Methodology for field sampling of archaeological sediments Site selection

In document Agriculture in Tongan Prehistory: An Archaeobotanical Perspective (Page 131-137)

During fieldwork in November 2011, sediments were sampled from open archaeological sites using flotation and bulk sampling for microbotanical analyses. Comparisons of identified taxa within sediments were made on a spatial scale across north-eastern Tongatapu, to provide a scope for the chronology of the introduction of crops and development of production systems. This incorporated test-pitting at two known sites on the south-eastern edge of the Fanga’Uta Lagoon, and one site at Heketa on the northern coast underneath the ‘esi or chiefly back-rest monolith. Talasiu (TO-Mu-2) and Leka (J17) are both located in the Mua region within the village of Lapaha and are now inland although Talasiu is located on the edge of an old palaeoshore (Dickinson 2007). The ‘esi (TO-Nt-2) is attributed to the construction phase at Heketa initiated by the 11th Tui Tonga within local traditions, and is considered contemporary

with the Ha’amonga trilithon of the same complex.

These sites had already been located, surveyed and sampled for archaeological material by Clark and others (2008), Golson (1957), McKern (1929), Poulsen (1987), and Spenneman (1986, 1989) in archaeological projects on Tongatapu. McKern (1929) carried out the first major archaeological survey of the Tongan archipelago as part of the Bayard Dominick Expedition of the Bernice P. Bishop Museum in 1920-21. Mapping and excavation of the tombs and enclosing ditch at Lapaha and Heketa was a particular focus of his work, and his labelling of the tombs has been used in all subsequent archaeological research. Langi Leka (J17) and the ‘esi at Heketa (TO-Nt-2) were among the monuments mapped by McKern. Golson (1957) carried out a limited survey of Tongatapu and offshore islets, exploring the potential for using ceramics as cultural markers. During this survey Golson excavated a section of a site named by

114 Spenneman TO-Mu-2 near the village of Talasiu, and reported on the potential of the pottery- bearing midden. Golson encouraged one of his students, Jens Poulsen, to further explore the ceramic typological sequence within Tongatapu. Poulsen (1964, 1967, 1987) subsequently spent a year carrying out a detailed program of excavation around the Fanga’Uta lagoon.

Following extensive archaeological research in the late 1960s by Davidson (1969, 1971), Groube (1971) and others that focused on late prehistoric sites and the construction of a Tongan culture history, Spenneman (1986) spent some months in Tonga attempting to locate post-Lapita sites in accordance with the research prerogatives of the Tongan Dark Age Project initiated at the ANU. He returned to the site at Talasiu (TO-Mu-2) and sampled the midden for ceramic and organic material for radiocarbon dating, reporting on the presence of decorated and plainware pottery within the same context. Most recently, Geoff Clark returned to Tongatapu in 2006–2008 to assess the construction sequences of the monumental sites of Heketa and Lapaha as part of two Australian Research Council (ARC) grants exploring the development of social hierarchy in Tongan prehistory.

These surveys of the archaeological landscape on Tongatapu and subsequent excavations have provided valuable insights into the nature and timing of cultural deposits. The sites were all noted to have intact deposits with little disturbance related to subsequent occupation. The morphology of the material culture and age ranges of dating material (shell and charcoal) also indicated that each of these sites represented a different chronological period within Tongan prehistory spanning over 2000 years. The presence of refuse such as shell, fishbone and charcoal within the deposits at the sites also pointed to the likelihood that the sites were where domestic practices such as cooking and rubbish disposal took place. These three sites were selected for my PhD research based on site preservation and the reasonable possibility that macrobotanical remains might be recovered that would allow archaeobotanical exploration of plant use in Tongan prehistory.

Field methods

High-resolution excavation techniques were employed at Talasiu, Leka and Heketa to gain information about subsistence practices and vegetation history at the three sites. Botanical, faunal and artefactual remains were collected from test-pits of around 50x50cm using simple bucket flotation and wet sieving. These techniques are recommended for use in tropical climates (Fairburn 2005b; Pearsall 2010), where access to water and technical equipment can be limited. Some further revision of these techniques was required due to issues that were encountered during fieldwork, and therefore a revised protocol was established that suited the conditions specific to Tongatapu during the 2011 season. The methodology can thus be broken down into four sequential activities:

115 2. Bucket flotation

3. Wet sieving

4. Sorting, identification and quantification Excavation:

Test pits of 50x50cm were marked out using string and excavated in 5cm arbitrary levels, referred to as ‘spits’. There was slight variation in the size of the test-pits where contemporary cultural deposits had been already sampled in previous test-pits at particular sites. If a change in stratigraphy was noticed during excavation, a new level was begun. All excavated material was placed into buckets for flotation with a water-proof label. Any observed charcoal concentrations were sampled in situ and wrapped in aluminium foil. The charcoal locations were then recorded on datasheets for each spit. The datasheets also included detailed information upon the soil texture, colour, sorting, and inclusions. Bulk soil samples of around 100g were retained from a 10cm-wide bulk soil sampling column that was measured out and marked on the test pit wall, with samples reserved for microbotanical analysis.

Bucket Flotation:

The excavated material from each level was weighed using a spring balance, and the volume was calculated by pouring the sediment into a bucket with volume marks (litres). This measurement was then recorded in a flotation logbook. The material was then divided into buckets, so that each bucket was filled to one-third (33%) of its volume. Water was added until the bucket was nine-tenths (90%) full, and the material was gently stirred by hand. This mixture was left to settle and deflocculate for around 5-10 minutes (after Fairburn 2005b).

A flotation sieve was set up by pegging muslin mesh into a bucket with large holes cut into the base. To recycle water, this rested on two cut branches that sat atop a plastic washing tub. The excavated material was then decanted into the muslin, allowing any material that was floating on the surface of the water to be separated from the heavy residue that had sunk to the bottom of the bucket. More water was then added to the heavy residue, and the process of settling and decanting was repeated at least twice until no more material was visible floating on the surface. The muslin was then unpegged from the lip of the bucket and tied to create a sack that held the ‘flot’ sample with a waterproof label showing the site, test-unit number, level number and date. Finally, these flot samples were placed onto a string line to dry in the sun, and then put into labelled aluminium foil envelopes in zip-lock bags.

Wet sieving:

The remaining heavy residue was taken to a new station for wet sieving through a 3mm mesh sieve. Again water recycling techniques were employed by keeping water in two large plastic washing tubs. The sieves were immersed in these tubs to just below the rim and agitated. Keeping the rim out of the water ensured that no larger material floating on the surface (such as leaves) could enter the sieve while being agitated. Once all dirt was removed from the sample,

116 the remaining material was placed onto a plastic rice sack with a water-proof label to dry in the sun.

Sorting, identification and quantification:

Some basic sorting of the dried material was carried out in the field. Heavy residue from each level was dry-sieved using a 6mm mesh sieve to create a large (>6mm) and small (<6mm) fraction. Each fraction was sorted and recorded separately. These fractions were placed onto trays and sorted using tweezers into basic material types: ceramics, lithic material, other artefacts, bone, shell, charcoal, seeds and other organics. The sorted fraction were then bagged and labelled. After being returned to the ANH laboratories at the Australian National University, further sorting and quantification of both the heavy residue and the flot samples was carried out. The heavy residue was sorted in much the same way as that in the field, and weights of the various artefacts and material types were recorded. The flot samples were also sorted into material types such as seeds, wood charcoal, charred root and tuber parenchyma, land snails, bone and insect remains. The weights and counts of these were then recorded on datasheets which were then digitised.

Figure 6.1 Map showing location of archaeological sites included in this study from Tongatapu

Site descriptions

Talasiu (TO-Mu-2)

All three sites were cultural midden deposits of varying sizes and concentrations. Talasiu (TO- Mu-2) in the Mua region is a late Lapita-associated site which was first excavated by Golson

117 (1957). Golson’s field-notes (1957) describe the density of shell and pottery within his excavation units, causing his team to halve the size of the original unit from 10ft² to 10x5ft. During survey, Spenneman (1986:38) noted that the midden site extended as far south as the Langi or monumental stone tombs (J18-J19) but was distinct from another nearby midden (TO- Mu-67). Spenneman sampled the midden and recorded a mix of decorated and undecorated pottery, stone flakes, adzes and adze fragments, and a coral abrader alongside some fragmented human remains.

Evaluation of the topography and geology of the Mua region indicates that the midden was originally deposited on the shore of the Fanga’Uta lagoon; however, later extensive land reclamation using coral rubble, limestone and soil in-filled the inter-tidal lagoon flats, and the site is now approximately 200m inland from the current lagoon shore, except to the north where an inlet cuts close to the site (Dickinson 2007). The midden stretches approximately 100m along the old palaeoshore (Spenneman 1986). The site could have been an attractive location for settlement in prehistory, with a small limestone solution channel and spring located near the site that provides a source of fresh water and feeds into the lagoon (Spenneman 1986; Valentin and Clark 2013). Several cultural strata are present at the site, the earliest of which is composed of many cooking and other domestic features cut or embedded into the reddish sterile clay. Above this is the deposition of a dense midden of near shore and inter-tidal shellfish and fish from the lagoon. Cut into this midden and the earlier domestic features are a number of complete and incomplete burials, believed to be mostly contemporary with the deposition of the midden from 2700-2500 cal BP (Valentin and Clark 2013).

The only test pit excavated at Talasiu (TP2) for this project was a 50x50cm test-unit that sampled the cultural midden, and cut into the sterile basal clay by 5cm. The whole unit was a total of 100cm in depth, with every 5cm of deposit excavated, processed and described as a separate level or spit. All excavated material from these spits was processed using flotation and wet-sieving to isolate botanical, faunal and cultural material. A total of 19 spits were excavated and processed for botanical remains. 100gm bulk samples were taken from each level within a 10cm sampling column marked out on the north-facing profile of TP2, where stratigraphic levels were most distinctive, for starch analysis.

Leka (J17)

Langi Leka (also known as Lekamakatuituioha or ‘Puipui’) is a four-tier monumental tomb or

langi (J17) built during the classic Tu’i Tonga chiefdom phase of Tongan prehistory, and is said to be associated with Tu’i Tonga Tulunga (McKern 1929:41). The langi is located about 400m inland from the Fanga’Uta lagoon in the community known today as Lapaha, within the Mua region of Tongatapu. Geophysical survey at Langi Leka indicates that the langi had been built on top of an older cultural deposit. Initial excavations were carried out in 2008 to assess the

118 nature and timing of this older occupation through test-pitting. These prior excavations also shed light on the construction sequence for the langi. Attempts were made to relocate TP1which had been abutting Langi Leka. It was believed that TP1 had been found, and TP2 was subsequently excavated parallel to this location further west. A 2x1m test pit was measured out and extended north from the base of the first tier of the langi. The red clay and coral construction rubble debris was recorded and then removed in bulk, so that the darker cultural material below could be excavated and processed in a 100x50cm test pit in the northern end of the excavation. It turned out this unit missed the more dense shell midden that pre-dated Langi Leka by about 5m. Despite this, seven 5cm levels were processed for botanical remains below 90cmbd through flotation and wet-sieving.

In a second attempt, another test 2x1m unit (TP3) was opened up approximately 2m to the east of TP2. TP3 also abutted the langi, and was sampled from 90cmbd within the cultural material using the same techniques employed for TP2, but this time from within a 100x50cm test pit in the southern end of the excavation. Cultural deposits were encountered in both TP2 and TP1, although these were not of the same density noted during previous excavations. TP3 was abandoned and back-filled after excavating to 110cmbd. TP1 was finally located after a system of shovel-test pitting along the base of Langi Leka was employed and the modern fill was quickly removed. The northern baulk was cleaned and chosen for sampling the cultural deposit observed at 95cmbd. A 25x50cm test unit was cut into the baulk, and material above 95cm was discarded. Sampling began below this in the dense cultural midden with every 5cm level excavated, processed and recorded in the same manner as other test pits made at Talasiu and Leka.

Heketa (TO-Nt-2)

The final site chosen for excavation and botanical sampling was located in the north-east at Heketa near the ‘esi’ or chiefly backrest known as Makafakinanga (TO-Nt-2) situated close to the Ha’amonga a Maui trilithon (TO-Nt-1). This complex is composed of nine stone structures (Nt-1 to 9) that represent a short period of monumental stone architecture construction, forming an early centre of the classic Tu’i Tonga chiefdom (Clark and Reepmeyer 2014) or state (Kirch 1994). Tongan traditions associate most of the structures with the 11th Tu’i Tonga, Tuitatui,

while an earth mound is tied to the 10th Tu’i Tonga, Momo (McKern 1929). Dates for the

construction of the stone architecture all fall within the 14th century (Clark and Reepmeyer

2014; Spenneman 2002), but several earlier deposits have been located underneath these structures dating to the 12th-14th century AD, indicating that the site may have also been a

significant place associated with early Tu’i Tonga Hikuleo for several centuries before this. In traditions the centre of the chiefdom was moved south to Lapaha by the 12th Tu’i Tonga

119 This original test pit (TP1) had been excavated in order to assess the construction sequence for the ‘esi or backrest monolith. The aim was to cut into the baulk of this original test pit to sample a cultural midden and charcoal deposit that predates the erection of the ‘esi which had been recorded and dated previously. With this in mind, a 50x50cm test pit (TP2) was marked out about 1.5m away from the ‘esi. The first test pit was excavated to 90cm below datum (bd), when the sterile clay was encountered. Each 5cm level was sampled for botanical, faunal and artefactual material through flotation and wet-sieving. Assessment of the profile and excavated material suggested that this test pit had only sampled the edge of the feature exposed in the original test pit.

Further test pits were then used to locate the looser and more mixed backfill of TP1, and then this backfill was removed to expose the original walls of the test pit. The final unit, TP3, was excavated into the cleaned face of the baulk between TP1 and TP2. TP3 was a 50x30cm test unit that was excavated to 105cmbd below datum, where the sterile clay was again encountered. All material from each 5cm level after 40cmbd was processed using flotation and wet-sieving. The material above this point was all sterile red clay which capped the cultural deposits and topsoil, and had already been sampled within TP2. This strategy allowed specific targeting of the cultural deposit within this test unit.

Stratigraphic descriptions

In document Agriculture in Tongan Prehistory: An Archaeobotanical Perspective (Page 131-137)