Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya

Quaternary International xxx (2017) 1e16

Contents lists available at ScienceDirect

Quaternary International journal homepage: www.elsevier.com/locate/quaint

Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya Mary E. Prendergast a, *, Amanuel Beyin b a b

Radcliffe Institute for Advanced Study, Harvard University, Cambridge, MA 02138, USA Department of Anthropology, University of Louisville, Louisville, KY 40292, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 19 January 2017 Received in revised form 16 March 2017 Accepted 24 April 2017 Available online xxx

During the African Humid Period (AHP; c. 15e5.5 ka), the rivers and lakes of much of the continent swelled due to changes in monsoonal rainfall driven by Earth's orbital precession. This period witnessed the growth of diverse fisher-forager communities, whose members adapted their settlement patterns and created new technologies in order to take advantage of aquatic resources. Around Lake Turkana in northern Kenya, numerous surface sites have been documented along former shorelines dating to the AHP. Relatively few have been excavated and dated however, and just three e all from the eastern basin e have published faunal analyses. Here, we present archaeofaunal assemblages from the Kalokol region of the western basin, where three sites with microlithic technology, bone harpoons, and radiocarbon dates falling within the AHP were excavated. We present a detailed taphonomic assessment of the fish assemblages and a comparison with both natural and anthropogenic, and ancient and modern, fish bone accumulations. Taxa identified at the Kalokol sites are discussed in terms of the occupants' possible fishing technologies and strategies, drawing on ethological and ethnographic data. Our analysis, combining our data with those published from the eastern basin, enables a broader discussion of how people may have responded to fluctuating AHP environments in the Turkana Basin. © 2017 Elsevier Ltd and INQUA. All rights reserved.

Keywords: Archaeology Zooarchaeology Fishers African humid period East Africa

1. Introduction Across much of the African continent, the transition from the terminal Pleistocene to early Holocene was marked by major environmental transformations, providing people with new economic opportunities and ultimately shaping social changes. The African Humid Period (AHP; c. 15e5.5 ka) led to the swelling of rivers and lakes across much of the northern half of the continent (DeMenocal et al., 2000). Perhaps the best-known effect of the AHP is the so-called “Green Sahara,” where aquatic and savanna fauna are beautifully captured in the rock art of today's deserts (Le Quellec, 1993). The AHP led to the expansion of lakes not only in the Sahara and Sahel, but also along the Rift Valley from Ethiopia to Tanzania (Street-Perrott et al., 1989; Tierney et al., 2011). Around many of these lakes, and along rivers such as the Niger and Nile, fishing-based communities emerged. Archaeological indicators of aquatic resource specialization across much of northern Africa e in

* Corresponding author. E-mail addresses: [email protected] (M.E. Prendergast), amanuel. [email protected] (A. Beyin).

the form of fish remains, barbed bone points, and early pottery e were once interpreted as evidence of a widespread “cultural complex” termed the “Aqualithic” (Sutton, 1974, 1977). In this view, abundant aquatic resources drove early Holocene sedentism, technological specialization, culinary innovation, and social change, just as agriculture drove the same in the Near East. While the “Aqualithic” hypothesis has been critiqued for imposing an inferred cultural homogeneity across a vast and archaeologically heterogeneous expanse, Sutton's ideas inspired new investigations and improved understandings of sites dating to the AHP. In the Turkana Basin of northern Kenya, excavations in the 1970s-80s produced new evidence for early Holocene occupations along former shorelines (Robbins, 1974, 1975; Phillipson, 1977; Angel et al., 1980; Robbins, 1984; Barthelme, 1985), and provided the basis for the first comprehensive study of prehistoric fishing in eastern Africa (Stewart, 1989). Since the publication of Stewart's (1989) zooarchaeological investigation of faunal assemblages from Turkana Basin sites, no further archaeological study has been made of Holocene-era fishing in the region. In this paper, we present an in-depth taphonomic and zooarchaeological analysis of faunal assemblages from three recently excavated sites in the Kalokol area of the western Turkana Basin. The sites are

http://dx.doi.org/10.1016/j.quaint.2017.04.022 1040-6182/© 2017 Elsevier Ltd and INQUA. All rights reserved.

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

2

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

characterized by microlithic technology, barbed bone points, and in one case, pottery, and have produced radiocarbon dates falling within the AHP. They provide an excellent point of comparison to the eastern basin sites, and these samples are jointly considered alongside published data from modern fishing camps and both modern and ancient natural fish bone accumulations. Collectively these datasets enable an assessment of the taphonomic history of the Kalokol sites. The present analysis, while limited by small samples and insufficient chronological resolution, sheds light on fishing-foraging adaptations at the Pleistocene-Holocene transition. This period has come into the spotlight recently due to the discovery of remains of 28 human individuals at Nataruk, c. 90 km south of our study area, where an early Holocene “massacre” is attributed to possible inter-group competition over resources
n Lahr et al., 2016). This suggestion e made in the absence (Mirazo of subsistence data e makes it all the more imperative to clarify early Holocene fisher-foragers’ economic strategies during a time of climatic fluctuation. 2. Environmental and cultural changes during the African Humid Period 2.1. Orbital and environmental shifts Increased humidity at the Pleistocene-Holocene transition was caused by shifts in Earth's orbital cycles, specifically precession of the equinoxes (DeMenocal et al., 2000). This describes a wobble in the Earth's axis that occurs in approximate 20,000-year cycles, effectively determining the annual timing of the Earth's most proximal pass of the Sun. In the early Holocene, precession led to an increase in Northern Hemisphere summer insolation. Consequent heating of the landmass changed monsoonal wind and rainfall patterns, with one study suggesting that the estimated 7% increased summertime radiation could have increased rainfalls by 17%e50% (Kutzbach and Liu, 1997). Ultimately, this led to the characteristic traits of the AHP: swelling of rivers, and expansion and in some cases connection of lakes; these effects last until the mid-Holocene (c. 5000 BP), when Northern Hemisphere insolation decreased and conditions became drier. These shifts are well documented through a wide range of geological, micro- and macro-fossil, sedimentary, and isotopic proxies (e.g., Gasse, 2000; Gasse et al., 2008; DeMenocal and Tierney, 2012), and can be correlated with key events in African prehistory (Kuper and € pelin, 2006). Kro Despite the overall humid conditions of the AHP, there were important millennial- or even decadal-scale shifts in rainfall, and consequently in lake levels, during the early Holocene. These were caused by minor shifts in the Earth's orbit and solar radiation, and at the decadal level, by the Indian Ocean dipole (Marchant et al., 2007). Some of the larger-scale fluctuations are documented in the Turkana Basin, where extensive geological work and other research has aimed to reconstruct lake-level history (e.g., Butzer et al., 1972; Owen et al., 1982; Garcin et al., 2012; Forman et al., 2014; Bloszies et al., 2015). Correlating these reconstructed lake levels with the archaeological evidence for human occupations along paleoshorelines is a priority of current research in the Tur
n Lahr et al., kana Basin (Ashley et al., 2011; Beyin, 2011; Mirazo 2016; Wright et al., 2015), and demands a level of chronological resolution currently lacking at early Holocene sites (Beyin et al. in press). 2.2. Cultural innovations during the AHP Some archaeological sites dating to the terminal Pleistocene and early Holocene across Saharan, Sahelian and Great Lakes Africa

appear to share certain traits in common (Fig. 1). Many of them are located near paleoshorelines of existing lakes (e.g., Chad, Turkana, and Victoria), within the catchments of paleolakes that no longer exist in the formerly “Green” Sahara, or within the catchments of major rivers (e.g., Niger, Nile). Many have archaeofaunal evidence for aquatic resource exploitation: abundant fish bones, as well as more limited samples of crocodile, hippopotamus, turtles, and/or water birds. Some of these animals, as well as those inhabiting savannas (e.g., giraffe, elephant), are depicted in Saharan rock art. Sites from a wide geographic range (Morocco to Botswana) and long chronological timespan have produced barbed bone points (Yellen, 1998). While these are usually interpreted as tips of fishing spears or harpoons, in one case (Daima, Chad Basin) a bone point was used as weapon (Connah, 1981), while in the recently reported “massacre” at Nataruk in the Turkana Basin, the use of bone points
n Lahr et al., 2016:399). as possible weapons is suggested (Mirazo Numerous sites dating to c. 9000-8000 BP in the Nile Valley and the Sahara-Sahel have early pottery, in addition to evidence for aquatic resource exploitation (Barich, 1987; Close, 1995; Mohammed-Ali and Khabir, 2003; Garcea, 2006; Barich, 2013).
cor, a Some but not all of these assemblages share “wavy-line” de trait which also appears on limited numbers of sherds in the Turkana Basin, possibly after c. 8000 BP (Robbins, 1972; Barthelme, 1985). The apparent similarities among early Holocene sites led Sutton (1974, 1977) to boldly put forward a hypothesis that they might belong to a shared “cultural complex”, termed “Aqualithic,” which he saw as a viable and equally important alternative to agriculture in resource-rich environments. Fishing, in particular, was argued to have enabled a kind of semi-permanent settlement leading to the development of some of the same technologies e grinding stones, pottery e seen in agricultural societies. Sutton advanced that “Aqualithic” groups across the continent (possibly linked through migrations) might not only share ceramic and bone technologies and subsistence strategies, but might also have a common ethnolinguistic identity, a hypothesis which he argued remained to be tested. Sutton's broader intention was to stimulate research, and he made it clear that he was laying out a proposition for future work and debate. In this sense, the “Aqualithic” papers were successful. For example, Sutton inspired productive discussions on the role of aquatic resources in the development of pottery (the “fish stew revolution”) and on the social changes that may have accompanied this development (e.g., Haaland, 1992, 1993, 1997, 2007, 2009). However, the hypothesis also generated criticism. Increased fieldwork led to larger samples with which to examine the notion of a widespread cultural complex (Holl, 2005). These samples have shown substantial regional variation in ceramic and barbed bone _ point traditions (e.g., Barich, 1987; Krzyzaniak et al., 1993; Yellen, 1998; Dale, 2007). Lithic analyses suggest local continuity from late Pleistocene to early Holocene technological traditions (e.g., Phillipson, 2005; Seitsonen, 2010). Consumption of freshwater fish also has a long history throughout much of the continent (e.g., Van Neer, 1986; Stewart, 1994; Plug and Mitchell, 2008; Trapani, 2008; Braun et al., 2010; Linseele and Zerboni, this volume). The shift to a large number of aquatic-focused settlements during the AHP may thus be a question of scale, with sites not only more numerous but also more visible to archaeologists today, since they are located on exposed paleoshorelines. In many areas the joining of lakes during the AHP would have also led to species exchanges, creating greater biodiversity and, by extension, potentially greater opportunities for aquatic resource exploitation (Stewart, 1989). Despite the importance of aquatic resources to AHP economic and social shifts, faunal analyses have been limited to a handful of well-researched areas, with the Nile River and its tributaries being

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

3

Fig. 1. Map of Africa, indicating the continent's present-day largest lakes and rivers (thus not including Saharan paleolakes), and showing areas with early to middle Holocene settlements containing evidence for aquatic exploitation. These sites have faunal assemblages with abundant fish or other aquatic animal remains (see discussion in text). Some but not all of these sites are associated with barbed bone points and/or wavy-line pottery. Site locations are based on reviews by Haaland (1993), Yellen (1998), Holl (2005), Phillipson (2005), and Barich (2013). See Fig. 2 for detail of Turkana Basin. Basemap from Natural Earth Data (www.naturalearthdata.com).

best known (Peters, 1991, 1995; Van Neer, 1994; Chaix, 2003; Van Neer, 2004; Linseele and Zerboni, this volume). In eastern Africa, fishing was an important component of AHP subsistence strategies in at least four basins: Lake Turkana (Stewart, 1989), Lake Nakuru (Leakey, 1931), Lake Victoria (Robertshaw et al., 1983; Prendergast and Lane, 2010), and Lake Rutanzige (ex-Edward) (Stewart, 1989); with the exception of Nakuru, all have witnessed zooarchaeological work. In the Turkana Basin, Stewart's (1989) pioneering faunal analyses laid out a number of criteria for distinguishing natural and anthropogenic sites, and she used ethological and ethnographic data to infer fishers' strategies and technologies. Her work, though limited by the small number of excavated sites (as opposed to surface collection) and few dates, remains influential.

3. Background to study area 3.1. Research history The Turkana Basin is best-known for numerous sites of paleoanthropological importance, as summarized by Harris et al. (2006). Although the region's later prehistory is sometimes overlooked, beginning in the 1970s a number of pioneering projects e mainly though not exclusively focused on the eastern basin e documented important environmental and cultural shifts from the terminal Pleistocene through the Holocene. These included early Holocene fishing settlements (Robbins, 1972, 1974; Phillipson, 1977; Angel et al., 1980; Barthelme, 1985), as well as the mid-Holocene arrival

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

4

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

of livestock herders (Marshall et al., 1984; Barthelme, 1985) and the related construction of megalithic pillar sites, with or without elaborate burials (Nelson, 1995). Later Holocene sites, created by people who were herding, fishing and/or foraging, were also documented (Robbins, 1980, 1984). The region's later prehistory has seen renewed interest in recent years, particularly in the western basin with two active research programs, the Later Prehistory of West Turkana Project (LPWT), of which the present research is a part, and the In Africa project. These efforts have led to new investigations of mid-Holocene pillar sites (Hildebrand et al., 2011; Hildebrand and Grillo, 2012; Grillo and Hildebrand, 2013), and to discovery of the aforementioned human remains at Nataruk, interpreted e not without controversy e
n Lahr et al., 2016; Cf. Stojanowski et al., as a massacre (Mirazo 2016). The eastern basin has also seen renewed interest in the later prehistory of the Koobi Fora area and especially on obsidian exchange networks (Ndiema et al., 2010, 2011; Ashley et al., 2011). Zooarchaeological research in the Turkana Basin has been very limited. The only faunal analyses for the AHP are those of Stewart (1989) at Koobi Fora sites (FxJj12, GaJi3, and several surface sites) and at Lowasera, both in the eastern basin (Fig. 2). While Lothagam in the western basin produced abundant fish remains (>7000 specimens), only a fraction of these were examined, and informal,

mainly non-quantitative data were published (Lynch and Robbins, 1972:52; Robbins, 1974:204; Stewart, 1989:155e158). 3.2. Environmental setting Lake Turkana is a large and shallow lake whose evaporation greatly exceeds local precipitation. It receives nearly all its water from the Omo River, and therefore fluctuations in precipitation in the Ethiopian highlands have major impacts on its water levels. The current lake lies at about 362 m asl, a significant reduction from AHP levels (Bloszies et al., 2015). During the AHP, the lake expanded and reconnected with the Nile multiple times, enabling a biomass exchange that helped build rich aquatic resources (Stewart, 1989). Deposits dating to this era are found in what is known as the Galana Boi Formation. Galana Boi deposits are characterized by poorly consolidated clayey silts, sands, and gravels, with abundant shell and fish bone inclusions, in particular the shells of Melanoides tuberculata. The Kalokol study area, at c. 420e460 m above sea level (asl), covers ~210 sq. km and is largely defined by a low semi-arid landscape and the Lothidok hills. These Miocene-era basalt hills are cut by the Kalokol River to form the Kadokorynang Gorge, where Galana Boi deposits are well-exposed. Elsewhere, these deposits are seen in isolated patches of shell-strewn gravelly beds. Today, the landscape is arid, with the river being seasonally dry, and the main vegetation consisting of acacia trees and shrubs, with some palms along the riverbanks. In the past, however, the lake level would have been at or covering much of the study area, and the Kalodir-Kokito basin to the west of the Lothidok range would have become a protected bay. Protected and relatively shallow waters may have been ideal for aquatic exploitation. Today, inshore areas, often with sedge stands (sought by fish for protection) are preferred fishing sites in the Turkana Basin (Gifford-Gonzalez et al., 1999). In these areas, other fauna (e.g., terrapin, crocodile, hippopotamus) may also be obtained, and it is possible to fish in calm, shallow waters with minimal technology. In deeper waters, by contrast, fishers' abilities can be compromised by wind and waves. 3.3. Study sites Initial reconnaissance was carried out in the Kalokol study area in 2009, drawing on reports of finds at the Dilit locality by Whitworth (1965). More formal surveys and excavations took place in 2010 and 2013, leading to the documentation of ten sites (Beyin, 2011) and to excavations at Dilit 01 (SASES code1: GcJh 6), Kokito 01 (GcJh 11) and Kokito 02 (GcJh 12) (Fig. 2). Details of the latter two sites are forthcoming (Beyin et al. in press). Dilit 01 (3
280 1200 N, 35
460 3300 E, 450 m asl) is the most intensively investigated of the three sites. It lies on a deflated pebble lag surface, and slopes gently down to the west, toward a small channel. Surface finds are abundant, and a 100 m2 systematic collection grid was laid on an eroding slope in the western part of the site. Nine units (A-I), totaling 22.5 m2 area and 20.6 m3 volume, were excavated. Stratigraphy in the largest excavation unit (A) features shell-strewn, gravelly sand, indicating beach deposits, as well as a distinct bed of hard concretions (8 cm thick) at c. 65 cm below surface. This layer signifies a dry episode, and appears to indicate an occupational hiatus, as no artifacts or bone remains were found associated with it. Lithics, fauna, and a small number of non-diagnostic pottery fragments are found both below and above this concreted deposit. Notably, 90% of the faunal assemblage was

Fig. 2. Map of the Turkana Basin with current and reconstructed shorelines and early Holocene archaeological sites. African Humid Period (AHP) maximum based on Bloszies et al. (2015).

1 SASES stands for Standardized Site Enumeration System for the Continent of Africa (Nelson, 1993).

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

5

Table 1 AMS (accelerator mass spectrometry) radiocarbon dates from the study sites. Calibrated using the IntCal13 curve (Reimer et al., 2013) in OxCal 4.2.4 (Bronk Ramsey, 2013), 2sigma range. OES, ostrich eggshell; bs, below surface; bp and BP, before present Site

Context

Material

uncal bp

cal BP

Lab Number

Reference

Dilit 01 Dilit 01 Dilit 01 Kokito 01 Kokito 01 Kokito 01 Kokito 01 Kokito 01 Kokito 01 Kokito 01 Kokito 02 Kokito 02 Kokito 02 Kokito 02

Unit Unit Unit Unit Unit Unit Unit Unit Unit Unit Unit Unit Unit Unit

OES beada OES beada OES bead Charcoal Charcoal Charcoal Shell Charcoal Charcoal Shell Charcoal Charcoal Charcoal Charcoal

6260 ± 30 6245 ± 30 31620 ± 450 9060 ± 30 9785 ± 35 14745 ± 50 9710 ± 35 9600 ± 25 11530 ± 25 10825 ± 40 11735 ± 30 11705 ± 25 11670 ± 25 11670 ± 35

7266e7030 7258e7025 36450e34660 10248e10193 11246e11177 18113e17760 11222e10896 11126e10776 13444e13295 12769e12680 13613e13452 13576e13458 13566e13441 13571e13435

Beta-458955 ISGS-A3246 ISGS-A3049 ISGS-A1715 ISGS-A1714 ISGS-A3752 ISGS-A3753 ISGS-A3721 ISGS-A3722 ISGS-A3754 ISGS-A3724 ISGS-A3720 ISGS-A3723 ISGS-A3028

present work present work present work Beyin et al. in Beyin et al. in Beyin et al. in Beyin et al. in Beyin et al. in Beyin et al. in Beyin et al. in Beyin et al. in Beyin et al. in Beyin et al. in Beyin et al. in

a

E, 6 cm bs E, 6 cm bs E, 55 cm bs A, 16 cm bs A, 25 cm bs D, 9 cm bs D, 9 cm bs D, 24 cm bs D, 40 cm bs D, 40 cm bs Bn, 19 cm bs Bn, 34 cm bs B, 30 cm bs B, 44 cm bs

press press press press press press press press press press press

These are two portions of the same OES bead, sent to two different laboratories.

recovered from the uppermost three spits, while 60% of the lithic assemblage was recovered from the lowermost three spits (70e100 cm below surface), preceding the drying event. Multiple occupational episodes, of a distinct character and/or occurring under different preservation conditions, are thus inferred. These occupations may have followed lake regression events, with the beach deposits then being flooded as the lake rose again. Artifacts are thus unlikely to be in primary position, and it is notable that no archaeological features such as hearths or pits could be observed. Three accelerator mass spectrometry (AMS) radiocarbon dates were obtained on ostrich eggshell beads (Table 1). While one of these beads was deemed to be too old to accurately represent the site, at 31620 ± 450 bp (36450-34660 cal BP; ISGS-A3049), the other bead, found just below the surface, was fragmented and sent to two different labs, each producing near-identical dates of 6245 ± 30 bp (7258-7025 cal BP; ISGS-A3246) and 6260 ± 30 bp (7266-7030 cal BP; Beta-458955). Further dating efforts at this site are ongoing. Kokito 01 (3
250 5600 N, 35
4501900 E, 447 m asl) extends over c. 3000 m2 and is the farthest from the present shoreline of the three excavated sites. A 9 m2 grid was established for systematic surface collection, and six units (A-F), totaling 15 m2 area and 9.7 m3 volume, were excavated. Stratigraphy indicates periods of inundation, with clayey sediments, abundant shell inclusions, and redoximorphic features. The site is thus interpreted as a backshore or lagoon-like setting, and may have experienced multiple episodes of inundation and regression like those described for Dilit 01; as at Dilit 01, no features such as hearths or pits could be observed. Artifacts found in both surface and subsurface contexts include barbed bone points, lithics, and abundant faunal remains. Seven AMS radiocarbon dates e five on charcoal and two on M. tuberculata shell e mainly fall between c. 13,000 and 10,500 cal BP, the expected time frame given the cultural material at Kokito 01. However, the dates show multiple stratigraphic inversions and present interpretive challenges. Currently, we offer two tentative interpretations of the site (Beyin et al. in press): the more conservative approach is that, given the mixed nature of the dated samples, we must take the latest date, 9060 ± 30 bp (10190e10245 cal BP, ISGS-A1715), as a terminus post quem for the site, and assume that this and other dated samples were brought in through inundation at some point after this date. An alternative interpretation is that the site was occupied multiple times between c. 13,000 and 10,200 cal BP, and that the dated samples were postdepositionally mixed, again possibly through inundation. In both scenarios, we consider that a single very early date of 14745 ± 50 bp (17760e18110 cal BP, ISGS-A3752) is unlikely to be

representative of the site. Based upon comparison to Kokito 02 (see below), we suggest that it is likely that most of the cultural material at Kokito 01 dates to the Pleistocene-Holocene transition (Beyin et al. in press). However, in the absence of direct dates on bone or pottery, it is difficult to be more precise than this. Kokito 02 (3
260 1800 N, 35
450 2200 E, 437 m asl) is found about 1 km north of Kokito 01, and is a fourth of its size. Stratigraphy is not unlike that of Kokito 01, and implies another backshore or lagoonlike setting with alternate flooding and drying. Judgmental surface collection took place, and two units (A and B), totaling 7.25 m2 area and 2.8 m3 volume, were excavated. Artifact densities were lower than at Kokito 01, with a modest faunal assemblage, very few lithics, and with surface occurrences of barbed bone points. One area with a relatively high concentration of charcoal has been interpreted as possible remains of a disintegrated hearth. Four charcoal samples, including one from this possible hearth area, were AMS radiocarbon dated, tightly clustering at c. 13,500 cal BP (Beyin et al. in press). 4. Methods Field methods will be detailed elsewhere. Units were excavated until reaching sterile sediments, using arbitrary 10 or 15 cm spits depending on artifact density (in sterile sediments, 20 cm spits were sometimes used). All sediments were dry-sieved, using a double-layered 4 mm mesh. Faunal remains were dry-brushed in the field and lightly brushed with water in the lab. Sampling strategies were employed at Kokito 01, where 87% of the faunal assemblage was analyzed, and at Dilit 01, where 61% was analyzed; at Kokito 02, the full assemblage was analyzed. At Kokito 01, Unit A and its extensions (A-N, A-W, A-NW), Unit D, and Unit F, and surface collection grid squares A-I were included in the faunal analysis; these were contexts determined during excavation to have the deepest and richest deposits, and these are the contexts from which radiocarbon dates were obtained. At Dilit 01, Unit A (and its extensions A-N and A-S), Unit D, Unit E (and its extension ENW), and Unit I were included; as at Kokito 01, these prioritizations were made based upon in-field observations of stratigraphy and sample size per unit, such that units that yielded larger samples were prioritized. For each studied context, all faunal remains were examined, and were recorded in the database and considered part of the NISP (Number of Identified Specimens) so long as they could be identified, minimally, to skeletal element or element group (e.g., vertebra, limb bone), and classified at least as either fish or tetrapod. This means that less diagnostic specimens e for example fish spine or

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

6

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

vertebral fragments, turtle carapace, and mammal limb shafts e are included in the NISP. The remaining unidentified specimens (“NID”) were separated into size categories using maximum dimension (0e10 mm, 10e20 mm, etc.) and tallied for each context. Tetrapod identification was challenging due to the paucity of reference material at the Turkana Basin Institute (TBI) Turkwel facility, where the analysis took place. Most identification had to rely on the analyst's own photographic collection, as well as on illustrated guides such as Walker (1985). Poor preservation, and a lack of quality illustrations for certain taxa and elements, made these resources insufficient. At minimum, mammalian material was identified to carcass size, using a system based on Brain (1981) but adapted to eastern African fauna. Fish remains, which dominate all three sites, were easier to identify due to their completeness and the availability of reference material recently collected by TBI researchers. Skeletal material was available for all fish taxa observed in the archaeological collections, with the exception of Protopterus aethiopicus (lungfish). For all taxa, many of the modern individuals in the collection are much smaller than their archaeological counterparts, inhibiting easy identification. For siluroids (catfishes), spine identification relied mainly on Gayet and Van Neer (1990). No attempts were made to identify the catfishes to species level, given that reference material was limited to one species per genus for Bagrus, Clarias, and Synodontis. Furthermore, catfish cranial shield fragments were rarely identified beyond the order Siluriformes; even though these can be diagnostic of genus, the analyst did not feel comfortable making such distinctions and in any case, such fragments are rarely informative for estimating MNI. For Cichlidae, two genera (Oreochromis and Tilapia) were available for comparison, but the analyst did not feel confident identifying archaeological tilapia to the genus level given their strong osteological similarities. Whenever possible, the skeletal element, portion, and side was recorded for each identified specimen. For fish remains, estimates of the Minimum Number of Elements (MNE) were used to compare relative representation of skeletal parts, following Stewart (1989, 1991): elements were grouped as “cranial” (including all bones anterior to the first vertebra, as well as cleithral bones), “axial” (vertebrae and ribs), or “epaxial” (pelvic girdle, bony lepidotrichia, and pterygiophores). Estimates of the Minimum Number of Individuals (MNI) considered element identification, estimated live body size, laterality, and the number of elements in a complete individual, which can vary for some taxa; the most conservative estimate was used. MNE and MNI estimates were calculated at the site level, rather than by individual contexts. Therefore, they should be taken as conservative (i.e., low) estimates. For mammalian limb bone shafts, breakage was recorded as green or diagenetic following Villa and Mahieu (1991), and the specimen was assigned to one of three types (<50%, >50%, or 100% of shaft circumference) following Bunn (1982). All identified tetrapod specimens were examined under strong oblique light

using a 20 hand lens, to seek bone surface modifications such as cut marks, percussion marks, or tooth marks, following criteria reviewed by Domínguez-Rodrigo et al. (2007). While it was not possible to examine each fish specimen individually, identified fish bones larger than 2 cm in greatest dimension, and all pectoral, dorsal, and anal spines, basioccipitals, and near-cranial vertebrae, were examined for cut marks or other possible marks. Surface preservation was recorded in multiple ways. Behrensmeyer's (1978) scoring system for subaerial weathering was used for tetrapods, but Stewart (1989:107) argues that this system does not apply well to fish bone and particularly not in the Turkana Basin's burial environments. Therefore, Stewart's (1989:108e109) stain index and weathering categories were also adopted, for both fish and tetrapods. Occurrences of root etching, burning, abrasion, biochemical pitting (the latter following Domínguez-Rodrigo and Barba, 2006), and the above-mentioned marks were also recorded. Finally, cortical preservation was given an overall score as “good” (100% of surface visible), “moderate” (>50% visible), or “bad” (<50% visible). Surfaces were often coated in burial matrix and difficult to see; such specimens were noted as “encrusted”. To estimate live body sizes of fish, measurements were taken using a Mitumoyo digital caliper at locations on skeletal elements specified by Van Neer (1989) and Van Neer and Lesur (2004). Power equations were then used to estimate lengths of Cichlidae and Clarias sp. following Van Neer and Lesur (2004), and of Synodontis sp. following Van Neer and Depraetere (2005). For Lates niloticus (Nile perch), power equations were calculated based on measurements collected by Van Neer (1989) and available via OsteoBase (http://osteobase.mnhn.fr/) (Table S1). 5. Results 5.1. Overview of surface and subsurface assemblages Table 2 summarizes the counts of identified and unidentified fauna within the analyzed contexts, and demonstrates the size of the Kokito 01 assemblage, relative to those of Dilit 01 and Kokito 02; however, densities of faunal remains are similar at all three sites (Fig. 3), and show sharp declines with increasing depth below surface. At all sites, about two-thirds of studied remains were not identifiable, indicating poor preservation and high degrees of fragmentation, particularly of tetrapod remains. The relative abundance of finds from surface collection ranges from 9% of NISP at Dilit 01 to 44% of NISP at Kokito 01; to some extent, these frequencies are shaped by both the size of surface collection grids, and the analytical sampling strategies at each site. Many of the surface finds may be derived from prehistoric occupations, having been exposed through deflation. However, in this paper they are treated with caution due to their uncertain provenience: we explicitly distinguish between surface and subsurface samples and we generally focus on excavated specimens, particularly considering

Table 2 Overview of analyzed assemblages. Site

Dilit 01 Kokito 01 Kokito 02 All sites

Surface

Subsurface

NISP

NISP

MNI

NTAXA

D

Combined surface and subsurface NISP

MNI

NID

NR

%NID

75 334 261 670

783 1073 337 2193

32 52 27 111

12 13 8

0.86 0.82 0.71

858 1449 598 2905

34 69 38 141

1502 2445 1316 5263

2360 3894 1914 8168

64% 63% 69%

NISP, Number of Identified Specimens; MNI, Minimum Number of Individuals; NTAXA ¼ number of individual taxa (taxonomic richness); D, Simpson's Dominance index; NID, Unidentified specimens; NR, Number of remains (NISP þ NID); %NID, Percent of remains unidentified. To ensure comparability with other datasets summarized by Gifford
P  n ðn 1Þ i i i Gonzalez et al. (1999), we used the “unbiased” version of D for finite samples, calculated as 1  . NðN1Þ Where N ¼ total MNI of all taxa in the sample, and n ¼ MNI of a particular taxon.

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

7

(generally fish) bones have migrated downward, while larger (generally mammalian) bones remained on the (potentially deflated) surface. However, it is unclear what processes would produce this pattern at Dilit 01 but not at the Kokito sites. On current evidence e in the absence of direct dates on faunal remains e we find it more plausible that the Dilit 01 surface finds are unrelated to the main occupation. 5.2. Bone preservation and surface modifications

Fig. 3. Densities of faunal remains at the studied sites, by depth below surface. Density is expressed as the total number of remains (NR) per cm of depth and per m2 of area. NR is distinct from the number of identified specimens (NISP) in that it includes unidentified specimens. Two counts of NR were averaged: that acquired during in-field counts by the excavation team, and that recorded during faunal analysis. These were found to be very similar in nearly all contexts.

the evidence (below) at Dilit 01 for comparatively recent (domestic) fauna on the surface and in immediately underlying contexts. At all sites and particularly at Dilit 01, removing the surface finds from the dataset substantially elevates the percentage of aquatic taxa in each assemblage (Fig. 4). What initially appears to be a major difference between the Dilit and Kokito sites, in terms of emphasis on terrestrial fauna, becomes relatively muted after excluding surface finds. Indeed, all sites bear evidence for specialized aquatic resource exploitation, ranging from 78 to 98% of subsurface NISP. We suggest that the abundance of terrestrial species in the Dilit 01 surface sample may represent events (anthropogenic or not) postdating the AHP-era occupations evidenced in the subsurface deposits. An alternative explanation is that smaller

Table 3 summarizes observations made on bone preservation and taphonomy. The Behrensmeyer (1978) system of scoring weathering was only applicable to small samples, since only tetrapod and some fish bones were large enough to enable observation of subaerial weathering. There were notable differences between Kokito 02, where weathering was severe, and the much better-preserved Kokito 01. Other differences were observed using Stewart's (1989) index developed for Turkana burial environments. Unfortunately, since this measure was only adopted mid-way through the analysis, not all of the bones were observed with this in mind. Most observed bones at Dilit 01 and all of those at Kokito 02 exhibited cracking and friability related to “clay-shattering,” caused by expansion and contracting of surrounding sediments (Stewart's, 1989 type A). By contrast, just 28% of all observed bones at Kokito 01 showed this pattern, while the remainder were fresh, with no apparent evidence of transport (type C). Also at Dilit 01, 8% of bones exhibited rolled edges and polished surfaces (type B). In many cases, bones exhibiting rounding are also chemically leached (resulting in soft, powdery surfaces and a lightweight feel); however, in other specimens, such as those pictured in Fig. 5a, the surfaces are firm, smooth, shiny, and pitted, which also resembles digested bone. Tentatively, given the presence of leached bone and the absence of extensive carnivore damage in the assemblage as a whole, we interpret the rolled bones as the result of water transport, though we cannot rule out other processes. By contrast, rolled or polished bone was not observed at the Kokito sites. Overall, Kokito 01 has generally good cortical preservation, with two-thirds of specimens having fully visible surfaces. By contrast, poor preservation was the norm at Dilit 01 (98%) and Kokito 02 (96%). At both sites, this is largely due to encrustation (Fig. 5b), particularly at Dilit 01, where 84% of specimens were coated in burial matrix. The lithic artifacts from both sites exhibit somewhat similar patterns to the

Fig. 4. Relative abundance of aquatic and terrestrial resources in surface (SF) and subsurface (SUB) contexts at the studied sites, based on NISP (Number of Identified Specimens). Aquatic fauna include fish, hippopotamus, crocodile, and turtles. Categories that do not enable aquatic/terrestrial distinctions (e.g., “large animal indet.”) are excluded from this analysis.

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

8

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

Table 3 Taphonomic observations at the studied sites. Bone surface modifications (NISP)a

Dilit 01 Kokito 01 Kokito 02

Weatheringb NISP (fraction of total observations)

CM

CM?

TM

ABR

BRN

CP

RT

WA

CR

NISP*

0e1

0 1 1

1 12 0

3 4 3

0 2 1

1 0 0

0 1 0

1 133 3

8 0 0

651 35 180

779 1067 337

4 (.8) 267 (.96)

Weatheringc NISP (fraction of total observations)

Cortical Scored NISP (fraction of total observations)

2e3

4e5

A

B

C

B

M

G

230 (.86) 44 (.28) 154 (1)

21 (.08)

7 (.03) 11 (.21)

1 (.2) 4 (.01) 41 (.79)

17 (.06) 115 (.72)

644 (.98) 69 (.15) 178 (.96)

8 (.01) 82 (.18) 7 (.04)

5 (.01) 301 (.67) 1 (.005)

a CM, cutmark; CM?, possible cutmark; TM, toothmark; ABR, sedimentary abrasion; BRN, burned; CP, biochemical pitting; RT, root etching, WA, water polishing, CR, encrusted bone. NISP* ¼ total NISP minus tooth NISP. b Five-stage scoring system developed by Behrensmeyer (1978), where 0 indicates no subaerial weathering and 5 indicates most severe stage. c System developed by Stewart (1989) specifically for the Turkana region, where A indicates bone that is friable, crumbling, and indicative of clay-shattering; B indicates bone that has been rolled, with rounded edges and superficial polish; and C indicates fresh, non-transported bone. d B, bad cortical preservation (<50% visible); M, moderate cortical preservation (>50% but <100% visible); G, good cortical preservation (100% visible).

Fig. 5. Examples of taphonomic observations made at the Kalokol sites: a: rolling/polishing of unidentified specimens at Dilit 01, Level 9, indicative of water action; b: Clarias pectoral spines from Dilit 01, Level 1, demonstrating encrustation; c: cut mark on a cranial bone of an unidentified tetrapod, likely mammalian, from Kokito 02, Level 3; d: carnivore tooth marks on a parasphenoid of Lates niloticus from Kokito 02, Level 3, with modern specimen to illustrate size.

fauna, in that those from the Kokito sites display low incidences of edge damage/rolling compared to the lithics from Dilit 01. Signs of human and carnivore modification are extremely rare at all three sites. Tooth marks, likely imparted by carnivores, were observed on just a handful of specimens (aquatic and/or terrestrial) at each site (Fig. 5d). Although many bones have dark staining due to the burial environment, just one specimen (near the surface of Dilit 01) had been burnt; it is possible, however, that additional

specimens were burnt but not detected due to the heavy concretions coating nearly all bones at this site. At Kokito 01, a large mammal limb bone shows a definitive mark, while 12 other specimens e three large mammal specimens, two crocodile specimens, a turtle humerus, and six fish bones e have possible cut marks; five of these are surface finds. At Kokito 02, just one specimen, a mammalian (?) cranial fragment, bore a cut mark (Fig. 5c). At Dilit 01, no definitive cut marks were identified, though one large

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

mammal specimen had a possible cut mark. The implications of these low human signals are discussed further below. 5.3. Taxonomic distribution Aquatic fauna consistently dominate the studied sites (Fig. 4). The vast majority of these, whether using NISP and MNI, are fish (Table S2). Other aquatic fauna include softshell turtle (Trionyx sp.; represented by numerous carapace fragments), other turtles, isolated specimens of crocodile (Crocodylus niloticus), and at Dilit 01, hippopotamus (Hippopotamus amphibius). While these aquatic tetrapods are too few to permit a useful site comparison, there are important differences in the relative abundances of fish taxa. Fish vary widely in terms of numbers of elements per individual and bone survivorship. For example, the order Perciformes e which in the Turkana Basin includes various species of tilapia (Cichlidae) and Nile perch (Lates niloticus) e have numerous durable vertebrae, spines, and cranial bones, and preserve well archaeologically. By contrast, the order Siluriformes e which here includes bagrid catfish (Bagrus spp.), air-breathing catfishes (Clarias spp.), and squeakers (Synodontis spp.) e tend to have durable and highly distinctive cranial bones and spines, but poor preservation of the vertebral column. Lungfish (Protopterus aethiopicus) are postcranially cartilaginous, but their durable cranial plates preserve well and are found in fossil and archaeological sites (Stewart, 1989; Prendergast, 2010). Therefore, estimates of MNI are preferred over NISP when examining relative abundances of fish taxa, despite the fact that MNI has its drawbacks e for example, MNI estimates can be overly conservative, or MNI can overstate the importance of taxa represented by isolated fragments. For comparison, both measures are illustrated in Fig. 6. All three sites have roughly similar indices of richness and diversity (Table 2), but the relative abundances of major fish groups differ between the Kokito sites and Dilit 01. At the Kokito sites,

9

Perciformes account for most of the MNI (61% at Kokito 01 and 73% at Kokito 02), and roughly 90% of NISP. There is a notable difference among the Perciformes at these two sites, with tilapia being three times more abundant at Kokito 02 than Kokito 01, where Nile perch dominates. Apart from a single lungfish at Kokito 01, catfishes make up the remainder of fish at the Kokito sites, mainly Synodontis, followed by Clarias and, at Kokito 01, Bagrus. By contrast, at Dilit 01, Siluriformes account for two-thirds of MNI, and are slightly more than half of NISP, a notable fact given the abovementioned features of siluroid preservation. The abundance of catfishes by NISP is due to high numbers of cranial shield fragments and broken spines. Unlike at the Kokito sites, Clarias sp. is particularly abundant at Dilit 01 (45% of MNI), followed by Synodontis sp. (23%); the remainder of fish are Nile perch (23%) and tilapia (9%). The implications of the contrast between the Perciformes-dominated Kokito sites, and Siluriformes-dominated Dilit 01, will be discussed further below. Terrestrial faunal remains were poorly preserved at all the studied sites. Few specimens could be identified with confidence beyond carcass size; most mammalian remains belonged to medium- to large-sized bovids, about the size of a topi or buffalo (Table S2). At Kokito 02, none of these could be identified beyond their size class; the only identifiable terrestrial faunal remains were two equid bones from the surface collection. At Kokito 01, a possible buffalo (Syncerus caffer) was identified from surface collection, and a dik-dik (Madoqua sp.) from a subsurface context. An equid, likely zebra (either Equus quagga or Equus greyvi), was also identified on the surface at Kokito 01. At Dilit 01, an equid distal tibia in the uppermost level of Unit A was identified as donkey (Equus asinus) based on its small size. This is almost certainly intrusive, as is the case for a single tooth of domestic cattle (Bos taurus) from the uppermost level of Unit I. The only other identifiable terrestrial fauna at Dilit 01 are multiple remains of a small alcelaphine, either topi (Damaliscus lunatus) or kongoni (Alcelaphus buselaphus), in both surface and subsurface contexts, and a single tooth of warthog

Fig. 6. Relative abundance of fish taxa at the studied sites, using NISP and MNI. Siluriformes include Clarias, Bagrus, and Synodontis, and Perciformes include perch and tilapia. Where NISP frequencies are 1%, names of taxa are not indicated (see Table S2).

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

10

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

(Phacochoerus sp.) from the surface. Given very limited mammalian samples, and the absence of definitive cutmarks, it is difficult to make inferences about hunting by the inhabitants of the Kokito and Dilit sites. 5.4. Fish skeletal part representation Relative abundances of cranial, axial, and epaxial elements were calculated for Perciformes and Siluriformes. To increase the number of modern (natural and anthropogenic), fossil (natural), and archaeological samples to which we could compare the Kalokol sites, epaxial elements are excluded from Fig. 7. It was expected that the ratio of cranial to axial elements would diverge substantially from that of the average fish, and this is indeed the case. For Perciformes at the Kalokol sites, axial elements are greatly overrepresented, whereas more fragile cranial elements are underrepresented, relative to their abundance in the average perciform fish. This is particularly true at Kokito 01 and Dilit 01. For

Siluriformes at the Kalokol sites, cranial elements are overrepresented relative to the average fish, particularly at the Kokito sites. For both fish orders, these trends are unsurprising: delicate perciform cranial elements tend to preserve poorly at archaeological sites, while robust cranial shields of siluroids hold up well over time, surviving even in Plio-Pleistocene sites (Stewart, 1994; Braun et al., 2010); by contrast, siluroid vertebrae are unlikely to survive. As discussed below, there are no evident contrasts among cranial:axial ratios at the Kalokol sites when compared against other Turkana Basin sites, modern fishing camps, and natural accumulations on beaches. 5.5. Fish body size estimates Live body sizes were estimated for Clarias sp, Cichlidae, Synodontis sp., and L. niloticus (Fig. 8). Under ideal conditions, the single most-abundant measured skeletal element would be used for each taxon to avoid double-counting and errors in size estimation that

Fig. 7. Representation of cranial and axial skeletal elements in Perciformes (top set) and Siluriformes (bottom set), using MNE and excluding epaxial elements to increase comparative samples. Average cranial:axial ratio in a typical fish is shown in black and white. Cross-hatched bars indicate natural accumulations collected by Stewart (1991) and as reported in Gifford-Gonzalez et al. (1999). Archaeological assemblages are the Kalokol sites (present work) and Lowasera, GaJi3, and FxJi12 (Stewart, 1989). Modern fishing camps are AS1, FC1, and Sites 06/10/20 (Stewart and Gifford-Gonzalez, 1994; Gifford-Gonzalez et al., 1999).

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

11

Fig. 8. Estimated body sizes at the studied sites for A: African catfish (Clarias sp.); B: tilapia (Cichlidae); C: squeaker and related catfishes (Synodontis sp.); D, Nile perch (Lates niloticus). Due to small sample sizes, all measured skeletal elements are combined here. Size estimates given as Total Length except for Synodontis, where Standard Length is used. For data see Table S3.

can arise from combining data from different elements. However, the extremely small numbers of measurable bones from the Kalokol sites made it necessary to analyze all measured skeletal elements together, and to use NISP rather than MNI; the reader is referred to Table S3 for the full dataset. Even with this aggregate, NISP-based dataset, the size reconstructions are limited, and the results presented here should be taken with appropriate caution. Possible interpretations of these data, in terms of human predation strategies and taphonomic factors, are discussed further below. Based on the limited available data, we see that most (though not all) of the fish represented at the Kalokol sites are larger than their modern-day counterparts' lengths at maturity (Froese and Pauly, 2016). For example, Clarias gariepinus matures at c. 300 mm standard length (SL). Most of the 13 measured Clarias specimens at Dilit 01 belong to much larger fish, measuring c. 550e850 mm (median 736 mm) in total length (TL).2 Synodontis schall matures on average at 210 mm SL; for this taxon, the distribution of sizes at the Kalokol sites (median SL 198 mm) roughly approximates a natural distribution. There are differences e based on small sample sizes e between Dilit 01, where Synodontis is generally larger (n ¼ 6, median SL ¼ 246 mm), compared with the slightly smaller individuals at Kokito sites (n ¼ 20 at Kokito 01, n ¼ 6 at Kokito 02, median SL at each site ¼ 189 mm). Among the Perciformes, mature fish are also the norm for tilapia, but the distribution of Nile perch at the Kalokol sites more closely approximates a natural distribution. Tilapia such as Oreochromis niloticus mature at 186 mm TL; the median TL for all

2 Standard Length (SL) is the measurement from tip of snout to base of caudal fin, whereas Total Length (TL) includes the caudal fin. While body size estimates were calculated as TL (except for Synodontis), published lengths at maturity often use either TL, SL, or an unspecified “length”.

measured Cichlidae specimens from Kalokol sites (n ¼ 18) is 333 mm, with no notable differences among sites. Nile perch (L. niloticus), the largest taxon in the Turkana Basin today, mature on average at 743 mm TL and can reach up to 2 m. By this standard, the Kalokol assemblages are not overly dominated by mature Nile perch individuals. Many must have been caught while immature; the median TLs at Kokito 01 (n ¼ 23, 479 mm), Kokito 02 (n ¼ 62, 712 mm), and Dilit 01 (n ¼ 22, 573 mm) are all under the average mature TL. However, each site also has multiple fish exceeding 1 m in length, and reaching up to 1.5 m at Kokito 02, where Nile perch are larger than at the other sites. 6. Discussion 6.1. Natural or anthropogenic accumulations? Given the low human signal e inferred from burning and cut marks e in the Kalokol assemblages and the paucity of evidence for carnivore activity, it is reasonable to inquire whether fish (and other taxa) arrived at the sites as food, or died and were deposited through natural processes. As noted above, there are reasons to suspect complex processes of site formation and deflation at all the Kalokol sites. All three are characterized by clayey sediments with redoximorphic features indicative of saturation, and by shellstrewn, gravelly sand (beach) sediments. These are interpreted as resulting from lake inundation and regression. At Dilit 01, as mentioned earlier, there are major differences in the vertical location of faunal remains (primarily in upper levels) and lithic artifacts (primarily in lower levels) that cannot be resolved without further chronological work. At Kokito 02, the tight clustering of dates has been interpreted as indicative of rapid site formation and a lack of disturbance, but reworking (leading to vertical dispersal of dated samples from a single event) cannot be ruled out. At Kokito 01,

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

12

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

stratigraphic inversions of radiocarbon-dated charcoal and lacustrine shell samples suggest either post-depositional mixing (perhaps through inundation) of occupational debris dating to c. 13e10.2 ka, or, in an alternate scenario, inundation after 10.2 ka would have led to the deposition of the lithic artifacts, faunal remains, and charcoal. In the first scenario, the site is presumed to be anthropogenic, while in the latter scenario, the origins of artifacts and ecofacts are unknown. We favor an at least partially anthropogenic interpretation for Kokito 01, and for the other Kalokol sites, on current evidence (Beyin et al. in press). Faunal remains are key to this interpretation. The association of faunal remains with lithic artifacts, charcoal, and other cultural material in the excavation units of the Kalokol sites hints at intentional accumulation. Nevertheless, numerous studies demonstrate that this assumption cannot be made uncritically, and illustrate the challenges of differentiating natural and anthropogenic accumulations, especially at previously-inundated ~ iz, 1992; sites (Stewart, 1989, 1991; Van Neer and Morales Mun Butler, 1993; Stewart and Gifford-Gonzalez, 1994; GiffordGonzalez et al., 1999). Zohar et al. (2001) summarize these findings and outline criteria that may be used to distinguish natural and anthropogenic assemblages. For example, while there are 33 genera in the modern Turkana Basin, just six inshore genera e Bagrus, Clarias, Synodontis, Oreochromis, Lates, and Labeo e regularly appear at human-created sites (Stewart and Gifford-Gonzalez, 1994). All but the last of these dominate the Kalokol assemblages. Densities of faunal remains (Fig. 3), which in the uppermost 50 cm of each site range from 1.4 to 12 bones/cm/m2, also suggest that these are not natural accumulations. Stewart (1991) notes that bone scatter densities in natural accumulations are low (0.06 bones/m2 in recent shoreline surface collections), compared to a recent fishing camp (13 bones/m2 in surface collection). While anthropogenic sites can have low or very patchy bone densities (Gifford-Gonzalez et al., 1999), natural accumulations are unlikely to exhibit high densities except under conditions of water transport. Apart from some polished/rolled specimens at Dilit 01, evidence for water transport (more extensive rolling and size sorting) is not evident at the Kalokol sites. Burning and cut marks should also be obvious indicators of human activity, but are notably rare at the Kalokol sites. Although not all fish bones were examined for bone surface modifications (see Methods), and poor cortical preservation limited surface visibility (particularly at Dilit 01 and Kokito 02), the paucity of cutmarked and burned bone is striking. Experiments suggest that cut marks on fish should be detectable, assuming similar prehistoric butchery practices (Willis et al., 2008); however experiments also show sharp declines in cut mark visibility and detection rates, following burial and postdepositional processes (Willis and Boehm, 2014). Low cut mark identification rates at the Kalokol sites are consistent with what has been documented in other eastern African fish bone analyses (Stewart, 1989; Prendergast, 2010; Prendergast and Lane, 2010; Quintana Morales, 2013), and could indicate routine under-identification of marks (related to taphonomic or analytical factors), and/or culinary preparations that do not require butchery. The absence of burned bone at the Kalokol sites, however, contrasts with most of the above-cited analyses. Fish braincase fragmentation is common in anthropogenic sites, and rare in natural accumulations (Stewart, 1991), though postdepositional processes, e.g., trampling or soil compaction, could also lead to fragmentation. Although a cranial fragmentation index (indicating the degree of completeness) was not recorded here, which would have enabled comparison to Stewart (1991), impressionistically, there are very few complete cranial elements for any taxon. Tentatively, this is further support for human agency; however it should be noted that diagenetic breakage was generally

quite common among the tetrapod remains, and the same can be assumed for fish remains. Skeletal element frequencies using MNE are often used to differentiate natural and anthropogenic accumulations. However as noted earlier, the cranial:axial ratios (Fig. 7) calculated for the Kalokol sites do not cluster in informative ways with any of the comparative samples. Rather, there is a total lack of resolution, suggesting that perhaps skeletal part representation is not the best indicator of anthropogenic versus natural agency in fish assemblage formation. However, we caution that the sample of modern and fossil natural scatters is limited (Stewart, 1991) and that with increased sample size, it is possible that these could be differentiated from both the archaeological sites (Stewart, 1989) and the modern fishing camps (Stewart, 1991; Stewart and GiffordGonzalez, 1994; Gifford-Gonzalez et al., 1999) used in the comparative analysis. Finally, fish body size reconstructions suggest overrepresentation (relative to their natural occurrence) of large, mature individuals of Clarias sp. and Cichlidae at all sites, as well as of Synodontis sp. at Dilit 01, and of Nile perch at Kokito 02. Caution is warranted in interpreting these data, given very small samples of measured bones, as well as possible taphonomic or analytical biases toward larger individuals. Postdepositional processes could bias the Kalokol assemblages toward larger bones, since smaller and more delicate fish bones tend to fragment easily; however, the presence of a few very small Synodontis individuals in the analyzed assemblages, and the regular occurrence of delicate skeletal elements of other taxa (e.g., ribs), suggests that the absence of small fish does not merely result from differential preservation. The mesh size used (4 mm) may also have led to the loss of bones of smaller individuals while sieving; however we expect that at least some skeletal elements of individuals in the 100e150 mm TL range would have been recovered using this size of sieve. Yet there is a near-absence of fish measuring <150 mm TL. The abundance of larger individuals might suggest fishers' selection of fish that would provide ample amounts of meat and fat. Gifford-Gonzalez et al. (1999:414) note that human-accumulated assemblages tend to be dominated by medium- to large-sized fish. By contrast, in natural beach scatters recorded in the Turkana Basin, smaller individuals tend to be more abundant (Stewart, 1991:18e19). Thus one interpretation of the body size data is that the Kalokol sites represent intentional accumulation rather than natural beach scatters. However, size alone cannot be used to determine site formation processes. Below we discuss additional interpretations of body size data. 6.2. Inferring fishing strategies: technologies, and locations and seasons of capture Fish body size might also be interpreted in terms of the location and season of capture (e.g., Van Neer, 1986; Van Neer, 2004; Van Neer and Lesur, 2004; Prendergast, 2010; Prendergast and Lane, 2010). Ethological literature on fish behavior, and ethnographic literature on fishing in sub-Saharan Africa, reviewed by Stewart (1989) and Prendergast (2008), provides a framework for thinking about possible prehistoric technologies and seasonal exploitation strategies. For example, tilapia and juvenile Nile perch prefer inshore habitats, but adult Nile perch are found in open, well-oxygenated waters, where they would be caught either by spearing or harpooning from rocky promontories or steep shorelines, or via hook-and-line fishing, possibly with boats. While Nile perch can be pursued year-round, they are caught with greater ease in the dry season, when waters are calm. By contrast, mature Clarias and tilapia spawn in upriver locations in the wet season (see below), and can thrive in shallow waters, e.g., in inshore locations,

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

or pools left by receding waters, during the dry season. Clarias are particularly adapted to saline, poorly oxygenated waters, where they are easily captured using spears, harpoons, baskets, or bare hands. Synodontis are adaptable and can thrive in both shallow and deeper waters; they are often caught during low lake stands with thrust baskets, but could be caught by the same open-water strategies used to capture Nile perch. While Synodontis are often considered open water fish, behavior is highly variable across the genus (Van Neer, 2004; see also Linseele and Zerboni, this volume), making them less informative in terms of determining fishing locations and strategies. At the Kalokol sites, the presence of large, mature individuals of Clarias could indicate capture just after the first rains of the long wet season, when this taxon migrates upriver and congregates in shallow pools for spawning (Stewart, 1994; Van Neer and Lesur, 2004). Clarias can be easily caught during this tightly-defined window of time, often using bare hands (Bruton, 1979; Jubb, 1967). The Kalokol River and nearby seasonal wadis could have been loci for this activity for the occupants of the Kalokol sites. Similarly, the presence of mature tilapia can also indicate predation during the rainy season, during which adults nest in shallow inshore or floodplain locations. However, both Clarias and tilapia could also have been caught in the dry season, especially in residual pools of the receding lake. In eastern Africa, fishers tend to revisit dry-season fishing locations each year, knowing that trapped individuals will be easy prey (as discussed by Stewart, 1989; Stewart, 1994). However, such individuals are, generally speaking, smaller than those captured in the spawning season (Van Neer and Lesur, 2004). Given the large size of most Clarias at the Kalokol sites, we believe wet-season exploitation of this taxon is more likely, but we caution that this is based on very small sample sizes; in the case of tilapia, body size reconstructions do not offer adequate resolution on this issue. Likewise, Nile perch and Synodontis show a wide range of body sizes at the Kalokol sites, without a strong bias toward larger individuals, though there are some larger perch at Kokito 02, and larger Synodontis at Dilit 01. The lack of a strong size trend might indicate that both of these taxa were caught using less selective capture methods, as might have occurred more frequently in deep, open water. Further information, especially more robust fish body size reconstructions from larger samples, is necessary in order to make stronger arguments about the seasonal timing of occupations at the Kalokol sites and their duration. Notably, the lithic assemblages from all three sites suggest dependence on local raw material, which might be interpreted as a sign of lower residential mobility. However, at least in the Kokito sites there is nothing in the artifactual or faunal assemblages that might suggest storage (e.g., in
n Lahr et al. (2016) alongside pottery), as is suggested by Mirazo reduced mobility as a possible cause of inter-group competition and violence. A few non-diagnostic potsherds were recovered from Dilit 01, but the sample size is inadequate for ascertaining the role of pottery at the site. Nevertheless, competition over key fishing localities could have been a source of conflict, even in the absence of reduced mobility or other signs of forager “complexity.” Ownership over strategic fishing spots has been inferred elsewhere in prehistoric eastern Africa (Dale et al., 2004; Prendergast, 2010), and is documented in ethnographic and historic literature on nonindustrial fishing communities elsewhere in the world (as reviewed by Acheson, 1981; see also Reid, 2015; among many others). 6.3. Taxonomic representation across space and time Finally, the Kalokol fauna offer an opportunity to compare

13

archaeological assemblages from the western and eastern sides of the basin, and also enable diachronic comparisons between past and present fish taxonomic abundances. Taxonomic compositions of both ancient and modern assemblages differ substantially from the diversity of the lake as surveyed in the 1970s via trawl catches reported by Hopson (1982) (Fig. 9). This initially suggests targeted capture of specific taxa by humans in both past and present. However, even natural accumulations (Stewart, 1991) are quite distinct from Hopson's surveys. Modern surveys can be problematic since they vary in terms of selectiveness of capture (trawl vs. line) and sampling locations. The Turkana Basin has also experienced marked fluctuations in biodiversity in recent decades, with some survey areas impacted by overfishing (Gownaris et al., 2015). Here, we use Hopson's (1982) inshore/littoral samples for the eastern and western sides of the basin, and we exclude taxa smaller than 350 mm TL following Stewart (1991), who argued that smaller fish are not typically found in either natural or archaeological assemblages. This comparison not only shows how the modern and ancient samples differ from one another, but also highlights key contrasts on either side of the basin. On the eastern side, nearly two-thirds of individuals in Hopson's (1982) survey were Cichlidae (tilapia), while another quarter were other taxa not found archaeologically (Hydrocynus and Alestes), except at low frequencies (3% of MNI each) at the Koobi Fora site of GaJi3 (Stewart, 1989). The absence of these two genera, however, could also be due to their small bones and the likelihood that commonly-used mesh sizes would not catch their remains. Nile perch is a minor taxon in the Hopson (1982) survey (2%), but can be up to half of fish represented (by skeletal MNI) in modern fishing camps (Gifford-Gonzalez et al., 1999). Clarias is also particularly popular at modern fishing camps, ranging from a fifth to two-thirds of skeletal MNI. Among ancient assemblages, Nile perch make up 20e45% of MNI at eastern basin archaeological sites, and similarly comprise about a third of MNI in natural beach scatters (Stewart, 1989, 1991). While targeted capture may explain the Nile perch's abundance in anthropogenic assemblages, other factors e such as differential preservation of its robust bones e may also be at work. Differential preservation could also help explain the abundance of siluroid catfishes, especially Clarias, represented by cranial fragments at eastern basin archaeological sites (Stewart, 1989). On the western side of the basin, Hopson's (1982) survey again shows Cichlidae to be the dominant taxonomic group (40% of inshore/littoral individuals), followed by taxa not represented archaeologically (mainly Hydrocynus and Alestes), and then Labeo, a member of the Cyprinidae family (carps). The absence of cyprinids at the Kalokol sites is notable, since two genera (Labeo and Barbus) have been identified in fossil and modern beach scatters on both sides of the lake (Stewart, 1991); archaeologically, however, these genera are reported only in moderate frequencies at Lowasera (4.5% of MNI) and GaJi3 (8% of MNI), and not at all at FxJj12 (Stewart, 1989). Despite the absence of adequate cyprinid reference material at TBI, routine under-identification of cyprinids seems unlikely, given the analyst's experience with identifying this family elsewhere in eastern Africa (Prendergast, 2010). Nile perch is rare (1%) in the Hopson (1982) western basin trawl survey, though it is notably more abundant in line-caught surveys (Gownaris et al., 2015). The siluroid catfishes e Synodontis, Clarias, and Bagrus e together form less than 10% of Hopson's (1982) survey, but are abundant in both modern (41% of MNI) and fossil (36% of MNI) natural accumulations (Stewart, 1991). Similar frequencies of siluroids are found at Kokito 01 (38% of MNI) and Kokito 02 (27% of MNI), and at Dilit 01, more than two-thirds of MNI are siluroids. While in the western basin Synodontis is generally more abundant than Clarias, in both modern and ancient assemblages, this is not

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

14

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

Fig. 9. Relative abundance of fish taxa in samples from the Turkana Basin. Total MNI of each sample (where available) in parentheses. Archaeological datasets (present work and Stewart, 1989): Kokito 01, Kokito 02, Dilit 01, Lowasera, GaJi3, FxJj12. Natural scatter datasets (Stewart, 1991): fossil beach W, fossil beach E, modern beach W, modern beach E. Modern fishing camps (Stewart and Gifford-Gonzalez, 1994; Gifford-Gonzalez et al., 1999): AS1, FC1, Site 20, Site 10, Site 06. Survey of Hopson (1982), including only littoral-inshore taxa, and only those taxa whose Total Length exceeds 35 cm (as reported by Stewart, 1991): modern trawl W, modern trawl E.

the case at Dilit 01, where Clarias is unusually abundant. While we caution that the Kalokol MNIs are small, this could suggest intersite differences in terms of resource availability or seasonality, as discussed above. 7. Conclusion The Pleistocene-Holocene transition across much of Africa was marked by dramatic climate change and the swelling of lakes and rivers, prompting people's technological innovations and the intensification of previously existing aquatic resource exploitation strategies. The zooarchaeological analysis presented here sheds light on this transition in the western Turkana basin, where until now there has been no quantitative faunal analysis for this period. This has severely constrained our understanding of early Holocene economies, and by extension the social pressures that could have
n Lahr et al., led to inter-group violence among foragers (Mirazo 2016). Our analysis indicates that fisher-foragers during this time frame specialized in the capture of a few key taxa (the same ones preferred by modern Turkana Basin fishers), sometimes well past maturity; this would initially suggest resource abundance, as expected for the AHP. While rich and predictable resources could

theoretically lead to storage and sedentism e and by extension to inter-group competition over resources e there is no archaeological evidence for this at the Kalokol sites, though lithic artifacts seem to indicate reduced mobility, evidenced by heavy emphases on local raw material and core maintenance. Likewise, competition over key fishing locales is simply not detectable from current evidence, though it is a possibility that invites further exploration. A potentially fruitful direction for the dataset presented in this paper is to correlate the occupations of the Kalokol sites with past lake levels, recently reconstructed for the area by Bloszies et al. (2015). Doing so requires a higher degree of chronological resolution than presently available, particularly for Dilit 01. With tight site-to-lake level correspondences, stronger inferences can be made about selection of fishing locales, and about the extent to which body size reconstructions reflect human predation strategies. Such an approach could also help explain the contrasts between the Dilit 01 site, where shallow-water fish are common, and the Kokito sites, where deep-water fish are more abundant. Ultimately, this analysis will contribute to an understanding of the effects of the AHP and fluctuating lake levels on fisher-foragers’ choices at the opening of the Holocene.

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

Acknowledgments We thank V. Linseele, K. Douglass, and A. Antonites for inviting us to contribute to this volume. Fieldwork was supported by grants to AB from the Wenner-Gren Foundation (Grant #8093), National Geographic Society (Grant #9263-13), and Turkana Basin Institute (TBI). MEP was supported while writing this paper by the WennerGren Foundation (Hunt Fellowship) and the Radcliffe Institute for Advanced Study. We thank J. Ekusi, N. Gownaris, K. Kimirei, M. Kirinya, M. Leakey, and F.K. Manthi for guidance prior to and during MEP's visit to TBI, and especially for efforts to expand the reference collection. We are grateful to D. Wright and E. Ndiema for sharing mapping information, to W. Van Neer for advice on fish body size estimation, and to K. Chritz for comments on a draft of this manuscript. We thank two anonymous reviewers and the guest editor for their constructive comments, which improved the manuscript. All errors or omissions are our own. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.quaint.2017.04.022. References Acheson, J.M., 1981. Anthropology of fishing. Annu. Rev. Anthropol. 10, 275e316. Angel, J.L., Phenice, T.W., Robbins, L.H., Lynch, B.M., 1980. Late Stone-Age Fishermen of Lothagam, Kenya, vol. 3. Michigan State University Museum, East Lansing, Mich. Anthropological Series. Ashley, G.M., Ndiema, E.K., Spencer, J.Q.G., Harris, J.W.K., Kiura, P.W., 2011. Paleoenvironmental context of archaeological sites, implications for subsistence strategies under Holocene climate change, northern Kenya. Geoarchaeology 26 (6), 809e837. http://dx.doi.org/10.1002/gea.20374. Barich, B.E. (Ed.), 1987. Archaeology and Environment in the Libyan Sahara. The Excavations in the Tadrart Acacus, 1978-1983, vol. 283. Oxford: B.A.R. International Series. Barich, B.E., 2013. Hunter-gatherer-fishers of the Sahara and the Sahel 12,000e4,000 Years ago. In: Mitchell, P., Lane, P. (Eds.), The Oxford Handbook of African Archaelogy. Oxford University Press, Oxford, pp. 445e460. Barthelme, J., 1985. Fisher-Hunters and Neolithic Pastoralists in East Turkana, Kenya, vol. 254. Oxford: B.A.R. International Series. Behrensmeyer, K., 1978. Taphonomic and ecologic information from bone weathering. Paleobiology 4, 150e162. Beyin, A., 2011. Recent archaeological survey and excavation around the greater Kalokol area, west side of Lake Turkana: preliminary findings. Nyame Akuma 75 (1), 40e50. Beyin, A., Prendergast, M.E., Grillo, K.M., Wang, H., in press. New radiocarbon dates for terminal pleistocene and early holocene settlements in West Turkana, Northern Kenya. Quat. Sci. Rev. http://dx.doi.org/10.1016/j.quascirev.2017.04. 012. Bloszies, C., Forman, S.L., Wright, D.K., 2015. Water level history for Lake Turkana, Kenya in the past 15,000 years and a variable transition from the African humid period to holocene aridity. Glob. Planet. Change 132, 64e76. Brain, C.K., 1981. The Hunters or the Hunted? an Introduction to African Cave Taphonomy. Univeristy of Chicago Press, Chicago. Braun, D.R., Harris, J.W.K., Levin, N., McCoy, J.T., Herries, A.I.R., Bamford, M.K., Bishop, L.C., Richmond, B.G., Mzalendo, K., 2010. Early hominin diet included diverse terrestrial and aquatic animals 1.95 Ma in East Turkana, Kenya. Proc. Natl. Acad. Sci. 107 (22), 10002. Bronk Ramsey, C., 2013. OxCal v. 4.2.4. Retrieved from http://c14.arch.ox.ac.uk/. Bruton, M.N., 1979. The breeding biology and early development of Clarias gariepinus (Pisces: Clariidae) in Lake Sibaya, South Africa, with a review of breeding in species of the subgenus Clarias (Clarias). Transvaal Zool. Soc. 35, 1e4. Bunn, H.T., 1982. Meat-eating and Human Evolution: Studies on the Diet and Subsistence Patterns of Plio-pleistocene Hominids in East Africa. PhD thesis. University of California (Berkeley). Butler, V.L., 1993. Natural versus cultural salmonid remains: origin of the dalles roadcut bones, Columbia River, Oregon, U.S.A. J. Archaeol. Sci. 20 (1), 1e24. http://dx.doi.org/10.1006/jasc.1993.1001. Butzer, K.W., Isaac, G.L., Richardson, J.L., Washbourn-Kamau, C., 1972. Radiocarbon dating of East African lake levels. Science 175 (4026), 1069e1076.
solithiques et ne
olithique de la zone du Nil Chaix, L., 2003. La faune des sites me Bleu (Soudan Central). Complutum 14, 373e396. Close, A.E., 1995. Few and Far between: Early Ceramics in North Africa. Washington, DC. Connah, G., 1981. Three Thousand Years in Africa: Man and His Environment in the

15

Lake Chad Region of Nigeria. Cambridge University Press, Cambridge. Dale, D., 2007. An Archaeological Investigation of the Kansyore, Later Stone Age Hunter-Gatherers in East Africa. PhD thesis. Washington University in St. Louis. Dale, D., Marshall, F., Pilgram, T., 2004. Delayed-return hunter-gatherers in Africa? Historic perspectives from the Okiek and archaeological perspectives from the Kansyore. In: Crothers, G.M. (Ed.), Hunters and Gatherers in Theory and Archaeology. Center for Archaeological Investigations, Carbondale, pp. 340e373. DeMenocal, P., Ortiz, J., Guilderson, T., Adkins, J., Sarnthein, M., Baker, L., Yarusinsky, M., 2000. Abrupt onset and termination of the African Humid Period: rapid climate responses to gradual insolation forcing. Quat. Sci. Rev. 19 (1), 347e361. DeMenocal, P., Tierney, J.E., 2012. Green Sahara: African humid periods paced by earth's orbital changes. Nat. Educ. Knowl. 3 (10), 12. Domínguez-Rodrigo, M., Barba, R., 2006. New estimates of tooth mark and percussion mark frequencies at the FLK Zinj site: the carnivore-hominid-carnivore hypothesis falsified. J. Hum. Evol. 50 (2), 170e194. Domínguez-Rodrigo, M., Barba, R., Egeland, C.P., 2007. Deconstructing Olduvai. A Taphonomic Study of the Bed I Sites. Springer-Verlag, New York. Forman, S.L., Wright, D.K., Bloszies, C., 2014. Variations in water level for Lake Turkana in the past 8500 years near Mt. Porr, Kenya and the transition from the African humid period to holocene aridity. Quat. Sci. Rev. 97, 84e101. Froese, R., Pauly, D. (Eds.), 2016. FishBase. World Wide Web electronic publication. www.fishbase.org. accessed January 2017. Garcea, E.A., 2006. Semi-permanent foragers in semi-arid environments of North Africa. World Archaeol. 38 (2), 197e219. Garcin, Y., Melnick, D., Strecker, M.R., Olago, D., Tiercelin, J.-J., 2012. East African mid-Holocene wetedry transition recorded in palaeo-shorelines of Lake Turkana, northern Kenya Rift. Earth Planet. Sci. Lett. 331e332, 322e334. Gasse, F., 2000. Hydrological changes in the African tropics since the Last Glacial maximum. Quat. Sci. Rev. 19 (1), 189e211.
, F., Vincens, A., Williams, M.A.J., Williamson, D., 2008. Climatic Gasse, F., Chalie patterns in Equatorial and Southern Africa from 30,000 to 10,000 years ago reconstructed from terrestrial and near-shore proxy data. Quat. Sci. Rev. 27, 2316e2340. res diagnostiques des e
pines de quelques Gayet, M., Van Neer, W., 1990. Caracte silures africains. J. Afr. Zool. 104, 241e252. Gifford-Gonzalez, D., Stewart, K.M., Rybczynski, N., 1999. Human activities and site formation at Modern Lake margin foraging camps in Kenya. J. Anthropol. Archaeol. 18 (4), 397e440. Gownaris, N.J., Pikitch, E.K., Ojwang, W.O., Michener, R., Kaufman, L., 2015. Predicting species' vulnerability in a massively perturbed system: the fishes of Lake Turkana, Kenya. PloS one 10 (5), e0127027. Grillo, K., Hildebrand, E., 2013. The context of early megalithic architecture in eastern Africa: the Turkana Basin c. 5000-4000 BP. Azania Archaeol. Res. Afr. 48 (2), 193e217. Haaland, R., 1992. Fish, pots and grain: Early and mid-Holocene adaptations in the central Sudan. Afr. Archaeol. Rev. 10, 43e64. Haaland, R., 1997. Emergence of sedentism: new ways of living, new ways of symbolizing. Antiquity 71, 374e385. Haaland, R., 2007. Porridge and pot, bread and oven: food ways and symbolism in Africa and the Near East from the Neolithic to the present. Camb. Archaeol. J. 17 (2), 165e182. Haaland, R., 2009. Aquatic resource utilization and the emergence of pottery during the Late palaeolithic and Mesolithic. A Global perspective from the Nile to China. In: Ostigard, T. (Ed.), Water, Culture and Identity. Comparing Past and Present Traditions in the Nile Basin Region. BRIC, Bergen. Haaland, R., 1993. Aqualithic sites of the middle Nile. Azania 28, 47e86. Harris, J.W.K., Leakey, M.G., Brown, F.H., 2006. A brief history of research at Koobi Fora, northern Kenya. Ethnohistory 53 (1), 35e69. Hildebrand, E.A., Grillo, K.M., 2012. Early herders and monumental sites in eastern Africa: dating and interpretation. Antiquity 86 (332), 338e352. Hildebrand, E.A., Shea, J.J., Grillo, K.M., 2011. Four middle holocene pillar sites in West Turkana, Kenya. Field Archaeol. 36 (3), 181e200. Holl, A.F.C., 2005. Holocene "aquatic" adaptations in North tropical Africa. In: Stahl, A.B. (Ed.), African Archaeology. A Critical Introduction. Blackwell, Malden, MA, pp. 174e186. Hopson, A.J., 1982. Lake Turkana : a Report on the Findings of the Lake Turkana Project, 1972-1975. Overseas Development Administration, London. Jubb, R.A., 1967. Freshwater fishes of Southern Africa. Capetown. Balkema. _ Krzyzaniak, L., Kobusiewicz, M., Alexander, J., 1993. Environmental Change and Human Culture in the Nile Basin and Northern Africa until the Second Mil
Archaeological Museum, Poznan
. lennium B.C. Poznan € pelin, S., 2006. Climate-controlled Holocene occupation in the Sahara: Kuper, R., Kro motor of Africa's evolution. Science 313 (5788), 803. Kutzbach, J.E., Liu, Z., 1997. Response of the African monsoon to orbital forcing and ocean feedbacks in the middle Holocene. Science 278 (5337), 440e443. Le Quellec, J.-L., 1993. Symbolisme et art rupestre au Sahara. L'Harmattan, Paris. Leakey, L.S.B., 1931. The Stone Age Cultures of Kenya Colony. Cambridge University Press, London. Linseele, V., Zerboni, A., this volume. Done with fish? A diachronic study of fishing in the Holocene Nile basin of Sudan. Quat. Int. Lynch, B.M., Robbins, L.H., 1972. Excavations at Lothagam 1975: a comparison between a typological and computer study of stone artifacts. Azania 12, 43e52. Marchant, R., Mumbi, C., Behera, S., Yamagata, T., 2007. The Indian Ocean dipole e

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022

16

M.E. Prendergast, A. Beyin / Quaternary International xxx (2017) 1e16

the unsung driver of climatic variability in East Africa. Afr. J. Ecol. 45 (1), 4e16. Marshall, F., Stewart, K.M., Barthelme, J., 1984. Early domestic stock at Dongodien in Northern Kenya. Azania 19, 120e127.
n Lahr, M., Rivera, F., Power, R.K., Mounier, A., Copsey, B., Crivellaro, F., Mirazo Edung, J.E., Maillo Fernandez, J.M., Kiarie, C., Lawrence, J., Leakey, A., Mbua, E., Miller, H., Muigai, A., Mukhongo, D.M., Van Baelen, A., Wood, R., Schwenninger, J.L., Grün, R., Achyuthan, H., Wilshaw, A., Foley, R.A., 2016. Intergroup violence among early Holocene hunter-gatherers of West Turkana, Kenya. Nature 529 (7586), 394e398. Mohammed-Ali, A., Khabir, A.-R., 2003. The wavy line and the dotted wavy line pottery in the prehistory of the Central Nile and the Sahara-Sahel Belt. Afr. Archaeol. Rev. 20 (1), 25e58. Ndiema, E.K., Braun, D.R., Dillian, C., 2010. Interaction and exchange across the transition to pastoralism, Lake Turkana, Kenya. In: Dillian, C., White, C.L. (Eds.), Trade and Exchange: Archaeological Studies from History and Prehistory. Springer, New York, pp. 95e110. Ndiema, E.K., Dillian, C., Braun, D.R., Harris, J.W.K., Kiura, P.W., 2011. Transport and subsistence patterns at the transition to pastoralism, Koobi Fora, Kenya. Archaeometry 53 (6), 1085e1098. Nelson, C.M., 1993. A standardized site enumeration system for the Continent of Africa. Nyame Akuma 40, 62e67. Nelson, C.M., 1995. The work of the Koobi Fora field School at the Jarigole pillar site. Kenya Past Present 27, 49e63. Owen, R.B., Barthelme, J.W., Renaut, R.W., Vincens, A., 1982. Palaeolimnology and archaeology of holocene deposits north-east of Lake Turkana, Kenya. Nature 298 (5874), 523. Peters, J., 1991. Mesolithic fishing along the Central Sudanese Nile and the Lower Atbara. Sahara 4, 33e39. Peters, J., 1995. Mesolithic subsistence between the 5th and the 6th Nile cataract: the archaeofaunas from Abu Darbein, El Damer and Aneibis (Sudan). In: Haaland, R., Magid, A.M. (Eds.), Aqualithic Sites along the Rivers Nile and Atbara, Sudan. Alma Mater, Bergen, pp. 172e219. Phillipson, D.W., 2005. African Archaeology, third ed. Cambridge University Press, Cambridge. Phillipson, D.W., 1977. Lowasera. Azania 12, 1e32. Plug, I., Mitchell, P., 2008. Fishing in the Lesotho highlands: 26,000 years of fish exploitation, with special reference to Sehonghong shelter. J. Afr. Archaeol. 6 (1), 33e55. Prendergast, M.E., 2008. Forager Variability and Transitions to Food Production in Secondary Settings: Kansyore and Pastoral Neolithic Economies in East Africa. Unpublished PhD thesis. Harvard University. Prendergast, M.E., 2010. Kansyore fisher-foragers and transitions to food production in East Africa: the view from Wadh Lang'o, Nyanza province. Azania Archaeol. Res. Afr. 45 (1), 83e111. Prendergast, M.E., Lane, P.J., 2010. Middle Holocene fishing strategies in East Africa: zooarchaeological analysis of Pundo, a Kansyore shell midden in northern Nyanza (Kenya). Int. J. Osteoarchaeol. 20 (1), 88e112. Quintana Morales, E., 2013. Reconstructing Swahili Foodways: the Archaeology of Fishing and Fish Consumption in Coastal East Africa, AD 500e1500. Unpublished PhD thesis. Universty of Bristol. Reid, J.L., 2015. The Sea Is My Country. The Maritime World of the Makahs, an Indigenous Borderlands People. Yale University Press, New Haven. Reimer, P.J., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Bronk Ramsey, C., Buck, C.E., Cheng, H., Edwards, R.L., Friedrich, M., Grootes, P.M., Guilderson, T.P.,
, C., Jeaton, T.J., Hoffman, D.L., Hogg, A.G., Haflidason, H., Hajdas, I., Hatte Hughen, K.A., Kaiser, K.F., Kromer, B., Manning, S.W., Niu, M., Reimer, R.W., Richards, D.A., Scott, E.M., Southon, J.R., Staff, R.A., Turney, C.S.M., van der Plicht, J., 2013. INTCAL13 and MARINE13 radiocarbon age calibration curves 0e50,000 years cal BP. Radiocarbon 55, 1869e1887. Robbins, L.H., 1980. Lopoy: a Late Stone-age Fishing and Pastoralist Settlement in the Lake Turkana Basin, Kenya, vol. 3. Michigan State University Museum, East Lansing, MI. Robbins, L.H., 1974. The Lothagam Site. Publications of the Museum, Michigan State University, East Lansing, MI. Anthropology Series. Robbins, L.H., 1984. Late prehistoric aquatic and pastoral adaptations West of Lake Turkana, Kenya. In: Clark, J.D., Brandt, S.A. (Eds.), Hunters and Farmers. University of California Press, Berkeley, pp. 206e212. Robbins, L.H., 1972. Archeology in the Turkana District, Kenya. Science 176 (4033),

359e366. Robbins, L.H., 1975. Bone artifacts from the Lake Rudolf basin, East Africa. Curr. Anthropol. 16 (4), 632e633. Robertshaw, P., Collett, D., Gifford, D., Mbae, N.B., 1983. Shell middens on the Shores of Lake Victoria. Azania 18, 1e43. Seitsonen, O., 2010. Lithics use at Kansyore sites in East Africa: technological organisation at four recently excavated sites in Nyanza Province, Kenya. Azania Archaeol. Res. Afr. 45 (1), 49e82. Stewart, K.M., 1989. Fishing Sites of North and East Africa in the Late Pleistocene and Holocene: Environmental Change and Human Adaptation, vol. 521. Oxford: B.A.R. International Series. Stewart, K.M., 1991. Modern fishbone assemblages at Lake Turkana, Kenya: a methodology to aid in recognition of hominid fish utilization. J. Archaeol. Sci. 18 (5), 579e603. Stewart, K.M., 1994. Early hominid utilisation of fish resources and implications for seasonality and behaviour. J. Hum. Evol. 27 (1e3), 229e245. Stewart, K.M., Gifford-Gonzalez, D., 1994. An ethnoarchaeological contribution to identifying hominid fish processing sites. J. Archaeol. Sci. 21 (2), 237e248. Stojanowski, C.M., Seidel, A.C., Fulginiti, L.C., Johnson, K.M., Buikstra, J.E., 2016. Contesting the massacre at Nataruk. Nature 539 (7630), E8eE10. Street-Perrott, F., Marchand, D., Roberts, N., Harrison, S., 1989. Global Lake-level Variations from 18,000 to 0 Years Ago: a Palaeoclimate Analysis. Prepared for United States Department of Energy Office of Energy Research. Office of Health and Environmental Research (Carbon Dioxide Research Program). Sutton, J.E.G., 1974. The aquatic civilization of middle Africa. J. Afr. Hist. 15, 527e546. Sutton, J.E.G., 1977. The African aqualithic. Antiquity 51, 25e34. Tierney, J.E., Lewis, S.C., Cook, B.I., Legrande, A.N., Schmidt, G.A., 2011. Model, proxy and isotopic perspectives on the East African humid period. Earth Planet. Sci. Lett. 307 (1), 103e112. Trapani, J., 2008. Quaternary fossil fish from the Kibish formation, Omo valley, Ethiopia. J. Hum. Evol. 55 (3), 521e530. Van Neer, W., 1986. Some notes on the fish remains from wadi Kubbaniya (Upper Egypt, Late palaeolithic). In: Brinkhuizen, D.C., Clason, A.T. (Eds.), Fish and Archaeology, pp. 103e113. Oxford: B.A.R. International Series.
ome
trie de la perche du Nil, Lates niloticus Van Neer, W., 1989. Contribution  a la oste
ologie animal pour l'arche
ologie. Se
rie A: (Linnaeus, 1758) (Fiches d'Oste Poissons). Van Neer, W., 1994. Fish remains from Late pleistocene and holocene archaeological sites near Khasm El Girba, Sudan. Archaeofauna 3, 115e126. Van Neer, W., 2004. Evolution of prehistoric fishing in the Nile Valley. J. Afr. Archaeol. 2 (2), 251e269. Van Neer, W., Lesur, J., 2004. The ancient fish fauna from Asa Koma (Djibouti) and modern osteometric data on 2 tilapiini and 2 Clarias catfish species. Doc. Archaeobiol. 2, 143e160. ~ iz, A., 1992. Fish middens: anthropogenic accumulations Van Neer, W., Morales Mun of fish remains and their bearing on archaeoichthyological analysis. J. Archaeol. Sci. 19 (6), 683e695. Van Neer, W., Depraetere, D., 2005. Pickled fish from the Egyptian Nile: osteological evidence from a Byzantine (Coptic) context at Shanhur. Rev. Paleobiol. 10, 159e170. Villa, P., Mahieu, E., 1991. Breakage patterns of human long bones. J. Hum. Evol. 21, 27e48. Walker, R., 1985. A Guide to Post-Cranial Bones of East African Animals. Hylochoerus Press, Norwich. Whitworth, T., 1965. Artifacts from Turkana, Northern Kenya. South Afr. Archaeol. Bull. 20, 75. Willis, L.M., Boehm, A.R., 2014. Fish bones, cut marks, and burial: implications for taphonomy and faunal analysis. J. Archaeol. Sci. 45, 20e25. Willis, L.M., Eren, M.I., Rick, T.C., 2008. Does butchering fish leave cut marks? J. Archaeol. Sci. 35 (5), 1438e1444. Wright, D.K., Forman, S.L., Kiura, P., Bloszies, C., Beyin, A., 2015. Lakeside view: sociocultural responses to changing water levels of Lake Turkana, Kenya. Afr. Archaeol. Rev. 32, 335e367. Yellen, J., 1998. Barbed bone points: tradition and continuity in Saharan and subSaharan Africa. Afr. Archaeol. Rev. 15, 173e198. Zohar, I., Dayan, T., Galili, E., Spanier, E., 2001. Fish processing during the early holocene: a taphonomic case study from Coastal Israel. J. Archaeol. Sci. 28 (10), 1041e1053.

Please cite this article in press as: Prendergast, M.E., Beyin, A., Fishing in a fluctuating landscape: terminal Pleistocene and early Holocene subsistence strategies in the Lake Turkana Basin, Kenya, Quaternary International (2017), http://dx.doi.org/10.1016/j.quaint.2017.04.022