[Following a portion of: Fuller, D.Q.(in press). "Non-human genetics,
agricultural origins and historical linguistics in South Asia" in The
Evolution and History of Human Population in South Asia, edited by M.
Petraglia and B. Allchin, Springer Press, Doetinchen, Netherlands)

On the Origins and Spread of Rice

Rice (Oryza sativa) is one of the most utilized crops of the world
today, but the complexities
of its early history remains largely unraveled. Rice is now
cultivated in a wide range of habitats
from temperate northern China and Korea to the eutropical areas of
Indonesia. It is grown as
broadcast sown crops on hillsides, often as part of extensive slash-
and-burn systems, and it is grown
in highly labor intensive, flooded 'paddy' lands in which seedlings
grown in one paddy are dug up
and individually replanted into another field. The assumption, which
is widespread in the literature,
that all Asian rice derived from a single domestication, somewhere in
the wild rice belt from eastern
India across northern Indo-China or South China (e.g. Chang 1995,
2000), has been based more on
the presumption of single origins for crops in general, coupled with
problematic archaeological
inferences. Starting with the assumption that rice was domesticated
once, there have been some
5
rather extreme attempts to relate East Asian and South Asian
archaeology, such as via comparisons
between Neolithic China (sixth through fourth millennium BC) and
Neolithic Kashmir (2500-1000
BC) (e.g. Van Driem 1998), even though the latter had agriculture
based on Near Eastern crops
(wheat, barley, lentils and peas) and not rice! More recently,
Kharakwal et al.'s (2004) attempt to
link cord-impressed ceramics with rice agriculture suggests hyper-
diffusionism based on superficial
similarities in ceramics, including the Jomon of Japan (which is non-
agricultural), parts of Neolithic
China of the early to mid-Holocene, and much later 4th to 2nd
millennium BC material from the
Ganges. All such hyperdiffusionist studies are flawed, not only
because they stretch archaeological
logic by drawing comparisons across such vast areas and time-spans,
but most importantly because
they fail to take into account what we already know from botany about
rice origins. Historical
linguists have been mistaken in trying to make sense of a vast array
of potential rice words on the
assumption of a single centre of rice origin from which such words
ought to originate (e.g. Mahdi
1998, Pejros and Snirelman 1998, Witzel 1999: 30-33). Less explicitly
reasoned attempts to link all
of South and East Asian rice into a single story, are the grand
narratives linking agriculture and
language spread, in which the spread of rice from the middle Yangzi
to India with demographically
expanding and migrating farmers is argued largely on the basis of
model assumptions rather than
archaeological evidence (e.g. Higham and Glover 1996, Bellwood 1996,
2005). Any attempt to
make a single narrative about Asian rice is already falsified by
phylogenetic evidence from rice
itself.
Asian rice, despite being lumped under the species name, Oryza sativa
(a Linnaean
convention in use since the 1750s), is composed of two distinct
phylogenetic species, indica and
japonica. This has long been suggested by plant breeding research, in
which hybridization between
these two cultivars is found to be difficult and imperfect, with the
majority of crosses between
indica and japonica cultivars being wholly or partly sterile (Wan and
Ikehashi 1997). As a result,
the botanical literature has had a persistent debate between
hypotheses of rapid divergence after a
single origin or two domestications (Chang 1989, 1995,Oka 1988, White
1989, Thompson 1996),
although it is the single origin that has tended to be assumed in
archaeological syntheses (e.g.
Glover and Higham 1996, Bellwood 1996, 2005, Bellwood and Diamond
2003, Higham 1998),
perhaps largely due to the influence of T. T. Chang (1989, 1995,
2000). There now is substantial
evidence for genetic distinctions between indica and japonica from a
range of data (Sato et al.
1990, Sano and Morishima 1992, Chen et al 1993a, 1993b, Cheng et al.
2003, Sato 2002). Most
significant is genetic evidence from the chloroplast (a plant
organelle like the mitochondria
inherited maternally) and nuclear DNA variants called SINEs. A
sequence deletion in the chloropast
DNA of indica cultivars links them with wild annual "O. rufipogon"
(i.e. O. nivara in the taxonomy
used here) (Chen et al. 1993a, 1993b, Cheng et al. 2003, for current
rice taxonomy see Vaughan
1989, 1994). Meanwhile, there are some seven SINEs that separate the
nivara-indica group from
the rufipogon-japonica (Cheng et al. 2003). Figure 1 shows the
phylogenetic model produced by
Cheng et al. (2003), in which the japonica cultivars form a very
tight group in relation to the
dispersed groupings of wild rufipogon types. By contrast, the
grouping of indica is looser and more
interspersed with wild nivara. This contrast might even suggest that
indica is composed of more
than one domestication event from wild nivara populations. On the
basis of the modern geography
of wild forms and cultivars at least one of these indica
domestications in likely to have occurred in
northern or eastern South Asia (Figure 2), while the japonica
domestication can be placed in
Southern China, probably the Yangzi basin.
The available archaeological evidence also suggests two distinct
centres of early rice
cultivation. In China, despite continuing controversies about the
antiquity of rice use, cultivation,
and domestication, it is widely accepted that rice cultivation was
underway in the Middle Yangzi,
and adjacent South China by the sixth millennium BC (e.g. Yan 2002,
Crawford and Shen 1998, Lu
1999, 2006, Cohen 2002, Crawford 2006). While rice spreads down the
Yangzi river and
northwards into parts of central China, and probably the Shandong
peninsula during this early
period, archaeological evidence from further north, south or the
upper Yangzi post-dates 3000 BC
6
(see Figure 2). In India, rice cultivation is quite widespread by ca.
2500 BC from the eastern
Harappan zone in the upper Ganges basin (e.g. At Kunal: Saraswat and
Pokharia 2003) and the
Swat valley in northern Pakistan (Costantini 1987) through the middle
Ganges (see Fuller 2002,
2003c, Saraswat 2004a, 2005). A few sites with evidence for rice
impressions in pottery (not
necessarily domesticated) date back to the fourth millennium BC
(Kunjhun II and Chopanimando),
while recent excavations at Lahuradewa have been suggested to put
rice cultivation back to as early
as ca. 7000 BC, based on an AMS on a piece of a charred mass of rice
(Saraswat 2004c, 2005, I.
Singh 2005, Tewari et al. 2003, 2005). It must be cautioned, however,
that criteria for recognizing
domesticated rice as opposed to wild gathered rice remains weak and
unsubstantiated, and the
presence of cultivation practices is unclear. The sample size is very
small, with less than a dozen
grains recovered from the first season of work. While further
research is needed, the recent
evidence from Lahuradewa indicates at the very least that foragers
were exploiting (wild) rice in the
Ganges plain from ca. 7000 BC and perhaps already producing some
ceramics at this date (and
undoubtedly by the Fourth Millennium BC) (cf. Tewari et al. 2005,
Saraswat 2005, I. Singh 2005).
Sometime after this cultivation began and selection for domesticated
rice, which may have taken
one or two millennia, had taken place by 3000-2500 BC (see
below, `The Ganges Neolithic'). It is
after this time when rice had spread towards the northwest in the
first half of the third millennium
BC, indicated by finds at Early Harappan Kunal and at Ghaleghay (see
Figure 2). Whether early
rice cultivation in Eastern India (e.g. Orissa) should be seen as
dispersal from this same centre or a
separate process, perhaps rather later, requires further
archaeobotanical investigation (see below,
`The Eastern Neolithic').