Index Introduction Subjects and Methods Hainan island Paori hamlet Subject households Data collection Results The long-term process: introduction of cash crops from 1982–2004 Inter-household variation in cash cropping, labor resources, and impact of cash cropping on labor hours and dietary intake Discussion Inter-household variation in adoption of cash cropping Impact of cash cropping on labor and dietary patterns Acknowledgments References

Introduction

In China, a planned economy functioned during the period of the ‘people’s commune’, from the late 1950s to the early 1980s. The entire land of a village, including paddy fields and gardens, was communally owned and collectively cultivated. Since the middle of the 1980s, when the system was transformed to a market economy, the government assigned agricultural land to individual households, depending on the number of each household’s members. People were encouraged to adopt cash cropping under the so-called ‘household responsibility’ system (Guan, 1995; Kawahara, 1999; Oi, 1999). Later on, each household was allowed to expand its cash crop gardens in the village communal land by its own efforts.

Hainan island is one of the regions where the development of a market economy was prioritized as part of the effort to improve poverty in rural areas. In some areas of this island, including the target hamlet of the present study, the production of banana and rubber started in the 1980s. This was followed, in the late 1990s, by litchi and longan, profitable cash crops that grow only in subtropical regions (Iwasa, 1984; Ke, 1999). The statistics of Hainan province indicated that, since the 1980s, per capita income has increased in parallel with the increase in cash crop production (Editorial Office of Hainan Province History, 1997). On the other hand, however, economic inequality among households has emerged, and has been gradually growing (Yao, 1999).

Based on a field survey in one hamlet of the Li ethnic group, we first investigated the process of how cash crops have been adopted in the target hamlet, from the time of transformation to a market economy in the 1980s until the present. Special attention was paid to clarify the factors that explain inter-household variation in the adoption of cash cropping. Secondly, we investigated the impact of the adoption of cash cropping on basic survival-related factors of households, such as labor hours and nutritional intake. We attempt to clarify the process of how economic inequality among households emerged and has grown in rural Chinese societies, and the possible consequences of such processes for the living conditions and diet of the people.

Subjects and Methods

Hainan island

Hainan island, with 34000 km² of land area in a tropical and subtropical monsoon environment, is located to the south of mainland China. The mean monthly temperature fluctuates slightly, from 23°C to 25°C, and annual precipitation is about 1500 mm, 85% of it concentrated from May to October (Gao et al., 1988).

Hainan island was inhabited by about seven million people in 2000. Historically, its indigenous populations were Li and Miao ethnic minorities. They have inhabited the inner areas for ten centuries, due to migrant populations of the Han people to the coastal regions (Wu, 1997). The Li people, one of 55 minority groups registered by the Chinese government, has been recognized as the first settlers of the island, and the total Li population was 1.2 million in 1999 (Hainan Province Bureau of Statistics, 1999).

Wuzhishan city, located in the mountainous inland of Hainan island, is the home of the Li people and functioned as the autonomous government of the Li and Miao ethnic minorities until 1987. Before the 1980s, Hainan island, especially its inner part, was one of the least developed regions in China, while large cities developed along its coast. For instance, the road network scarcely extended to inland areas until the 1970s. Since the 1980s, when the economic system began to transform in China, various developmental projects have been launched, first in the eastern coastal regions and then expanding to the inner parts of the island (Liang et al., 2003). In 1988, the entire island—previously a part of Guangdong province—attained provincial status. The new Hainan province was designated as a special economic zone (SEZ) by the Chinese government, for the promotion of economic development, especially in the tourism and agriculture sectors. Since then, huge amounts of money have been invested in various projects by enterprises of mainland China and foreign countries. Since the 1990s, Hainan province has been one of the most rapidly developing provinces in China.

Paori hamlet

In our fieldwork, we targeted the Paori hamlet of the Li people in Fanyang township. The township is located northwest of Wuzhishan city at an altitude of about 200 m above sea level; the inhabitants are speakers of the Ha dialect of the Li language. A provincial road (No. 29) runs along the Chang Hua Jiang river that flows from east to west through this township. The settlements and paddy fields have been established in the flat zone between this river and the mountainous hinterland.

According to the records of the traditional food production system of the Li people before the 1950s, the Paori people subsisted on rice cultivation in paddy fields (two crops a year), slash-and-burn cultivation, and hunting and gathering (von Stübel, 1937; Odaka, 1944; South-Central University for Nationalities, 1992). The introduction of new technologies, such as cropping of hybrid rice strains, an improved irrigation system, chemical pesticides, and fertilizers (Editorial Office of Hainan Province History, 1997), brought about a remarkable increase of per capita annual production of rice from 130 kg in 1952 (South-Central University for Nationalities, 1992) to 240 kg in the early 1970s, and then to 480 kg in 2000 (Jiang, 2004). The annual per capita consumption of rice was 240 kg in 2000, with an annual per capita surplus of 240 kg (Jiang, 2004).

In addition, cash cropping was introduced to this village. During our survey period of 2000–2004, the villagers depended mainly on rice cultivation in paddy fields and cash cropping in the surrounding mountainous areas for their livelihood. With double rice cultivation (two crops per year), the first cultivation cycle begins at the end of January and ends in May, and the second begins at the end of July and ends in October. In the beginning and last two weeks of each cycle, the villagers spend much of their time on rice cultivation; in this paper, we call this period the ‘busy farming season’. In the periods other than these busy times, people spend most of their working hours on cash cropping; we call this period the ‘cash crop season’.

Subject households

In October 2003, the authors conducted an extensive survey of all 37 households of the Paori hamlet, and collected data on their demographic composition, economic status, land use, and subsistence activities. Of these, 11 households, located in the central block of the hamlet, were targeted for study of labor hours and dietary intake. Although socioeconomic conditions differed to a considerable extent from one household to another, such inter-household differences did not appear to relate to geographical blocks. In fact, the average number of adult members, the major labor force, did not show a significant difference between the 11 households (3.3 ± 1.2, mean and standard deviations) and all 37 households (3.0 ± 1.0). Neither did the cash crop garden areas differ; averages and standard deviations were 1.8 ± 1.3 ha for the former and 1.6 ± 1.0 ha for the latter.

Data collection

The fieldwork was conducted for a total period of 12 months: from October–November 2000, February–March 2001, June–September 2001, July–August 2002, October–December 2003, and from November 2004 to February 2005. One of the authors (H.W.J.) lived in the hamlet to establish a rapport with the people. Because the people usually speak the Li language in daily life (although they also speak Han Chinese for communication with outside people), data were collected using the Li language, partly complemented by Han Chinese.

Reconstruction of the long-term process

Demographic change and adoption of cash cropping between 1982 and 2004 were reconstructed for the 11 households. This was done on the basis of interviews with people who were 40 years or older in 2004 (i.e. 18 years or older in 1982) and by reference to the statistical records of agricultural production of the hamlet and Fanyang township. The data were cross-checked in order to clarify the long-term processes of economic system transformation and demographic change that occurred between 1982 and 2004.

Time allocation and food consumption survey

Seven-day time allocation and food consumption surveys were conducted during the ‘cash crop season’ of November 2003, December 2003 (only time allocation survey), and November 2004. Every household was visited every 90 minutes (from 0700 to 2030) to record the activities of adults (24 males and 12 females, aged between 18 and 65 years). The 6804 observations (36 adults × 9 observations/day × 7 days × 3 times) were classified into three types of categorized activities: cash cropping, other subsistence activities, and others. ‘Other subsistence activities’ include maintenance of paddy fields, gathering wild vegetables, and caring for domesticated animals such as pigs. The foods brought to/from the houses or consumed/discarded/wasted in the households were weighed by using a portable digital scale and recorded during observation hours. Food consumption was quantified at the household level. The nutrient intake of a household was adjusted to that of an adult male with a mean body weight of 53.1 kg (n = 24), assuming that household meals were proportionally distributed to individuals according to each individual’s energy requirements (FAO/WHO/UNU, 1985; see Umezaki et al., 1999, for method of calculation). The Chinese food composition tables were used to estimate energy and protein contents, as well as the proportion of edible parts (Institute of Nutrition and Food Hygiene, Chinese Academy of Preventive Medicine, 1991).

Land use survey

We investigated land use patterns of the subject households in 2004, using a geo-referenced QUICKBIRD satellite image (60-cm resolution panchromatic image, acquired on December 3, 2003) which we linked to field data as described below. First, one of the authors (H.W.J.) and a landowner visited his plot(s), and mapped the boundary of each ‘parcel’ (a portion or whole of paddy field or cash crop garden) on the printout of the satellite image. A GPS device (GPS Pathfinder Pocket, Trimble, Inc.) was used, if needed, to delineate the exact shape and location of each parcel. Then each parcel was labeled with a serial number. This process was repeated with all the household heads (n = 37), and, in total, 747 such parcels were identified. The 747 parcels served as the units of land use, since size and shape of these parcels appeared to remain unchanged for twenty years (from 1985 through 2004). Then the land use types of each unit for the years 1986, 1995, and 2004 were reconstructed through interviews with the landowners.

On the satellite image displayed on the geographic information system software (ArcGIS 9.0, ESRI, Inc), the boundaries of cash crop gardens, paddy fields, and other types of agricultural fields, which were visually identifiable, were traced, and vector files of each of these were made. By applying an ArcGIS algorism of area calculation (ArcGIS manual: Using ArcGIS Spatial Analyst, ESRI inc.), we calculated the size of the cash crop gardens in 1986, 1995 and 2004.

Cash crop gardens were classified into two stages: ‘harvested gardens’ refers to those that had already produced crops, while ‘unharvested gardens’ refers to those prior to harvest. Figure 1 shows the location of all the cash crop gardens in the Paori hamlet in 2004. In the present study, only data for the 11 households of the central block were used.

View Details

Figure 1. Land use map of the Paori hamlet in 2004. Land use was categorized into: ‘harvested’ cash crop gardens, ‘unharvested’ cash crop gardens, paddy fields, and others (i.e. residential area and graveyard).

Statistical analysis

The relationship between the area of cash crop gardens of households in 1995 and the availability of labor resources in a household in 1985–1994 (i.e. total number of adults in 1985–1994 [adult-years]) was examined by calculating Spearman’s correlation coefficient. Similarly, the relationship between the area of cash crop gardens of households in 2004 and the availability of labor resources in a household in 1995–2004 were investigated.

The impact of a household’s adoption of cash cropping on their basic survival-related factors was investigated for the year 2004. This was done by correlation analysis (Spearman’s correlation coefficient) between the area of ‘harvested’ cash crop gardens in 2004 (which were supposed to provide the income for obtaining protein-rich foods) and dietary intake (energy and protein), and between the area of cash crop gardens in 2004 (‘unharvested’, ‘harvested’, or total) and labor hours. The statistical significance level was set at 0.10, because the sample size was small (n = 11). All statistical analysis was performed by SPSS (version 11.5).

Results

The long-term process: introduction of cash crops from 1982–2004

Table 1 shows the total number of adult-years, area of paddy fields, and area of cash crop gardens for the 11 households in 1986, 1995, and 2004. After the transformation of the economic system in the early 1980s, banana was planted as the first cash crop in the Paori hamlet. Since bananas readily reproduce via shoots, the villagers could easily obtain new seedlings. Several households had started to plant bananas in their gardens by 1984. Part of their crop was sold to the market. However, banana gardens were gradually abandoned because the people could earn only a small amount of money from them.

In 1985, rubber was introduced to the village. A household (A in Table 1) decided to purchase 1200 rubber seedlings at the cost of 1200 Chinese yuan, equivalent to the price of 8000 kg of rice in 1985 (1 kg of pork cost 1.6 yuan at that time), to plant them in their slash-and-burn gardens where rice and cassava had been grown the previous year. The area of the rubber gardens was 21760 m². Of the remaining households, six planted rubber, as the township government provided free seedlings in 1987. The number of seedlings provided by the government varied with the number of holes people prepared in their gardens. However, the rubber gardens of three households (households F, G, and I) were seriously damaged by buffaloes that were reared around the village, because the owners of these gardens did not take elaborate measures such as fencing or digging drainage systems to protect the seedlings. As a result, the rubber gardens started at this time were mostly abandoned.

The success of household A in 1995 changed the people’s attitude toward cash cropping. The head of household A, who first planted rubber in his gardens, started tapping and earned a considerable amount of money from selling rubber. He used the profit from rubber to purchase protein-rich foods for his family’s daily diet, and also to pay for his son to get professional training at an agricultural college in the urban area. These developments gave the people of the Paori hamlet a clear image of success that rubber cultivation might bring. The area of rubber gardens rapidly increased, and people started to manage the rubber gardens intensively.

Another epochal event occurred in 1998. A person in a neighboring village succeeded in harvesting the longan and litchi fruits in his gardens (ca. 20000 m²), and earned an enormous sum of money for the Paori people (according to the informants, each tree produced 400 to 600 yuan; 1 yuan was equivalent to 0.12 US dollars in 2004). The local government had supported him to demonstrate a ‘model case’ of fruit cultivation to the people of Fanyang township. In fact, the person later received an official commendation as a ‘model farmer’ (Lao Dong Mo Fan) from the Chinese central government. He traveled to Beijing to attend the award ceremony. His success shocked the people of the township. Many people in Paori started to devote more effort to the growing of litchi and longan. As the result of this, almost all the secondary forests of the village were converted to cash crop gardens of litchi, longan, and rubber. The people spent longer hours on cash cropping than before, while surplus rice from the paddy fields was used for investment in cash cropping rather than for purchasing protein-rich foods such as meat and fish.

During the period between 2000 and 2004, when we were collecting data in the Paori hamlet, people were intensively cultivating cash crops. The exception was household K, which had no cash crop gardens throughout the the early 1980s up to the present (see Table 1). According to the head of household K, they had paddy fields that were large enough to feed six household members, and three of them, females, had worked away from home since 1996. Although remittance from the three females was negligible, the three adults who remained in the hamlet were able to produce a sufficient amount of surplus rice to brew ‘rice wine’ or to trade for protein-rich foods. He was satisfied with his life.

At any rate, almost all parts of the communal land of the village had been converted to cash crop gardens by 2004 (see Figure 1). Most gardens for rubber (planted before 1995) and banana have been ‘harvested’ by 2004, while most of the gardens planted with longan, litchi, and rubber (planted after 1995) were ‘unharvested’. That is, the people are able to earn money from the former but not from the latter. According to the results of the land use survey of all the households of Paori in 2004, of the 55.6 ha of cash crop gardens, 17.7 ha (32%) were ‘harvested’ and 37.9 ha (68%) were ‘unharvested.’

In summary, the people’s attitude to cash cropping changed after the two epochal events of 1995 and 1998. Since then, almost all the households adopted cash cropping. Next, we analyze inter-household variation separately for the two periods: before 1995 and after 1995.

Inter-household variation in cash cropping, labor resources, and impact of cash cropping on labor hours and dietary intake

Spearman’s correlation coefficients were calculated for the 11 households to investigate the association between a household’s availability of labor resources (i.e. the number of adults in a household, aged between 18 and 65 years) and its area of cash crop gardens (Figure 2a, Figure 2b). There was a significant correlation between the total area of cash crop gardens in 2004 and the total number of adult-years in 1995–2004 (r = 0.62, P = 0.04), but not between the total area of cash crop gardens in 1995 and the total number of adult-years in 1985–1994. The labor resource of a household correlated with the area of cash crop gardens made after the epochal event of 1995, but not before it.

View Details

Figure 2. Scatter plots between the availability of labor resource and total area of cash crop gardens for the 11 Paori hamlet households. (a) Horizontal axis, total number of adult-years in 1985–1994; vertical axis, total area of cash crop gardens in 1995. (b) Horizontal axis, total number of adult-years in 1995–2004; vertical axis, total area of cash crop gardens in 2004. Spearman’s correlation coefficient is indicated in each figure.

Table 2 shows the results of the time allocation and food consumption surveys of the 11 households in 2004, and the area of cash crop gardens for each household in 2004. A household on average had 7634 m² of ‘harvested’ and 10064 m² of ‘unharvested’ cash crop gardens. The daily time of an adult spent on cash cropping ranged from 1.5 to 5.3 hr with a mean of 3.2 hr. The mean total labor hours of an adult were 4.6 hr a day. The mean energy intake was 9.9 MJ (2370 kcal) per day per adult male. The basal metabolic rate (BMR) of an adult male weighing 53.1 kg (the mean weight of the target adult males of the food consumption survey) was 6.2 MJ (1480 kcal) (FAO/WHO/UNU, 1985), and thus, the mean energy intake of the subjects was 1.6 times the BMR. According to the criteria of FAO/WHO/UNU (1985), the activity level of the Paori people fell within the range of low (1.55 BMR) and moderate (1.78 BMR). Therefore, the level of energy consumption was judged adequate.

The mean level of protein intake was 68.2 g. We set net protein utilization (NPU) at 65%, which is equivalent to the amino acid score of rice, the major source of protein for the Paori people (FAO/WHO/UNU, 1985; The Resources Council of the Science and Technology Agency of Japan, 1997). On this assumption, the safe level of protein intake was calculated at 62.0 g for an adult male of mean body weight. Thus, the mean protein intake of the subject households exceeded the safe level. Of the 11 households, however, protein intake was below the safe level in four households.

The area of cash crop gardens partly explained inter-household variation of labor hours and dietary intake. The protein intake level of a household correlated positively with area of ‘harvested’ cash crop gardens (Spearman’s correlation coefficient, r = 0.55, P = 0.08, n = 11, Figure 3a). The time spent on cash cropping correlated positively with area of ‘unharvested’ cash crop gardens (Spearman’s correlation coefficient, r = 0.75, P = 0.01, n = 11, Figure 3b).

View Details

Figure 3. Scatter plots between the area of cash crop gardens and the variables for the basic components of human survival, in 2004, for the 11 Paori hamlet households. (a) Horizontal axis, area of ‘harvested’ cash crop gardens; vertical axis, per capita intake of protein. (b) Horizontal axis, area of ‘unharvested’ cash crop gardens; vertical axis, time spent on cash cropping. Spearman’s correlation coefficient is indicated in each figure.

Discussion

Inter-household variation in adoption of cash cropping

Previous studies that investigated rural development in China mostly used statistical data which were gathered together by the administrative unit and neglect internal variation (Lin, 1992; Yao, 1999). Although these studies reported on the impact of developmental policy or program at the regional level, and the increase in inequality between regions under different programs (Yao, 1999; Liu and Yin, 2004; Yao et al., 2004), the dynamics of economic development at the household or village level have largely been neglected. Since the household or village is the direct target of developmental policies and programs, in order to reach an adequate understanding of the factors that generate inequality at the macro (e.g. regional) level, clarification of the process through which people adopt developmental programs at the micro (i.e. household/village) level is indispensable. In the present study, we shed light on this issue by reconstructing the process of cash cropping adoption in a Li hamlet during the period of progressive economic transformation which began in the 1980s.

One of the important findings of the present study is that two epochal events (the first instance of cash cropping success in 1995, and the start of longan harvest in 1998 by a model farmer in a neighboring village), through which people realized the economic value of cash cropping, played an essential role in initiating economic transformation at the hamlet level. Before these events, most of the households did not make a serious effort to plant and manage cash crops, despite the fact that seedlings and technologies were provided by the government. In those days, the area of cash crop gardens was not related with the availability of labor resources in a household, but rather depended on individual household decisions. In contrast, after the epoch-making events, almost all the households of the hamlet started to make cash crop gardens as large as possible. The total adult-years of a household for the period 1995–2004 was found to determine the inter-household variation of total cash crop garden area observed in 2004.

A similar process was observed among Okinawan settlers in eastern Bolivia (Ohtsuka, 1980). Households initially adopted diversified agricultural strategies in their new environment, and then the predominant activities of the settlers switched from the homeland type, consisting of rice cultivation and pig-raising, to the Bolivia-appropriate type, i.e. cattle raising. In the process of this switch, a few households initially adopted the Bolivia-appropriate type and then the remaining households followed suit. Such ‘pioneer’ households played a critical role in recognizing subsistence strategies that were better adapted to the new environment. In the case of the present study, in facing economic transformation, the households that first planted rubber or litchi/longan played a similar pioneering role.

An understanding of such processes involved in the adoption of new crops or technologies will contribute to governmental implementation of effective rural developmental strategies in developing countries. Farmers may better accept new crops or technologies if they have a clear image of the prospects of success. In fact, the UNU/GEF program in Yunnan province of China adopted this strategy. In this program, several innovative farmers were selected from the community to demonstrate walnut and chestnut agroforestry (Dao et al., 2002). It was reported that the other households in the community accepted the new crops readily because they actually saw how successful the innovators were.

In the case of the present study, after the epochal events, almost all the households of the hamlet started to make cash crop gardens. Thereafter, availability of household labor resources was the key factor that determined the scale of cash cropping of each household. A similar relationship between household labor resources and cash cropping was reported by Johnson (1988) in the Papua New Guinea highlands. She emphasized the importance of the availability of labor resources (i.e. the number of women) in economic development and the dynamic relationship between household structure and economic development. Paori and the Papua New Guinea highlands had some common characteristics regarding the initiation of cash cropping; in neither region did farmers have to invest money for seedlings because these were provided by the local government to encourage rural development. This is in contrast with the case reported for the Brazilian Amazon (Smith et al., 1996). Smith et al. (1996) suggested that insufficient financial aid appeared to be the major constraint to the expansion of cash cropping. In Paori, after the government stopped providing free seedlings in 2002, a situation similar to that of the Brazilian Amazon has emerged regarding cash cropping of litchi/longan. Farmers now have to spend money not only on seedlings but also on chemical fertilizers, pesticides and herbicides. The cost of maintenance (expenditure on agrochemicals) of the litchi/longan gardens was higher in comparison with that for banana or rubber (e.g. 30000 yuan/hectare/year for longan; 3000 yuan/hectare/year for rubber). Money for the investment was extracted from selling the surplus production of rice to some extent, and/or from the income of already ‘harvested’ gardens of rubber and banana. The area of longan/litchi gardens of the year 2004 correlated with the average area of ‘harvested’ gardens of rubber and banana/cassava in 1995–2004 (Spearman’s correlation coefficient, r = 0.53, P = 0.10). The households that succeeded in cash cropping were able to afford further investment in their gardens. This prompted an increase of inter-household variation in cash cropping in the Paori hamlet.

Impact of cash cropping on labor and dietary patterns

We begin with an examination of the relationship between labor hours (sum of time spent on cash cropping and on other subsistence) and dietary intake in the year 2004, in order to find the effects of cash crops on people’s livelihood. Of the 52 weeks in a year, eight weeks were regarded as the ‘busy-farming season’ and the remaining 44 weeks as the ‘cash crop season’. On average, an adult spent 9.5 hr on subsistence (i.e. cash cropping and other subsistence activities) during the ‘busy-farming season’ (Jiang, unpublished data), while he/she spent 4.6 hr during the ‘cash crop season’. The average time spent on farming per day was, therefore, estimated as 5.4 hr, the weighted average of the two seasons, in an adult for the whole year. Since quantitative data on labor hours were not available for other Chinese populations, the labor hours of the Paori people were compared with those of the people in Indonesia engaged in a similar type of agriculture consisting of irrigated cultivation of rice and cash cropping. Time spent in farming per day was reported as 6.2 hr among the Sundanese in West Java (Moji, 1980), 7.9 hr among the Kari Loro in central Java (White, cited in Minge-Klevana, 1980), 6.7–7.1 hr among the Kabupaten in central Java (Hart, cited in Moji, 1980), and 6.3 hr among Javanese in East Java (Edmundson, cited in Moji, 1980). Thus, although a part of the variation might be attributable to methodological differences, the labor hours of the Paori people were shorter in comparison with those of any of these Javanese populations. Moreover, the time spent on farming in the Paori people was located at the lower end of the range of the human populations reviewed by Minge-Klevana (1980).

Among the possible determinants of labor hours, labor productivity is the most likely factor influencing labor hours among different societies. In Hainan island, newly introduced technologies (i.e. maintenance of the irrigation system, employment of agrochemicals, etc.) and high-yield varieties have drastically increased labor and land productivity of rice since the 1980s (Editorial Office of Hainan Province History, 1997), alleviating the workload of rice cultivation in paddy fields. The labor hours spent for cash cropping may have had increased along with the acceptance of cash cropping (as inferred from the cross-sectional data shown in Figure 3b), but have been counterbalanced by reduced labor hours spent on rice cultivation. The annual productivity of rice reached 6750 kg per ha in Paori in 2000, more than double the standard yield in Java in 1979 (3240 kg per ha; Biro Pusat Statistik, cited in Moji, 1987). Both the mean energy and protein intake exceeded the recommendation of FAO/WHO/UNU (1985). The high productivity of rice allowed the Paori people to produce a surplus, which contributed to maintaining food security irrespective of income from cash cropping.

Yet, the cash cropping had some effects on the variation in protein intake among the Paori households. The correlation analysis suggested that the households with larger ‘harvested’ cash crop gardens tended to consume more protein. The households who adopted cash cropping before the epochal event of 1995 were already earning money from their cash crops. The variation in protein intake stemmed mainly from the availability of meat and separable fat of pork and fish, which were purchased from motorbike peddlers who occasionally visited the hamlet. The households with larger ‘harvested’ gardens used the income from cash crops for such purchases. Thus, nutritional improvement brought about by cash crop introduction may be progressing, but only in a part of the hamlet.

Dewey (1979) demonstrated that, in many societies of developing countries, agricultural development, which is defined as transformation from an agricultural system primarily based on self-sufficiency to commercial agriculture, has led to a deterioration of the nutritional status of the people, rather than an improvement. Such deterioration was usually observed when local food crops were replaced by cash crops (e.g. Hughes and Hunter, 1970). For example, in a society in Nigeria that introduced rubber growing, the farmers switched their staple crops from yam, maize, rice, beans, and groundnut to cassava to cope with the high labor demand for rubber. The switch of staple crops resulted in the emergence of nutritional deficiency among the people (Indusogie, cited in Dewey, 1979).

In the case of the present study, the intensification and maintenance of rice cultivation, partly determined by government policy, greatly contributed to the overall food security of the people. The nutritional status of Paori people as a whole may improve gradually when all the households start to earn money from cash crops. The future impact of cash cropping on their living condition is unpredictable, however, since the price of cash crops varies with the world market situation, and since production fluctuates with climatic conditions. A long-term follow-up study on the dynamic process of people’s socioeconomic adjustment to the changing outside world is essential for an in-depth understanding of the inequality issues in rural societies under economic transformation.

Acknowledgments

This study was conducted as a joint project between the National Museum of History (Japan), the Central University for Nationalities (China), and the Hainan Provincial Institute for Ethnology (China), with financial support from the Japan Society for the Promotion of Science (to R. Ohtsuka), the Ministry of Education, Culture, Sports, Science and Technology (to T. Shinohara), and the Nissan Science Foundation (to T. Shinohara). The authors thank T. Shinohara, M. Nishitani, C. Watanabe, Sheng-Min Yang, Jing-Yu Liang, Xiao-Jun Dong, and other project members for their support and encouragement. The authors also thank the people of the Paori hamlet for their hospitality and kindness during the fieldwork.