Partner in the Padi Fields: The Water-Buffalo

Water-buffalo are so closely associated with wet rice cultivation that it is difficult to see how an efficient wet rice (sawah) economy could function without them. Their broad splaying hooves spread their weight out in swampy ground and they plod through the soft ricefields without sinking in as cattle would, hauling ploughs behind them and at the same time puddling the soil. If they need to spend long hours soaking in ponds or streams, that is a small price to pay for their services.

Asian buffaloes (Bubalus arnee), of which the water-buffalo is the domestic form, are restricted to floodplain and deltaic regions. Genuinely wild representatives still occur in Assam, especially along the Brahmaputra River; in the Mahanadi Delta extending inland to Bastar district; and on the borders of Nepal (Map 1). Until the turn of the century they also lived in the Sunderbans of Bengal, and the Mughal Emperors hunted them in the Indus Valley. Wild-living buffaloes in Sri Lanka, on the upper Chindwin, in the Chao Phraya Valley of Thailand and (until at least the 1920s) in the Irrawaddy Delta of Burma are also probably truly wild, although it remains possible that they may be feral. Those of Vietnam, Cambodia, Lampung, the Miri River in Sarawak and the Baluran National Park in Java are almost certainly feral; a wild buffalo was present in Java up to the early Holocene (remains occur in the Sampung Cave) but has since become locally extinct.

Map 1. The distribution, within the past century, of wild Asian buffalo (Bubalus arnee).

Map 1. The distribution, within the past century, of wild Asian buffalo (Bubalus arnee).

The distributional areas marked represent Sri Lanka, the Bastar region, Mahanadi Delta, southeast Nepal, Sunderbans, Brahmaputra Valley, upper Chindwin, Irrawaddy Delta, and Chao Phraya Valley. Evidence also suggests that wild buffalo existed in southern China within the last millennium.

Names available for subspecies of the wild Asian buffalo, if they should prove distinguishable, are:

Bubalus arnee arnee — Bengal
Bubalus arnee fulvus — Upper Assam
Bubalus arnee septentrionalis — Sunderbans
Bubalus arnee migona — Yala, Sri Lanka

Other populations which may be distinct are so far unnamed.

The name available for the domestic buffalo (water-buffalo) is Bubalus bubalis. It is not really, of course, a different species from the wild buffalo, but for a variety of reasons it seems useful to maintain the fiction that domestic species are different from their wild relatives/ancestors (Corbet and Clutton-Brock 1984). There are two general breed-groups of the domestic buffalo: swamp and river buffaloes (Mason 1974a). Swamp buffaloes are bred in Southeast Asia and China, northeastern India and also Sri Lanka. They are heavily built, with simple crescentic horns, and are grey with one or two white stripes on the throat, and white legs below the knees and hocks. They are indispensable for ploughing and other traction and their meat is eaten, but they give little or no milk.

River buffaloes, typical for the Indian subcontinent and parts of the Middle East and Europe, are longer-bodied and longer-legged than swamp buffaloes. The sacrum is more prominent than the withers, the horns curve back from a strongly convex forehead and are often tightly curled, and the colour is black, without white markings. There are numerous other consistent differences in conformation and in the skeleton between the two breed-groups; whether they truly prefer to bathe in swamps and rivers respectively I could not say. They also differ in chromosome number: swamp buffaloes have 48 chromosomes, river buffaloes 50 (Mason 1974b).

Swamp buffaloes differ little from each other wherever they live. Those of Sumba have exceptionally long, outswept horns. In Tanah Toraja, Sulawesi, they are very large and often piebald. In Thailand, Yunnan and South Sulawesi there are high frequencies of albinism. However, there are no true breeds. On the other hand, river buffaloes have given rise to a number of highly specialized breeds such as the Murrah, which is an excellent milker, and they have been exported to Egypt, Brazil and the West Indies. They are also replacing swamp buffaloes in parts of Thailand, Malaysia and the Philippines. The replacement by river buffaloes of the more primitive swamp type seems to have been going on for a long time; we not only have the oddity of the swamp type surviving in a peripheral region such as Sri Lanka, but the Toda buffaloes of the Nilgiri Hills are of swamp type, separated from the swamp buffaloes of Southeast Asia by most of the Indian subcontinent where only river buffaloes are used. Prehistoric depictions of domestic buffaloes with the characteristic crescentic horns of the swamp form are known from Mohenjo-Daro and from Ur, dating from the mid-third to early second millennia BC.

The wild Indian buffalo resembles an enormous version of the domestic swamp buffalo and is surely its direct ancestor, although its chromosomes are unknown. Mean skull lengths for different wild populations are as follows (in millimetres, followed by sample size in brackets):





603 (13)

585 (12)

Bastar, Orissa

574 (2)

555 (3)


556 (5)

545 (8)


557 (4)


Sri Lanka

539 (6)


In comparison, domestic buffaloes have skull lengths around 450-500 mm, the river breeds tending to be smaller than most swamp buffaloes.

In order to get some idea of which wild population most resembled the domestic ones in skull form, and so forms the most suitable candidate for their ancestor (always assuming that river and swamp types do have a common domestic ancestor), I have undertaken a discriminant analysis on craniometric variables. The variables used were greatest skull length, biorbital breadth, postorbital breadth, occipital breadth (greatest), occipital breadth (constriction), breadth of horn base, nasal breadth posterior, nasal breadth anterior, nasal length, and basal skull length. It should be explained that not all measurements were available for every skull.

The results are shown in Figure 1. The first discriminant function (horizontal) accounts for 67.5 per cent of the total variance and is in part at least dependent on size, but also contrasts wide nasal tip with narrow nasal base and emphasizes relatively slender horn bases. The second function, which accounts for 14.6 per cent of total variance, contrasts a broad occipital constriction and broad nasals with a narrow postorbital constriction and short nasals. No other function accounted for more than 8 per cent of the total variance.

In Figure 1, means and one-standard-deviation circles have been plotted for all geographic samples, and individual specimens from other regions have been plotted separately. Assam, Nepal and Thailand are well separated from Bihar and the domestic samples, with individual specimens from central India (Bastar) and Sri Lanka falling between. The fact that the Bihar/Orissa sample (centring on the Mahanadi delta) is the only wild one whose dispersion widely overlaps that of the domestic samples suggests that, if skull form is any guide, this is the best bet for a wild ancestor. Did this important component of the wetrice complex come from that region of India? This would be surprising, given that this is well outside the Austronesian area and that the oldest putative domestic buffaloes come from Neolithic sites in southern China, although northeastern India is within the Austroasiatic (Munda-speaking) area. In what follows, we will see whether there is any analogy for such a distribution and inferred place of origin.

Figure 1. Discriminant analysis of cranial measurements in Asian buffaloes. The circles represent one-standard-deviation limits of samples from Assam, Thailand, Nepal and Bihar/Orissa (wild) and swamp and river samples (domestic).

Figure 1. Discriminant analysis of cranial measurements in Asian buffaloes. The circles represent one-standard-deviation limits of samples from Assam, Thailand, Nepal and Bihar/Orissa (wild) and swamp and river samples (domestic).