I. A new approach, using a combination of glass and mineral compositions (feldspar, augite, hornblende, and oxides) produced successful geochemical fingerprints for all ten major Bed I (∼2.03-1.79 Ma) tuffs. These fingerprints make available a reliable means for correlating specific tuffs between the well-dated "Junction" sites, such as FLK and HWK, and less well-dated sites at the basin margins. The new correlations provide a high-resolution stratigraphic framework for Bed I and correct previous miscorrelations in the west of the basin. Olduvai Hominin 65, from western Olduvai, was recovered from a level between Tuff IC and the Ng’eju Tuff, and therefore dates to 1.79-1.84 Ma.     

Soricidae (Mammalia) from the Early Pleistocene of Olduvai Gorge, Tanzania

Soricid remains collected from Bed I of Olduvai Gorge are described. The great majority of the specimens are mandibles. A survey of the mandibles of living African species revealed many differences in characters of the lower teeth and jaw that can be used for identification. On the basis of these characters, nine species are distinguished in the Olduvai collection, of which six are well enough preserved to permit a discussion of their relationships to living species. Three new species and one new subspecies are described. All the Olduvai shrews differ in some respects from their nearest living relatives; three species are close to species from Makapansgat, Swartkrans and Sterkfontein, RSA, though there appear to be slight differences. A change in the soricid fauna takes place within Bed I, interpreted as due to increasing aridity.     

The puzzle from JK2—A femur and a tibial fragment (O.H.34) from Olduvai Gorge,Tanzania

During excavations in Bed III, Olduvai Gorge, Tanzania in 1962, a slender fossil femur and part of a tibial shaft were recovered. Their strange appearance resulted in their neglect for many years. Anatomical examination now confirms that these bones are hominid, probably hominine, yet distorted in form. The distortion does not appear to be the result of pathology but due to damage and abrasion while in the deposits; deposits dated at approximately 1 million years B.P. O.H. 34 is the first hominid to be recovered from Bed III, Olduvai Gorge.     

Faunal assemblage seriation of southern African Pliocene and Pleistocene fossil deposits

Fossil assemblages from the Pliocene and Pleistocene of southern Africa were seriated in order to give a better idea of their relative chronology. Time-sensitive mammals were selected for calculation of the Faunal Resemblance Index among 17 site units. On the basis of a logistical seriation and subsequent site analysis, the following sequence of sites was deemed most probable: Makapansgat Member 3, Makapansgat Member 4, Taung Dart deposits, Sterkfontein Member 4 and Taung Hrdli?ka deposits, Sterkfontein Member 5 (in part) and Kromdraai B, Kromdraai A and Swartkrans Member 1, Swartkrans Member 2, Swartkrans Member 3, Plovers Lake, Cornelia, Elandsfontein Main Site, Cave of Hearths Acheulian levels, Florisbad and Equus Cave and Klasies River Mouth. © 1995 Wiley-Liss, Inc.     

Role of carnivores in the accumulation of the Sterkfontein Member 4 hominid assemblage: a taphonomic reassessment of the complete hominid fossil sample (1936-1999)

New taphonomic data on the Sterkfontein Member 4 (South Africa) fossil hominid assemblage are presented. The previous estimate of hominid individuals represented in the deposit (45) is increased to 87. New minimum numbers of hominid skeletal elements are provided, and incidences of bone surface damage inflicted by prehistoric biological agents are summarized. The hominid sample from Member 4 is composed predominately of gnathic remains and has a paucity of postcrania. This dearth of postcrania limits, to some extent, inferences about the formation of the Sterkfontein assemblage. However, carnivore tooth marks on some fossil specimens and an overall broad similarity in patterns of skeletal part representation between Sterkfontein and primate bone assemblages created by extant carnivores suggest that carnivores did have some involvement in the accumulation of the fossil hominid assemblage. Thus, this study provides support for the “carnivore‐collecting hypothesis” of Brain (Brain [ 1981 ] The Hunters or the Hunted? Chicago: University of Chicago Press), which implicates large carnivores as prominent collecting agents of hominid body parts in Sterkfontein Member 4. Evidence of bone surface damage is, however, too scant to make confident inferences about specific carnivore taxon/taxa involved in hominid bone collection at the site. Am J Phys Anthropol, 2004. © 2004 Wiley‐Liss, Inc.     

Long-distance carcass transport at Olduvai Gorge? A quantitative examination of Bed I skeletal element abundances

Relative abundances of skeletal elements at Plio-Pleistocene archaeological sites have long been interpreted to represent selective transport of portions of large prey. Models from optimal foraging theory suggest that the degree of carcass transport selectivity reflects transport constraints, particularly transport distance. A quantitative analysis of skeletal element abundances in five bone assemblages from Bed I, Olduvai Gorge, indicates that within the subset of elements most likely to resist attritional processes, there is no evidence for preferential transport of small or large mammals. The results suggest relatively low carcass transport costs and are most consistent with site formation models favoring short-distance carcass transport. The data are also consistent with the possibility that hominins were not responsible for transporting bones at some sites. Several Bed I assemblages, with the exception of FLK-Zinjanthropus, lack evidence of a functional relationship between flaked stone artifacts and the faunal remains, such as cut-marks or percussion-marks on bone. In conjunction with the skeletal part data, this suggests that hominin involvement with the bone assemblages was minimal at all sites but FLK-Zinjanthropus. The patterning at Bed I contrasts strongly with Middle Stone Age and Middle Paleolithic assemblages, which provide clear evidence for selective transport, suggesting higher transport costs and longer transport distances.     

Environments and hominin activities across the FLK Peninsula during Zinjanthropus times (1.84 Ma), Olduvai Gorge, Tanzania

We establish through 13 excavations the landscape context and nature of hominin activities across the Zinjanthropus land surface from which the Leakeys recovered the FLK 22 and FLK NN 1 paleoanthropological assemblages. The land surface was created by fluvial incision of the eastern margin of paleo-Lake Olduvai following a major lake withdrawal. Erosion was uneven, leaving a peninsula bounded by a river channel, the FLK Fault, and a freshwater wetland. This FLK Peninsula supported groves of trees that attracted hominins and carnivores, and that preserved the dense concentrations of carcass remains and stone tools they left behind, including those at FLK 22. Some carcasses appear to have been acquired at the ecotone of the Peninsula and Wetland, where another dense artifact and bone assemblage accumulated. A lesser topographic high at the edge of a Typha marsh in the Wetland was the site of FLK NN 1 and a scatter of large stone tools used possibly for rootstock processing.Our landscape reconstruction delimits the vegetation mosaic indicated by previous work and provides a topographical explanation for the existence of FLK 22 and FLK NN 1. Both are unexpected if the FLK area was the flat, featureless lake margin terrain typical of lake basins similar to paleo-Olduvai. The results show that the Leakeys’ sites were not isolated occupation floors but rather parts of a land surface utilized intensively by hominins. Although commonly considered to have been home bases, their likely high predation risk, evidenced by large carnivore feeding traces and the remains of four hominin individuals, suggests visits to them were brief and limited to feeding. Finally, stratigraphic observations confirm that FLK NN 3 accumulated on an older land surface, refuting the hypothesis that the OH 8 foot found there is the same individual as the OH 35 leg from FLK 22.     

Vegetation during UMBI and deposition of Tuff IF at Olduvai Gorge, Tanzania (ca. 1.8 Ma) based on phytoliths and plant remains

As part of ongoing research at Olduvai Gorge, Tanzania, to determine the detailed paleoenvironmental setting during Bed I and Bed II times and occupation of the basin by early hominins, we present the results of phytolith analyses of Tuff IF which is the uppermost unit of Bed I. Phytoliths were identified in most of the levels and localities on the eastern paleolake margin, but there are not always sufficient numbers of identifiable morphologies to infer the specific type of vegetation due to dissolution. Some surge surfaces and reworked tuff surfaces were vegetated between successive ash falls, as indicated by root-markings and the presence of a variety of phytolith morphotypes. Dicotyledonous wood/bark types were dominant except at the FLK N site just above Tuff IF when monocots are dominant and for the palm-dominated sample from the reworked channel cutting down into Tuff IF at FLK N. The area between the two fault scarps bounding the HWK Compartment, approximately 1 km wide, was vegetated at various time intervals between some of the surges and during the reworking of the Tuff. By lowermost Bed II times the eastern margin was fully vegetated again. Climate and tectonic activity probably controlled the fluctuating lake levels but locally the paleorelief and drainage were probably the controlling factors for the vegetation changes. These data support a scenario of small groups of hominins making brief visits to the paleolake during uppermost Bed I times, followed by a more desirable vegetative environment during lowermost Bed II times.