Coastal Subsistence, Maritime Trade, and the Colonization of Small Offshore Islands in Eastern African Prehistory

The Journal of Island and Coastal Archaeology

ISSN: 1556-4894 (Print) 1556-1828 (Online) Journal homepage: http://www.tandfonline.com/loi/uica20

Alison Crowther, Patrick Faulkner, Mary E. Prendergast, Eréndira M. Quintana Morales, Mark Horton, Edwin Wilmsen, Anna M. Kotarba-Morley, Annalisa Christie, Nik Petek, Ruth Tibesasa, Katerina Douka, Llorenç Picornell-Gelabert, Xavier Carah & Nicole Boivin

To cite this article: Alison Crowther, Patrick Faulkner, Mary E. Prendergast, Eréndira M. Quintana Morales, Mark Horton, Edwin Wilmsen, Anna M. Kotarba-Morley, Annalisa Christie, Nik Petek, Ruth Tibesasa, Katerina Douka, Llorenç Picornell-Gelabert, Xavier Carah & Nicole Boivin (2016) Coastal Subsistence, Maritime Trade, and the Colonization of Small Offshore Islands in Eastern African Prehistory, The Journal of Island and Coastal Archaeology, 11:2, 211-237, DOI: 10.1080/15564894.2016.1188334

To link to this article: http://dx.doi.org/10.1080/15564894.2016.1188334

© 2016 The Author(s). Published with Published online: 09 Jun 2016. license by Taylor & Francis Group, LLC© Alison Crowther, Patrick Faulkner, Mary E. Prendergast, Eréndira M. Quintana Morales, Mark Horton, Edwin Wilmsen, Anna M. Submit your article to this journal Article views: 330 Kotarba-Morley, Annalisa Christie, Nik Petek, Ruth Tibesasa, Katerina Douka, Llorenç Picornell-Gelabert, Xavier Carah, and Nicole Boivin View related articles View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=uica20

Download by: [UQ Library] Date: 19 August 2016, At: 23:49 Journal of Island and Coastal Archaeology, 11:211–237, 2016 Copyright © 2016 Alison Crowther, Patrick Faulkner, Mary E. Prendergast, Erendira´ M. Quintana Morales, Mark Horton, Edwin Wilmsen, Anna M. Kotarba-Morley, Annalisa Christie, Nik Petek, Ruth Tibesasa, Katerina Douka, Llorenc¸ Picornell-Gelabert, Xavier Carah, and Nicole Boivin ISSN: 1556-4894 print / 1556-1828 online DOI: 10.1080/15564894.2016.1188334

Coastal Subsistence, Maritime Trade, and the Colonization of Small Offshore Islands in Eastern African Prehistory

Alison Crowther,1 Patrick Faulkner,2 Mary E. Prendergast,3 Erendira´ M. Quintana Morales,4 Mark Horton,5 Edwin Wilmsen,6 Anna M. Kotarba-Morley,7 Annalisa Christie,8 Nik Petek,9 Ruth Tibesasa,10 Katerina Douka,11 Llorenc¸ Picornell-Gelabert,12 Xavier Carah,1 and Nicole Boivin13,14 1School of Social Science, University of Queensland, Brisbane, Queensland, Australia 2Department of Archaeology, School of Philosophical and Historical Inquiry, University of Sydney, Sydney, New South Wales, Australia 3Department of Sociology and Anthropology, St. Louis University in Madrid, Madrid, Spain 4Department of Anthropology, Rice University, Houston, Texas, USA 5Department of Archaeology and Anthropology, University of Bristol, Bristol, UK 6School of Geography, Archaeology and Environmental Studies, University of the Witwatersrand, Johannesburg, South Africa 7Centre for Archaeological Science, School of Earth and Environmental Sciences, University of Wollongong, Wollongong, New South Wales, Australia 8Sainsbury Research Unit for the Arts of Africa, Oceania and the Americas, University of East Anglia, East Anglia, UK 9Department of Archaeology and Ancient History, Uppsala University, Uppsala, Sweden 10Department of Archaeology and Anthropology, University of Pretoria, Pretoria, South Africa 11Research Laboratory for Archaeology and the History of Art, University of Oxford, Oxford, UK

Received 6 October 2015; accepted 7 May 2016. Address correspondence to Alison Crowther, School of Social Science, University of Queensland, Bris- bane, QLD 4072, Australia. E-mail: [email protected] Color versions of one or more ﬁgures in this article are available online at www.tandfonline.com/uica This is an Open Access article distributed under the terms of the Creative Commons Attribution- NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way.

211 Alison Crowther et al.

12Department of Historical Sciences and Theory of Art, University of the Balearic Islands, Palma, Spain 13School of Archaeology, University of Oxford, Oxford, UK 14Max Planck Institute for the Science of Human History, Jena, Germany

ABSTRACT

Recent archaeological research has firmly established eastern Africa’s offshore islands as important localities for understanding the region’s pre-Swahili maritime adaptations and early Indian Ocean trade connections. While the importance of the sea and small offshore islands to the development of urbanized and mercantile Swahili societies has long been recognized, the formative stages of island colonization—and in particular the processes by which migrating Iron Age groups essentially became “maritime”—are still relatively poorly understood. Here we present the results of recent archaeological fieldwork in the Mafia Archipelago, which aims to understand these early adaptations and situate them within a longer-term trajectory of island settlement and pre-Swahili cultural developments. We focus on the results of zooarchaeological, archaeobotanical, and material culture studies relating to early subsistence and trade on this island to explore the changing significance of marine resources to the local economy. We also discuss the implications of these maritime adaptations for the development of local and long-distance Indian Ocean trade networks.

Keywords ﬁshing, Iron Age, Late Holocene, Maﬁa Archipelago, maritime adaptation, Pre-Swahili

INTRODUCTION long-term processes by which some early terrestrial groups—which in the late Holocene While much research in Africa has focused included farmers, pastoralists, and hunter- on the significance of maritime adaptations gatherers—essentially became ‘maritime’. for models of earlier modern human be- Archaeological research over the past two havior and out-of-Africa migrations, with decades clearly points to the offshore islands particular emphasis on the archaeologi- as key loci for understanding early maritime cal evidence from southern Africa (e.g., adaptations by the region’s foraging and Marean 2014), the development of coastal farming communities, while the presence economies during the late Holocene has of Indian Ocean trade goods at several sites received much less attention (see Blench offers tantalizing, but as yet unconfirmed 2012; Mitchell 2004). In eastern Africa, the links, to precocious maritime trade (see importance of the sea—and in particular the Boivin et al. 2013 for a recent review). Many small offshore islands that dot the coastline of these sites remain poorly dated, however, from Lamu in the north to Mafia and Kilwa and there are still major archaeological gaps, in the south (Figure 1)—to the development particularly in the subsistence records of of maritime Swahili trading societies of early sites, that prevent proper understand- the second millennium AD has long been ing of these formative coastal adaptations recognized (e.g., Fleisher et al. 2015; Horton and maritime developments. and Middleton 2000). What is less well Archaeological research on eastern understood, however, are the formative Africa’s smaller islands has also been partially stages of these coastal societies, and the eclipsed by studies focused on the larger

212 VOLUME 11 • ISSUE 2 • 2016 Small Islands in Eastern African Prehistory

Figure 1. a) Map of the eastern African coast showing sites mentioned in text; b) map of the Maﬁa Archipelago showing the location of the Juani Primary School site on Juani Island.

archipelagoes, particularly Zanzibar, which ological investigations at the Juani Primary were major hubs of Swahili and colonial pe- School site in the Mafia Archipelago, which riod trade. Yet multi-disciplinary research was one of the few island groups off the east- in other regions of the world, such as the ern African coast to be colonized during this PacificandCaribbean,hasdemonstratedthat initial Iron Age phase. In contrast, the first se- smaller offshore islands are often colonized cure evidence of the settlement of the Lamu earlier than those with larger landmasses ow- andZanzibarArchipelagoes,aswellthemore ing to their greater ecological diversity, par- distant Comoros, dates to no earlier than the ticularly resource-rich marine habitats (see seventh–eighthcenturiesAD(e.g.,Crowther Keegan et al. 2008 for discussion). We also et al. 2015; Fleisher and LaViolette 2013; Hor- know that colonization of small islands can ton 1996b; Wright et al. 1984, 1992). Our have dramatic impacts on endemic flora and excavations at the Juani Primary School site fauna (Masse et al. 2006; Rick and Erlandson comprise the first attempt to systematically 2008; Weisler 2003). Comparative datasets recover and analyze high-resolution zooar- for island colonization, resource exploita- chaeological and archaeobotanical evidence tion, and the long-term ecological impacts from any coastal EIA site, allowing us to un- of human occupation of small islands are derstand more fully how these early farming needed for eastern Africa, to contribute to groups adapted socially and economically to these global debates. their new littoral setting. The presence of The aim of this article is to examine a later, Middle–Late Iron Age (MIA–LIA, c. evidence for coastal adaptations in eastern seventh–fifteenth centuries AD) occupation Africa during the Early Iron Age (EIA, c. first– at the site also provides a useful point of com- sixth centuries AD), when pottery-using, parison for situating these early adaptations iron-working, farming groups—to whom the within a longer-term trajectory of island set- ancestors of the Swahili have been traced— tlement, subsistence, and maritime trade. first migrated to the eastern African coast and The emerging picture suggests that began to settle the small offshore islands. We while some EIA populations shifted rapidly focus on the results of our recent archae- to a forager-fisher lifestyle during the ini-

JOURNAL OF ISLAND & COASTAL ARCHAEOLOGY 213 Alison Crowther et al.

tial stages of island and coastal occupation, blefrommostelementsofSwahililife.”While the intense focus on marine resources was the emergence of a complex and urbanized only a short-term adaptation, with a broad- maritime-oriented culture, featuring strong ening of the subsistence base occurring in engagement with the wider Indian Ocean the later MIA period. We suggest that this world, is undoubtedly unique to the Swahili marine foraging adaptation may have been era, we would argue from new evidence a colonization-phase strategy. Local trade that the Swahili were not the first maritime- connections in this early phase, based for focused culture on the eastern African coast. example on the production and exchange of We propose that the Swahili culture needs to marine shell disk beads, may have supported be situated within a long-term record of mar- the longer-distance Indian Ocean trade net- itime adaptation across the region, one that is works that emerged during later periods and neither straightforwardly cumulative nor de- became a defining feature of Swahili mar- velopmentallycontinuous.Understandingof itime economy and identity. this longer-term record is hampered by limited material evidence and a reliance on minimal and ambiguous textual evidence that is only beginning to be remedied by systematic ARCHAEOLOGICAL CONTEXT: archaeological research. MARITIME TRADE AND ADAPTATION In addressing the maritime orientation ON THE EASTERN AFRICAN COAST of pre-Swahili groups, it is important to de- fine and distinguish among “coastal use,” Eastern Africa’s Swahili coast occupies a nar- “coastal adaptations,” and “maritime adap- row strip of land extending some 3000 km tations.” For example, Beaton (1995:801– from the north coast of Somalia to south- 802) and Marean (2014:20) distinguish be- ern Mozambique, incorporating a number tween coastal use, where marine resources of small offshore Indian Ocean islands, in- are exploited systematically, regularly, and cluding the Lamu, Zanzibar, Kilwa, and Mafia recurrently, but do not necessarily trans- Archipelagoes, as well as the more dis- form lifeways, and coastal adaptations, tant Comoros (Horton and Middleton 2000) where economies, mobility, and settlement (Figure1).TheSwahiliregionistodaycharac- structures have been transformed to re- terized by a distinctive coastal culture, which volve around marine environments and the emerged in the last millennium from indige- inter-tidal zone in particular. Building on nous roots (Fleisher et al. 2015). The Swahili these points, Marean (2014) views maritime exploited the monsoon winds and ocean cur- adaptations as those incorporating marine rents to ply the coast and islands in their resource use via boat technology capable sewn boats, dhows, and outrigger canoes, of long-distance transport, deep-sea fishing, acquired raw commodities from the African and whaling. Fleisher et al. (2015) extend interior for local, inter-regional, and interna- this type of definition to include not just tional trade, and developed a coastal-focused maritime resource exploitation, technology, culture defined by complex social, commer- trade, and communication, but also the no- cial, and urbanized features. tion of socio-cultural structures influenced The question of when the Swahili be- by the maritime environment. came maritime has recently attracted archae- These are important definitions to bear ological attention. One proposal is that this in mind, as the archaeological record of the was a relatively late phenomenon dating to Swahili coast shows a long history of occu- the last 1,000 years (Fleisher et al. 2015). pation extending back to the Pleistocene, While not denying the earlier coastal adap- when hunter-gatherer groups using a mi- tations of pre- and proto-Swahili groups, crolithic Later Stone Age technology fre- Fleisher and colleagues (2015:102) focus on quented many of the limestone caves in the the significant re-orientation of settlement coastal uplands (Helm 2000b; Shipton et al. patterns, subsistence, domestic spaces, and 2013). These groups only occasionally ex- material culture in this period as marking the ploited coastal resources such as fish and point when “maritimity becomes inextrica- marine shells, the latter not only for eco-

214 VOLUME 11 • ISSUE 2 • 2016 Small Islands in Eastern African Prehistory

nomic, but also social purposes (for exam- Iron Age groups. Archaeological studies of ple, the production of cowrie [Cypraeidae ceramics and settlement patterns have yet spp.] shell beads). Despite claims for Later to provide resolution on interpretations of Stone Age groups on Juani as well as on direct continuity (with MIA culture develop- the island of Zanzibar (Chami 2004, 2009), ing from its EIA predescessor; e.g., Chami re-excavation and re-dating of two key sites 1994; Helm 2000b) versus discontinuity be- has failed to replicate evidence for island col- tween culturally discrete populations (Col- onization prior to the Iron Age. Rather, in lett 1985; Horton 1996a, 1996b; see also the case of Zanzibar, Later Stone Age oc- Nkirote M’Mbogori 2015). cupation predates the postglacial sea-level The EIA also broadly coincides with the rise that led to the formation of both this first documentary evidence of Indian Ocean island and the Mafia Archipelago by the early trade in eastern Africa. The first century CE Holocene (Crowther et al. 2014; Prendergast Greco-Roman text, the Periplus Maris Ery- et al. 2016; Shipton et al. in press). While thraei (‘The Periplus of the Erythraen Sea’; occupation of the coast thus seems limited Casson 1989) describes a thriving trade on prior to the Iron Age, it is possible that sites the eastern African coast, involving a num- might have been inundated by rising seas. ber of small trading centers that extend as Currently, the first clear evidence of perma- far south as the (so far unidentified) empo- nentandsustaineduseofthecoast,reflecting rium of “Rhapta.” Rhapta was described as a specific coastal adaptation, appears in the lying opposite the also unidentified island of first millennium AD after the arrival of Iron “Menouthias”—thought to be either Pemba, Age groups. The dominant paradigm is that Zanzibar, or Mafia. Rhapta would thus be lo- these Iron Age groups spread to the coast cated on the mainland anywhere from the as part of the broader dispersal of Bantu- Pangani River to the Rufiji Delta (Baxter speakingfarmersintosub-SaharanAfrica(see 1944; Chami 1999b; Datoo 1970; Horton de Maret 2013; Holden 2002; Russell et al. 1996a:451; Horton and Middleton 2000:33). 2014). Here, traders from the Arabian peninsula In coastal eastern Africa, three major brought metal objects, including spears, chronological and cultural sub-divisions of knives, small awls, and different types of the Iron Age are broadly recognized1:the “glass stones” (perhaps rock crystal or glass EIA or pre-Swahili period (c. first–sixth cen- beads),aswellasgrainandwine,inexchange turies AD), traced archaeologically by the for ivory, rhinoceros horn, tortoise shell, presence of bevel-rimmed Kwale ware ce- and possibly nautilus shell (Casson 1989). Al- ramics; the MIA or proto-Swahili period though a number of coastal and near-coastal (c. seventh–tenth centuries AD) associated easternAfricansiteshavebeenidentifiedthat with triangular-incised Early Tana Tradi- date to the EIA, none of these contain any se- tion ceramics and the florescence of In- cure evidence of long-distance trade connec- dian Ocean trade; and the Late Iron Age or tions(Boivinetal.2013;Wood2011).Rather, Swahili period (c. eleventh–fifteenth cen- the majority of archaeological evidence for turies CE), which saw the transformation early Indian Ocean trade in this region dates of many coastal villages into urban cen- from the MIA onwards, when the offshore ters ruled by a cosmopolitan merchant elite islands in particular became major commer- (e.g., Chami 1994; Fleisher and Wynne-Jones cial loci, as seen at large trading ports such as 2011; Helm 2000a; Helm et al. 2012; Hor- Shanga and Manda in the Lamu Archipelago, ton 1996b; LaViolette 2008). Prior to the TumbeonPemba,UngujaUkuuonZanzibar, emergence of urban Swahili “stone towns” in and Kilwa in the Kilwa Archipelago (Chittick the eleventh century, early coastal commu- 1974, 1984; Fleisher and LaViolette 2013; nities lived in relatively small, mixed farm- Horton 1996b; Juma 2004). ing settlements dominated by wattle and Evidence from eastern Africa’s offshore daub architecture. There is still some de- islands is particularly crucial in resolving bate about the chronological and cultural re- questions surrounding the region’s early lationship between these Early and Middle coastal subsistence, adaptations, and Indian

JOURNAL OF ISLAND & COASTAL ARCHAEOLOGY 215 Alison Crowther et al.

Ocean trade, given their central role in MIA ological substrate comprises upraised coral and later Swahili maritime trade, as well as, limestone that has emerged as rocky cliffs more fundamentally, the necessity of mar- along the east coast, though the north- itime seafaring technologies for successful ern and western sides of the island have island colonization. However, identified EIA more sheltered mangrove-lined coastlines sites are relatively rare on the eastern African and intertidal flats. The island’s vegetation littoral(Chami2003),mostlikelyasaresultof is currently a mosaic of agricultural areas, sampling or preservation bias rather than ac- secondary forests, tidal mangrove swamps, tual distribution patterns. Rather, the major- scrubby coastal moorlands, and degraded fal- ity of known EIA sites are located in the main- low bush, although the presence of rem- land hinterland, particularly in the Kwale, nant patches of evergreen lowland coastal Usambara, Pare, and Taita Hills of southern forest (found all along eastern Africa’s coast Kenya and northern Tanzania (Chami 1992; and on the offshore islands; Burgess and Collett 1985; Soper 1967; Walz 2010) (see Clarke 2000) suggests that forest environ- Figure 1), as opposed to on the islands. This ments might have once been more domi- distribution provides few opportunities to nant (Greenway et al. 1988). The mangrove examine the early phases of coastal colo- swamps provide a rich array of resources, in- nization and adaptation by Iron Age groups. cluding fish, mollusks, and crustaceans, and One of the main exceptions is the Mafia wood for charcoal, craft, and building ma- Archipelago, located some 20 km off the terial (Christie 2011). The island also has a central Tanzanian coast, where a number rich marine and littoral fauna supported by of EIA sites have been reported since the a range of tropical marine habitats including late 1990s (Chami 1999a, 2000, 2004; Chami coral reefs, sea grass beds, mangroves, and and Msemwa 1997a; see also Chami and inter-tidal flats. The mammalian fauna today Msemwa 1997b). While preliminary archae- is much more depauperate than that of other ological investigations have been conducted continental eastern African islands and re- on a small number of these sites, none have markably lacking in endemics (Walsh 2007). been subjected to the kinds of systematic Communities living on Juani today practice a study needed to fully assess the role of the mix of rain-fed agriculture (focused on the sea to early society and economy on the east- cultivation of rice, Oryza sativa; cassava, ern African coast. As a result, the region cur- Manihot esculenta;bananas,Musa spp.;and rently lacks an empirically informed model maize, Zea mays), animal husbandry (mainly for island colonization, coastal subsistence, the keeping of cattle, Bos sp.; goats, Capra and the emergence of maritime adaptations hircus; and chickens, Gallus gallus), fishing, during the EIA. and shellfish gathering. The island lacks any large rivers, but has several seasonal and in- termittent streams (Greenway et al. 1988). JUANI PRIMARY SCHOOL SITE, MAFIA Groundwater aquifers tapped from surface ARCHIPELAGO wells provide the main water source for the island’s inhabitants. In order to address questions surrounding The Juani Primary School site, also pre-Swahili maritime subsistence and trade, known by its local name, Kisima Jumbe, was we conducted new archaeological excava- first investigated in the 1990s by Felix Chami, tions at the Juani Primary School site in the whose excavations revealed the remains of Mafia Archipelago, located off the central a large earth-and-thatch village that was rich coast of Tanzania (Figure 1). As its name sug- in EIA Kwale ceramics and evidence for the gests, the site is located within the grounds exploitationofmarineresourcessuchasmol- of a local primary school at the central north- lusks, fish, and possibly shark (Chami 1999a, ern end of Juani Island, which is a small islet 2000, 2004). Chami’s excavations, however, situated to the east of Mafia Island. did not involve any archaeobotanical anal- Juani, like all islands in the Mafia ysis to document the role of agriculture Archipelago, is very flat and low lying. Its ge- relative to marine and other food sources,

216 VOLUME 11 • ISSUE 2 • 2016 Small Islands in Eastern African Prehistory

and only a small archaeofaunal assemblage were included in the analysis and examined was reported, focusing mainly on mollusks, for signs of modifications. with largely unidentified fish and no tetrapod Archaeobotanical samples (analyzed by fauna. The University of Oxford’s Sealinks AC) were sieved into size fractions and the Project returned to the site in 2012 with the ≥1 mm fractions were scanned for charred aim of recovering higher resolution faunal, remains (seeds, chaff, etc.) using a stereomi- archaeobotanical, and material culture data, croscope at 10–40 × magnification. Taxo- and situating these within a radiocarbon- nomic identifications of crop remains were based chronological framework for Iron Age made using published criteria (e.g., Giblin occupation at the site. and Fuller 2011; Manning et al. 2011) and Our excavations focused on an area in botanical reference collections at Univer- the northern periphery of the site (Figure 2), sity College London. Nine AMS radiocarbon where a shovel test pit survey located the dates were obtained on identified charred richest and potentially deepest deposits in seeds and charcoal fragments to establish an aslightlyraisedareameasuringabout50m absolute chronology for each of the main across. This area was also where Chami’s occupation phases identified in our excava- Trench 1 was excavated in 2000 (Chami tions. Radiocarbon dated charcoal samples 2000). Here we placed four trenches, each 2 were identified by LPG using standard char- × 2 m in size and approximately 20 m apart. coal analysis procedures, with reference to Trencheswereexcavatedincontrolledstrati- woodanatomyatlasesofflorafromAfricaand graphic sequence using the single context adjacent regions (Fahn and Werker 1986; Fa- method, with thicker contexts being sub- solo 1939; Neumann et al. 2007). Where pos- divided into smaller arbitrary units to obtain sible, charcoal specimens identified to the tighter vertical control for sampling. Bulk Rhizophoraceae (mangrove) family were tar- samples of at least 30–60 liters per context geted for dating, as species within this group were processed by flotation (0.3 mm mesh) generally do not form large girth trees that to recover charred plant remains. All re- may have greater inbuilt age errors. maining excavated deposits were either dry- (3 mm) or wet-sieved (1 mm, after flotation). The archaeomalacological samples (ana- OCCUPATION PHASES AND DATING lyzed by PF) were identified using a range of published sources (Abbott and Dance 1998; The general stratigraphic sequence at the Carpenter and Niem 1998; Richmond 2011; Juani Primary School site was similar across Robin 2008, 2011; Rowson 2007; Rowson all trenches, comprising an upper topsoil et al. 2010) and the nomenclature standard- layer with mixed modern and ancient ma- ized with reference to the World Register terials, underlain by a series of reddish- of Marine Species (WoRMS Editorial Board brown sandy silts that contained the bulk 2016). The Minimum Number of Individu- of the archaeological finds, transitioning als (MNI) was recorded based on a range of to a fine yellowish-white beach sand con- taxon-specific non-repetitive elements (fol- taining little to no cultural materials (Fig- lowing Harris et al. 2015), in addition to ure 3). Within the archaeological sequence, noting the presence of taxa that did not re- we identified two main occupation phases. tain diagnostic features necessary for inclu- The first phase belonged to the EIA and sioninMNIcounts.Tetrapodfauna(analyzed contained a rich assemblage of Kwale ce- by MEP) were identified with reference to ramics (including some near-complete ves- the collections of the National Museums of sels; Figure 4a–g) in association with a dense Kenya (NMK) in Nairobi. All identified speci- midden deposit and very small quantities mens were examined using a 20 × hand lens of slag from iron working. All diagnostic under strong oblique light to identify sur- sherds found in this horizon were Kwale (n face modifications. Fish remains (analyzed =455),thoughtheremainingnon-diagnostic by EQM) were also identified using the os- sherds (n = 2836) are probably also Kwale teology collection at the NMK. All elements based on similar fabric and surface treat-

JOURNAL OF ISLAND & COASTAL ARCHAEOLOGY 217

218

Figure 2. Plan of the Juani Primary School site, showing the location of the four excavated trenches (JS12-03 to JS12-06) at the north-western boundary of the site, where stratiﬁed Kwale-bearing deposits were located. A large pottery scatter was also located at the far eastern edge of the site, near two present-day mosques (one abandoned), but this contained only later ceramics. Small Islands in Eastern African Prehistory

Figure 3. a) Drawing of trench JS12-03 east section showing representative site stratigraphy and main cultural layers; b) photograph of trench JS12-04 north section showing the dense shell layer (inset), most likely representing a natural accumulation from a storm surge or similar event. Scale = 50 cm. ments. No sherds belonging to any other con- Late Iron Age phase containing a small num- temporaneous ceramic tradition—eastern ber of Tana Tradition ceramics (n = 31) African or imported—were identiﬁed. (Figure 4h) as well as some Kwale sherds. While we cannot entirely exclude the possi- Given the ∼250–400 year hiatus in the ra- bility that the site’s occupants were hunter- diocarbon dates between the EIA and MIA gatherers who traded or adopted pottery layers,itseemslikelythattheKwalesherds from Iron Age groups on the mainland, have been incorporated post-depositionally the large quantity of high-quality decorated into these later layers. The MIA–LIA pottery Kwale sherds (>40 kg in total), found in assemblage was not as rich as that from the combination with evidence of iron-working EIA layers, yielding only 35.9% (24.4 kg) of as well as daub structures, are strongly sug- the entire assemblage by weight. gestive of an Iron Age cultural afﬁliation. In trenches JS12-04 and -05, as well as the The Kwale layer was overlain by a Middle– southeastcornerofJS12-03(context306K),a

JOURNAL OF ISLAND & COASTAL ARCHAEOLOGY 219 Alison Crowther et al.

Figure 4. Examples of local ceramics recovered from the Juani Primary School site: a–g) Early Iron Age Kwale; h) Middle Iron Age Tana tradition. (Contexts: a–b: 304H; c–d: 405E; e–f: 406F; g: 506J; h: 303E).

5–8 cm thick shell layer was found (Figure 3), deposits (e.g., Attenbrow 1992; Henderson composed almost entirely of densely com- et al. 2002), we currently interpret this layer pacted, highly fragmented, and very small as most likely representing water-borne de- (mostly 3–15 mm) marine (e.g., Batillariidae position via a storm surge or similar natural spp. Cerithiidae spp., Columbellidae spp.) event. and terrestrial (e.g., Cerastidae spp., Subu- Both occupations at the Juani Primary linidae spp.) gastropods, and small articu- School site were probably relatively small-

lated bivalves. Although this context con- scale. Evidence of daub from all layers points tained some larger, economic taxa (such totheformerpresenceinourexcavationarea as Anadara antiquata, Fasciolariidae sp., of small earth-and-thatch structures typical Modiolus sp., and Strombidae spp.), the vast of Iron Age village settlements and, while majority of the small marine shell taxa recov- not conclusive, are suggestive of permanent, ered from this layer has not been recorded year-round habitation rather than seasonal previously as economic species in the reoccupation. Although Christie (2011:139) gion (see below; Christie 2011:314). The notes that many of the more common mol- layer contained some Tana Tradition as well lusk taxa within the Mafia Archipelago are as plain ceramics, which may have been more abundant during particular seasons, mixed in. In trenches JS12-03 and JS12-04, with few exceptions (e.g., Pinctada spp.) the shell layer contained little to no other mollusks would have been available for ex- cultural material, while in trench JS12-05 (in ploitation throughout the year, a factor also which the layer was thin, and where there highlighted by Msemwa (1994:174). Simi- also appeared to have been the most strati- larly, the terrestrial mammals and fish re- graphic mixing, based on anomalous radio- covered at the site would have been avail- carbon dates on charred seeds, see discus- able year-round; however, the small sample sion below) it also bore daub fragments, sizes do not permit seasonality analysis. This shell disk beads, and a glass piece. This evidence, combined with the pottery- and layer is also delineated by clear stratigraphic shell-rich midden deposits we uncovered, changes, which in addition to the paucity of suggests our excavations were in a domes- cultural material and significant amounts of tic occupation area. Very small quantities of non-economic mollusk taxa and fragmented slag were also found (<50 g in total from all shell, also include a shift from fine brown four trenches), indicating that iron-working silts in the adjacent contexts to coarse yellow activities were occurring in the vicinity, but sand in its sedimentary matrix. Pending fur- that the area of our excavation was not the ther analysis, in line with criteria developed main locus of these activities. The site’s iron- for distinguishing natural and cultural shell rich lateritic soils would have provided the

220 VOLUME 11 • ISSUE 2 • 2016 Small Islands in Eastern African Prehistory

Table 1. AMS radiocarbon dates from the Juani primary school site.

Calibrated age Trench, rangeb (95.4% Cultural context Material, taxon Lab no.a14C age BP probability) phase

JS12-03, 303F Charcoal, Wk-40967 1153 ± 25 AD 885–990 MIA–LIA Rhizophoraceae JS12-03, 304G Charcoal, Wk-40966 1612 ± 21 AD 420–535 EIA Rhizophoraceae JS12-03, 304H Charcoal, Wk-40965 1604 ± 20 AD 420–540 EIA Rhizophoraceae JS12-05, 503E Charred seed, Wk-40939 1173 ± 20 AD 875–980 MIA–LIAc Vigna cf. unguiculata JS12-05, 505H Charred seed, Wk-40938 1184 ± 21 AD 775–975 MIA–LIAc Vigna cf. unguiculata JS12-05, 505I Charred seed, Wk-40937 1181 + 20 AD 775–975 MIA–LIAc Vigna sp. JS12-05, 506J Charcoal, Wk-40968 1646 ± 23 AD 385–525 EIA Rhizophoraceae JS12-06, 604E Charcoal, Wk-40970 1002 ± 21 AD 1020–1150 MIA–LIA

Ziziphus sp. JS12-06, 607I Charcoal, Wk-40969 930 ± 24 AD 1040–1210 MIA–LIA Rhizophoraceae

aAll radiocarbon dates were obtained from the Waikato Radiocarbon Laboratory, New Zealand using standard pretreatment methods. bRadiocarbon dates were calibrated using mixed southern (SHCal13, 70%) and northern (IntCal13, 30%) hemisphere calibration curves to account for the effects of the intertropical convergence zone (ITCZ) on the study area. Calibrations were calculated using OxCal v4.2.4 at 95.4% probability (2σ ) and were rounded to 5 years. cSeed samples were recovered from EIA contexts; direct dating demonstrates that these are intrusive from overlying MIA layers. raw material for on-site iron production. Our otanical and material culture studies relat- radiocarbon dates suggest the Kwale occu- ing to subsistence and trade from these two pation may have been quite short, possibly occupation phases (EIA and MIA–LIA). In just one or two generations, with all three addition to exploring the changing signifi- dates clustering between cal AD 385–540 cance of marine resources to the local econ- (Table 1). The later phase appears slightly omy, we also discuss the implications of longer in duration, with dates spanning ca. these maritime adaptations for the develop- cal AD 880–1200, though these fall into two ment of local and long-distance Indian Ocean main clusters around cal AD 775–990 and trade networks. cal AD 1020–1210 (Table 1), and could in fact reflect two discrete occupations. These later layers are nonetheless considered here EVIDENCE OF MARITIME SUBSISTENCE as one phase in light of their shared cultural ADAPTATIONS affiliation. The following discussion focuses on the While direct comparison of the relative im- results of our zooarchaeological, archaeob- portance of different food sources to the

JOURNAL OF ISLAND & COASTAL ARCHAEOLOGY 221 Alison Crowther et al.

diet of the Juani occupants is not possible anthropogenic modifications potentially re- without isotopic studies of human remains, lated to ornament manufacture are evident. their relative contributions can nonetheless While much of the shell assemblage is be broadly estimated based on the overall still under analysis, owing to its size, the pre- abundance of different plant and animal re- liminary results from Trench 3 (JS12-03) are mains in the archaeological record. What reported here. The condition of the shell stands out from our data is the overwhelm- is moderate, and although they cannot be ing dominance of marine resources, such as quantified at this stage, fragmentation and mollusks and fish, at the Juani Primary School dissolution are apparent to varying degrees site, particularly during the EIA. In compar- throughout the deposit. As the focus here ison, terrestrial food sources were quantita- is on comparing trends in mollusk exploita- tively minor, with plant and animal domes- tion and discard between cultural phases, ticates being entirely or nearly absent from the natural shell layer noted in the discus- the EIA layers. While this pattern may reflect sion above (context 306K) has been ex- differential deposition and preservation, the cluded so as not to skew the data available general importance of marine resources to for the Middle–Late Iron Age. The remain- EIA subsistence at the site is nonetheless ing JS12-03 molluskan assemblage, compris- apparent. ing a total of 89 identified invertebrate categories (e.g., species, genus, or family), is Shellfish Remains represented by a total of 1838 specimens by MNI, with 1510 MNI in the EIA and 328 MNI By far the most abundant type of sub- in the MIA–LIA contexts. The EIA is domi- sistence evidence recovered from the Juani nated by Nerita spp. (40.9% MNI), which de- Primary School site, for both the Early and creases markedly in abundance in the MIA– Middle–Late Iron Age occupation phases, LIA (14.0% MNI) (Figure 5a). The distribu- was shellfish. Previous research on Juani tion of the Cypraeidae spp. (Cypraea and

Island has also highlighted the abundance Monetaria spp.) and the sandflat bivalve At- of molluskan faunal remains in Iron Age actodea striata demonstrate the inverse pat- and later sites (Chami 2004; Christie 2011), tern, increasing from the EIA (1.7% and 1.9% with more than 15 taxa recorded, including MNI respectively) into the MIA–LIA contexts Anadara antiquata, Cypraea tigris, Mon- (12.5% and 14.6% MNI respectively). Similar etaria annulus, Pleuroploca trapezium, patterns of increasing abundance from the Nerita textilis, Chicoreus ramosus, Tere- EIA to the MIA–LIA are also seen in a num- bralia palustris,andVolema pyrum as be- ber of other taxa (albeit to a lesser degree), ing the most frequently represented (Chami notably the sandflat bivalve Anadara anti- 2000:212, 2004:90; Christie 2011:290). The quata, the mangrove gastropod Terebralia Juani Primary School molluskan faunal as- palustris, and the intertidal/shallow subtidal semblage was not only relatively large (com- Strombidae spp. gastropods (Figure 5a). prising approximately 31.3 kg of shell in total Comparison of the EIA and MIA–LIA from the four excavation trenches; JS12-03 = phases at Juani elucidates some important 16.9 kg, JS12-04 = 4.7 kg, JS12-05 = 7.7 kg, trends. The density of molluskan remains JS12-06 = 2.1 kg), but also very rich, sig- (based on the calculation of approximate nificantly expanding on the number of taxa deposit or phase volume and indicated by noted in previous research, and represent- MNI/m3 of excavated deposit) is compara- ing a range of species from marine, terres- tively high in the EIA, dropping sharply into trial, and freshwater habitats. It includes taxa the MIA–LIA (Figure 5b). Several measures used for economic purposes as well as those are used here to examine potential shifts commonly collected for decorative or orna- in species diversity through time. Species mental functions, such as Cypraeidae spp. richness is calculated by the number of (cowrie), though none of the latter were taxa present within the sample population, pierced to suggest that they functioned as here using both the number of invertebrate beads. Analyses are ongoing to assess if other categories, as well as the number of taxa

222 VOLUME 11 • ISSUE 2 • 2016 Small Islands in Eastern African Prehistory

Figure 5. Shell data from trench JS12-03, comparing EIA and MIA–LIA phases, as well as comparative data from the MIA sites of Unguja Ukuu and Fukuchani on Zanzibar: a) abundance of dominant taxa by % minimum number of individuals (MNI) (JS12-03 only); b) density (MNI/m3); c) species richness (count of invertebrate categories and NTAXA); and d) species diversity (Simpson’s 1-D and Simpson’s E).

(NTAXA),wherebythesecategoriesarecom- to that seen in molluskan density, decreasing bined to the highest taxonomic level to avoid slightly from the EIA into the MIA–LIA. In counting overlapping taxa (Figure 5c). Simp- comparison, diversity as measured by Simp- son’s Index (1-D) and Shannon’s measure of son’s 1-D and Shannon’s E slightly increases evenness (E) are common measures of het- from the EIA into the MIA–LIA relative to erogeneity, both of which take into account this decrease in both density and richness. species richness and relative abundance This indicates a comparatively reduced (Magurran 2004) to indicate the distribution focus on harvesting nearshore mollusks, but of taxa across populations (dominance or even with a slight reduction in the number evenness) (Figure 5d). Interestingly given of species exploited there would appear to the early dominance of Nerita spp., assem- be more similar proportions in the range blage richness demonstrates a similar trend of mollusk taxa exploited, particularly with

JOURNAL OF ISLAND & COASTAL ARCHAEOLOGY 223 Alison Crowther et al.

the reduction in abundance of the dominant ered to be likely shark vertebrae based on Nerita spp., from the EIA into the MIA–LIA. the presence of well-defined pairs of dorsal Comparison of the Juani data with pre- and ventral foramina (Kozuch and Fitzger- liminary analyses of Sealinks assemblages ald 1989). Although more specific identifica- from two MIA sites on Zanzibar—Fukuchani tions have not been possible at this stage, the (a small wattle-and-daub trading village, possibility that there may be shark remains ca. AD 600–900) and Unguja Ukuu (a during an EIA occupation is particularly im- large Indian Ocean trading port, ca. AD portant because other evidence of shark ex- 650–1000)—also highlight similarities in the ploitation occurs in late first millennium CE lower density of molluskan material dis- sites (e.g., Badenhorst et al. 2011; Quintana carded during the MIA compared to the Juani Morales and Prendergast in press; but see EIA (Figure 5b). While Fukuchani and Unguja Shipton et al. in press) with a marked in- Ukuudemonstratelocation-specifictrendsin creaseinsharkexploitationandoffshorefish- assemblage richness, diversity, and evenness ing in the Swahili region in the second mil- (likely linked to regional resource availability lennium CE (e.g., Horton 1996b:380; Horton and abundance or local preference), the den- and Mudida 1993:679)—a factor advanced sity pattern across all three MIA sites suggests by Fleisher et al. (2015) as evidence of the that the reduced focus on shellfish gathering late development of a fully maritime way of may have been a widespread economic shift life on the eastern African coast. Previous across the broader coastal region in this later discussions (Anderson et al. 2013) about the period. interpretation of offshore fishing strategies from fish remains stress the importance of Fish Remains precisely identifying the species involved, since some families, such as the requiem Alongside the foraging of shellfish, the sharks (Carcharhinidae), include nearshore and offshore species. Further analysis may Juani Primary School site produced evidence for fishing, with fish dominating the EIA ver- clarify whether the Chondrichthyes speci- tebrate assemblage from all four trenches. mens from Juani belong to species found Previous excavations at the site (Chami closer to shore or in the open sea, signaling 2004) report small quantities of fish, includ- different fishing strategies. There was no ev- ing 12 “shark fish”, without further details. idence of fishing equipment in the cultural Analysis of the fish remains recovered from assemblages from either period to add fur- the Sealinks excavations produced an assem- ther clarification. blage of 292 specimens attributed to family In the MIA–LIA layers at Juani, there is a or genus. Only 40% of the total weight of ex- marked decrease in the abundance of identi- cavated fish bone was identified due to poor fied fish remains (43 NISP) compared to the preservation. Surface modifications on the EIAdeposits(249NISP)thatmayindicateless fish bones were minimal—six elements iden- intensive fish exploitation in the later period, tified to at least family had signs of burning. withsimilartaxaexploitedoverall.However, The majority of identified taxa live near or this pattern is not mirrored at other MIA sites around coral and bay habitats—Lethrinidae with comparable datasets of systematically (emperors), Serranidae (groupers), Scari- collected fish remains. Large quantities of dae (parrotfish), and Balistidae (triggerfish) fish remains from the Sealinks excavations (Figure 6a)—indicating a preference for at Fukuchani and Unguja Ukuu demonstrate nearshore fishing. that fishing was a key economic strategy at A number of vertebrae of cartilaginous these MIA settlements (Figure 6b). fish(Chondrichthyes,e.g.,sharks,rays)were also found in both EIA and MIA–LIA deposits Tetrapod Fauna (35 and 5 NISP [number of identified specimens], respectively). Approximately half of In contrast to the fish and shellfish pat- the cartilaginous fish vertebrae in each phase terns, evidence of domesticated crops (dis- (18 and 3 NISP, respectively) are consid- cussed below) and animals, long considered

224 VOLUME 11 • ISSUE 2 • 2016 Small Islands in Eastern African Prehistory

Figure 6. a) Fish remains (NISP) at the Juani Primary School site (JS), comparing EIA and MIA–LIA phases; b) density of fish remains (NISP/m3) at JS compared to the MIA sites of Fukuchani (FK) and Unguja Ukuu (UU) on Zanzibar. key components of the Iron Age “Bantu” age. This resulted in a low identification rate: package, was near absent from the EIA layers only 94 specimens were identifiable to taxon at the Juani Primary School site (Figure 7a). or group across all phases, leaving 73% of the Wenote,though,thattetrapodvertebrate tetrapod assemblage (by weight) unidenti- remains were generally not well preserved fiable. This mirrors the findings of Chami’s in the Juani assemblages, due to high de- (2004) excavation, from which not a sin- grees of both diagenetic and recent break- gle tetrapod is reported. In our excavation,

JOURNAL OF ISLAND & COASTAL ARCHAEOLOGY 225 Alison Crowther et al.

Figure 7. a) Tetrapod remains (NISP) at Juani Primary School (JS) comparing EIA and MIA–LIA phases; b) wild versus domesticate assemblage compositions for JS compared to Fukuchani and Unguja Ukuu on Zanzibar, excluding birds and specimens only identiﬁed to size class, since these could not be distinguished as wild or domestic; c) frequency of domesticates at Early and Middle Iron Age sites on the Swahili coast, expressed as %NISP. Total NISP, shown after each site name, includes terrestrial and avian tetrapods, but excludes human remains, microfauna, and specimens identiﬁed at such a general level that it was not possible to distinguish between wild and domestic taxa (e.g., “small mammal,” “bird”). For Manda, the faunal assemblage was reported as percentages, without providing total NISP. For Chibuene, only the early (MIA) phase is included here. References for comparative data: Badenhorst et al. 2011; Chami 1994; Chittick 1984; Crowther et al. 2014; Helm 2000a, 2000b; Horton 1996b, in press; Juma 2004. Note that Unguja Ukuu is represented twice: Once from Horton’s (in press) and once from Juma’s (2004) excavations; for the latter, only fauna from Period I are included.

226 VOLUME 11 • ISSUE 2 • 2016 Small Islands in Eastern African Prehistory

both the EIA and MIA–LIA tetrapod assem- phase. Again, the poor surface preservation blages were dominated by wild taxa, particu- and apparent absence of marks means it is larly blue duiker (Cephalophus monticola), difficult to determine if these were related to while green sea turtle (Chelonia mydas) was human subsistence. However, Nile monitor also present in the EIA. Duiker remains ap- and hyrax have been eaten on Pemba, Zanz- peared to belong mainly to subadult individ- ibar, and/or Tumbatu Islands in recent times, uals, whose fragile bones were often highly and monkeys are frequently trapped as ver- fragmented; broken cancellous bone that min on Mafia (Walsh 2007). None of these likely derived from sea turtle also dominated are consumed as foods on Mafia today. the large amount of unidentified bone. A single duiker-sized limb bone was cutmarked, and three specimens also likely belonging Charred Macrobotanical Remains to duiker were burnt, but otherwise there is little that can be gleaned about butchery Despite intensive archaeobotanical sam- or consumption practices, due to the matrix pling (33 samples comprising 1732 liters of that obscured surface modifications on most sediment were processed, including 1030 specimens. liters from EIA layers), only a small quan- Domesticates were absent except for a tity of identifiable crop remains was recov- single, highly fragmented tentative caprine ered at the Juani site (Table 2). Despite the tooth and a caprine-sized mammal bone overall low recovery rates, however, signifi- identified in the EIA layers; given the ab- cant differences were detected between the sence of caprine-sized wild bovids on Mafia, EIA and MIA–LIA phases, with all crop re- these are probable domesticates. However, mains appearing to derive from the latter. Al- considering their fragmentation, caution is though a small number of seeds were recov- stressed. No domestic fauna were identi- ered from the upper EIA layers, direct AMS fied in the MIA–LIA contexts, though sam- dates on a sample of these demonstrate that ple sizes are quite small. This pattern con- they have been stratigraphically displaced trasts with that seen at the contemporane- from the overlying MIA deposits (Tables 1 ous MIA trading port of Unguja Ukuu on and 2). Zanzibar, where domestic livestock and wild The absence of crops in the EIA layers at bovids are equally prevalent in the assem- Juani is at odds with direct evidence for the blage; however, it is similar to the pattern at spread of the full suite of African cereals and the smaller MIA site of Fukuchani, also on pulses—sorghum (Sorghum bicolor), pearl Zanzibar, where livestock are few and suni millet (Pennisetum glaucum), finger millet and duiker are abundant (Figure 7b). Unguja (Eleusine coracana), and cowpea (Vigna Ukuu, however, is much larger and more cos- unguiculata)—to at least as far as Rwanda in mopolitan than either Juani or Fukuchani, so the eastern African interior during this same it is unsurprising that it has more domesti- period (Giblin and Fuller 2011). Sorghum cates. Similarly, urban MIA sites of the Lamu andpearlmillethavealsobeenreportedfrom Archipelago, such as Shanga and Manda, EIA sites in Mikindani on the southern Tan- have mainly domestic fauna at this time zanian coast (Pawlowicz 2011). This com- (Chittick 1984; Horton 1996b) (Figure 7c). bined evidence demonstrates that migrating In contrast, hunting likely remained key to early farming communities most likely car- subsistence in some smaller village sites in ried crops with them to eastern Africa, in- the MIA, and this was likely the case at the cluding to the coastal region. The absence Juani Primary School site. of any EIA crop evidence from Juani there- Additional potential food taxa are repre- fore strongly suggests a major reconfigura- sented at the Juani Primary School by only a tion of this subsistence strategy, either upon few specimens: these include Nile monitor reaching the offshore islands or before, with (Varanus niloticus)andtreehyrax(Dendro- the abandonment of crops (and possibly also hyrax validus) in the EIA phase, and a small livestock) in preference for a maritime forag- monkey (Cercopithecus sp.) in the MIA–LIA ing strategy.

JOURNAL OF ISLAND & COASTAL ARCHAEOLOGY 227 Alison Crowther et al.

Table 2. Archaeological seeds remains identiﬁed at the Juani primary school site in each occupation phase.

Identiﬁed plant taxa MIA–LIA EIA

Liters of sediment processed and analyzed 180a 1030 Crops Sorghum bicolor (sorghum), seed 1 Vigna cf. unguiculata (cf. cowpea), cotyledon 1 2b Vigna sp. (probably cowpea), cotyledon 7b Adansonia digitata (baobab), seed/seed fragment 3 cf. Adansonia digitata (baobab), testa fragment 4 cf. Adansonia sp. (baobab)/Gossypium sp. (cotton), testa fragment 4 Weeds/otherc Brachiaria sp., seed pd Ficus sp., seed p Dicot, cf. Acacia sp., seed p Aizoaceae, seed p Euphorbiaceae, seed p Poaceae-type, tuber p Portulacaceae type, seed p Rhamnaceae, seed p Unidentiﬁed fruit stones p

aOnly 180 liters of the 802 liters processed from the MIA–LIA layers have been analyzed to date. bDirect radiocarbon dates on three seeds recovered from EIA contexts indicate they are intrusive from overlying MIA layers. Given that all such seeds were recovered from within 10–15 cm under the MIA layers, we infer that all crop remains are likely to be from this later phase. cPreliminary identifications and quantifications of weed/other taxa; many seeds remain unidentified owing to limited local reference material. dp = present.

During the MIA, on the other hand, we evidence of culinary features commonly as- see evidence at the Juani site of crops such as sociated with cereal processing in the form cowpea, sorghum, and baobab (Adansonia of three ceramic-lined earth ovens, referred digitata), all of which are of African origin, to locally as mofa. These are widespread but were originally domesticated in regions across coastal eastern Africa in the MIA–LIA outside of eastern Africa (Boivin et al. 2013; (Chittick 1974; Fleisher and LaViolette 2013; Giblin and Fuller 2011) and were thus intro- Horton 1996b) and are thought to have been duced to the island. Currently, the earliest used for baking millet bread. Dates on char- evidence for all these domesticated crops in coal from the ﬁll of these ovens place their coastal eastern Africa is in MIA contexts as- use in the eleventh–twelfth century cal AD sociated with Tana Tradition pottery, where (Table 1: Wk-40969, Wk-40970). they have been identiﬁed at various hinter- Our preliminary analyses of the charred land, coastal, and island sites across the re- wood assemblages (targeted at identifying gion (Crowther et al. in press; Helm et al. samples for radiocarbon dating, see Table 1) 2012; Walshaw 2010), including at the site indicate the presence of mangrove (Rhi- of Ukunju Cave on Juani, located 1.5 km zophoraceae) taxa in both the Early and from the school (Crowther et al. 2014). At Middle Iron Age layers, providing further the Juani Primary School site, we also see evidence of the incorporation of coastal

228 VOLUME 11 • ISSUE 2 • 2016 Small Islands in Eastern African Prehistory

resources into the economy. Presumably and non-existent for food crops, suggesting these taxa were used for fuel/firewood, these were not key components of EIA sub- though we might also speculate upon other sistence at the site. In the following MIA likely uses including for construction, tools, period, as populations across the broader andfurniture,whicharecommonuses coastal and offshore island region became of mangrove by Swahili communities to- larger and more established, they also be- day. Further anthracological analysis (so far came more reliant on food production (agri- largely under-developed in eastern Africa; cultureanddomesticatedanimals).Although see Schmidt 1997) would be useful to ex- hunting and trapping continued, livestock— plore the economic uses of mangrove and together with marine resources—came to other woody taxa at the site. dominate many of the larger MIA–LIA sites in the Zanzibar and Lamu Archipelagos, as shown by a recent review of Swahili zooar- ISLAND COLONIZATION, SUBSISTENCE chaeology (Quintana Morales and Prender- STRATEGY-SWITCHING, AND A gastinpress)(seealsoFigure7c).Bycontrast, NON-LINEAR EMERGENCE OF at Juani Primary School, we see the rapid MARITIMITY transformation in the EIA of inland Bantu- speaking agriculturalists into coastal fisher- In many respects, the overall patterns appar- foragers with an effective subsistence econ- ent in the Juani archaeological assemblage omy based on shellfish gathering and marine appear more consistent with an adaptive fishing as opposed to agriculture and animal strategyfacilitatingoccupationofcoastaland husbandry. island environments, rather than a long-term As noted above, direct comparison of linear trajectory of increasing maritimity. In the dietary contributions of the various plant eastern Africa, it has often been assumed that and animal resources discussed here are dif- marine resources were secondary in impor- ficult in the absence of isotopic analyses, tance to terrestrial foods in coastal diets until however the data presented here highlight at least the seventh century AD (i.e., MIA), the shifting abundance, density and, impor- with earlier coastal communities practic- tantly, species diversity within and between ing only opportunistic harvesting of marine the plant, fish, tetrapod, and invertebrate products (Breen and Lane 2003:475) rather assemblages through time. In combination than extensive or even systematic exploita- with evidence from Sealinks excavations at tion (as in the Beaton [1995] and Marean Unguja Ukuu and Fukuchani, these data pro- [2014] definitions highlighted above). The vide an indication of relative dietary con- archaeological data from Juani is nonetheless tributions from the EIA into the MIA–LIA. indicativeofafairlyrapidlocaladaptationtoa Future research involving a comparison of nearshore marine environment involving an the Juani faunal assemblages through meat initial systematic focus on wild, readily avail- weight estimation would provide another able resources (especially fishing and shell- line of evidence regarding the contribution fish gathering) in place of agriculture and do- of the marine and terrestrial resources to the mesticates, and reflecting a specific coastal diet. adaptation during this phase of island colo- While evidence remains limited, we sug- nization and occupation (similar to that pro- gest that this form of opportunistic subsis- posed for many of the Remote Pacific islands tence adaptation enabled both colonization and island southeast Asia (e.g., Anderson and of offshore islands and perhaps the expan- O’Connor 2008; Bulbeck 2008). sion of populations down the coast, with The Juani EIA midden assemblages are significant food production (crops, domesti- clearly dominated by marine foods, partic- cates) being re-introduced once population ularly mollusks but also fish. Wild terres- sizes increased and connections to the main- trial fauna appear to have been a secondary land were more established. As noted above, food source. Remains of domesticates are the apparent reduction in marine resource rare and unconfirmed in the case of animals, use at the Juani site from the EIA into the

JOURNAL OF ISLAND & COASTAL ARCHAEOLOGY 229 Alison Crowther et al.

MIA also highlights that there was not a long- clear in the Juani EIA faunal assemblage. The term, linear trajectory towards a maritime Periplus further describes that the coastal adaptation or increasing “maritimity.” Occu- forager-fisher communities of the region also pation of offshore islands and use of coastal had baskets for catching fish, instead of resources occurred as a result of fluid and dy- nets (Casson 1989:60, PME 15), which “they namic behaviors, with the EIA perhaps rep- fasten[ed] across the channel opening be- resenting a short-term adaptation during col- tween the breakers” (Schoff 1912:28). Pos- onization, which appears to have been fol- sibly similar traps, called mgono,werestill lowed by a decreasing emphasis on marine being employed on Pemba in the 1920s (In- resources during the MIA, and as Fleisher grams1931:64;Schoff1912:95plate;seealso et al. (2015) argue, a subsequent intensifica- Quintana Morales and Horton 2014). tion in maritime adaptations after AD 1000. Despite the obvious parallels between Flexibility in the adoption of coastal and the small-scale, maritime-adapted societies maritime adaptations (i.e., strategy switch- described in the Periplus and those revealed ing) would also have facilitated broader pro- in the Juani excavations, our high-resolution cesses of coastal dispersal, in particular with recovery methods at the site failed to pro- maritime subsistence adaptation potentially vide any clear archaeological evidence for enabling the rapid movement of EIA popula- the thriving trade that the Periplus also ref- tions down the eastern African coast and into erences. The vast majority of foreign trade southernAfrica,wheremanycoastalEIAsites goods recovered from the Juani site, includ- similarly feature large shell middens (e.g., ing most of the glass beads and all glass Morais 1988; Sinclair et al. 1993). pieces and imported pottery, was found in the MIA–LIA layers and thus dates to the main Evidence of EIA Maritime Trade period of intensive and well-documented In- dian Ocean trade. The only exception to this trend in our excavations was a very small The Periplus of the Erythraean Sea, = cited above, is notable for its discussions number of glass beads (n 2) that were re- of Classical era trade on the eastern African covered from the uppermost EIA layer in coast, but also for its mention of the maritime trenchJS12-03.Given thatthis layerof trench societies that inhabited the region. While 3 was stratigraphically superimposed by MIA it describes communities some 300 years deposits that also contained glass beads = earlier than those we have found at Juani, (n 8), we are cautious about their EIA as- the similarities with the kind of lifestyle ev- sociation. Like the crop remains also found idenced in the EIA archaeology at the Juani in these upper EIA layers, but radiocarbon Primary School are striking. The coastal east- dated to the MIA, it is likely that these small ern African societies briefly described in the beads moved down from the overlying sandy Periplusareobservedtopossessbothdugout deposits. Chemical testing of the glass beads canoes and sewn boats that are “used for fish- is underway to trace their types and prove- ing and catching turtles” (Casson 1989:59– niences, which may shed further light on this 60, PME 15). While direct evidence for the chronological issue. use of boats is lacking from the Juani site, Notably, all other occurrences of for- such evidence is unlikely to preserve. In- eign trade goods (glass beads, glass pieces, stead, we can draw on the findings of pos- and imported ceramics, usually in extremely sible offshore resources like shark in the fau- small quantities) at coastal EIA sites in east- nal assemblage from the site to suggest that ern Africa also derive from deposits that im- boats were used in the EIA. Indeed, the ini- mediately underlie Middle or Late Iron Age tial colonization of Juani Island, probably not occupations (e.g., Chami 1999b, 2003). Re- long before the Juani Primary School EIA site viewing this broader evidence, it is difficult was established, would have demanded at toreconcilethearchaeologicalrecord—thus least the kinds of simple boat technology de- far silent on EIA inter-regional trade—with scribedinthePeriplus.Meanwhile,evidence the Periplus’ descriptions of long-distance for both fishing and capturing sea turtles is trade. While this could be the result of a num-

230 VOLUME 11 • ISSUE 2 • 2016 Small Islands in Eastern African Prehistory

ber of factors, including preservation and ar- It is noteworthy, however, that a single shell chaeological visibility, we also note that our disk bead was found with Kwale pottery on radiocarbon dates place the EIA occupation Kwale Island, located ca. 35 km northwest on Juani in the late fourth century AD at the of Juani, but owing to the lack of evidence earliest,somethreecenturiesaftertheevents elsewhere for EIA shell bead manufacture, described in the Periplus. The identity of was considered by the excavators to be in- these early traders thus remains a mystery, trusive (Chami and Msemwa 1997b:55). All though it has been suggested that they might other known examples of mollusk shell disk not have been iron-working communities as beads in the region date from the MIA period, widely thought, but perhaps mobile foragers when the production of this bead type was and/or pastoralists that occupied the coast a major household industry at pre-Swahili before the arrival of Iron Age groups (see sites, particularly on the offshore islands also Boivin et al. 2013). (e.g.,Flexneretal.2008;Horton1996b;Juma While evidence for long-distance trade 2004). The Juani assemblage thus provides was lacking, the EIA deposits at the Juani site the first clear evidence that this practice ex- did, however, feature a large assemblage of tends back chronologically to the EIA, and mollusk shell beads (Figure 8), including 60 mayhintattheformationofa‘maritime’iden- finished disks, two drilled disk roughouts, tity among these colonizing populations and and nine undrilled disk blanks, pointing to the attendant socio-cultural transformations their local manufacture. In addition, a further associated with their maritime adaptations. 19 finished disks, two drilled disk roughouts, Second, the presence of these shell 33 possible disk blanks, and a single cylinder beadsraisesquestionsastotheirroleinlo- were found in the MIA–LIA layers. Owing to cal trade and society. Shell bead production varying degrees of corrosion, most identify- during the MIA is widely viewed as a mecha- ing shell landmarks have been eroded or ob- nism of local rather than long-distance Indian scured, precluding the positive taxonomic Ocean trade—the beads are thought to have identification of the majority of the assem- been produced by coastal pre-Swahili “mid- blage.However,severalexampleswereiden- dlemen” for trade with communities in the tified as marine bivalves, probably Anadara eastern African hinterland and interior, in ex- sp.Thecylinderbeadappearstobemadeona change for the raw materials that were cov- large cowrie shell. Just over half of the assem- eted by Arab and Indian traders (e.g., Fleisher blage (n = 71, 56.3%) was recovered from and LaViolette 2013; Horton 1996b). Some EIA layers, including in the lowermost con- have argued that this local trade may have texts. Accepting that a small amount of mix- tapped into networks that formed much ear- ing may have occurred between the upper lier, such as during the EIA, which in turn EIA and lower MIA layers, it is still likely that mayhaveexploitedtheestablishednetworks a significant proportion of the assemblage of indigenous foragers and pastoralists who is from the EIA occupation phase. Interest- occupied the region for thousands of years ingly, none of the bead grinders commonly prior (e.g., Boivin et al. 2013; Shipton et al. found on Proto-Swahili sites that are thought 2013; Wright 2005). We propose that the to be used for shell disk bead manufacture Juani beads may have also been manufac- (e.g., Flexner et al. 2008; Horton 1996b), tured for local trade and exchange, though were found at Juani, suggesting that a differ- we suggest that a key motivation was to ent manufacturing process might have been form and maintain social networks (possi- used. bly similar to social gifting and alliance main- The finding of these shell beads along- tenance seen in the Torres Strait (e.g., Mc- side evidence of their local manufacture in Niven 1998, 2015) rather than solely to ac- the EIA is significant for several reasons. First, quire novel trade goods (though this too they are the only known mollusk shell disk might have been a purpose). Archaeological beads securely associated with EIA Kwale oc- research in other regions, such as the Pacific, cupation, making them also among the earli- has shown that such networks were cru- est examples of this bead type on the coast. cial to ensuring the successful exploration

JOURNAL OF ISLAND & COASTAL ARCHAEOLOGY 231 Alison Crowther et al.

Figure 8. Mollusk shell disk beads from Early Iron Age layers, showing different stages of manufacture: a) undrilled blanks; b) drilled roughout; and c) ﬁnished beads. (Contexts: a: 304G; b–c: 406F). and colonization of new island landscapes CONCLUSION (e.g., Kirch 1988), and they may have played a similar role for migrating eastern African The near absence of systematic zooarchaeo- Iron Age groups. Until more systematic re- logical research on pre-Swahili assemblages, search is conducted at other EIA sites on the in particular from smaller offshore islands coast and islands to establish if shell beads where ready access to rich littoral ecosys- were traded in this way, our proposal re- tems may have encouraged or facilitated a mains hypothetical, but we note that it is more marine-focused lifestyle (see Keegan consistent with evidence of local interaction et al. 2008) has constrained models for the among mainland EIA communities who ap- development of maritime adaptations in this pear to have traded iron, copal (resin), and region. While the analysis of the Juani as- othergoods(Chami1992;seealsodiscussion semblage offers only the ﬁrst step in address- in Crowther et al. 2015). It is also notewor- ing this lacuna, the results nonetheless imply thy that pierced cowrie shells have also been that maritime-focused societies were poten- found at EIA sites in Rwanda, demonstrating tially established as early as the EIA on east- evidence of coast–interior links at this time ern Africa’s small islands, but also that an ini- (Giblin et al. 2010). tial focus on marine resources subsequently

232 VOLUME 11 • ISSUE 2 • 2016 Small Islands in Eastern African Prehistory

shifted in favor of a more mixed agricultural- chaeological science and systematically col- hunting-foraging-fishing lifestyle in the MIA. lected data on early subsistence to challenge The emergence of the maritime-oriented so- and transform our understanding of the im- cieties of the later Swahili period was accord- portance of small offshore islands in eastern ingly the result of an adaptive process that African prehistory. was neither linear nor continuous. Instead, early coast-dwelling peoples adopted components of a maritime-oriented lifestyle in a ACKNOWLEDGEMENTS fluidanddynamicwaythatinvolvedanactive response to new environmental opportuni- We thank Scott M. Fitzpatrick and Aaron ties and challenges. Maritime subsistence, Poteate for their invitation to present in colonization, trade, and expansion were all the 2015 SAA session, “Life in the Diminu- possibilities that were suitable at particular tive Realm: Human Adaptations to Smaller times and places. The flexibility in adaptive Island Environments,” where an earlier behaviorsthatweobservemayhaveenabled, version of this article was presented. We and encouraged, an expansion of coastally also thank Jon Erlandson and two anony- settled EIA populations thousands of kilome- mous reviewers for their comments, which ters south. This hypothesis, however, will helped improve the article. We are particu- remain speculative until further research and larlygratefulfortheassistanceofCatherine basic data collection is carried out. Lyakurwa (Division of Antiquities, Tanza- This study also re-asserts the need for nia), Cosmas, Mariam, and the people of greater consideration of small islands in re- Juani who helped with the excavations and gional models for Iron Age settlement, trade, welcomed us to their island. and migration. Our data highlight the possibility that Early Iron Age settlers may have preferentiallytargetedsmalloffshoreislands, FUNDING such as those in the Mafia archipelago, owing to their ecological diversity, including The Sealinks Project is funded through a access to rich marine resources suitable for grant to Nicole Boivin from the European Re- supporting subsistence, trade, and other so- search Council (Starter Grant 206148) under cial activities. Future research should also the “Ideas” specific Programme of the 7th seek to understand the long-term environ- Framework Programme (FP7). Additional mental impacts of these activities on the re- funding includes: postdoctoral fellowships gion’s small islands, which although perhaps from the British Academy (2010–2013) and ecologicallyandgeographicallyattractivefor the University of Queensland (2015–2017) early settlement and trade, are typically also to Alison Crowther; a Fondation Fyssen Post- more susceptible to over-exploitation than doctoral Grant to Erendira´ Quintana Morales; islands with larger landmasses (e.g., Masse and a postdoctoral fellowship from the Gov- et al. 2006; Weisler 2003). The compara- ernment of the Balearic Islands (Conselleria tively higher-intensity exploitation of marine d’Educacio)´ and the European Social Fund resources in the EIA and potential resulting to Llorenç Picornell-Gelabert. Fieldwork was resource depression might well have played carried out under COSTECH Research Per- a role in the EIA-MIA economic transition de- mit no. 2012-303-ER-2011-85 and the Divi- scribed here, though this hypothesis cannot sion of Antiquities (Ministry of Natural Re- be addressed without additional research. sources and Tourism, Tanzania) excavation Such research would clarify the potential for license no. EA.402/605/01 issued to Alison human impacts on nearshore environments Crowther, and with the support and permis- during the early phase of island coloniza- sion of the Regional Administrative Secretary tion and occupation (e.g., Butler 2001; Gio- (Dar es Salaam and Coast/Pwani), the District vas et al. 2010; Giovas et al. 2016; Morrison Administrative Secretary (Mafia), and the Vil- and Hunt 2007). The findings presented here lage Executive Officer (Mtendaji wa Kijiji, demonstrate the exciting potential for ar- Juani).

JOURNAL OF ISLAND & COASTAL ARCHAEOLOGY 233 Alison Crowther et al.

END NOTE Carpenter, K. E. and V. H. Niem (eds.). 1998. The Living Marine Resources of the Western 1. While the use of labels such as “Early-,” Central Pacific: Volume 1. Seaweeds, Corals, Bivalves and Gastropods. Rome: FAO Species “Middle-,” and “Later Iron Age” on the east- Identification Guide for Fishery Purposes, Food ern African coast is seen by some as problem- and Agriculture Organization of the United Na- aticowingtotheirimpliedculturalassociations tions. (e.g., Chami 1994), we employ them here for Casson, L. 1989. Periplus Maris Erythraei: chronological and terminological simplicity. Text with Introduction, Translation, and Commentary. Princeton: Princeton University Press. REFERENCES Chami, F. A. 1992. Limbo: Early Iron Working in south-eastern Tanzania Azania: Archaeological Abbott, R. T. and S. P. Dance. 1998. Compendium Research in Africa 27:45–52. of Seashells: A Color Guide to More Than 4,200 Chami, F. A. 1994. The Tanzanian Coast in the of the World’s Marine Shells. El Cajon, CA: First Millennium AD. Uppsala: Societas Ar- Odyssey Publishers. chaeologica Upsaliensis, Uppsala University. Anderson, A. and S. O’Connor. 2008. Indo-Pacific Chami, F. A. 1999a. The Early Iron Age on Mafia migration and colonization—Introduction. island and its relationship with the mainland. Asian Perspectives 47:1–11. Azania: Archaeological Research in Africa Anderson, A., S. O’Connor, R. Ono, G. Bailey, and 34:1–10. J.M.Erlandson.2013.Inshoreoroffshore?Boat- Chami, F. A. 1999b. Roman beads from the Rufiji ing and fishing in the Pleistocene. Antiquity delta, Tanzania: First incontrovertible archaeo- 87:879–895. logical link with the Periplus. Current Anthro- Attenbrow, V. 1992. Shell bed or shell midden. pology 40:237–241. Australian Archaeology 34:3–21. Chami, F. A. 2000. Further archaeological re- Badenhorst, S., P. Sinclair, A. Ekblom, and I. Plug. search on Mafia island. Azania: Archaeological 2011. Faunal remains from Chibuene, an Iron Research in Africa 35:208–214. Age coastal trading station in central Mozam- Chami, F. A. 2003. Early ironworking communi- bique. Southern African Humanities 23:1–15. ties on the East African coast: Excavations at Baxter,H.C.1944.Pangani:Thetradecentreofan- Kivinja, Tanzania. In East African Archaeol- cient history. Tanganyika Notes and Records ogy: Foragers, Potters, Smiths and Traders (S. 17:15–25. KusimbaandC.Kusimba,eds.):87–97.Philadel- Beaton, J. M. 1995. The transition on the coastal phia: University of Pennsylvania Press. fringe of Greater Australia. Antiquity 69:798– Chami, F. A. 2004. The archaeology of the Mafia 806. archipelago, Tanzania. In African Archaeology Blench, R. 2012. Two vanished African maritime Network: Reports and Views (F. A. Chami, G. traditions and a parallel from South America. Pwiti, and C. Radimilahy, eds.):73–101. Dar es African Archaeological Review 29:273–292. Salaam: Dar es Salaam University Press. Boivin, N., A. Crowther, R. Helm, and D. Q. Fuller. Chami, F. A. 2009. Zanzibar and the Swahili 2013. East Africa and Madagascar in the In- Coast from c.30,000 Years Ago. Dar es Salaam: dianOceanworld.Journal of World Prehistory E&D Vision. 26:213–281. Chami, F. A. and P. J. Msemwa. 1997a. A new Breen, C. and P. Lane. 2003. Archaeological ap- look at culture and trade on the Azanian coast. proaches to East Africa’s changing seascapes. Current Anthropology 38:673–677. World Archaeology 35:469–489. Chami, F. A. and P. J. Msemwa. 1997b. The exca- Bulbeck, D. 2008. An integrated perspective on vation of Kwale Island, south of Dar es Salaam, the Austronesian diaspora: The switch from Tanzania. Nyame Akuma 48:45–56. cereal agriculture to maritime foraging in the Chittick, N. 1974. Kilwa: An Islamic Trading City colonisation of island Southeast Asia. Aus- on the East African Coast. Nairobi: British In- tralian Archaeology 67:31–52. stitute in Eastern Africa. Burgess, N. D. and G. P. Clarke. 2000. Coastal Chittick, N. 1984. Manda: Excavations at an Is- Forests of Eastern Africa. Gland: IUCN. land Port on the Kenya Coast. Nairobi: British Butler, V. L. 2001. Changing fish use on Mangaia, Institute in Eastern Africa. southern Cook Islands: Resource depression Christie, A. C. 2011. Exploring the Social Con- and the prey choice model. International Jour- text of Maritime Exploitation in the Mafia nal of Osteoarchaeology 11:88–100. Archipelago, Tanzania: An Archaeological

234 VOLUME 11 • ISSUE 2 • 2016 Small Islands in Eastern African Prehistory

Perspective. Ph.D. Dissertation. York: Univer- hold economy: Analysis of an assemblage from sity of York. Tumbe, Pemba Island, Tanzania, 7th–10th cen- Collett, D. P. 1985. The Spread of Early Iron- turies AD. Journal of African Archaeology Producing Communities in Eastern and 6:161–181. Southern Africa. Ph.D. Dissertation. Cam- Giblin, J., A. Clement, and J. Humphries. 2010. bridge: University of Cambridge. An Urewe burial in Rwanda: Exchange, health, Crowther, A., M. C. Horton, A. Kotarba-Morley, M. wealthandviolencec.AD400.Azania: Archae- Prendergast, E. Quintana Morales, M. Wood, C. ological Research in Africa 45:276–297. Shipton, D. Q. Fuller, R. Tibesasa, W. Mills, and Giblin, J. and D. Q. Fuller. 2011. First and sec- N. Boivin. 2014. Iron Age agriculture, fishing ond millennium AD agriculture in Rwanda: and trade in the Mafia Archipelago, Tanzania: Archaeobotanical finds and radiocarbon dates New evidence from Ukunju Cave. Azania: Ar- from seven sites. Vegetation History and Ar- chaeological Research in Africa 49:21–44. chaeobotany 20:253–265. Crowther, A., L. Lucas, R. Helm, M. C. Horton, Giovas, C. M., S. M. Fitzpatrick, M. Clark, and M. C. Shipton, H. T. Wright, S. Walshaw, et al. Abed. 2010. Evidence for size increase in an in press. Ancient crops provide first archae- exploited mollusc: Humped conch (Strombus ological signature of the westward Austrone- gibberulus) at Chelechol ra Orrak, Palau from sian expansion. Proceedings of the National ca. 3000–0 BP. Journal of Archaeological Sci- Academy of the Sciences. ence 37:2788–2798. Crowther, A., M. A. Veall, N. Boivin, M. C. Hor- Giovas, C. M., S. M. Fitzpatrick, O. Kataoka, and M. ton, A. Kotarba-Morley, D. Q. Fuller, T. Fenn, Clark. 2016. Prey body size and anthropogenic O. Haji, and C. Matheson. 2015. Use of Zanz- resource depression: The decline of prehis- ibar copal (Hymenaea verrucosa Gaertn.) as toric fishing at Chelechol ra Orrak, Palau. Jour- incense at Unguja Ukuu, Tanzania in the 7-8th nal of Anthropological Archaeology 41:132– century CE: Chemical insights into trade and 146. Indian Ocean interactions. Journal of Archae- Greenway, P. J., W. A. Rodgers, R. J. Wingfield, ological Science 53:374–390. and L. B. Mwasumbi. 1988. The vegetation of Datoo, B. A. 1970. Rhapta: The location and im- Mafia island, Tanzania. Kirkia 13:197–238.

portance of East Africa’s first port. Azania: Ar- Harris, M., M. Weisler, and P. Faulkner. 2015. A chaeological Research in Africa 5:65–75. refined protocol for calculating MNI in archaeo- De Maret P. 2013. Archaeologies of the Bantu logical molluscan shell assemblages: A Marshall expansion. In The Oxford Handbook of Islands case study. Journal of Archaeological African Archaeology (P. Mitchell and P. Science 57:168–179. Lane, eds.):627–643. Oxford: Oxford Univer- Helm, R. 2000a. Recent archaeological research sity Press. on the iron-working, farming communities of Fahn, A. and E. Werker. 1986. Wood Anatomy coastal Kenya. Azania: Archaeological Re- and Identification of Trees and Shrubs from search in Africa 35:183–189. Israel and Adjacent Regions. Jerusalem: The Helm, R. 2000b. Conflicting Histories: The Ar- Israel Academy of Sciences and Humanities. chaeology of the Iron-Working, Farming Com- Fasolo, U. 1939. Atlante Micrografico dei legni muniites in the Central and Southern Coast dell’Africa Orientale Italiana.Firenze:Erbario Region of Kenya. Ph.D. Dissertation. Bristol: Coloniale. University of Bristol. Fleisher, J., P. Lane, A. LaViolette, M. C. Horton, Helm, R., A. Crowther, C. Shipton, A. Tengeza, E. Pollard, E. Quintana Morales, T. Vernet, A. D. Q. Fuller, and N. Boivin. 2012. Exploring C. Christie, and S. Wynne-Jones. 2015. When agriculture, interaction and trade on the east- did the Swahili become maritime? American ern African littoral: Preliminary results from Anthropologist 117:100–115. Kenya. Azania: Archaeological Research in Fleisher, J. and A. LaViolette. 2013. The early Africa 47:39–63. Swahili trade village of Tumbe, Pemba Island, Henderson, W. G., L. C. Anderson, and C. R. Tanzania, AD 600–950. Antiquity 87:1151– McGimsey. 2002. Distinguishing natural and ar- 1168. chaeological deposits: Stratigraphy, taxonomy, Fleisher, J. and S. Wynne-Jones. 2011. Ceram- and taphonomy of Holocene shell-rich accu- ics and the early Swahili: Deconstructing the mulations from the Louisiana chenier plain. Early Tana Tradition. African Archaeological Palaios 17:192–205. Review 28:245–278. Holden,C.J.2002.Bantulanguagetreesreflectthe Flexner, J. L., J. Fleisher, and A. LaViolette. spread of farming across sub-Saharan Africa: a 2008. Bead grinders and early Swahili house- maximum-parsimony analysis. Proceedings of

JOURNAL OF ISLAND & COASTAL ARCHAEOLOGY 235 Alison Crowther et al.

the Royal Society Biological Sciences Series B Masse, W. B., J. Liston, J. Carucci, and J. S. Athens. 269:793–799. 2006. Evaluating the effects of climate change Horton, M. C. 1996a. Early maritime trade on environment, resources depletion, and cul- and settlement along the coasts of eastern ture in the Palau Islands between AD 1200 Africa. In The Indian Ocean in Antiquity and 1600. Quaternary International 151:133– (J. Reade, ed.):439–459. London: Kegan Paul 143. International. McNiven,I.J.1998.Enmityandamity:Reconsider- Horton,M.C.1996b.Shanga: The Archaeology of ing stone-headed club (gabagaba) procurement a Muslim Trading Community on the Coast of and trade in Torres Strait. Oceania 69:94–115. East Africa. Nairobi: British Institute in Eastern McNiven, I. J. 2015. Ascendancy of the Tudul- Africa. gal, central Torres Strait: Socio-political ma- Horton, M. C. In press. Zanzibar and Pemba, The nipulation and domination of an archipelago Archaeology of an Indian Ocean Archipelago. polity. Journal of Anthropological Archaeol- London: British Institute in Eastern Africa. ogy 39:164–180. Horton,M.C.andJ.Middleton.2000.The Swahili: Mitchell, P. 2004. Towards a comparative archae- The Social Landscape of a Mercantile Society. ology of Africa’s islands. Journal of African Ar- Cambridge: Blackwell. chaeology 2:229–250. Horton, M. C. and N. Mudida. 1993. Exploitation Morais, J. 1988. The Early Farming Communi- of marine resources: Evidence for the origin ties of Southern Mozambique. Uppsala: Upp- of the Swahili communities of East Africa. In sala University Department of Archaeology and The Archaeology of Africa: Food, Metals and Ancient History. Towns (T. Shaw, P. Sinclair, B. Andah, and A. Morrison, A. E. and T. L. Hunt. 2007. Human im- Okpoko, eds.):673–693. London: Routledge. pacts on the nearshore environment: An ar- Ingrams, W. H. 1931. Zanzibar: Its History and chaeological case study from Kaua’i, Hawaiian its People. London: Witherby. Islands. Paciﬁc Science 61:325–345. Juma,A.2004.Unguja Ukuu on Zanzibar: An Ar- Msemwa, P. J. 1994. An Ethnoarchaeological chaeological Study of Early Urbanism. Upp- Study of Shellﬁsh Collection in a Complex Ur- sala: Uppsala University Department of Archae- ban Setting. Ph.D. Dissertation.NewYork:

ology and Ancient History. Brown University. Keegan, W. F., S. M. Fitzpatrick, K. S. Sealey, M. Neumann, K., W. Schoch, P. Detienne,´ F. H. J. LeFebvre, and P. T. Sinelli. 2008. The role of Schweingruber, and H. Richter. 2007. Woods small islands in marine subsistence strategies: of the Sahara and the Sahel. Bern: Verlag Paul Case studies from the Caribbean. Human Ecol- Haupt. ogy 36(5):635–654. Nkirote M’Mbogori, F. 2015. Population and Ce- Kirch, P. V. 1988. Long-distance exchange and is- ramic Traditions: Revisiting the Tana ware land colonization: The Lapita case. Norwegian of coastal Kenya (7th–14th century AD).Ox- Archaeological Review 21:103–117. ford: Archaeopress. Kozuch, L. and C. Fitzgerald. 1989. A guide to Pawlowicz, M. C. 2011. Finding their Place in identifying shark centra from southeastern ar- the Swahili World: An Archaeological Explo- chaeological sites. Southeastern Archaeology ration of Southern Tanzania. Ph.D. Disserta- 8:146–157. tion. Charlottesville: University of Virginia. LaViolette, A. 2008. Swahili cosmopolitanism in Prendergast, M., H. Rouby, P. Punnwong, R. Africa and the Indian Ocean world, AD 600– Marchant, A. Crowther, N. Kourampas, C. Ship- 1500. Archaeologies 4:24–49. ton,M.Walsh,K.Lambeck,andN.Boivin.2016. Magurran, A. E. 2004. Measuring Biological Di- Continental island formation and the archaeol- versity. Oxford: Blackwell Publishing. ogy of defaunation on Zanzibar, eastern Africa. Manning, K., R. Pelling, T. Higham, J.-L. Schwen- PLoS ONE. doi:10.1371/journal.pone.0149565 ninger, and D. Q. Fuller. 2011. 4500-year old Quintana Morales, E. M. and M. C. Horton. 2014. domesticated pearl millet (Pennisetum glau- Fishing and ﬁsh consumption in the Swahili cum) from the Tilemsi Valley, Mali: New in- communities of East Africa, 700–1400 CE. In- sights into an alternative cereal domestication ternet Archaeology 37. doi:10.11141/ia.37.3 pathway. Journal of Archaeological Science Quintana Morales, E. M. and M. E. Prendergast. 38:312–322. In press. Animals and their uses in the Swahili Marean, C. W. 2014. The origins and signiﬁcance world. In The Swahili World (S. Wynne-Jones of coastal resource use in Africa and western and A. LaViolette, eds.). London: Routledge. Eurasia. Journal of Human Evolution 77:17– Richmond, M. D. 2011. A Field Guide to the 40. Seashores of Eastern Africa and the Western

236 VOLUME 11 • ISSUE 2 • 2016 Small Islands in Eastern African Prehistory

Indian Ocean Islands. Dar es Salaam: SIDA, P. Sinclair, B. Andah, and A. Okpoko, eds.):409– WIOMSA. 431. London: Routledge. Rick, T. C. and J. M. Erlandson. 2008. Human Im- Soper, R. 1967. Kwale: An Early Iron Age site in pacts on Ancient Marine Ecosystems: A global south-eastern Kenya. Azania: Archaeological perspective. Berkeley: University of California Research in Africa 2:1–17. Press. Walsh, M. 2007. Island subsistence: Hunting, trap- Robin, A. 2008. Encyclopedia of Marine Gas- ping and the translocation of wildlife in the tropods. Hackenheim, Germany: ConchBooks. Western Indian Ocean. Azania: Archaeologi- Robin,A.2011.EncyclopediaofMarineBivalves, cal Research in Africa 42:83–113. Including Scaphopods, Polyplacophora and Walshaw, S. 2010. Converting to rice: Urbaniza- Cephalopods. Hackenheim, Germany: Conch- tion, Islamization and crops on Pemba Island, Books. Tanzania, ad 700–1500. World Archaeology Rowson,B.2007.LandmolluscsofZanzibarIsland 42:137–154. (Unguja), Tanzania, including a new species of Walz, J. 2010. Route to a Regional Past: An Gulella (Pulmonata: Streptaxidae). Journal of Archaeology of the Lower Pangani (Ruvu) Conchology 39(4):425–466. Basin, Tanzania, 500–1900 CE. Ph.D. Disser- Rowson,B.,B.H.Warren,andC.F.Ngereza.2010. tation. Gainesville: University of Florida. Terrestrial molluscs of Pemba Island, Zanzibar, Weisler, M. I. 2003. Contraction of the southeast Tanzania, and its status as an “oceanic” island. Polynesian interaction sphere and resource de- ZooKeys 70:1–39. pression on Temoe Atoll. New Zealand Jour- Russell, T., F. Silva, and J. Steele. 2014. Modelling nal of Archaeology 25:57–88. the spread of farming in the Bantu-speaking re- Wood, M. 2011. Interconnections: Glass beads gions of Africa: An archaeology-based phylo- and trade in southern and eastern Africa and geography. PLoS ONE 9:e87854. the Indian Ocean—7th to 16th centuries AD. Schmidt, P. R. 1997. Archaeological views on the Uppsala: African and Comparative Archaeology history of landscape change in East Africa. Jour- Department of Archaeology and Ancient His- nal of African History 38:393–421. tory Uppsala University. Schoff, W. H. 1912. The Periplus of the Ery- WoRMS Editorial Board. 2016. World Reg-

thraean Sea: Travel and Trade in the Indian ister of Marine Species. Available from OceanbyaMerchantoftheFirstCentury. Lon- http://www.marinespecies.org at VLIZ don: Longmans, Green, and Co. Wright, D. K. 2005. New perspectives on early Shipton, C., A. Crowther, M. Prendergast, N. regional interaction networks of East African Kourampas, M. C. Horton, K. Douka, J.-L. trade: A view from Tsavo National Park, Kenya. Schwenninger, et al. In press. Reinvestigation African Archaeological Review 22:111– of Kuumbi Cave, Zanzibar, reveals Later Stone 140. Age coastal habitation, early Holocene aban- Wright, H. T., J. E. Knudstad, L. W. Johnson, donment, and Iron Age reoccupation. Azania: and R. W. Redding. 1992. Early Islam, oceanic Archaeological Research in Africa. trade and town development on Nzwani: The Shipton, C., R. Helm, N. Boivin, A. Crowther, P. Comorian archipelago in the XIth-XVth cen- Austin, and D. Q. Fuller. 2013. Intersections, turies AD. Azania: Archaeological Research in networks and the genesis of social complexity Africa 27:81–128. on the Nyali Coast of East Africa. African Ar- Wright,H.T.,C.Sinopoli,L.Wojnaroski,E.S.Hoff- chaeological Review 30:427–453. man, S. L. Scott, R. W. Redding, and S. M. Good- Sinclair, P., J. Morais, L. Adamowicz, and R. T. man. 1984. Early seafarers of the Comoro Is- Duarte. 1993. A perspective on archaeological lands: The Dembeni phase of the IXth-Xth cen- research in Mozambique. In The Archaeology turies AD. Azania: Archaeological Research in of Africa: Food, Metals and Towns (T. Shaw, Africa 19:13–59.

JOURNAL OF ISLAND & COASTAL ARCHAEOLOGY 237