A Cross-Cultural Historical Analysis of Subsistence Change

This paper reports on a comparative study of changes in subsistence patterns in small-scale communities resulting from contact with global economic forces. The local communities are represented by 87 Standard Cross-Cultural Sample Societies ("Standard Sample") (Mur-dock and White 1969) distributed across four major world geographic areas. This study describes the patterns and general characteristics of the societies on a new set of coded variables representing change processes. The variable set includes agricultural and nonagricultural intensification, the addition of new crops and animals, changes in settlement patterns and expansion, catastrophic loss, changes in trade and wage labor, and the date of ethnographic observation. The findings of the study are both method-ological and substantive. The method-ological results focus on sampling prob-lems for comparative studies as well as the dates of observation set for sample societies. The substantive findings indicate that variables are regionally clustered, with two basic patterns emerging for the way subsistence systems are affected and changed in different world regions. The research reported here is part of a larger project, the University of Califor-nia-Irvine "World-Systems and Ethnological Theory" project. The larger study was designed to explore the political, economic, social, and ecological impact of the world economic system on small-scale societies, as well as to describe trade net-works and political and economic relationships between small- and large-scale societies.

dock and White 1969) distributed across four major world geographic areas. This study describes the patterns and general characteristics of the societies on a new set of coded variables representing change processes. The variable set includes agricultural and nonagricultural intensification, the addition of new crops and animals, changes in settlement patterns and expansion, catastrophic loss, changes in trade and wage labor, and the date of ethnographic observation. The findings of the study are both methodological and substantive. The methodological results focus on sampling problems for comparative studies as well as the dates of observation set for sample societies. The substantive findings indicate that variables are regionally clustered, with two basic patterns emerging for the way subsistence systems are affected and changed in different world regions.
The research reported here is part of a larger project, the University of California-Irvine "World-Systems and Ethnological Theory" project. The larger study was designed to explore the political, economic, social, and ecological impact of the world economic system on small-scale societies, as well as to describe trade networks and political and economic relationships between small-and large-scale societies.

Subsistence Economy in Anthropological Research
Subsistence economy has formed t cornerstone of much anthropological search, particularly in studies using hol cultural methods. Anthropological the ries about the relationship between su sistence and other aspects of culture ten to have one of two emphases. They focu either on the adaptations of individual s cieties to their environments, or on t constraints that subsistence economies place on other elements of society or culture.
Subsistence economy has been shown to have good predictive power in studies of socialization (Barry, Child, and Bacon 1959), the sexual division of labor (Boserup 1970;Bradley 1987;Burton and White 1984;Ember 1983;Martin and Voorhies 1975;Nerlove 1974;Sanday 1973), and features of social structure (Goody 1976). Psychological anthropologists, such as Whiting and Whiting (1975) for example, formulate a model for psychological research that begins with the culture's "maintenance systems." Cultural materialists and traditional Marxists include subsistence within infrastructure. In the cultural-materialist perspective, infrastructure "probabilistically determines" other aspects of the economy as well as behavior and mentality (Harris 1979:55). Because of the importance of subsistence to materialist theory, types of food-getting activities con-Some comparativists have used subsistence economy as a part of social evolutionary theory building, particularly for creating scales with which to predict cultural complexity or cultural evolution. Subsistence economy was a popular index of cultural evolution around the turn of the century (e.g., Steinmetz 1899;Nieboer 1910;and Hobhouse, Wheeler, and Ginsberg 1915). More recently, Lomax and Arensberg (1977) used type of subsistence economy as a measure of cultural evolution. However, comparativists generally do not consider subsistence type to be an especially precise measure of cultural evolution (Levinson and Malone 1980:33;Naroll 1978:421). Most recent cross-cultural investigations into cultural evolution do not rely on subsistence types (e.g., Carneiro 1970) or use them only peripherally (Freeman 1957;Tatje and Naroll 1970).
In economic anthropology, subsistence economy is often distinguished from market-oriented activity. In contrast, most cross-culturalists tend to include marketoriented food production under the subsistence rubric. Barry, Child, and Bacon (1959:53) defined subsistence as "what general types of food-getting are predominant." This broad definition includes both production for use and production for exchange. Murdock and Morrow's (1970) definition of subsistence economy includes not only the acquisition (or "derivation") of food, but also its transport, preservation, and storage. Subsistence economy is defined in this report as the acquisition, transport, and storage of all products of agriculture, herding, hunting, fishing, and gathering, for use or exchange, and also includes wage labor and trade.
Past analyses have tended to emphasize change in subsistence through local evolutionary adaptation. In contrast, the current study looks at changes in subsistence economy as a consequence of larger systems of trade and political economy.

Description of the Project
The Sample The data were collected as part of the "World-Systems and Ethnological The-ory" project, a study of societies from the Standard Sample. The Standard Sample consists of 186 societies and includes many of the best-documented and widely known cultures in the anthropological literature. Each Standard Sample society has a locational pinpoint focusing on a particular society or unit within a society. Each culture also has a focal date. These range from 2,000 years ago to 20 years ago.
The 87 societies included in the current study were selected from the larger group of 186 via stratified random sampling and have pinpoint dates of ethnographic observation ranging from 1750 to 1965.' The geographic distribution of the sample includes 38 societies from Africa and Eurasia, 17 from Oceania, 18 from Northeast Asia and North America, and 14 from Central and South America.

Coding
Coding of the subsistence variables began as part of the larger project, then developed into a separate study as it became apparent that changes in subsistence systems due to world-system contact were widespread. A separate codebook was developed by Moore and Bradley (1986), which focused specifically on changes in subsistence systems. Ethnographers' statements about the nature of change in indigenous forms of subsistence, as well as the causes and consequences of such changes, were recorded in narrative form.
A codebook was then developed that encompassed a majority of the categories that had emerged through the reading.

Variable Definitions and Frequencies
Fourteen variables were coded as indicators of change processes. They are discussed here along with their frequencies.
The sample, the codebook for the variables, and the codes for societies are presented in the Appendix.
Whether change had occurred within 100 years prior to the designated date ofethnographic observation. The most important piece of information coded is whether or not changes in subsistence patterns occurred in the society within 100 years before the date of observation. Changes earlier than the present study. If change occurr was noted whether it was due to worldsystem contact. Seventy-three societies, or 84%, had experienced changes in their subsistence systems. In all cases, the changes were due to world-system contact. In 14 societies, the subsistence economy was unaffected.2 Extent ofchange. This describes whether changes in the subsistence system were partial or total, that is, if the change in the system was so extensive that the entire subsistence system changed. Of the 73 societies where change had occurred, the extent of change was partial in 65 societies and total in 8.
Agricultural intensification. Agricultural intensification is defined as any increase in agricultural inputs per hectare. Boserup (1965) sees agricultural intensification as a consequence of population pressure, while other theorists see it as resulting from microscopic processes such as the policies of the state. The agricultural intensification variable is defined to include increased production of an existing crop, introduction of new agricultural technology, changing from shifting to permanent cultivation, or change in the diversity of crops (e.g., a move to monocultivation). Agricultural intensification occurred in 29 of 73 societies.
Nonagricultural intensification. This variable is defined to include intensification of any nonagricultural activity, including herding, hunting, fishing, or gathering. Hunting and fishing people are very sensitive to technological changes, such as the introduction of rifles, which increase profit-oriented production. These types of changes upset ecological balances and lead to rapid depletion of resources (Curtin 1984;Wolf 1982). Intensification in nonagricultural modes of subsistence oc- Date of ethnographic observation. This is the Standard Sample pinpoint date for each society assigned by Murdock and White (1969 known as correspondence analysis (Greenacre 1984;Hoffman and Franke 1986), dual scaling (Nishisato 1980), and canonical analysis (Gittins 1984).3 The set of societies where change had occurred served as the basis for the scaling analysis. Only societies with complete data on all variables were used. A binary matrix of 63 societies by 13 indicators of change served as the input data matrix.4 The input matrix as well as the optimal scores for societies are listed in the Appendix. Figure 1 presents the two-dimensional optimal structure displaying the interrelationships of change processes.5 The first dimension contrasts two distinct clusters  The evidence in this study of regional differences in patterns of change is strongly linked to a number of wellknown historical themes. The Northeast Asian-North American change system, which includes intensification of hunting, gathering, and fishing as well as loss and destruction of the subsistence system, has been discussed by Curtin (1984) and Wolf (1982). In this study, the pattern includes a high proportion of societies in which trade was introduced and more than half that experienced changes in settlement patterns. The Pawnee, for example, experienced intensification of buffalo hunting in the 18th and 19th centuries as a result of trade with beaver trappers (White 1983). Similar events transformed a group of Pacific Northwest cultures including the Ainu, Aleut, Eyak, Gros Ventre, and Comanche. Contrastingly, the Continental Old World, Central America (including Mexico), and South America contain regions of agrarian civilizations with long histories of external penetration and contact, and show characteristic patterns of agricultural intensification (e.g., the Tiv, Lepcha, Vietnamese, Otoro Nuba, and Goajiro) and of 3.00 modifications of pastoral subsistence (Abipon, Rwala, Kzak, Lapp, and Andamans). In Africa, for example, after the British pacification of the Nuba Hills, the Otoro Nuba were able to expand farming from the hills and gardens into the plains. This entailed increased demand for labor as well as the introduction of new crops and agricultural intensification (Nadel 1947). In Oceania, a region with a history of first contacts in the 19th century, the pattern of agricultural intensification is marked by the introduction of trade, wage labor, and expansion, and gains support from historical accounts of labor recruiting and trade in copra and sandalwood (Fox 1977;Hezel 1983). Palau and Truk, for example, experienced intensification of coconut production and fishing, introduction and increases in wage labor, as well as the introduction of trade. In both societies the changes were due to German and Japanese colonial policies.
One interesting trend in the data is that intensification of activities such as hunting, fishing, and foraging, along with destruction of subsistence systems, were reported during the 19th century, while agricultural intensification and expansion were reported during the 20th century. This finding lends support to theorists such as Curtin (1984) and Wolf (1982) about the temporal nature of the world economy, with the 19th century being marked by expansion of the world system into nonagricultural regions and the concomitant destruction of indigenous human and animal life, while the 20th century is characterized more by agricultural intensification.

Conclusions
The findings of this study are both substantive and methodological. Three con clusions are interesting and worth noting variable "subsistence economy" m continue to be used cautiously. Whet subsistence economy is used as an ind pendent or a dependent variable, it is aspect of Standard Sample societies t is subject to rapid change. As we ha shown, 84% of Standard Sample soci ties' subsistence systems changed as a r sult of contact prior to the designa date of observation. This is despite fact that the Standard Sample was cr ated with the goal of setting the dates observation prior to the time of major culturative contact (Murdock and Wh 1969). It might be argued quite correct that all cross-cultural variables are subject to change. However, subsistence systems seem to be especially sensitive to the effects of contact, and what is coded at the time of observation is more apt to be a phenomenon in flux rather than a static aspect of a unified social system. The current study contributes two new independent variables, which directly capture change processes.
2. The temporal clusterings of the indicators of change raise questions about cross-cultural sampling. Cross-cultural samples are generally uncontrolled for time. This problem might be corrected by using samples that are stratified across time as well as space. Another solution, and one that this study is moving toward, would be to study the societies in samples such as the Standard Sample at different time pinpoints. The strong temporal trends in the data may indicate temporal differences in the nature of the world economy; however, it is just as likely that samples are temporally biased. This might happen one of two ways. First, the focus of anthropological interests has shifted from an emphasis on strictly nonagricultural societies to include peasant societies as the discipline has matured. Second, foraging societies have been more subject to destruction and extinction and thus have been less available for study and description during the 20th century.
3. The strong regional clustering of change processes is a reminder of Galton's observation that societies in close proximity to one another share common features, on the basis of shared historical processes. Societies are not isolated from one another in time and space, nor are they discrete independent units. We have attempted to show how the historical dimension may be incorporated in crosscultural research.

Notes
Acknowledgments. This project was funded by NSF grant BNS-83-04782 to Michael Burton and Douglas White and by NSF grant BNS-85-07685 to Douglas White and Michael Burton. Candice Bradley was responsible for supervising the coding team. We are grateful to Kim Romney, Frank Cancian, and Jim Ferguson for careful readings of an earlier draft of this report.
'See Bradley (1987) for details of the sampling procedures.
They do not differ from the remainder of the sample in mean date of ethnographic observation or in regional distribution.
3In more formal terms, optimal scaling is a principal-components technique, which extracts maximally correlated principal components for both the rows and columns of a matrix. Optimal scores are computed which maximally differentiate cases. The optimal scores for each dimension are also independent.
4The 13 variables used include dichotomized date of observation (la), "extent of change" (3), and "trade increased" (14). See the Appendix. The variable "was there change" (2) is not included in the scaling, since all societies used for scaling had experienced change.
"5The optimal scaling revealed that the twodimensional solution was appropriate. The two dimensions account for approximately 40% of the variance of the original data. No attempt was made to test the significance of the dimensions. Instead, the dimensions are used as a heuristic to explore the patterns among change processes and societies.
"6The significance of the regional clustering was formally tested by a one-way analysis of the variance of the mean scores by regional groups on dimensions 1 and 2. The results were highly significant for both dimensions (dimension 1, df3,59, F = 19.752; dimension 2, df3,59, F = 6.64). The correlation ratio (eta squared) for dimension 1 is .50 and for dimension 2, .25.