The Spread of Agriculture to Southwest China

A review of Adaptation and Invention During the Spread of Agriculture to Southwest China, by Jade D’Alpoim Guedes.

The research presented in this dissertation focuses on the ways in which humans adapt to novel environment and how they modify agricultural strategies to suit ecological niches not previously amenable to agricultural production.

To address these issues, this research seeks to explain why it took more than 3000 years for agricultural food production to spread from the Middle and Lower Yangzi River valley to the foothills and highlands of Southwest China. This study therefore explores of the details – the ecological, technological, and social variables – that made agricultural food production possible in some settings but utterly impractical in others, and how the nature of these possibilities changed through time.

Methodologically and theoretically, the author frames this approach in the logic of decisions about subsistence strategies, specifically the evolutionary end economic logic of human behavioral ecology (specifically the relationship between agricultural risk and human survival). To establish the cost structure behind these decisions, the author uses ecological niche modeling. This practice makes clear the differences between the costs and constraints of agricultural food production in different areas using different crops. In particular, the author pays considerable attention to differences in the nature of agricultural subsistence between lowlands and highlands, and how the ecological attributes of different areas either promoted or prevented the use of different crops. The model is considerably more complex than this short explanation, but the details of which are far too numerous and nuanced to recount in a single review.

The method applied in this dissertation focuses on the insights generated through constraints on plant phenology, specifically the cumulative heat requirements of plant growth, much of which the author establishes with data collected through controlled agronomic experimentation from around the world. The resulting models of spatial variation in growing potential provide insights into the barriers and the possibilities of prehistoric food production, and illustrate how both external (e.g. hemispheric climate change) and internal change (e.g. habitat alteration or technological innovation) in the controlling variables might promote or constrain the diffusion of agricultural products by altering the economic risks associated with producing them, and therefore the decision to adopt them or not. The author argues that it is “necessary to construct models…capable of dealing with the constraints faced by ancient agriculturalists if we are to understand how and why humans adapted and invested in agricultural systems during the spread of agriculture” (p. 37). The models of agricultural potential are then compared with detailed archaeobotanical data on plant use throughout the spread zone.

The modeling process Guedes employs focuses on the constraints of practicing agriculture (specifically plant-based agriculture) in novel environments. To do this she uses “ecological niche modeling” to evaluate the productive capacity of different settings, and asks whether or not these areas possess the requisite attributes for stable or predictable growth and maturation of specific plant taxa. Questions about stability and predictability are framed in the context of risk, meaning inter-annual variation in the expected productivity of the harvest, which is assessed using different measures of climate. Utility (or rather, the economic feasibility) of plant-based agriculture is further assessed by an examination of the labor investments required for plant cultivation in different settings.

This approach to ecological niche modeling evaluates the relationship between climate variables and species distributions, and in so doing harnesses a body of literature that explores the effects of past and present climate change on range shifts in different taxa. The author is careful to distinguish between correlative and mechanistic models for understanding species distributions. While the former may be useful for understanding extant taxa in contemporary settings, she argues this method is generally unsuitable to studies of prehistoric agriculture because the archives of the past are simply too spotty to correlate with prehistoric environmental variables (which are also spotty) for any predictive power. Instead the author focuses on mechanistic models tuned to an understanding of the factors (such as temperature, precipitation, frost injury, and competition) that limit or promote the physiological dimensions of plant growth, survival, and reproduction. From such an understanding, the modeler establishes the “fundamental niche” for the existence of the plant (following G.E. Hutchinson’s 1957 notion of an n-dimensional hypervolume that provides for the existence of any organism), which can be altered by processes such as human niche construction to create a “realized niche” which is then used to illustrate the potential distribution of the agro-economic species.

Though the author does admit that the mechanistic approach to ecological niche modeling has its complications, it may be the only way to model the prehistoric distributions of agriculturally feasible ecosystems. As more is learned about the phenology of ancient domesticates, and as the resolution of the paleoclimate reconstructions increases and improves, it’s possible that some of the fundamental niches established for the plants in this study, and therefore the hypothetical spatial distribution of their cultivation, might change. But as a means for marshaling current data, and for generating archaeologically testable hypotheses, this approach is both valuable, and valorous.

In this study of agricultural feasibility in southwest China the mechanistic modeling focuses on constructing a thermal niche, based primarily on an ecological metric known as the Growing Degree Day (GDD) itself a function of the minimum and maximum temperatures under which an organism will grow, and a cumulative account of mean daily temperatures over any duration relevant to the growth of that organism in a given location. The GDD is a measure of heat accumulation sometimes used to establish rates of plant growth and development.

If an annual account of maximum and minimum daily temperatures are available for a given area, and the internal constrains of the plant are known (or experimentally derived), one can establish the GDD for it in a place where prehistoric use is purported. In this study, calculation of GDD for individual plant species is based solely on contemporary weather data interpolated over a three-dimensional landscape. This interpolated surface was then combined with a simple method for evaluating “risk” which looked at the probability of crop failure, based on the annual GDD for each crop, in each area, again calculated from contemporary weather data and interpolated and draped over a three-dimensional landscape model. The author rightly notes that in the absence of daily temperature data from prehistoric settings, or models capable of retrodicting them, the data collected from the mid to late 20th are sufficient for a first-order analysis, and that’s exactly what this is.

To test the modeled species distributions based on the mechanistically modeled thermal niches for each taxon, the author collected prehistoric plant remains from archaeological sites in different settings, and at vastly different elevations, throughout southwest China. The following is an outline of the conclusions derived from a comparison of the archaeobotanical data to the expectations of the ecological niche modeling.

While most research in southern China has focused on the origins and diffusion of rice-based agriculture, this research suggests that rice AND millet were essential features of the spread of agricultural subsistence to southwest China. This leads both the author, and the reader, to wonder about the role of millets in the spread of the essentially rice-based agricultural program to Southeast Asia more generally.

The first form of agriculture to appear in southwest China was based on millets. The relative importance of each type of millet is, in part determined by the constraints of temperature and precipitation on growth, with foxtail millets tolerant to a wider range of higher elevation habitats. For the most part, the relative importance of foxtail (in the archaeological record) corresponds to the GIS model of risk-assessment, with foxtail millet dominating in colder, high altitude environments where the likelihood of crop failure is less than that of broomcorn (which does better at lower elevation).

Though the appearance of millet agriculture is contemporaneous with the appearance of the Majiayao Neolithic ceramic tradition in western Sichuan, ca. 5500 BP (a tradition usually linked to intensive, sedentary agriculturalists), the author suggests that millets may have been added to the subsistence economy of pre-existing hunter-gatherers in the region because the short growing season of these summer grasses may have been easily incorporated into a residentially mobile seasonal cycle organized around the procurement of spatially segregated wild resources.  In suggesting this, she echoes previous assertions about the early adoption of millets by hunter-gatherers further north (e.g. L. Barton, C. Morgan and R. L. Bettinger, “Harvests for the hunters: the origins of food production in arid northern China.” The SAA Archaeological Record 9 (2009), pp. 28-31; idem, “The origins of food production in North China: a different kind of agricultural revolution.” Evolutionary Anthropology, p. 19, pp. 9-21). Both the seasonality of this pattern and the nature of human mobility throughout the region are ripe for future research, as is the nature of the contact (co-existence, competition) between foragers and farmers, from the lowlands of Sichuan to the highlands of the Tibetan Plateau. Furthermore, and perhaps for similar reasons, millets were likely the first agricultural package to move into highland Yunnan and Guizhou and ultimately into mainland Southeast Asia (though the author admits that these final assertions are based more on the implication of the models, than on current archaeobotanical data).

Though millets constitute the earliest archaeological evidence for agriculture in some places (e.g. the Three Gorges area), the earliest evidence for agricultural subsistence on the Chengdu Plain is rice. While rice may have been domesticated as early as 8000 BP in the middle Yangzi drainage, rice-based agriculture did not take hold in southwest China until after 5000 BP. The earliest evidence for this is at Baodun, and the rice appearing here is fully domesticated (suggesting the arrival of a well-developed agricultural economy). Importantly, the author contends that the spread of rice to the Chengdu plain was delayed until the evolution of a phenotype adapted to cooler conditions. This assertion stems both from the expectations of the ecological niche modeling (which suggests that the region has the requisite “growing degree days” to support production of temperate rice varieties), but also from the recovery of charred rice remains with measurements consistent with those of temperate varies of rice found in contemporary and prehistoric northern China.

The importance of millet goes beyond its early appearance, and its potential relevance to mobile hunter-gatherers. Foxtail millet (in particular) is far better suited to cold, arid conditions, grows much faster, and demands less labor than does rice. The author suggests these characteristics may have enabled lowland rice farmers (or mixed rice-millet farmers) to expand into the uplands of southwest China, namely the foothills of the Three Gorges, Yunnan, and Guizhou. Furthermore, she links the mixed rice-and-millet strategy to increasing annual yields, reduced agricultural risk, human population growth, and increasing social complexity in the lowlands of southwest China.

Archaeobotany suggests that the prehistoric subsistence strategies in the Chengdu Plain were relatively stable, but the archaeological records of the highlands (e.g. Yunnan-Guizhou) are marked by change, perhaps a result of the recurrent adjustments and modifications required to meet the amplitude of environmental volatility in higher elevations and the effect it had on the availability and productivity of food. In such places, the relative importance of the two millets, rice, wild plants, and eventually wheat and barley fluctuated in response to change in local environments, and the caloric demands of the human groups inhabiting them.

In the long haul, while millets, which are more tolerant of water stress and low temperature than rice, may have enabled the early expansion of agricultural subsistence into upland areas, it was barley (and to a lesser degree, wheat), with its short growing season, low requirements of GDD and high tolerance of low temperature and frost tolerance that enabled agricultural expansion, and relative stability, in the highlands of Tibet. This is to say that the ecological niche modeling suggests that wheat and barley were better suited to the highlands of southwest China than were millets (and certainly rice). Both the phenology of the plants, and the ecological niches in which people attempted to grow them dictated where and when people could adopt them. And the archaeobotanical record tentatively supports this.

The bread-and-butter of archaeological investigations hinge on observations about change in the spatial and temporal distribution of scant remains, which are then employed to illustrate change in the behavior of people. Archaeologists differ on the causes of change, and the past century and a half has seen considerable variation in the popularity of using mass migration or the diffusion of ideas as explanations for change itself. This tradition of scholarship ranges from one extreme that refuses to see change as a function of either (migration or diffusion) in favor of in-situ developments, to another extreme that explains most everything we see in the world as a product arriving (in one way or another) from some glorious epicenter of origin (imagine the “Garden of Eden”). Both extremes are preposterous. This Dissertation represents a concerted effort to understand the processes of change with a novel approach to modeling possibilities, and by tuning the modeling effort to the ever-expanding catalog of carefully collected archaeological data. It is a first-rate exploration of how the practice and products of agriculture diversified and evolved in unlikely settings. This careful, modeled approach, backed by meticulous collection of archaeological data, may prove profitable in other parts of the world where agricultural subsistence appeared considerably later than it did in neighboring regions.

Loukas Barton, PhD
Department of Anthropology
University of Pittsburgh

Primary Sources

Primary archaeobotanical data from Southwest China

Dissertation Information
Department of Anthropology, Harvard University, Cambridge, MA. 558 pp. March 2013. Primary Advisors: Rowan K. Flad, Richard Meadow.

Image: Photograph by Jade D’Alpoim Guedes, with millet.

Leave a Reply