By : Rathnasiri Premathilake I. Introduction : Domesticated rice ( Oryza sativa) is one of the world’s most important crops t...
By : Rathnasiri Premathilake
I. Introduction : Domesticated rice (Oryza sativa) is one of
the world’s most important crops today. It is believed that early humans
associated ancestors of domesticated rice (wild rice) before the appearance of domesticated
rice in human cultures of South Asia from the early Pleistocene onwards (1).
But, understanding the antiquity of the human usage of wild rice in the archaeological
context remains in dispute due to lack of studies (1, 2-6). The available rice
data indicate that several independent geographical origins of rice domestication
from their wild ancestors appear to have occurred in East and/or Southern Asia
(7-13). Currently, the Yangtze valley in China has yielded the earliest evidence
for the intensive use of wild rice during the Late Glacial (20 ka), with a
transition to domestication early in the Holocene, around 9 ka (11), and
evidence from the Ganges plains in North India in
dictates the use of wild rice from the post glacial time (15 ka), with a transition
to domestication around 8 ka (12). In this paper, we report well-preserved rice
phytolith evidence from the late Pleistocene archaeological context at the Fahien
rock shelter in Sri Lanka which is indicative of the intensive use of wild rice
species.
II. Fahien rock Shelter : Fahien rock shelter, one of the largest rock sheltersin Sri Lanka, is situated at E80° 12' 55" N6° 38' 55" at 130 m asl in Yatagampitiya village, near Bulathsinhala in the Kalutara District, southwest Sri Lanka (Fig. 1). The rock shelter has great potential for understanding the late Pleistocene humans and environment. It is a complex of interconnected rock shelters developed in coarse crystalline gneiss rock faces (14). The mouth has a width of 30 m and average height of 20 m above the floor. The interior is about 10 m in depth and slopes down from west to east. The present day climate is warm-humid monsoonal, with an average rainfall around 3000-4000 mm/yr and a mean annual temperature at sea level about 26-27°C(15). The landscape around the rock shelter is characterized by disturbed lowland rainforest with paddy fields present in the slightly incised valley below the rock shelter(16.20).
Fig. 1: Location of the Fahien rock shelter in the
southwestern Sri Lanka. Beli-lena and Batadomba-lena
are excavated prehistoric rock shelters
III. Material and Methods : a) Site Stratigraphy In 1968, W. J. Wijeyapala, the former Director
General in the Department of Archaeological Survey of
Sri Lanka (DAS) first examined Fahien and excavated
over several seasons between 1986 and 2012 under the
direction of the DAS. The depiction of litho logical
succession with archaeological contexts at Fahien was
made according to the standard tool, Harris Matrix.
Excavation (4 x 5 m) located in the east of the main
chamber of the Fahien rock shelter has indicated the
potential for understanding of the archaeological
stratigraphies (16-20). The excavation penetrated ca.
2.40 m of heterogeneous clast-rich loam sediments
showing 5 major layers, 10 archaeological phases and
approximately 250 archaeological contexts (Table 1).
The bio-stratigraphy comprises of human bones, animal
remains, burnt and unburnt shells, shell beads,
charcoal, plant remains and coprolites. Human remains
include several internments; some coated with red
ochre and are associated with the first microlith and
osseous technologies anywhere in South Asia. The
stratigraphy also contains palaeo-floors, postholes,
excavated pits and preserved hearths.
Renewed excavation at the rock shelter yielded
a secure chrono-stratigraphy for the earliest modern
human habitation deposits (18-21). This work led to the
site being recognized as having, to date, the oldest
archaeological sequence in Sri Lanka. Well preserved, in
situ charcoal, charred wood, shells and sediment
samples were taken from the sections for 14C and OSL
dating. Twenty six (26) radiocarbon dates wereproduced using Accelerator Mass Spectrometry at the
CHRONO centre, Queens University, Belfast and the
Beta Analytic Laboratory in USA. They were calibrated
using Calib 6.11 (22). Two sediment samples from the
context 92 were processed using OSL and indicate that
the base of the sequence is between 39.9 ± 2.5 ka
(SUTL2327) and 22.0 ± 1.3 ka (SUTL2326) in age.
Twenty six AMS dates obtained from charred seeds,
charred wood, charcoal, Canarium cf. zylanicum nut and
freshwater shells indicate that the period of sequence
formation was between 47.80 ka and 3.85 ka. The most
significant late Pleistocene archaeological evidence,
which includes the oldest microlith toolkits known to
South Asia associated with the contexts from 87 to 92
are dated to between 47.80 ka and 28.5 ka (Fig. 2)
Fig. 2: Late Pleistocene stratigraphy of the rock shelter (Y). Excavated area is marked (X)
b) Sediment processing for phytolith extraction Twelve 30x10x8 cm monoliths were taken from
the southern profile of the excavated area (Fig. 2). These
covered five major layers (L1-L5) including the
described archaeological phases (I-X) (Table 1, 18-19).
From these monoliths, seventeen subsamples were
selected. Eleven subsamples (C-44 to C-86) of early to
late Holocene age were taken from L2 and L3. Six of the
subsamples (C-81 to C-92) were taken from L3/L4, L5
covering the late Pleistocene age (Fig. 2).The current
work considers the late Pleistocene sample only.
Phytolith extraction was made according the standard
procedure (23)
c) Phytolith taxonomy and taphonomy
Classifications and taxonomic identification of
phytoliths from archaeological samples were made
using a modern and archaeological phytolith collections
housed at the Laboratory for Palaeoecology,
Postgraduate Institute of Archaeology, University of
Kelaniya, Sri Lanka and at the French Institute of
Pondicherry(IFP), India. In order to precisely identify rice
taxa from the archaeological samples, comparative
knowledge from multiple reference samples of phytoliths
from a wide range of the parts (e.g. seeds/husks and
leaves) of rice plants growing in different ecological and
environmental contexts, both in Sri Lanka and Southern
India was used. In this procedure, the most common
morphological characteristics, general occurrence and
appearance of key archaeological phytoliths were
comparatively studied (1, 8-9, 13, 24-27).
IV. Results : a) Phytoliths from the Late Pleistocene samples (47.80-
12.15 ka)
In this paper, we mainly highlight the rice
phytolith records from late Pleistocene samples, while
the detailed phytolith records from the sequence studied
will be published elsewhere. The summary of the
phytolith assemblages are shown in Fig.3.All samples
contained high counts of well-preserved phytoliths.A few samples yielded pitted, displayed a few relatively large
micro-channels, mineralized micro-structures and
broken phytoliths. More than 54 phytolith morphotypes
(monocot and dicots) were identified. In this
composition, wild rice ( spp)and banana phytoliths
are extremely abundant in the samples. Phytoliths from
rice leaves (e.g. bulliforms and elongates) and
seeds/husks/glumes (e.g. double and single peaked
morphotypes) are closely comparable with modern
material from the leaves and seeds/husks/glumes of
from Sri Lanka and South India. In addition, phytoliths
from several other economic plants, e.g. Palmae,
Artocarpus cf. nobilis (wild breadfruit), Duriosp and
Poaceae (wild grasses) are foundin all samples.
Burseraceae (Canarium sp. nut) and Cyperaceae
phytolithsare found in several samples. Freshwater
diatomsare common in many samples. Few
brackish/marine diatomsare limited to few samples.
V. Discussion :a) Reliability of the late Pleistocene samples and
phytoliths evidence Fahienrockshelter sediments are
heterogeneous and the archaeological analysis of the
ca. 250 contexts indicates complex sedimentary
processes (18-20, 28). All dates of the late Pleistocene
samples are in good strait graphic order. The
chronology indicates that a significant depositional
hiatuses occur within the excavated sequence between
the late Pleistocene and early Holocene. Multiple
hiatuses extended only from 29.9ka (C-87) to 12.5 ka,
marked by the reduction of phytolith sum(Fig. 2).These
hiatuses can be explained by alternating periods of
desiccation and erosion of the rock shelter sediments.
The condition of the desiccation corresponds to the
number of severe millennial to multi century-scale dry
climatic cycles (e.g. arid/semi-arid) due to monsoon
failures identified from the peat and sedimentary
archives in Southern Asia between 24 ka and 8.1 ka (29-
35). The records suggest that the impact of climate,
environmental conditions including humans was the
dominant factor for forming the litho-stratigraphy
through the late Pleistocene.
Understanding phytolith taphonomy (a-e,
described below) is essential for interpreting therock
shelter phytoliths. The presence of phytoliths in the rock
shelter sediments provides information about the
depositional processes in several ways (a) in situ plant
decay leading to phytoliths deposition on surfaces (b)
alluvial or colluvial re-deposition of phytoliths along with
their associated sediments (c) wind deposition (d)
cultural deposition of phytoliths through plant materials
used by the occupants of the rockshelter for food and
other cultural purposes and (e) fossilization. The lack of
living plants taxa such as rice, banana and Palmaein the rock shelter suggests a minimal input of in situ
deposition of phytoliths. Abundant phytoliths from these
taxa in the samples suggest that alluvial and/or colluvial
processes (effect of horizontal and vertical movements)
may have played a limited role in the phytolith redeposition. Due to the presence of drip line, the impact
of rain water penetration into the rock shelter is minimal.
Wind deposition is rare due to the particular
geomorphology of the rock shelter in the humid tropical
environment. This is clearly confirmed by the highly
variable phytolith counts/sum in all the samples (Fig. 3).
Except for very uppermost parts of the sequence (e.g.
Holocene samples), a lack of the post-depositional
disruptions through roof falling, vertical cracks and
animal burrows indicate limited vertical movement of
phytoliths in the late Pleistocene samples (18-19).
Indeed, the lack of evidence of root penetration from
plants and the lack of organic litter deposits within the
clay- and silt-rich, highly-compacted and multi-layered
sediments do not interfere with phytoliths buried at
much deeper stratigraphic levels through multiple reworking events and bioturbation. More homogeneous
distribution of the smallest phytoliths (3-10 µm) from wild
banana seeds and from Bombacaceae and fine-grained
sediments in all the samples suggest the minimal
impact of illuviation of clay minerals as common
process in the rock shelter stratigraphy (28, 36-37). All
these minimized sources of biases indicate that spatial
and temporal fidelity ishigh in the late Pleistocene
samples (38).
The main process, therefore, of phytolith
deposition in the rock shelter is most likely to have been
through human or animal vectors. However, animals
such as bats, birds, and insects in the vicinity of the rock
shelter environment are very unlikely to play a role in the
phytolith deposition reported. This agrees with highly
variable phytolith counts through the sequence studied
(39). It is inferred that humans are the most likely agents
for phytolith deposition in the rock shelter, - the materials
from economic/anthropic plants such as rice and
banana and breadfruit brought into the rock shelter are
from the plants commonly grown in disturbed lowland
rainforest near to the rock shelter (most possibly within a
few kilometers at most). Abundant phytoliths from
monocotyledonous taxa (e.g. Poaceae/grasses and
Cyperaceae/sedges) identified as anthropic taxa in this
context, probably associated the rice plants brought by
humans. The significant occurrence of freshwater and
brackish-marine diatom species throughout is not
surprising in habitation deposits and is consistent with a
number of human activities. In the majority of samples,
abundantwith rice seeds/husks and leaf phytoliths
together with the lack of taphonomic markers (e.g.
breakage, corrosion, microchannels, regulation,
dissolution pits, mineralized microstructures, cut marks
and pitted patterns) indicate excellent preservation
conditions and selective distribution of phytoliths from rice used by rockshelter occupants. This suggests high
phytolith compositional fidelity in the samples.
The well-preserved phytoliths suggest that they
were directly subjected to the processes of diagenesis,
i.e., physical and chemical impact on phytoliths due to
the long-time environmental burial (or buried for a long
time) and permanent incorporation into the rock shelter
sediments (37, 40, 41-54). Alkaline conditions are also
thought to contribute to phytolith dissolution processes
(47, 53, 55-56) due to the increase in solubility of silica
at pH > 7.8. This impact on the iron (Fe) rich finegrained Fahienrock shelter sediments is limited as
indicated by pH measurements (6.5-7.3) in all the
sediment samples studied (57). Facetate and
sclereidsphytoliths from woody dicotyledonous (e.g.
forest taxa) are rare in the Late Pleistocene, possibly
due to dissolution (44), and/or less incorporation of
phytoliths from woody materials. We report that facetate
and sclereidsare were unlikely to be preserved in much
older samples (1).
b) Late Pleistocene wild rice exploitation
Late Pleistocene deposits yielded
archaeological records (Table 1) and high amounts of
phytoliths from economic plants. Phytolith records from
the wild rice species, together with number of other
economic plants (e.g. wild banana and breadfruit)
suggests that rock shelter occupants have usedwild rice
plants, most probably for food and also for various other
purposes, e.g. fuel, rituals medicines and artifacts.The
methodological constrains used for rice identification
confirm that phytoliths were from Oryza nivara and/or O.
rufipogon, but the criteria used herewith, cannot fully
separate the rufipogon, perennial from nivara annual
ecotype.
c) Ecology of wild rice
Understanding the evidence related to ecology
of wild rice provides an opportunity to explore the
relationship between human activity and the presence of
wild rice in the late Pleistocene. The ecology of the wild
rice species clearly indicates differing modes of human
exploitation of wild rice for food from the prehistoric
period in South Asia (58-61). The latter works suggest
that O. nivara, which commonly grows in drylands and
has a large-scale seed production could have been
easily used by prehistoric hunter-gatherers without any
serious cultivation whereas O. rufipogon, which is
prominently grown in aquatic habitats has a much lower
seed production during the very early stage of plant
domestication, i.e. late Pleistocene (59-60, 62-64). O.
nivara rice tends to grow in relatively small isolated
patches and not in stands of genetically uniform
populations.
Prior to the Toba volcanic event ca. 74 ka years
ago, humans at Jwalapuram, Locality 22, Southern Asia
lived in a mixed woodland and grassland mosaic. This
was followed by cooler and possibly drier conditions
after the eruption (65). The elevated δ13C and δ18O from
Batadomba-lenarock shelter (Fig. 1) faunal remains (66)
and Himalayan ice cores(67) indicate that lowland
rainforest of Southern Asia were more open with
decreased canopy cover during the period 36-29 ka.
This has been linked to decreased rainfall and
temperature (68). Prior to the Last Glacial maximum
(LGM), humid environments appear to have prevailed in
the Indian subcontinent (69-71). Paleoclimatic records
suggest that atmospheric cooling by 3-4 °C occurred in
the Tropics (72), with a remarkable drop in precipitation
during the LGM much greater than during any of the
earlier (middle Pleistocene) glaciations (66).
Palaeoclimate data from Sri Lanka suggest a drier LGM
punctuated by climatic ameliorations in short bursts (73-
74).It is obvious that climatic fluctuations that includes
prevailing prominent dry conditions in the Late
Quaternary may have resulted in a number of
climatically adapted wild rice populations (61,75-76).
Rock shelter occupants could have easily modified O.
nivara populations leading to more reliable wild grains
for human use, especially when they were already widely
growing in ideal habitats associated with prolonged dry
conditions long before the domesticated forms arose (1,
60, 77-79). During early rice exploitation, it is worth
noting that high micro charcoal, phytolith and poll
encore records indicate regular anthropogenic burning
from the Terminal (14.5-13 ka) and end of the
Pleistocene through early-middle Holocene in the
archaeological sites from Ganges plains in north India
(80-81). Several sites in the Yangzte valley in China,
dating from 17 ka through the Terminal Pleistocene
yielded O. nivara phytolithsin association with humans
(11, 82-85). Similarly, multi-proxy investigations (e.g.
pollen, phytolith, charcoal, mineral magnetics, stable
carbon and diatom) in the Horton Plains, Sri Lanka,
suggest anthropogenic burning and disturbance in
association with south west monsoon fluctuations from
17.5 ka through the late glacial time (31,35,75). In this
ecological regime, phytoliths from Oryzasppwere
reported in association with dry climate between 15.9-
13.8 ka. All those records indicate that wild rice was
present in human economies through the late
Pleistocene to the Holocene in South and East Asia.
This indicates that the rice species exploited by Fahien
rock shelter occupants (i.e. late Pleistocene huntergatherers) was more likely O. nivara than O. rufipogon,
adopting to the ecological/habitant, e.g. dry and mixed
woodland and grassland conditions prevailed in the late
Pleistocene (Fig. 4). The antiquity of human use of wild
rice species, O. nivara at Fahien is remarkably as early
as 48 ka, compared to the tradition of rice foraging in
known Asian sites (77,86).
VI. Conclusion : Little is known of the use of wild rice in
prehistoric Sri Lanka. Investigations from the archaeological sequence at Fahien rock shelter in south
western Sri Lanka, dated to 47.80-3.87 ka provide
phytolith evidence suggesting the use of wild rice, most
possibly Oryza cf.nivara, with several other wild plant
resources, e.g. banana, breadfruitand a number of
species from Palmae. The rock shelter provides the
oldest evidence for the wild rice associated late
Pleistocene human rainforest occupation among the
archaeological sites in Southeast Asia, Melanesia and
South Asia
Read Full Paper
Read Full Paper
About Author : Rathnasiri Premathilake
Postgraduate Institute of Archaeology,
University of Kelaniya, 407, Baudhaloka Mawatha, Colombo 7, Sri Lanka
