Keywords: SHELLFISH GATHERING, WATER TURBIDITY, SEA SURFACE TEMPERATURES, ... Introduction. The study of shell middens world-wide has been a.
Journal of Archaeological Science (1997) 24, 1031–1044
Changes in Shellfish Species Composition and Mean Shell Size from a Late-Holocene Record of the West Coast of Southern Africa Antonieta Jerardino Department of Archaeology, University of Cape Town, Rondebosch 7700, South Africa (Received 20 May 1996, revised manuscript accepted 21 January 1997) This paper presents and discusses the results of a systematic analysis of the shellfish assemblage recovered from Pancho’s Kitchen Midden, a site located in the western Cape coast of South Africa. Several hypotheses are considered in an attempt to explain the observed changes in shellfish frequencies and shell size. Changes in species composition in this particular sequence are mostly attributable to changes in collecting practices. Shellfish gathering was generally undertaken within the mid and low-intertidal zones throughout the Late Holocene, with slightly more frequent collections within the upper subtidal in the last 600 years. Changes in water turbidity, on the other hand, seem to explain most of the variability in the mean sizes of black mussels. Nevertheless, intense collection of large numbers of black mussels seems to have been an additional factor in bringing mussel sizes to a minimum around 3000 . Subsequent recovery of mean mussel sizes appears to be the result of several factors: improvement in marine productivity and low water turbidity, greater access to large individuals inhabiting the low intertidal and upper subtidal, and/or relatively low levels of shellfish exploitation by people. Changes in sea surface temperature of the magnitude reported here do not appear related to changes in species composition and shell size. The comparison of this shellfish sequence with others from nearby sites, and analysed in a similar fashion, would be essential for the reconstruction of pre-Colonial shellfish-gathering practices at the western Cape. ? 1997 Academic Press Limited Keywords: SHELLFISH GATHERING, WATER TURBIDITY, SEA SURFACE TEMPERATURES, INTERTIDAL, WESTERN CAPE, SOUTH AFRICA.
Introduction
T
he study of shell middens world-wide has been a key component in the reconstruction of past subsistence economies in coastal and riverine environments. The analysis of shell middens was pioneered in the 1840s by Danish scholars, whose reports stimulated subsequent research in the early 1900s on shell heaps located in North America and the rest of the world (Waselkov, 1987; Trigger, 1992). Whilst the initial research on shell middens mostly involved the study of rates of accumulation, site features, vertebrate remains, burials, and artefacts found at such sites, the systematic analysis of variation in shellfish composition and size through time has only become a major focus of research during the last 2 decades (Stein, 1992). In South Africa, the earliest systematic archaeological excavations of a coastal site (Goodwin, 1938) included the description of the dominant marine molluscan species, an aspect that was instrumental in the characterization of the excavated deposits and stratigraphical sequence. Changing frequencies of shellfish species and differences in shell size were noticed during subsequent excavations of coastal caves and open sites in southern Africa. These later
observations included the quantification of several parameters, such as frequency of species, mean sizes and degree of fragmentation of the shell material (Maggs & Speed, 1967; Klein, 1972; Voigt, 1973, 1975; Parkington, 1976a, b; Schweitzer, 1979). The patterning shown by these variables in the shellfish assemblages of several coastal sites located on the southern and western Cape has been interpreted in a variety of ways. Some of these alternative hypotheses relate to each other, and others are mutually exclusive. A brief summary of these interpretations is presented below. Environmental factors, such as fluctuations in sea surface temperatures and/or re-configuration of the shoreline morphology due to sea level changes, have been one of the most favoured explanations for changing frequencies of shellfish species (Klein, 1972; Voigt, 1975, 1982; Schweitzer, 1979; Buchanan et al., 1984; Buchanan, 1988; Thackeray, 1988; Yates, 1989; Jerardino, 1993). The practice of different shellfish collecting habits through time has also received considerable attention from those attempting to explain changes in frequency of shellfish species and shell size, as both of these variables are differentially represented across the vertical zonation of most southern African shorelines (Maggs & Speed, 1967; Voigt, 1973; Avery, 1974; Buchanan et al., 1984; Thackeray, 1988).
1031 0345–4403/97/111031+14 $25.00/0/as970182
? 1997 Academic Press Limited
1032 A. Jerardino
Variations in the relative proportion of mollusc species in archaeological deposits have also been interpreted in a few instances as changes in dietary preferences, which would have determined specific collecting practices (Voigt, 1973, 1975). Toxic red tides are a recurrent and seasonal phenomenon on the west coast (Grindley & Nel, 1970; Dale & Yentch, 1978; Horstman, 1981), and filter-feeders are much more likely than grazers such as limpets to retain the microorganisms responsible for such events. Aware of these observations and their implications for hunter–gatherer settlement and diet on the coast, Parkington (1976a, 1981) suggested that changes from a limpet- to mussel-dominated assemblage (and vice versa) could largely reflect changes in the season of occupation of coastal sites. Other scholars have also noticed the variability in shellfish species composition of broadly contemporary shell middens situated at varying distances from the shoreline and have suggested that a criterion of maximum returns for the least effort informed human selection of marine mollusc species (Buchanan et al., 1984; Buchanan, 1988; Parkington et al., 1988). Noticing the relatively small mean sizes of the two most common limpet species present in west-coast shell middens, when compared to specimens measured on the present and nearby shorelines, several authors have interpreted this as a result of intense limpet exploitation during the pre-colonial past (Parkington, 1976b; Robertshaw, 1977; Buchanan et al., 1978). Klein (1989: 332) suggested that this difference is likely to be the result of demographical changes, which are discernible in the past by comparing limpet measurements from a Middle Stone Age site with those obtained from a number of Later Stone Age sites. Archaeological excavations in the Elands Bay area over the last 15 years have generated a large and diverse set of observations on shellfish frequencies and shell size, with no similar precedent in southern Africa (Horwitz, 1979; Buchanan, 1988; Yates, 1989). As a result of this work, it was recognized ‘‘that patterns of shellfish gathering have been far from stable during the Holocene’’ and that there was a need to formulate ‘‘hypotheses relating shellfish-gathering to general reconstructions of settlement and settlement change that can be tested against accumulating archaeological observations’’ (Buchanan et al., 1984: 128). Unfortunately, except for Yates’ (1989) work, few attempts have been made to compare systematically and in detail the shellfish sequences from several sites in this area. Moreover, the analyses and discussions regarding changes in shell mean sizes have focused on limpet species (Patella spp) to the exclusion of black mussels (Choromytilus meridionalis) (Parkington, 1976b; Buchanan et al., 1984; Parkington et al., 1992), the latter being dominant in most coastal sites. Furthermore, changes in oceanographical conditions at the west coast, which might have affected the growth and availability of shellfish in the past, have also been
recently described (Cohen et al., 1992; Miller et al., 1995). Clearly, there is a need to integrate these latter observations into the discussion of the local shellfish sequences, a task that has not been performed until very recently. The aim of this paper is to document in detail the variability in the mussel-dominated shellfish assemblage from one coastal site located in the Elands Bay area, and to test several competing hypotheses (of behavioural and environmental relevance) that might explain the observed patterns. In doing so, a first step towards a more systematic analysis and comparison of local shellfish sequences is presented.
Ecological Setting of Elands Bay Shoreline The most important physical feature influencing west-coast environments is the Benguela Current (Figure 1). Surface waters flow adjacent to the coast in a north–north-westerly direction, moving progressively off-shore towards latitude 20)S. On average, coastal sea-surface temperatures in the Benguela Current reach lowest value (8–10)C) during late spring and summer months (October–February) in association with upwelling events. Relatively warmer sea-surface temperatures (11–14)C) are observed during the rest of the year (Shannon, 1985). One of the three upwelling cells reported south of latitude 29)S is located just south of the Elands Bay area (Shannon, 1985) on the northern edge of Cape Columbine (32–33)S). The upwelling coastal waters bring enough nutrients to the surface to support a diverse and highly productive marine environment (Branch & Griffiths, 1988). Today, long stretches of sandy beach dominate the open coastline north and south of Elands Bay. A number of isolated sandstone rocky reefs of various lengths protrude along this shore (Figure 1). Throughout the year, but mostly between late spring and early autumn, several species of marine birds and seal pups are washed ashore on these sandy beaches (Parkington, 1981; Avery, 1990). Most impressive is the presence of large quantities of giant kelp (Ecklonia maxima and Laminaria pallida) and several species of invertebrates in various stages of preservation after heavy winter storms. Whale and dolphin strandings have also been observed on sandy west-coast beaches (Best, 1982). Intertidal subsurface colonies of white mussels (Donax serra) occur in these sandy beaches in great quantities. Two species of gastropods from the genus Bullia also occur frequently (Bally, 1987). Small haul-outs of seals (Arctocephallus pusillus) have been observed on Baboon Point (Figure 1) with an intermittent appearance in the last decade. Permanent and migrant colonies of marine birds can be observed along this stretch of coastline, the most visible being cormorants (Phalacrocorax spp), Cape gannet (Morus capensis), gulls (Larus spp), terns (Sterna spp), and Jackass penguins (Spheniscus demersus) (Avery, 1990). Nevertheless, the greatest
Shellfish in Southern African Coastal Middens 1033
18°20' E
Rocky shore
0
km
2
Sandy beach Southern Africa C ela ngu Be ent urr
EBO
Bay Reefs
Verlorenvlei
Baboon Point EBC Lamberts Bay
SC TC
Atlantic Ocean N
32°20' S Elands Bay
Mussel Point PKM Contour interval = 40 m
Figure 1. Geographical location of Elands Bay, showing sites mentioned in text. PKM, Pancho’s Kitchen Midden; MTM, Mike Taylor’s Midden; TC, Tortoise Cave; EBC, Elands Bay Cave; EBO, Elands Bay Open; SC, Spring Cave.
marine biomass is found on the productive rocky reefs in the form of kelp forest (E. maxima), marine fish, crayfish (Jasus lalandii), large intertidal and subtidal mussel beds (C. meridionalis and Aulacomya alter, ribbed mussels), and extensive, though patchy, intertidal colonies of limpets (Patella spp) and seaweed gardens (Kilburn & Rippey, 1982; Rebelo, 1982; Buchanan, 1988). The shellfish community of the rocky shores between Lamberts Bay and Elands Bay closely follows the reported distributions of species richness and biomass for equivalent shores elsewhere on the west coast (McQuaid & Branch, 1984, 1985; McQuaid, Branch & Crowe, 1985). South of Elands Bay particularly, the flat exposed reef of Mussel Point supports a large biomass of filter-feeders dominated by black mussels. The reefs located inside the Bay, in contrast, are sheltered, with a total estimated shellfish biomass one order of magnitude smaller than that of Mussel Point, and dominated by limpets. The west-facing shore of Baboon Point is steep and exposed to wave action. Its rock surface is extensive but largely bare of shellfish, the accessible biomass being equivalent to 25% of that on the Bay reefs. The molluscs inhabiting this reef are dominated by mussels (C. meridionalis and A. ater) and limpets (Rebelo, 1982; Parkington et al., 1988). The degree to which mollusc species of economic interest were accessible to pre-colonial people seems to have depended to a large extent on their vertical distribution in the rocky shoreline. Two of the most common species found in shell middens of the west coast, P. granatina and P. granularis, have their
distribution in the mid-intertidal, an area of easy access to humans. In contrast, other less-common species in archaeological contexts have different distributions. P. argenvillei usually inhabits semi-vertical exposed rocks of the low-intertidal, and P. cochlear colonies form a narrow belt along the upper subtidal on the seaward side of low-angle reefs exposed to wave action. P. barbara and P. miniata are mostly found in low-tide pools and in the subtidal, and P. compressa is adapted to cling exclusively to the stipes of giant kelp, thus remaining submerged most of the time. Black mussels can be frequently found on exposed shorelines. This species of mussel is represented in the mid-intertidal by mussel beds of small-sized individuals which are easy to gather, whereas large individuals clump together in the low-intertidal and subtidal. Ribbed mussels, however, inhabit the low-intertidal, and are most frequently encountered in the subtidal among kelp beds (Kilburn & Rippey, 1982).
Palaeoenvironmental and Archaeological Background Sea surface temperatures Most of the open coast mollusc species on the west coast perform physiologically more efficiently under conditions of relatively low temperatures (Griffiths, 1981; Kilburn & Rippey, 1982). Their relative abundance on the shores and mean body size are affected by sea surface temperatures (SST) (Griffiths, 1981; Branch & Griffiths, 1988; Kilburn & Rippey, 1982). Thus, SST
1034 A. Jerardino
changes in the past shall be considered when explaining changes in species composition and mean shell size. In the case of black mussels and limpets, relatively warm SST would slow growth rates. The only Late-Holocene sea surface temperature record for the south-western Cape has been obtained from P. granatina shells found in archaeological shell middens located in the study area (Cohen et al., 1992; Cohen, 1993; Jerardino, 1995a). Lowest average sea surface temperatures are reached at c. 4200 , from 3500 to 2300 , and again between 600 and 400 . Relatively warmer temperatures were established at 4330 and c. 750 . Sea level changes At about 6000 , sea level reached a maximum of 2–3 m above present sea level (Flemming, 1977; Miller et al., 1995). Subsequently, there was a rapid and relatively short regression around 4200 , to more-orless the present sea level (Jerardino, 1993). These events were followed by a transgression of an order of 2 m more than today at 4000–3800 (Yates et al., 1986; Jerardino, 1993). Later, sea level fell once again, reaching a minimum close to the present between 3500 and 2800 . Some time around 2600 , sea level either stabilized or recovered, perhaps reaching a small, or short-lived, positive elevation relative to the present at about 1800 . Subsequently, sea level has remained stable at its present datum (Jerardino, 1993). This trajectory of Mid- to Late-Holocene sea level changes is also mirrored in other localities around southern African and Antarctic shorelines (Berkman, 1992; Jerardino, 1995a). A 2-m rise in sea level in Elands Bay around 6000 and again between 4000 and 3800 would have ensured complete inundation of low-lying and exposed rocky reefs, such as Mussel Point and those located within the Bay. In contrast, the area accessible for shellfish gathering on the high and steep rocky shore of Baboon Point would have decreased minimally as a consequence of such sea-level rise. These changes in sea level, and related accessibility to rocky reefs, would have also limited the range of mean sizes of shellfish available for collection. Recent research on the west coast of South Africa (G. M. Branch, pers. comm.) showed evidence of significantly larger mean sizes of limpets in sheltered areas than those found in more exposed circumstances. It was proposed that limpets on wave-beaten shorelines had limited time for feeding and therefore less energy for growth. On the other hand, significantly larger mean sizes of black mussels are found on exposed shores than on sheltered reefs (Griffiths, 1981; Branch & Griffiths, 1988). Similar trends in sizes of limpets and black mussels have also been noted in the Elands Bay area (Rebelo, 1982). Thus, it is suggested that any change in the locality of shellfish collection (from an exposed reef to a sheltered one and vice versa) by
groups of people settling at a particular site would affect the species composition and mean sizes of some species in the archaeological record. As shown elsewhere (Jerardino, 1993), this can occur as a consequence of minor fluctuations in sea level. Water turbidity Growth rates of the filter-feeding black mussel can be slowed by high levels of inorganic sediments present in the immediate environment. In open-coast natural environments, energy expenditure involved in the separation of nutritional from non-nutritional particles is greatest when high levels of water turbidity are reached. Such energy expenditure is ultimately to the detriment of energy spent on growth and can result in small-sized animals, if other variables remain constant (Griffiths, 1980). Coastal sedimentation is dependent on several factors, of which sea-level change is one of the most important. Generally, net on-shore aeolian sediment movement and dune formation take place during the falling of sea level, particularly in arid and semi-arid areas. On the other hand, off-shore sediment transport and erosion of the beach profile are characteristic of transgressive events (Chappell, 1982; Orford, 1987). Recent geoarchaeological studies at the Elands Bay area (Miller et al., 1993) show that substantial aeolian sand transport and dune formation took place between 4000 and 2400 , a period characterized by general falling sea levels. Consequently, black mussel sizes are expected to be smaller overall during periods of falling sea levels. Changes in the intensity of shellfish exploitation in the Elands Bay area during pre-colonial times A large number of caves, shelters and open sites have been recorded for the Elands Bay and adjacent coastal areas (Parkington et al., 1988; SARU data base, Department of Archaeology, University of Cape Town). The vast majority of these sites consist of deposits dominated by marine shell remains. The rate at which these deposits were accumulated in the past has clearly changed since the present shoreline was more or less defined at about 8000 (Jerardino, 1996; Jerardino & Yates, 1996). First signs of relatively fast accumulation of shell midden deposits and intensive shellfish consumption date to around 3500 (Jerardino, 1995b; Jerardino & Yates, 1996). Subsequently, between 3000 and 2000 , shellfish exploitation was greater than at any other time during the Holocene, and perhaps before that period. During this millennium, enormous shell middens (named ‘‘megamiddens’’) containing tons of black mussel shells and few faunal and cultural remains accumulated immediately behind rocky platforms (Buchanan, 1988; Parkington et al., 1988). The overall hunter–gatherer dietary mix, as reconstructed from
Shellfish in Southern African Coastal Middens 1035 Pancho's Kitchen Midden
North-west section
N7
M7
L7
0 2
1 3 4
5
6 1m Unexcavated
Gravel Calcrete
1. 570 ± 20 BP 2. 880 ± 50 BP 3. 2640 ± 60 BP 0
1 Shelly Sandy
4. 2940 ± 20 5. 3060 ± 60 6. 3570 ± 60 2
Rock
BP BP BP
3m Burrow
Root cast Figure 2. Stratigraphical sequence at Pancho’s Kitchen Midden.
isotopic measurements on skeletons buried along stretches of the west coast and from archaeological food waste, was more marine between 3000 and 2000 than either before or after (Sealy & Van der Merwe, 1988; Lee-Thorp, Sealy & Van der Merwe, 1989; Jerardino, 1996; Jerardino & Yates, 1996). Megamiddens ceased to accumulate about 2000 , apparently coinciding with the arrival of pastoralism to the south-western Cape (Parkington et al., 1988; Sealy & Yates, 1994). The scale of shellfish exploitation was dramatically reduced during the last 2000 years and pre-colonial diet was predominantly of terrestrial origin. The changes in shellfish exploitation described above seem likely to have been the outcome of a succession of different subsistence strategies and related changes in settlement patterns and population increase (Jerardino, 1996).
Case Study Pancho’s Kitchen Midden is a shell midden in the Elands Bay area (Figure 1). This site was recently excavated to obtain a more complete cultural sequence for the area as a whole (Jerardino, 1996) and with the explicit purpose of undertaking detailed analyses of the abundant molluscan remains present. Location, stratigraphy and dating Pancho’s Kitchen Midden is a small rocky overhang with a large talus situated at an altitude of 35 m above
present sea level in the surroundings of Waterkloof (35)20*22+S, 18)20*E). Of the site, 3 m2 were excavated to bedrock and another 3 m2 were opened to a depth of 20–30 cm. Most of the excavated material was sieved over a 3·3-mm mesh, although shell samples were sieved over a 1·5-mm mesh. Stratigraphically, Pancho’s Kitchen Midden is a superimposed series of relatively dense shell middens separated by sandier partings (Figure 2). Shell overwhelmingly dominates all other components. Seven stratigraphical layers were observed, and six radiocarbon dates on charcoal samples have been obtained for this sequence. All radiocarbon dates reported for this site are expressed as years . The basal occupation of Pancho’s Kitchen Midden (Layer 7) dates to 3570&60 (Pta-5743) and is characterized by two shell lenses in a 15-cm thick light brown sand and gravel matrix. Faunal remains in this layer are relatively frequent. Subsequent visits to the site have been dated to 3060&60 (Pta-5923) and stratigraphically defined as Layer 6. This layer is characterized by a 10-cm thick, grey lens of dense shell. Immediately above this deposit, Layer 5 appears substantially sandier, light brown in colour, and with minor shell lenses. This layer has not yet been dated directly, although a date of c. 3000 can be assumed according to the radiocarbon dates obtained from archaeological material deposited immediately above and below Layer 5. A date of 2940&20 (Pta-5990) was obtained from the next occupation layer. These deposits (Layer 4) are characterized by a distinctive 15–20-cm thick and very dark, dense layer of shell. The
1036 A. Jerardino
subsequent and last pre-pottery period occupation at the site (Layer 3) is represented by a moderately dense layer of highly fragmented and relatively poorly preserved shell. The matrix is sandy and dark brown in colour. This layer was dated to 2640&60 (Pta5602). After an apparent 1500-year hiatus, Pancho’s Kitchen Midden was visited once again during the last 1000 years. Layers 1 and 2 date to 570&20 (Pta-5605) and 880&50 (Pta-5921) respectively, and are represented by 10–15 cm and 5–10 cm thick subsurface deposits of sand and dense shell. Shell fishing from Pancho’s Kitchen Midden The nearest rocky reef is Mussel Point, 1·5 km from this site. It is likely that shellfish were collected mostly at this reef as extensive mussel beds are easily accessible here during low tides. The facts that this reef is by far the nearest to Pancho’s Kitchen Midden and that the shellfish assemblage closely resembles the present reef shellfish composition, support this assumption. Moreover, the occupation of Pancho’s Kitchen Midden took place after the nearby shoreline had remained largely unchanged, so mollusc colonies at Mussel Point were accessible for collection during its entire occupation. According to the stratigraphical sequence and associated radiocarbon dates, at least half of the shellrich deposits at this site were accumulated between c. 3100 and c. 2600 . This observation neatly coincides with the general temporal pattern of increased shellfish collection between 3000 and 2000 in the Elands Bay area. It is also important to note that changes in SST, sea level and water turbidity were under-way during the occupation of Pancho’s Kitchen Midden. Thus, it is likely that human predation, and/or changes in the marine environment (physical and biological) were responsible to a large extent for the changes observed in the shellfish record of Pancho’s Kitchen Midden. Consequently, the study of the marine molluscan fauna present at this site offers an excellent opportunity to investigate a range of behavioural and environmental variables in relation to this record.
Methods Several shell bulk samples were obtained from contiguous squares for each stratigraphical layer to control for variability across space. Shell remains (whole specimens and fragments) were sorted in the laboratory and identified wherever possible to generic or specific level, whereupon minimum number of individuals (MNI) and weights were established. For bivalves, such as black mussel, the highest number of either left or right hinges was used to determine the minimum number of individuals, for limpets and whelks the number of apices was counted. Barnacles were only quantified by weight, since shellfish assemblages from other sites have been quantified in
Table 1. Summary information of shell samples analysed from Pancho’s Kitchen Midden Layer 1 2 3 4 5 6 7
No. samples analysed
Total weight (kg)
5 3 5 5 3 6 5
60·8 18·4 41·8 38·0 23·4 46·9 25·8
this way, and comparison with these is important in this study. The number of samples and total shell weight analysed from each stratigraphical layer fluctuated between 3 and 6 and from 18·4 to 60·8 kg, respectively (Table 1). Mean percentages and associated range were calculated by using the arcsin transformation (Sokal & Rohlf, 1981). Size observations were obtained for black mussel by measuring the maximum prismatic band width of both left and right valves. The prismatic band is present along one of the borders of the shell, situated opposite to the anterior retractor scar (Kilburn & Rippey, 1982: 159). This method allows the retrieval of metrical data from broken black mussel shells (Buchanan, 1985), as the maximum prismatic band width remains proportional to the length of the shell throughout the size range of the shells. Specimens with band widths equal to or less than 4·5 mm were considered subadults and quantified separately. For either left or right valves, no less than 469 measurements were obtained for each stratigraphical layer. Size observations on limpet shells were obtained by measuring the total lengths of unbroken shells. The statistical significance of any changes in the mean sizes of black mussel shells was tested using analysis of variance (ANOVA) (Sokal & Rohlf, 1981) using the Statistical Analysis System (SAS, 1985). The relative measure of palaeoturbidity of littoral waters was established by quantifying marine sediments (water-worn shell and pebbles) present in each stratigraphical layer of Pancho’s Kitchen Midden and retained by a 1·5-mm mesh during excavations (Jerardino, 1996). These water-worn shells and pebbles are securely identifiable as of marine origin and comprise part of the coarsest sediment fraction available on intertidal reefs. It is assumed that changes in coastal sediment budget would affect quantities of both the finest fraction (sands) and coarsest fraction of marine sediments in similar ways. Water-worn shells are worn and polished shell particles derived from marine intertidal species (e.g. black mussel, white mussel, Burnupena spp and barnacles), ranging in size between approximately 2 and 20 mm. Water-worn pebbles are worn and polished particles of mineral origin, most frequently including those of quartzitic source. The size of water-worn pebbles ranges from 2 to 15 mm.
Shellfish in Southern African Coastal Middens 1037
Mussel shell
Byssus
Sediments
Figure 3. Representation of a black mussel (Choromytilus meridionalis) and its byssus content (sediments).
These coarse sediments were almost certainly brought to the site as part of the byssus contents of mussels collected from nearby rocky reefs (Figure 3). The presence of byssus contents in archaeological deposits has also been recognized elsewhere (Waselkov, 1987: 100), and personal observation on recently collected mussels also supports this suggestion. Water-worn shells and pebbles were observed clinging to mussel byssus, and only a few sand grains were found trapped inside these bivalves. These fine fraction sediments, however, are not part of this study. No water-worn shells or pebbles were observed clinging to limpets or whelks which, together with mussels, dominate the shellfish assemblage of shell middens in the research area. The possibility that the sediments under study here could have been transported to sites on people’s feet and clothing can be discounted, as similar temporal trends in the frequency of such sediments are found in several sites in the Elands Bay and Lamberts Bay areas (Jerardino, 1996). A more random distribution of water-worn shell and pebble frequencies through time would be expected if clothing was a major transporting agent as clothing can change with the season of occupation and fashion, both of which are expected to be variable through time. People’s feet are also unlikely to transport the main proportion of water-worn shells and pebbles reaching sites. Feet often keep submerged and wet during the
collection of shellfish, a condition that precludes the clinging of sediments (within the size range of waterworn shells and pebbles) for long before people return to their camp sites. Thus, it is likely that the presence of coarse marine sediments in shell midden deposits is the result of byssus contents. Nevertheless, further research into this matter should confirm or negate this hypothesis. In this study, the abundance of marine sediments is expressed as weight of sediments per kilogram of black mussel shells. For the purpose of interpreting marine sediment abundances, it is also assumed that the amount of sediment trapped in the byssus threads is proportional to the amount of sediment present in the immediate and contemporary marine-aquatic environment. Thus, high ratios indicate large numbers of water-worn shells and pebbles clinging to mussel byssus, due to relatively large amounts of mineral and organic-derived sediments present in the immediate marine environment. Low ratios indicate the opposite. It also seems reasonable to assume that most of the black mussels were collected in the past at approximately the same water depth, as beds would have been easily accessible from the mid- to low-intertidal zones of Mussel Point during low tides. Average palaeo-SST were established by analysing P. granatina shells of each stratigraphical layer via oxygen isotope analysis following Cohen’s (1993) laboratory procedures and shell-size requirements. Observations on Late-Holocene SST previously obtained by Cohen (1993) were not used in this study. This is mainly due to the fact that strict contemporaneity of Cohen’s data with Pancho’s Kitchen Midden sequence is not warranted. It is possible that short-term SST fluctuations that might be observable at Pancho’s Kitchen Midden are not likely to be reflected by Cohen’s observations, thus the need to establish a SST record directly from Pancho’s Kitchen Midden.
Results Mean weight frequencies (%) and associated ranges of the most frequent groups of mollusc species are shown in Table 2 and represented in Figure 4. Mean MNI frequencies were also calculated, but since barnacles only were weighed these observations do not allow for comparison among species and therefore are not presented here. Clearly, black mussel is the dominant species throughout Pancho’s Kitchen Midden occupation, with frequencies ranging between 96 and nearly 99% from Layer 7 to Layer 2 (3600–c. 900 ). In Layers 1 and 2, black mussel frequencies show a moderate drop to 74 and 91%, respectively. Whelks (Nucella spp, Burnupena spp and Argobuccinum pustulosum), and four species of the genus Patella, as well as barnacle remains (mostly Austromegabalanus cylindricus)
1038 A. Jerardino Table 2. Percentage frequencies of C. meridionalis, Patella spp, whelks and barnacles in Pancho’s Kitchen Midden sequence. The category ‘‘whelks’’ includes individuals from the genera Nucella, Burnupena and Argobuccinum Mean (%)
Upper range
Lower range
73·8 91·7 98·9 98·2 97·2 98·5 96·5
80·6 94·6 99·3 98·6 98·0 99·0 97·5
66·5 88·1 98·5 97·9 96·2 97·9 95·4
Patella spp 1 2 3 4 5 6 7
8·6 6·9 0·1 0·0 0·1 0·1 0·8
11·6 10·7 0·2 0·0 — — 1·8
6·10 3·90 0·09 0·00 — — 0·20
Whelks 1 2 3 4 5 6 7
9·1 4·7 0·7 1·5 2·2 0·9 2·2
12·4 6·9 1·0 1·9 3·1 1·6 2·6
5·3 2·9 0·4 1·1 1·5 0·5 1·8
Barnacle 1 2 3 4 5 6 7
8·10 2·30 0·17 0·17 0·45 0·40 0·32
11·00 2·80 0·22 0·25 0·63 0·51 0·47
5·50 1·80 0·13 0·10 0·30 0·28 0·20
Layer C. meridionalis 1 2 3 4 5 6 7
Weight (%)
100
increase in frequency in the two topmost layers, but without dominating the shellfish assemblage. Details of the frequency of species from the genus Patella present at Pancho’s Kitchen Midden are provided in Table 3. Mean sizes of both left and right valves’ prismatic bands show the same pattern of change (Table 4). Figure 5(a) shows mean width, 95% confidence intervals and standard deviations of black mussel prismatic bands from left valves for each stratigraphical layer. ANOVA results show that the mean sizes of black mussel increased and/or decreased significantly from one layer to the next (P