Macrobotanical assemblage breakdown
Overall, macrobotanical preservation within all of the sites and test units was relatively good, with small quantities of a range of plant material recovered. However, these remains only composed up to 0.46% of the total pre-processing material weights removed from these units (including soil, shell and other matrices). Within these assemblages, the largest quantities of macrobotanical material were recovered from Talasiu TP2, with a total weight of 60.8g. The smallest assemblage was recovered from Leka TP3, however only three spits were processed from this test unit.
The quantification of the macrobotanical assemblage was primarily based on the separation and analysis of three main components: coconut, other endocarp, and wood charcoal, from which vegetative storage parenchyma was separated later. Wood charcoal composed the bulk of the macrobotanical assemblages from almost all the test units at Talasiu (TO-Mu-2), Leka (J17) and the Heketa (TO-Nt-2), making up around 78% of the collective weight (see Table 8.1 and Figure 8.1). Only Leka TP4 varied from this trend, with wood charcoal making up only 26% of the macrobotanical assemblage from this test unit. In particular, Talasiu TP2 had the greatest amount and distribution of non-endocarp charcoal (53.4g). For this reason, this test unit was selected as a case study for the separation and identification of vegetative storage parenchyma. Only very small quantities of parenchyma (<1g) were found in any given sampled deposit. Aside from wood charcoal, non-coconut endocarp was the second most common
139 macrobotanical material recovered and quantified, composing around 17% of the total assemblage. Coconut endocarp was recovered from every test unit, but only in small quantities and so made up only 5.8% of the overall assemblage.
The quantification of macrobotanical materials is analysed here in terms of the rates of recovery from flotation, wet-sieving and in situ collection, and the overall quantities, as well as the distribution of these quantities within each test unit. This quantification indicated both the success of various methods for extraction of particular plant remains, as well as the effects of site taphonomy and formation processes upon archaeobotanical assemblage composition.
Table 8.1 Summary of total macrobotanical assemblages from all sites and test-pits
Figure 8.1 Composition of overall macrobotanical assemblage in terms of abundance
Coconut endocarp
Basic sorting processes highlighted an interesting pattern in the macrobotanical extraction process. Coconut endocarp was extracted exclusively during the wet-sieving stage of archaeobotanical processing in the field. This is most likely because of the higher density of coconut endocarp, which does not float as well as wood charcoal, or other varieties of endocarp. Another possibility is that coconut endocarp does not disaggregate easily from clay soils during the initial deflocculation process before flotation. No experimental tests were carried out to test either of these ideas. These coconut endocarp fractions were separated primarily to enable selection of samples for AMS dating.
When all of the test units are compared directly, the highest quantities of coconut endocarp were observed within Talasiu TP2 and Heketa TP3 with a total of 1.6g in each test unit. However, when these quantities are considered as a fraction of the total macrobotanical weight for each test unit, the highest percentages were extracted from Leka TP3 (10.7%) and TP4 (19.5%) (see Table 8.2). The remaining test units contained coconut endocarp fractions that Botanical material Talasiu Leka TP2 Leka TP3 Leka TP4 Heketa TP2 Heketa TP3 Total (g) % of assemblage
Coconut (g) 1.6 0.6 0.1 0.7 1.6 1.2 5.8 4.832125302
Endocarp (g) 5.8 0.3 2.3 4.1 2.6 5.1 20.2 16.82912605
Wood charcoal/parenchyma (g) 53.37 8.47 2.4 1.72 4 24.07 94.03 78.33874865
Total (g) 60.77 9.37 4.8 6.52 8.2 30.37 120.03
140 were less than 10% of the entire macrobotanical weight, with Leka TP3 containing the smallest percentage.
Recovered coconut endocarp was distributed throughout the test unit from Talasiu, with quantities found in every stratigraphic layer, aside from the sterile clay at the base, and almost every spit. Only one stratigraphic layer was processed in TP2 from Leka. This layer was Layer 8, which was a mid-brown silty clay cultural layer with charcoal, and this contained coconut endocarp in the bottom 20cm (105-125cm below surface) before sterile clay was encountered. Leka TP3 only contained coconut endocarp in this same cultural layer in the top 5cm (90- 95cmbs). In contrast, coconut endocarp was extracted from all sampled stratigraphic layers in Leka TP4 (Layers 4-6), but not every spit in these. Small quantities were recovered from the lower 15cm (110-125cmbs) of Layer 4, a mid-brown silty clay layer similar to that observed in TP2 and TP3, and the top 5cm (125-130cmbs) and middle (134-140cmbs) of Layer 5 in a more dense shell deposit. Coconut endocarp was also recovered from the bottom 10cm (155- 170cmbs) of this test unit in the mixed clay and shell deposit of Layer 6. At Heketa, coconut was extracted from all stratigraphic layers in TP2, but was not observed within the bottom 10cm of this test unit (90-100cmbs). Coconut endocarp extracted from TP3 was mostly concentrated in the lower 35cm (70-105cmbs) of this test unit, in the base of Layer 3 and throughout Layers 4-5.
Table 8.2 Quantification of coconut endocarp from all sites and test-pits
Other endocarp
Both wood charcoal and non-coconut endocarp composed a large percentage of the macrobotanical remains from all sites and test units. Endocarp was successfully extracted using all three archaeobotanical techniques employed during fieldwork. Wet-sieving and in situ
collection resulted in the extraction of endocarp from all test-units, whilst flotation had more variable rates of recovery. This could be the result of issues with the flotation of endocarp already highlighted with regard to coconut endocarp extraction. This fraction was separated as a preliminary step towards the selection of samples for AMS dating. These endocarp fractions remain unidentified because of the lack of a comprehensive comparative collection. Some
Aleurites moluccana was observed but was not separated from other endocarp due to a focus within this study on the identification of parenchyma within the macrobotanical remains.
Site TP Wet-sieved (g) In-situ (g) Flot (g) Total % of total macrobotanical weight
Talasiu 2 1.6 0 0 1.6 2.6 Leka 2 0.6 0 0 0.6 6.4 Leka 3 0.1 0 0 0.1 2.1 Leka 4 0.7 0 0 0.7 10.7 Heketa 2 1.6 0 0 1.6 19.5 Heketa 3 1.2 0 0 1.2 4.0 Total 5.8 0 0 5.8 45.3
141 The quantification of other endocarp followed the same pattern as coconut endocarp, whereby the highest quantities were extracted from Talasiu TP2 (5.8g) and Heketa TP3 (5.1g) (see Table 8.3). When quantities of endocarp are considered as percentages of the total weight of macrobotanical material extracted from these test units, Leka TP4 clearly has the highest percentage of other endocarp (62.9%), while Leka TP2 has the lowest (3.2%) by a significant margin.
Talasiu test unit TP2 again has recovered endocarp distributed throughout all strata. Only two spits did not contain any non-coconut endocarp, at 0-5cmbs and 80-85cmbs. The three test units at Leka have variable distributions of endocarp throughout the observed stratigraphic layers. Endocarp was recovered from the top (90-95cmbs), middle (105-110cmbs) and bottom (120-125cmbs) spits of Layer 5 in TP2. In contrast, TP3 had non-coconut endocarp in all three sampled spits of the same stratigraphic layer, while TP4 had small quantities of endocarp throughout the test unit apart from the top 5cm (95-100cmbs). Non-coconut endocarp was found throughout all strata within TP2 and TP3 at Heketa, but was absent from the bottom 5cm (100- 105cmbs) of TP3.
Table 8.3 Quantification of other endocarp from all test units
Wood charcoal and parenchyma
The separation of monocot and dicot wood and parenchymatous charcoal from coconut and other endocarp was the final stage in the quantification of macrobotanical material. Because of the similarity of vegetative storage parenchyma to wood charcoal, any plant-derived charred material that was not endocarp was separated first and weighed as a single fraction. Because of the abundance of charcoal from Talasiu TP2, this unit was used as a test for identifiable parenchymatous material. Wet-sieving and flotation techniques were by far the most successful method for recovering both wood charcoal and parenchyma. The vesicular cellular structure and specific gravity of wood charcoal enables this material to float relatively easily in water, a fact which has been utilised by archaeobotanists for many years. The success of wet-sieving, as a follow-up to flotation steps, is most likely the result of the inability of technicians to easily deflocculate clay sediments to release wood charcoal in the field using bucket flotation methods. The highest quantities of wood charcoal (including parenchyma) were recovered from test unit TP2 at Talasiu with a total of 53.4g (see Table 8.4). TP3 at Heketa had the second greatest quantity (24.1g) of wood charcoal, while the remaining test units all had less than 10g.
Spit TP Wet-sieved (g) In-situ (g) Flot (g) Total % of total macrobotanical weight
Talasiu 2 3.6 0.9 1.3 5.8 9.5 Leka 2 0.2 0.1 0 0.3 3.2 Leka 3 0.6 1.4 0.3 2.3 47.9 Leka 4 2.4 1.7 0 4.1 62.9 Heketa 2 1.6 0.3 0.7 2.6 31.7 Heketa 3 3.9 0.7 0.5 5.1 16.8 Total 12.3 5.1 2.8 20.2 172.0
142 Despite these figures, when these quantities are viewed as percentages of the whole macrobotanical assemblages from each of these test units, Leka TP2 has the highest percentage of wood charcoal, with 90.4%. These statistics indicate that the assemblages from Talasiu TP2, Leka TP2 and TP3, and Heketa TP3 are all dominated by wood charcoal. In contrast only 26.4% of recovered charcoal is represented by wood charcoal in Leka TP4, while Heketa TP3 has just under half (48.8%).
The distribution of these quantities within the test units from the sites of Talasiu, Leka and Heketa are variable; however, at least small amounts of wood charcoal were present in every spit and stratigraphic layer sampled. In Talasiu TP2, most charcoal was concentrated around Layer 3 (40-75cm) where it is likely that larger fragments were preserved in the matrix of loose shell and yellow red clay. Very little charcoal was recovered from Layers 1, 2 and 5 in the more compact shell midden deposits. Leka TP2, TP3 and TP4, and Heketa TP2 had relatively even distribution of wood charcoal in the test units, with small amounts (<2g) in every spit and stratigraphic layer. In contrast, Heketa TP3 had the largest concentration of charcoal within Layer 4 (80-90cmbs) with 12.5g, while the remaining strata had less than 5g.
Table 8.4 Quantification of wood charcoal and parenchyma from all test units Parenchyma distribution and identification: Talasiu TP2 case study
Within this section the macrobotanical assemblage from test unit TP2 at Talasiu (TO-Mu-2) was chosen to conduct a case study for the separation and identification of vegetative storage parenchyma. Due to the documented difficulty of separating these charred remains from wood charcoal (Hather 2000), it was decided to analyse only one of the six test units excavated during the 2011 field season. Talasiu TP2 was chosen for the abundance of charcoal in the macrobotanical assemblage of this unit, on the assumption that the preservation of any vegetative storage parenchyma was likely to be the greatest.
Parenchyma was first separated from wood charcoal based on ground tissue morphology, such as consistent cell shapes that are usually rounded or angular; the presence of distinctive vascular bundles or tissues, usually with very few rays dividing these tissues. This initial step enabled the presence of vegetative and non-vegetative parenchyma (i.e. fruits) in samples to be recorded. The distribution of these remains in the test unit varied, although at least one fragment of charred parenchyma was extracted from each stratigraphic layer. The disturbed
Spit TP Wet-sieved (g) In-situ (g) Flot (g) Total % of total macrobotanical weight
Talasiu 2 23.7 17.9 11.8 53.4 87.8 Leka 2 3.5 4.5 0.5 8.5 90.4 Leka 3 0.8 1.6 0.0 2.4 50.0 Leka 4 1.0 0.3 0.5 1.7 26.4 Heketa 2 2.0 1.1 0.9 4.0 48.8 Heketa 3 22.4 1.2 0.5 24.1 79.3 Total 53.4 26.6 14.1 94.0 382.6
143 midden within Layer 1, and very compact shell midden deposit in Layer 2 both contained three fragments. The largest overall quantity of parenchyma was recovered from the loose large shell matrix composing Layer 3 (see Table 8.5), with a total of five fragments. Layer 4 below this contained another two fragments, while the basal cultural layer of reddish clay and small crushed shell only contained one fragment.
Each fragment was identified to species where possible, using a range of attributes of ground tissue and vascular tissue morphology, and the identification flowchart key created for the reference collection (Chapter 5). Most fragments could be identified with a moderate degree of confidence, and were therefore labelled with the prefix ‘cf.’, indicating that most morphological features matched the written descriptions and resembled the reference SEM images.
A number of different families are represented within the identified parenchyma assemblage. These included Araceae, Dioscoreaceae, Moraceae, Musaceae, and Zingiberaceae.
Four fragments were identified as cf. Colocasia esculenta, or the common taro and were extracted from Layers 2 and 3 to a maximum depth of 65cmbs. Two species belonging to the
Dioscorea or yam genus, Dioscorea alata and Dioscorea nummularia, were found in Layers 3 and 4, with a total of four identified fragments. A single fragment identified as cf. Artocarpus altilis (breadfruit) fruit/flesh was found in Layer 2 between 25-30cmbs. Interestingly, a single fragment of parenchyma from Layer 3 was identified as belonging to the Musaceae family, indicating that the flesh of this fruit was either intentionally cooked or discarded into a fire and incorporated into the ash. Members of the Ginger family Zingiberaceae, such as Zingiber zerumbet or Curcuma longa were often cultivated and eaten in Tonga or used for medicinal and ornamental purposes. The antiquity of this use is demonstrated by the recovery of parenchyma identified confidently to this family within the archaeobotanical record at Talasiu. The remaining fragments could only be identified as ‘root-derived’ (Layer 3), or were left classified as ‘unidentifiable’ (Layers 1 and 5).
144
Table 8.5 Distribution and identification of parenchyma extracted from Talasiu TP2
Microbotanical analysis
Extraction, quantification and distribution
Microbotanical assemblage breakdown
Starch was successfully extracted from all test units sampled for microbotanical remains from Talasiu (TO-Mu-2), Leka (J17) and Heketa (TO-Nt-2). Due to time constraints, only four of the six excavated test units were sampled, with a total of 39 bulk soil samples processed. These test units included Talasiu TP2, Leka TP2 and TP4, and Heketa TP3, and were chosen based on site taphonomy and the likelihood of microbotanical preservation. Of these, Talasiu TP2 contained the largest quantity (count) of starch grains, but this is biased by the large number of samples analysed from this site. Every spit or level was sampled from Talasiu TP2, whilst every second spit was sampled from the remaining test units. This alteration of the microbotanical subsampling procedure was made after Talasiu TP2 was completely processed, and counts revealed that there was very little variation in starch distribution within the test unit apart from a spike at the very base of the cultural deposits. Due to time constraints, it was deemed appropriate to process every second sample and halve the overall processing time.
With this in mind, it may be appropriate to compare the quantities of starch from each of the test units as an average based on the number of processed samples (see Table 8.6). When these figures are calculated, it becomes clear that Leka TP2 had the highest quantity of starch per sample, with an average of 93 grains. Leka TP3 followed with an average of 31.4 grains per
Depth Spit Layer Parenchyma Identification
0-5 1 1 5-10 2 1 10-15 3 1 x Unidentifiable Zingiberaceae 15-20 4 1 x Zingiberaceae 20-25 5 2 x cf. Colocasia esculenta
25-30 6 2 x cf. Artocarpus altilis fruit
30-35 7 2 x cf. Colocasia esculenta 35-40 8 3 40-45 9 3 x cf. Dioscorea alata cf. Musaceae 45-50 10 3 x cf. Colocasia esculenta 50-55 11 3 x cf. Dioscorea alata 55-60 12 3 x Root-derived 60-65 13 3 x cf. Colocasia esculenta 65-70 14 4 70-75 15 4 x cf. Dioscorea alata 75-80 16 4 x cf. Dioscorea nummularia 80-85 17 5 85-90 18 5 x Unidentifiable 90-95 19 5 95-100 Control 6
145 sample, just slightly higher than Talasiu TP2 which contained an average of 22.5 grains per sample. The lowest average derived from TP3 at Heketa (9 grains). Considering that AMS dating indicates that this site was the most recently occupied, these figures would indicate that either plant food was not being processed or discarded in these cultural deposits at Heketa, or that taphonomic factors such as soil pH levels or enzymatic activity heavily impacted starch preservation. These quantities from each test unit will be briefly discussed with regard to the distribution of starch throughout the sampled cultural deposits below.
Table 8.6 Overall quantities (counts) of starch extracted from all sampled test units at Talasiu (TO-Mu-2)
Talasiu (TO-Mu-2)
Test unit TP2 from Talasiu contained a total of 449 starch grains, extracted from 20 processed 3gm bulk soil samples (see Table 8.7). When this quantity is broken down into the distribution of starch grains within each recorded stratigraphic layer, some patterning emerges. The disturbed midden at the top of the test unit (0-20cmbs) which composes Layer 1 has a total of 107 starch grains from four samples. Layer 2 below this has the smallest quantity of starch grains, with a total of 47 grains from three samples. Layer 3 has the largest number of samples (n=6), however this deposit has a quantity of starch (96 grains) that falls in the middle of the range when compared to the other strata. Layer 4 had the second smallest quantity of starch with 64 grains from three samples. Finally, Layer 5 at the base of the cultural deposits had the largest quantity, with a total of 128 grains extracted from three samples. The control sample taken from the sterile clay below the shell midden had seven starch grains, which most likely derived from the transition between these two deposits.
Site Total samples analysed Total count Average/sample
Talasiu TP2 20 449 22.5
Leka TP2 4 372 93.0
Leka TP4 8 251 31.4
Heketa TP3 7 63 9.0
146
Table 8.7 Distribution of starch counts within Talasiu TP2
Leka (J17)
Two test units were sampled from excavations at Leka, TP2 and TP4. Leka TP3 was excluded from this analysis as stratigraphic comparisons revealed that the cultural deposit sampled within this test unit was already sampled in TP2. TP2 had a total starch count of 372 starch grains extracted from four 3gm bulk soil samples. These were distributed relatively evenly throughout the test unit, which seems reasonable considering these four samples were taken from the same stratigraphic layer (Layer 7) (see Table 8.8). Slightly fewer starch grains were recovered in the lower 10cm of the unit, above the sterile clay. This is most likely a result of sampling as these figures are not significantly different from the samples above, especially when considered in comparison with other samples from other test units.
Table 8.8 Distribution of starch counts within Leka TP2
Leka TP4 had a different starch distribution pattern. Most starch was located in Layer 4 in two samples from 95-110cmbs, where the 100 grain maximum count was reached for both samples. Layer 4 is possibly the same as Layer 7 in TP2, but the distance between these two units makes this correlation difficult to prove. However, this would explain the high quantity of
Depth Spit Layer Starch count
0-5 1 1 11 5-10 2 1 19 10-15 3 1 34 15-20 4 1 43 20-25 5 2 11 25-30 6 2 11 30-35 7 2 25 35-40 8 3 14 40-45 9 3 5 45-50 10 3 20 50-55 11 3 6 55-60 12 3 46 60-65 13 3 5 65-70 14 4 16 70-75 15 4 10 75-80 16 4 38 80-85 17 5 18 85-90 18 5 24 90-95 19 5 86 95-100 Control 6 7 Totals 449
Depth Spit Layer Starch count
90-95 1 7 102
100-105 3 7 100
110-115 5 7 82
120-125 7 7 88
147 starch extracted from these deposits, especially in comparison with other deposits in TP4. Below Layer 4, starch quantities drop off significantly in Layers 5-6. Layer 5 has a total of 30 starch grains from three samples, while Layer 6 only has five grains from two samples.
Table 8.9 Distribution of starch counts within Leka TP4
Heketa (TO-Nt-2)
Only one test unit from Heketa was sampled for microbotanical remains. A decision was made to sample TP3, and leave TP2 due to both time constraints, and the fact that stratigraphic comparison indicated that the deposit sampled in TP2 (Layer 4) was most likely the same as Layer 3 in TP3. As mentioned earlier, TP3 had the smallest amount of extracted starch with a