Small is big: the microfossil perspective on human-plant interaction.

T he archaeological record contains only scattered and incomplete clues to the scope and complexity of past human behavior, and archaeologists must develop every possible source of useful information. Although much overused of late, the truism that absence of evidence is not evidence of absence holds particular force in archaeology and nowhere more than in the prehistory of human–plant interaction. Until recently, the reconstruction of plant use and domestication has been biased by a necessary dependence on macrobotanical remains preserved in arid environments or through serendipitous carbonization. Microbotanical analyses have proven increasingly valuable in balancing the record. Over the last few decades, phytolith (plant opal silica bodies) and starch grain studies have contributed significantly to the multiproxy reconstruction of past environments, human use of plants, and the pathways of plant domestication (e.g., refs. 1 and 2). This observation is especially true in humid regions, where conditions do not favor the preservation of macrobotanical remains, and in areas where tubers were important sources of plant foods; tubers are soft structures that often are consumed in their entirety, and they tend to leave few macro remains even under optimal preservation conditions. Working in the humid tropics of Central America, Dickau et al. (3) demonstrate in this issue of PNAS the power of archaeological starch grain analysis by using this proxy to elucidate the differential movement and adoption of both seed and root crops in western and central Panama beginning 7,800 years ago. Although starch grains and phytoliths have been recognized by scientists for nearly two centuries, the systematic study of these plant microfossils for archaeological purposes dates only to the last three decades (1, 4). Increasingly, archaeobotanists use phytoliths and/or starch grains to identify plants of origin at low taxonomic scale, to distinguish between domesticated and wild species, to differentiate between plant organs producing the microfossils, and to associate plants directly with human activity by recovering starch and phytoliths from artifacts and even human teeth. Recent research by myself and colleagues (5) at the site of Waynuna, in the southern Peruvian highlands, illustrates all of these data types. Although we excavated the site to study Terminal Pleistocene [13,000–11,400 calibrated years before present (cal BP)] coast–highland interaction, Waynuna proved to be Late Preceramic (4,000–3,600 cal BP) in age and thus unsuitable to answer the original research questions. Microfossil analyses turned this ‘‘failure’’ into a source of important new information on the history of Andean plant use. Although the excavations recovered no plant parts visible to the naked eye, analysis of starch grains from midden sediments and stone tools demonstrated the regional presence and processing of domesticated seed (maize, Zea mays) and tuber (potato, Solanum sp.; arrowroot, Maranta sp.) crops a millennium earlier than previous records based on macrobotanical remains had shown (5). Phytoliths provided independent confirmation of the starch identifications. The discovery of both cob and leaf phytoliths in the principal house floor deposit showed that these organs had deteriorated on or near the site and therefore supported an inference that the site’s inhabitants farmed the corn. Detailed examination of the starch grains found that both hard and soft endosperm varieties of maize were present and revealed damage to maize grains typical of grinding rather than natural deterioration. Phytoliths and starch grains have led to the identification of early agriculture elsewhere in the New World. On the Ecuadorian coast, Piperno and Stothert (6) found phytoliths of wild squashes and gourds (Cucurbita spp.) at one site dating to the Terminal Pleistocene and domesticated Cucurbita phytoliths dating to the earliest Holocene at the same site and one other. Mean size distinguished wild from domesticated specimens; these results are supported by studies of modern wild and domesticated squash phytoliths. In these sites, phytoliths provided an additional avenue of information as a source of datable organic material. As they form, phytoliths incorporate bits of the parent plant material, which is then preserved from contamination by the silica envelope of the phytolith. With a large enough sample, accelerator mass spectrometer radiocarbon dates can be successfully attained directly from the microfossils. Working in mid-Holocene sites of the Ecuadorian coastal zone, Pearsall et al. (7) have amplified our understanding of early maize types at Real Alto, a site dating to almost 6,000 years ago. Starch grains indicated both hard and soft endosperm varieties, as at Waynuna a millennium later, whereas phytolith morphology pointed to maize that ‘‘combined primitive and derived traits in a way that is not observed in living maize varieties’’ (ref. 7, p. 438). Recent microfossil studies at either extreme of the American continents have shed light on different human–plant interactions. Zarrillo and Kooyman (8) used starch grains from two late pre-European stone tools to document long-distance movement of foodstuffs such as maize in the Canadian plains. In Uruguay, Iriarte et al. (9) analyzed both starch grains and phytoliths from the 5,000to 4,000-yearold site of Los Ajos, finding a number of different crop plants such as maize and beans (Phaseolus sp.) (9). In the context of regional prehistory, the site is unexpectedly complex in architecture and, as the microfossils demonstrate, unexpectedly early in the adoption of agriculture. Macrobotanical remains of domesticated plants were not recovered at the site. The new data push back the frontier of agriculture and settled village lifeways in the southern cone of South America. Plant microfossils have led to similar advances in other parts of the world. Recent research by Denham et al. (10) at Kuk Swamp in New Guinea used starch grains from tools and phytoliths from site sediments as key parts of a multiproxy effort to track plant use over the Holocene (11,400 cal BP to present). Bananas (Musa sp.) and the tuber taro (Colocasia esculenta) were intensively exploited by 10,000 cal BP, and the domestication and cultivation of banana had occurred by 7,000 years ago. These results challenge the earlier conception of New Guinea as a passive recipient of domesticated plants and the idea of agriculture brought by Austronesian speakers rather than an independent center of plant domestication. At Ohala II in Israel, a well dated Upper Paleolithic site ( 23,500–22,500 cal BP), Piperno et al. (11) showed that wild barley seeds were being processed with a stone