The early Austronesians began their ethnolinguistic career as subtropical coastal and riverine peoples with a Neolithic economy based on cereal and tuber cultivation and a set of domesticated animals. Their ethnographic descendants in island Southeast Asia managed to create for themselves a much wider range of subsistence economies, including rainforest foraging and collection-for-trade; sea nomadism (see Sather, this volume); varied forms of both irrigated and rainfed rice cultivation; shifting cultivation of cereals, fruits and tubers; and even palm exploitation (Fox 1977). Underlying causes for specific transformations can only be hypothesized, but such hypotheses will not proceed very far unless it is recognized that the early An colonists and their descendants must have been influenced by two kinds of landscape, the environmental and the human.

In terms of environmental factors, the Austronesian expansion would never have begun at all had its early participants been unable to cross the gaps of about 120 km between China and Taiwan, and between Taiwan and the Batan Islands. Zorc (1994) has recently pointed out how difficult it is to reconstruct terms for boats and navigation for Proto-Austronesian (as opposed to Proto-Malayo-Polynesian) and he suggests that much An seafaring terminology might have been lost in Taiwan owing to the heavy overlay of Chinese settlement in coastal regions. If so, the Proto-Austronesian vocabulary would once have contained more seafaring terms than it does now. Therefore, the obvious place to seek archaeological evidence for innovations in boat construction and navigation^[4] may be amongst the hundreds of small islands which flank the coasts of Zhejiang and Fujian Provinces. To my knowledge these islands have produced few archaeological remains, apart from the fifth millennium BC pottery assemblage from the Fuguodun shellmound on Quemoy (Chang 1977; see also Chang 1992 for a recent update). However, if the Austronesians ever required a maritime “nursery”, it might have been here.

A second environmental factor which affected the results of An expansion would have been the gradual shift from a seasonal sub-tropical climate into an ever-wet equatorial one as colonists moved southwards from Taiwan and the Philippines towards the Equator. Eventually, of course, the settlement of Java and southeastern Indonesia placed Austronesians back into a zone of markedly seasonal climate. But in the equatorial islands proper — Sumatra, Borneo, central Sulawesi, Halmahera — the prevailing climatic conditions probably promoted an economic reliance on tubers and fruits rather than the major cereals, rice and millet. I have argued this before in several publications (e.g. Bellwood 1980a, 1985) and see no reason to change my views, which do offer a convenient explanation as to why cereals should have been absent in Oceanic economies beyond western Micronesia; the northern New Guinea passageway into Oceania is completely equatorial.

Concerning the pre-existing human landscape of what is now western Austronesia an interesting and varied pattern can be reconstructed. Most of Island Southeast Asia was probably inhabited by foragers in pre-Austronesian times, and in the rainforests of Sundaland there probably would have been only very sparse settlement, mostly focused in coastal regions, many of which are now of course beneath sea level as a result of the postglacial ice melt (Bellwood 1990a). Denser populations doubtless existed in the more seasonal environments of Java, the Lesser Sundas and the central and northern Philippines, as witnessed in the Philippines by the existence of Agta foragers of a Melanesian-related physical type (albeit now An-speaking; Reid 1990) to the present. Indeed, in order to settle the Wallacean islands (Philippines, Sulawesi, Lesser Sundas, Moluccas), as also Australia and New Guinea, the original Pleistocene colonists must have had some degree of seafaring capacity, even if rudimentary, by at least 40,000 years ago. Did the Austronesians learn a number of seafaring skills from them, together with perhaps the uses of a number of equatorial crop plants such as breadfruit and coconut, which first appear in An linguistic history in the PMP vocabulary? I have already referred to Zorc’s observation that rather little seafaring terminology is reconstructible for Proto-Austronesian. While it would be unwise to deny Proto-Austronesians the knowledge at least of canoes, it is worth remembering that much of the early expansion of the Austronesian-speaking peoples was through Wallacea, especially the Philippines and Sulawesi with their manifold satellite islands. It is amongst the more watery topography of Late Pleistocene Wallacea, rather than land-bridged Sundaland, that one might expect pre-Austronesian maritime traditions to have flourished and to have been transmitted to later arrivals (Irwin 1992).

Outside the islands of Southeast Asia the early Austronesians met with far stronger cultural resistance, both to the west and east. A different world again of course awaited colonists in the Pacific Islands beyond the Solomons — empty and inviting, with no attested pre-Austronesian populations at all (see following chapter).

The archaeological record now makes it clear that Neolithic populations occupied the whole region of Mainland Southeast Asia, including Peninsular Malaysia, by at least 2000 BC (Higham 1989; Bellwood 1993). The relevant archaeological affinities of all these cultures lie within the mainland region itself rather than across to the islands; perhaps it can be hypothesized that the majority of the agricultural populations of Thailand, Indochina and Peninsular Malaysia at this time spoke languages related most closely to the modern Austroasiatic family. In southern Vietnam the ancestors of the Austronesian Chamic-speaking peoples probably had to intrude into a landscape already peopled quite densely by such pre-existing cultivators. The Austronesian expansion into Peninsular Malaysia surely occurred long after the expansion of Austroasiatic cultivators (early speakers of Aslian languages) southwards from Thailand during the third millennium BC. Indeed, Austroasiatic-speaking cultivators might originally have colonized onwards quite rapidly into parts of Sumatra and western Borneo if one takes a broad view of the pollen evidence^[5] and some linguistic substratum hints (see Adelaar, this volume). The pre-existence of such agricultural populations in at least Malaya and Vietnam can explain why the mainland of Southeast Asia was only a region of small-scale prehistoric settlement by Austronesians, just as, under rather different cultural circumstances, was New Guinea.

As on the mainland of Southeast Asia, so in New Guinea and western Melanesia the Austronesian colonists also met a major level of resistance — biological, cultural and linguistic. The existence of primary agricultural development in New Guinea is no longer in doubt, even if it is unclear where it was focused (highlands or coasts?) and exactly how the participating economies functioned (Golson and Gardner 1990). Cultivation of plants, especially taro, is likely, but many coastal populations might have lifted their population densities to outsider-resistant levels by other forms of tree exploitation (sago and Canarium) or exchange of foodstuffs. Many of these peoples, like their Wallacean cousins, also knew how to cross sea, as witnessed by the Late Pleistocene discovery and distribution of New Britain obsidian (Allen et al. 1989). However, their seafaring skills were perhaps more limited than those of the later Austronesian settlers who, at about 1000 BC, were able to colonize far into western Polynesia and to trade Talasea obsidian from New Britain across a distance of 6500 kilometres from northern Borneo in the west to Fiji in the east (Bellwood 1989). This probably makes Talasea obsidian the furthest-distributed commodity of the whole Neolithic world.