AGRICULTURAL SYSTEMS Agricultural Systems 94 (2007) 694–703 www.elsevier.com/locate/agsy Resource integration for multiple beneﬁts: Multifunctionality of integrated farming systems in Northeast Thailand a,* Prasnee Tipraqsa , Eric T. Craswell b, Andrew D. Noble c, Dietrich Schmidt-Vogt d a Chiang Mai University, Unit for Social and Environmental Research (USER), P.O. Box 144, Chiang Mai 50202, Thailand b University of Bonn, Global Water System Project, Walter-Flex Strasse 3, 53113 Bonn, Germany c International Water Management Institute (IWMI), c/o WorldFish Center, P.O. Box 500 GPO, 10670 Penang, Malaysia d Asian Institute of Technology (AIT), School of Environment, Resources and Development, P.O. Box 4, Klong Luang, Pathumthani 12120, Thailand Received 7 July 2006; received in revised form 7 February 2007; accepted 20 February 2007 Abstract Resource degradation in rice farming systems in Thailand endangers food security, but the systems may become more sustainable by combining them with aquaculture and livestock farm enterprises by capitalization of their synergies in resource use and re-use, i.e. by adopting integrated farming systems. Most empirical studies that assess this potential have focused on a few speciﬁc aspects, but not on the multiple social, economic, and ecological functions of resource integration. This study uses the framework of multifunction agri- culture to assess the performance of integrated farming systems in Thailand and compares its performance with that of ‘normal-rice’ or non-integrated farming systems. Surveys were conducted in Khon Kaen province of Northeast Thailand using a combination of quan- titative and qualitative survey methods. Integrated farming systems were found to outperform the normal or commercial farming systems in all four dimensions of a multi- functional agriculture: food security, environmental functions, economic functions, and social functions. The ﬁndings support the notion that diversiﬁcation and integration of resources on farms is feasible in both economic and ecological terms. The analyses shows that integrated farming does not, however, diminish the need for external inputs. High start-up cost might constrain farmers from switching to integrated farming and from exploiting the beneﬁts of resource integration. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Lowland rainfed agriculture; Northeast Thailand; Sustainability; Resource integration; Biodiversity 1. Introduction frequently identiﬁed as major constraints to crop produc- tivity in Northeast Thailand (Noble et al., 2000; Wijnhoud Northeast Thailand is a region where smallholder farm- et al., 2001). Several development agencies have stressed ing prevails and where farmers operate under conditions of that this endangers food security at all scales (ADB, environmental constraints and rapid economic changes. 1998). Past development eﬀorts have aimed at large The region’s agricultural productivity and per capita increases in food production through the cultivation of a income is much below the national average (OAE, 2003). few high yielding varieties and intensive use of mineral fer- A declining soil fertility and a low water availability are tilizers. Farm households rich in resources did beneﬁt to some degree from these green revolution technologies. * Corresponding author. Address: Chiang Mai University, Unit for Yet for resource-poor farmers, the reliance on a few Social and Environmental Research (USER), P.O. Box 144, Chiang Mai improved varieties, high levels of external inputs, and the 50202, Thailand. Tel.: +66 83 3012608; fax: +66 53 854347. ineﬃcient use of those inputs has made them vulnerable E-mail addresses: email@example.com (P. Tipraqsa), eric.craswell@ to the vagaries in weather and markets, with increasing uni-bonn.de (E.T. Craswell), firstname.lastname@example.org (A.D. Noble), schmidt@ ait.ac.th (D. Schmidt-Vogt). debt levels as a result. 0308-521X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.agsy.2007.02.009 P. Tipraqsa et al. / Agricultural Systems 94 (2007) 694–703 695 To stop soil fertility decline and to regain productivity, serving biodiversity, and supporting the socio-economic farmers have organized themselves in groups, some of viability of rural areas. Four principles were considered which have adopted integrated farming as a strategy. The fundamental to agriculture and the agro-food sector in gen- basic idea behind integrated farming is that species diversi- eral: food security, environmental functions, economic ﬁcation and resource integration can contribute to regain- functions, and social functions (OECD, 2001). Multifunc- ing productivity on resource-poor farms (Smyth and tional agriculture aims to strengthen the mutual synergies Dumanski, 1993; Konboon et al., 2001; Devendra and between these four principles (OECD, 2001). Thomas, 2002a,b; Prein, 2002; Halwart et al., 2006). This In OECD countries such as Germany, the UK, Switzer- move needs to be seen within the larger context of an land, Norway, Australia, and Japan the concept has since emerging policy change which favours diversiﬁcation over become a shared policy objective, although it is sometimes intensiﬁcation as a strategy to achieve not only productiv- (mis)used to justify government support to agriculture, ity increases, but also livelihood and environmental sus- which has come under pressure in international trade nego- tainability (Conway and Barbier, 1991; Naegel, 1994). It tiations (Potter and Burney, 2002). The concept is, how- has been motivated by concerns over agricultural resources ever, rarely applied in developing countries, like decline and the experience of the Thai economic crisis in Thailand, where the focus is still very much on the primary 1997 (Coxhead and Plangpraphan, 1999). In Thailand, this function of agriculture as a supplier of food and ﬁbre. policy change is promoted by institutions at various levels, This study was designed to address the four principles of including farmer associations at the grassroots-level in the multifunctional agriculture in the case of Northeast Thai- northeast (Ruaysoongnern and Penning de Vries, 2005) land by testing the following hypotheses. and the King’s Policy of Self Suﬃciency at the government level (Suwanraks, 2000). 1. Food security functions: Resource integration increases This study compares the performance of the integrated food availability. farming system (IFS) with that of the non-integrated farm- 2. Environmental functions: Resource integration improves ing system, which we denote as the commercial farming the quality of resources (soil, water, and trees). system (CFS). That is not to say that the IFS is not com- 3. Economic functions: Resource integration improves the mercial, but whereas the CFS has a clear focus on produc- economic returns of farms. ing rice for the market, the IFS – through diversiﬁcation 4. Social functions: Resource integration is a practice well and resource integration – pursues multiple objectives, such acceptable to the local community. as food production for the household, the maintenance of natural resources for food security and the well-being of household members, and the support for local 2. Materials and methods communities. Studies on the performance of IFS have been conducted 2.1. Study area in many countries and all have shown that synergies between farm enterprises increase productivity (Talpaz The study area is located in the Huai Nong Ian water and Tsur, 1982; Alsagoﬀ et al., 1990; Dalsgaard and Oﬁ- catchment in the Waeng Yai and Chonnabot districts, cial, 1997; Gomiero et al., 1999; Berg, 2002; Jamu and Khon Kaen province. The catchment, covering an area of Piedrahita, 2002; Frei and Becker, 2005; Pant et al., 285 km2, was selected because of the presence of a group 2005). Most studies have focused on the sustainability of of innovative farmers who are practising integrated farm- the IFS in terms of productivity and economic viability. ing and have organized themselves in a farmer network. Yet, farm households practising IFS have a multitude of The average slope of the catchment is 0–2%. Soils with ﬁne objectives and these should ideally be assessed integrally sandy to very ﬁne loamy textures are most prevalent in the (Paris, 2002). There is a clear need for studies that examine catchment. The bimodal distribution of rainfall is inﬂu- in a comprehensive manner the full range of resources that enced by the southwest monsoon and tropical cyclones are vital for agricultural systems. Using the concept of mul- from the South China Sea. Water resources are inﬂuenced tifunctional agriculture widens the focus to include envi- by the hydrological characteristics of the Chi River basin. ronmental and social services in addition to crop yields Floods occur every few years in the lower catchment near and proﬁts (OECD, 2001). This paper presents an assess- the Chi River. Drought in the catchment results from the ment of IFS that considers a wide range of both socio-eco- uneven distribution of rainfall. The forest type consists of nomic and biophysical dimensions while using the dry dipterocarp, riverine, riparian forests, and plantations. framework of multifunctional agriculture. The concept of multifunctionality was introduced in 2.2. Sample selection Western countries in the 1980s and originated with enter- prising farmers and scientists (Vereijken, 1997, 2003). The A pre-survey of the area showed that 21 farmers in the role of agriculture was recognized to go beyond the mere Huai Nong Ian catchment area were practising integrated supply of food and ﬁbre, and to include its role in shaping farming. It was therefore decided to use a purposive the rural landscape, sustaining renewable resources, pre- (non-probability) sample (Dillon and Hardaker, 1993), 696 P. Tipraqsa et al. / Agricultural Systems 94 (2007) 694–703 also because the objective was to gain an in-depth knowl- been to evaluate the sustainability of the member farms for edge about the farms, including their vegetation, soils, which a list of indicators has been developed by the farmers resource ﬂows, land management, production, and income themselves. Variables for this study were selected by com- for which each farm would have to be visited frequently. bining this list of farmer indicators with other indicators The pre-survey also showed that the soil, vegetation, from literature. Seven proxy variables were selected to and water resources varied through the catchment. The assess the four dimensions of a multifunctional agriculture, catchment area was therefore divided into three parts: which are shown in Table 1 and explained below (see Tip- upper area, middle area, and lower area to control for raqsa, 2005 for details). the variation in biophysical resources. Eight integrated From the perspective of the farmers practising inte- farms were selected as being the most representative for grated farming, food security is achieved when ‘growing integrated farming in the area. Three of these were selected everything you eat and eating everything you grow’, which from the lower area (which had six integrated farms in reﬂects the King’s philosophy of self-suﬃciency (Suwanraks, total), two from the middle area (four farms in total), 2000). Following this perspective the following two indica- and three from the upper area of the catchment (11 farms tors of food security were identiﬁed: (1) Richness of species in total). This selection was based on a pre-survey among used for food, which was calculated by counting the number the integrated farms in the catchment and discussions with of edible crop and animal species on the farm (Lightfoot extension agents, members of the farmer network, and ﬁeld and Pullin, 1994). This reﬂects the diversity of available practitioners from research and NGOs. Where possible, food products and because this diversity also relates to cul- integrated farms were selected that had already several ture, it can also be taken as an indicator of social functions. years of experience in integrated farming, for these were (2) The share of home-produced food, which was calculated more likely to show changes in the resource base. as the value of home-produced food that the household Each of the eight integrated farms was paired with an consumed (valued at market prices) as a percentage of adjacent commercial farm with similar soil type, natural the total value of food expenditures, measured over a vegetation, water resources, farm size, and which had con- one-year period (Dillon and Hardaker, 1993). This reﬂects verted natural forest into agricultural land at about the the level of self-suﬃciency and as such is also an economic same time. All selected households were cooperative and indicator. willing to participate in the research, especially those who Inherently poor soils and low water availability reduce had adopted integrated farming showed interest in receiv- the eﬀectiveness of mineral fertilizers and chemical inputs. ing feedback from the researcher. In low-input agriculture, the integration of trees and Each farm was visited about seven times between water reservoirs into the farm has been suggested by some December 2002 and April 2003 to collect in-depth informa- scientists as a more suitable alternative to a high reliance tion using several techniques. First, a general questionnaire on external inputs (Noble et al., 2000; Viyakorn, 2001; was designed to investigate the structural characteristics of Craswell, 2002). To capture these aspects of vegetation, the households and practices in the farm. This was fol- lowed by several rounds of informal interviews using adjusted guideline questions. Socioeconomic data were col- Table 1 lected using direct observation and semi-structured inter- Selected proxy variables for assessing the four principles of multifunc- views (Galpin et al., 2000). Data were validated using a tional agriculture triangulation strategy (Johnson, 1997), which involves tak- Indicator Principles of multifunctional agriculture ing data from more than one person in the same household Food Environment Economic Social to improve the accuracy of information and to get a better security understanding of the context and possibly divergent per- 1. Richness of food species · · spectives. Soil organic matter and soil texture were ana- (number) lysed from 45 soil samples taken at each farm (Daly, 2. Share of home produced · · 1992). These samples were taken about one month after food (%) at the end of harvest season (December–January) to avoid 3. Tree growtha · 4. Soil organic matter (%) · errors from applying fertilizer or soil amendments. The 5. Share of months in dry · farm resources were analyzed using farm mapping and season irrigated (%) farm resource surveys. Finally, vegetation was studied 6. Agricultural · · using a line-intercept (Bonham, 1989), which was con- productivityb ducted a few months after the rainy season because the veg- 7. Richness of species for · social purposes etation has then completed its annual growth cycle. (number) a Includes stem height (m), distribution of tree size and social position, 2.3. Selection of variables stem density (stem haÀ1) and basal area (m2 haÀ1) b Includes agricultural output (USD), land productivity (USD haÀ1) and Integrated farms in the study area have organized them- labour productivity (man-day haÀ1); 1 USD = 43.73 THB (December 2, selves in a network; one of the activities of the network has 2002). P. Tipraqsa et al. / Agricultural Systems 94 (2007) 694–703 697 soils, and water, the following three indicators were The study will return to this point in the discussion. It was selected: also argued in the above that the diversity of food species (3) Growth performance of tree communities, which was produced on the farm is an indicator of its social functions assessed from the vertical characteristics of the tree stand as they relate to the local culture. One additional variable structure using a line-intercept method (Bonham, 1989). suggested by the farmers was to count the number of spe- A sample plot size of 10 · 80 m (800 m2) was used to cover cies used for social purposes other than food, as this also the diversity and stand structure of the vegetation in the relates to the basic idea that diversiﬁcation contributes to farm area (Smitinand et al., 1994). The measurement the sustainability of the farm. (7) The richness of species included woody trees, shrubs, and woody perennials with used for social purposes was calculated as the number of a minimum height of 1.30 m and with a diameter P5 cm species used for medicine, local rituals, the making of tools, measured at 1.30 m height. The growth performance of and the shading of houses. the tree communities was assessed on the basis of stand These seven proxy variables were used to assess the per- height (m), distribution of tree size and crown social posi- formance of the integrated farms. Two complementary tion (following the classiﬁcation of Dawkins, 1987), stem methods were used: a comparison of farms based on farm density (stem haÀ1), and basal area (m2 haÀ1) (Lamprecht types (IFS vs. CFS) and a comparison between farms based and Pancel, 1993). The important value index (IVI) was on the number of synergies between enterprises on the used in addition to rank the species and to identify the farm. These synergies are deﬁned as ﬂows of biological most dominant species in the farm. material between the various enterprises on the farm. These (4) Soil organic matter (SOM) was measured at 15 loca- enterprises included paddy rice, vegetables, cattle, pig, tions on each farm and at three depths at each location: 0– poultry, aquaculture, mushrooms, and trees. Synergies, 10 cm, 10–20 cm, and 20–30 cm following Daly (1992) and for example, include the use of manure and compost to fer- using the modiﬁed Walkley–Black method (Nelson and tilize vegetables and rice, rice bran and leaves to feed pigs, Sommers, 1996). Each sample was analysed separately in and manure to feed ﬁsh. The count of these synergies was the laboratory. An average value over all 45 samples was used as a measure of integration level; for each identiﬁed used in the analysis because statistical analyses showed synergy the integration level was incremented by one. This no signiﬁcant diﬀerences in SOM at diﬀerent depths or indicator also helped to distinguish an integrated farm locations on the same farm. from a diversiﬁed farm as the latter can comprise many (5) The share of months irrigated in the dry season was enterprises but might have a low integration level if calculated from the number of months that production resources are not recycled on the farm. activities were irrigated from water reservoirs (farm ponds The results are presented in two parts. The ﬁrst part and wells) in the dry season and expressed as a percentage gives a general overview of the two farming systems using of the total number of months in the dry season (Garcia, descriptive statistics while the second part tests the 1997). hypotheses. The economic function of the farm refers to its ability to convert inputs such as land and labour into crop and 3. Comparison of commercial with integrated farming livestock products, which can be expressed as output/ systems input ratios, i.e. partial productivity indicators. (6) Agri- cultural productivity was assessed using two such ratios. Farm households live close to each other in each village First, land productivity was calculated as the quotient of with their ﬁelds scattered around at an average distance total agricultural output (USD) and total farm area from the farmstead of about 3 km. Fields were converted (ha). Second, labour productivity was calculated as the from natural forest about 30 years ago while the eight quotient of total agricultural output and the amount of farms in the IFS converted from commercial to integrated labour supply (man-days). Agricultural output included farming about 10 years ago. Fig. 1 illustrates the spatial both crop and animal production (Dillon and Hardaker, characteristics of resource use with an example of farm 1993; Dillon and McConnell, 1997), in which crop output maps of one integrated and one commercial farm from was calculated as the sum of total rice, vegetable, and the upper area of the catchment. Six management units perennial production and animal output was the sum of can be identiﬁed: paddy rice ﬁelds, vegetable plots, farm cattle, pig, poultry, and ﬁsh production in the year of ponds, woodlands, farm buildings, and open land. A com- the survey. parison of all farm maps between the two systems showed Agriculture is widely perceived to be the basis of rural that the main diﬀerence in management units is that all livelihoods, maintaining the rural communities and their integrated farms have woodlands and farm ponds, neither culture. For agriculture to perform this social function, it of which the commercial farms have. ﬁrst of all needs to give an economic return, which is indi- This is further conﬁrmed by Table 2, which compares cated by the productivity of the farms. During the inter- the socio-economic and biophysical characteristics of the views, farmers stated that one objective of integrated IFS and the CFS using mean values. Irrigation facilities farming is to reduce the migration of people from rural were classiﬁed into four types according to structure and to urban areas by creating more employment on the farm. location: farm ponds with closed outlets, farm pond with 698 P. Tipraqsa et al. / Agricultural Systems 94 (2007) 694–703 Fig. 1. Comparison of spatial characteristics of two representative farms of the IFS (left) and CFS (right). Note: IFS = Integrated farming system. CFS = Commercial farming system. Table 2 Synergies between enterprises in the integrated farms Comparison of farm resource characteristics by farm type, means included the feeding of crop residues to pigs, poultry, and Characteristic IFS CFS ﬁsh; the feeding of solid waste from poultry to carnivorous Family members (persons) 5 (2.17) 4 (1.16) ﬁsh (catﬁsh); and the application of solid waste from cattle, Labour force (15–65 years, persons) 4 (1.49) 2 (0.53) pigs, and poultry to vegetable and tree plots. The direct use Persons employed on the farm 3 (1.51) 2 (0.74) of animal waste was widely practised on all integrated Persons employed oﬀ the farm 0 (1.06) 0 (0.00) farms. Persons not employed 1 (1.30) 1 (0.52) In the second part of the results, the hypotheses are Distance to the local market (km) 5.5 (1.0) 5.6 (1.2) Farm area (ha) 3.86 (1.40) 2.73 (1.97) tested. Fig. 2 summarizes these results using a radar graph. Sandy fraction in soil (proportion) 0.44 (0.22) 0.43 (0.21) The ﬁgure compares the performance of IFS with that of Cars (number) 0.50 (0.53) 0.00 (0.52) CFS according to the proxies introduced in Section 2. Each Two-wheeled tractors (number) 0.38 (0.52) 0.88 (0.35) value shows the performance of one system as a fraction of Small water pumps (number) 0.88 (0.35) 0.25 (0.46) the maximum performance of both systems. The relatively Irrigation facilities (number)* 1.50 (0.53) 0.38 (0.52) large area between the two lines shows that the diﬀerence Notes: Standard deviation in brackets. between the farming systems is substantial. The graph * Includes farm ponds, wells, and irrigation canals. IFS = integrated farming system. CFS = commercial farming system. open outlets, wells, and oﬀ-farm water sources associated with irrigation projects. All eight integrated farms together had 16 ponds with closed outlets, but none of the commer- cial farms had a pond of this type. The table also shows that the average farm size of the integrated farms (3.86 ha) was larger than that of the com- mercial farms (2.73 ha) and integrated farms also used more labour than the commercial farms. There was no sig- niﬁcant diﬀerence in soil texture between the two farming systems as this was most strongly determined by the loca- tion in the catchment: the upper catchment was dominated by light textured sandy soils while the lower catchment was dominated by heavy textured soils, which may be associ- Fig. 2. Radar graph comparing the IFS with the CFS in eight aspects ated with the redistribution of sediment (silt and clay) (Lightfoot et al., 1993; Pitcher and Preikshot, 2001; Dey et al., 2006). Note: material between upper and lower catchment. IFS = integrated farming system. CFS = commercial farming system. P. Tipraqsa et al. / Agricultural Systems 94 (2007) 694–703 699 Table 3 Descriptive statistics for diﬀerences between the integrated and commercial farming systems, with mean and medians, and statistical tests for signiﬁcance Variable Mean Median IFS CFS IFS CFS ** ** 1. Richness of food species (number) 38 (10) 17 (2) 41 18 ** ** 2. Share of home produced food (%) 68 (14) 33 (9) 71.75 31.67 3. Stem density (stem haÀ1) 463 (320) 35 (17) ** 497 34 ** * 4. Soil organic matter (%) 0.98 (0.73) 0.64 (0.43) 0.69 0.51 ** ** 5. Share of irrigation months (%) 39 (4) 14 (19) 38 0 6. Land productivity (USD haÀ1) 1449 (628) 1307 (548) 1328 1354 7. Labour productivity (USD man-dayÀ1) 3.4 (0.52) 2.7 (0.24) 2.87 2.57 ** * 8. Richness of species for social purposes (number) 35 (14) 18 (2) 37 19 The signiﬁcance level refers to a t-test for diﬀerence in means and a rank-sum test for diﬀerence in medians. IFS = integrated farming system. CFS = commercial farming system. * P < 0.05. ** P < 0.01. clearly shows that the IFS outperform the CFS in every average tree density of 35 stems haÀ1. There was no evi- aspect although not signiﬁcantly for land and labour pro- dence of seedling recruitment in between these scattered ductivity. Table 3 compares the same indicators using trees due to annual cropping activities around the base of means and medians and each is discussed in the following. the trees. The rank-sum and t-tests indicate that the median and mean of tree basal area and stem density were signiﬁ- 3.1. Richness of food species cant greater for the IFS as compared with the CFS. Three species are particularly useful for farm households Table 3 shows a signiﬁcantly greater number of food as they are used for lopping, fodder, ﬁrewood, and species in the IFS. This is conﬁrmed by a positive correla- cutting for timber. The IVI reﬂects the dominance of tion between integration level and the log of richness of Azadirachta indica (neem) on integrated farms and food species (r = 0.091, p < 0.01). Flacourtia indica (governor’s plum) on commercial farms. Azadirachta indica, which dominated on integrated farms, 3.2. Share of home-produced food is a multipurpose tree species. The analysis of the size classes with respect to the diam- The food expenditures were signiﬁcantly greater for the eter at 1.3 m height showed that the trees on integrated IFS (257 USD yearÀ1) than for the CFS (197 USD yearÀ1). farms had a higher number of stems but were mainly small Yet if expressed in per capita terms then the food expendi- (73.3% were in the class of 0–10 cm diameter), whereas on tures are about equal for both farming systems the commercial farms trees were relatively fewer but with (66 USD capitaÀ1 for the IFS and 73 USD capitaÀ1 for larger stems. the CFS). The rank-sum and t-test are consistent and show The analysis of the distribution of tree stems in each that the share of home-produced food is signiﬁcantly class of crown social position showed that the middle sto- greater for the IFS than for the CFS (IFS = 68% and ries (classes 3 and 4) of crown social position of trees on CFS = 33%). This is further conﬁrmed by a linear regres- integrated farms appeared to be denser. Most trees (60%) sion between the log of share of home-produced food on commercial farms were in a higher social position class and the integration level, which shows a positive coeﬃcient and had a higher average height but lower stem density. (r = 0.097, p < 0.01). Together with a greater number of food species in the IFS, these results suggest that the IFS 3.4. Soil organic matter is more secure in the supply of food than the CFS. The rank-sum and t-tests indicate that the median and 3.3. Tree growth mean values for soil organic matter were signiﬁcantly greater on integrated farms than on commercial farms. The tree communities on integrated farms were com- The relationships between tree stem density as aﬀected by posed of three vertical layers: tree layer, shrub layer, and integration level; soil organic matter as aﬀected by tree ground cover. The height of the tree layer ranged from 4 stem density; and soil organic matter as aﬀected by integra- to 7 m with an average density of 463 stems haÀ1 and a tion level were further explored. The results showed mutual basal area of 0.62 m2 haÀ1. Trees were generally more correlations between SOM, integration level, and tree stem evenly distributed on the commercial farms. This resulted density. The integration level on a farm showed a clear in a more open tree cover dominated by relatively more positive correlation with stem density (r = 0.698, mature trees. The tree height ranged from 10 to 13 m with p < 0.05). This suggests a greater number of trees available stands having an average basal area of 0.06 m2 haÀ1 and an on the farm with increasing integration level. The analyses 700 P. Tipraqsa et al. / Agricultural Systems 94 (2007) 694–703 further showed a positive correlation between soil organic farms (rice: 2783 kg haÀ1 for the IFS and 2500 kg haÀ1 for matter and stem density (r = 0.581, p < 0.05), and soil CFS; vegetables: 6617 for IFS and for CFS 2223 kg haÀ1: organic matter and the integration level (r = 0.500, fruits: 1893 for IFS and 30 kg haÀ1 for CFS). p < 0.05). These results suggest that soil organic matter It could be argued that the higher total output and crop content on the farm is positively associated with an increas- yields at the IFS relate to their greater average farm size ing integration level, though a more detailed analysis and labour supply rather than to the integration of would be needed to test whether this is because of a greater resources. Multivariate regression was therefore used to number of trees or because of other resource ﬂows on the assess the impact of integrated farming on productivity farm. while statistically controlling for the eﬀects of farm size, labour supply, and external input use. Four alternative 3.5. Share of months in dry season irrigated regression models were used to check the robustness of the results. One model examines the total output as a func- The irrigation facilities provided water for about 39% of tion of arable farm land, labour supply, external input use, the dry season (from October to April) on integrated farms, and integration level (Model A). The three other models but only for 14% at commercial farms. The rank-sum and are modiﬁcations of the ﬁrst one in which the total output t-tests indicated that the median and mean number of irri- is either divided by the arable farm land (Model B) or the gation months was signiﬁcantly higher on integrated farms labour supply (Model C). Arable land and external input than on commercial farms. use were separately included in the last model for these This diﬀerence is due to the establishment of farm ponds were highly correlated (Models C1 and C2). on the IFS, which provided an average of three months of Parameter estimates and signiﬁcance levels are presented additional irrigation water. Irrigation canals existed on in Table 4 and the four models explored are speciﬁed below both farm types and used water from the Chi River, which the table. The explained variance is high for all models, all was distributed evenly over all farms. Wells were predom- signs are positive as expected, and all parameters are less inantly used for domestic purposes, and only occasionally than unity – indicating diminishing returns to each input. to irrigate the surrounding vegetable gardens. The most important observation is that the integration The number of irrigation months correlated positively level is positive and signiﬁcant in each of the models. This with the farm gross income and number of food species. clearly shows that additional synergies between enterprises These results suggest that farms with ponds were more pro- improve total agricultural output, agricultural land pro- ductive and had more alternative production activities, e.g., ductivity, and agricultural labour productivity while con- raising ﬁsh, and irrigating vegetable plots. trolling for the eﬀects of farm size, external input use, and labour use. 3.6. Agricultural productivity 3.7. Richness of species used for social purposes The total output from the integrated farms (3480 USD per farm) was signiﬁcantly above that of the commercial The farms in the IFS used an average of 35 species for farms (2006 USD per farm, p < 0.01). Concerning the yield social purposes while the farms in the CFS used 19 species of individual crops, it was found that rice, vegetable and on average. The t-test and non-parametric tests (Table 3) perennial yields were signiﬁcantly greater on the integrated conﬁrmed that this diﬀerence was signiﬁcant. Table 4 Multivariate regression results estimating the eﬀect of resource integration on farm production and productivity Agricultural production Land productivity Labour productivity Model A Model B Model C1 Model C2 Area (ha) 0.167 – 0.499** – Labour (man-day) 0.263* 0.123* – – Inputs (USD) – 0.886** – 0.873** Integration level 0.064** 0.033** 0.043** 0.033** Constant 4.900** 1.159** 3.286** 1.198** R2 0.83 0.98 0.90 0.99 F-value (signiﬁcance) 19.26 164.98 58.91 512.03 (P < 0.001) (P < 0.001) (P < 0.001) (P < 0.001) Model A: ln(Output) = ln(Area) + ln(Labour) + Integration Level. Model B: ln(Output/area) = ln(Labour/area) + ln(Input/area) + Integration Level. Model C1: ln(Output/labour) = ln(Area/Labour) + Integration Level. Model C2: ln(Output/labour) = ln(Inputs/Labour) + Integration Level. * P < 0.05. ** P < 0.01. P. Tipraqsa et al. / Agricultural Systems 94 (2007) 694–703 701 4. Discussion high cost of purchased inputs such as animals and seeds. External inputs are therefore a necessary ingredient in a This study suggested that the framework of multifunc- regenerative agriculture. This ﬁnding does not support tional agriculture is useful as it allows the assessment of the notion of eliminating external inputs in the IFS. the performance of both economic returns and ecosystem Because of high start-up costs, commercial farms might services in farming systems. The idea behind integrated be constrained in their decision to switch to integrated farming is to provide multiple beneﬁts to the farm house- farming. Many farms already have high levels of debt holds and hence only looking at economic returns or crop and large new loans might not only be diﬃcult to get but yields would not be enough. also bring substantial risk to the farm households. The study relied on a purposive sample of eight inte- One of the reasons stated by farmers for adopting inte- grated farms that were matched with eight adjacent com- grated farming is to create an opportunity for their chil- mercial farms. Each farm was visited about seven times dren to continue working on the farm; farmers also while soils, vegetation, and resource ﬂows were studied suggested this as an economic indicator for the study intensively. One alternative would have been to collect less (see Section 2.3). Enterprises such as vegetable growing, in-depth information but to include all 21 integrated farms ﬁsh ponds and animal herding are relatively labour inten- in the catchment or to take a large stratiﬁed random sam- sive. This is also one reason why integrated farms do not ple from a larger area in Northeast Thailand. We believe, use as much mechanisation as their commercial counter- however, that the high quality of data provided by the parts (Table 2): a tractor would reduce the need for their small purposive sample outweighs the gains of a large ran- children to work on the farm. A decline in agricultural dom sample with unknown data errors, though the optimal labour – as is commonly reported for Thai agriculture balance between sample size and data quality is of course (Lightfoot et al., 1983; Hussain and Doane, 1995; Chalam- debatable. wong, 2001) – could, however, prevent farmers from One diﬃculty in applying this conceptual framework is a exploiting the beneﬁts from resource integration, as some possible bias in the selection of appropriate indicators for labour-intensive enterprises, such as animal herding, might diﬀerent dimensions. While the study chose to look at have to be abandoned when labour becomes scarce. The enhancing food security in terms of amounts and richness alternative would be for integrated farms to adopt more of species used for food and the share of home-produced mechanisation; although this would contradict the idea food, it considered neither the nutritional status of individ- of self-suﬃciency, the analysis has shown that integrated ual household members nor food collected from the vicin- farms already use as many external inputs as commercial ity of the farm, such as products from the communal native farms do. forest, which also inﬂuences food consumption. Neverthe- less, it may be hypothesised that household members of 5. Conclusion IFS are better nourished because of a more diverse diet, all other things being equal. Similarly, there is little reason The study showed that the assessment of integrated to believe that food collected from the vicinity would bias farming systems can be expanded to include food security, the results since the farms are located in the same areas environmental, and social functions in addition to the eco- and all households have equal access to the communal nomic functions that usually get most attention. The areas. framework of multifunctional agriculture oﬀers a useful The study considered the practice of resource integra- tool for studying these multiple functions in a country such tion as the main indicator distinguishing integrated from as Thailand. non-integrated agriculture. For each farm, the level of inte- Based on a carefully selected sample of eight integrated gration was calculated by counting the synergies between and eight commercial farms this study showed that the enterprises on the farm. This simple count gives equal integrated farming system outperforms the commercial weight to all enterprises, though they are not all equally farming system in all four dimensions of a multifunctional important to the farm household. We therefore suggest agriculture: it gives a more secure supply of food, it improving this measure by applying weights to each enter- improves the resource base, creates higher economic prise; for instance, by multiplying each count by the enter- returns, and better matches the social needs of agriculture prise’s proportion in the on-farm labour use, or by its as a supplier of materials for food, medicines, local rituals, proportion in the farm gross margin. This could not be tools, and shading. done in the present study as such detailed data at the enter- Past development eﬀorts in the Northeast aimed at prise level had not been collected. increasing food production by introducing high yielding The practice of integrated farming enables the farm varieties of rice in combination with mineral fertilizers. households in this study area to increase agricultural pro- The results of this study do not support such policy but duction while not depleting their natural resource base. rather suggest a two-pronged strategy of diversiﬁcation However, during the interviews farmers explained that and resource integration. A reduction in the use of external the success of the IFS relates to a high start-up cost for inputs, as suggested by the King’s Policy of Self Suﬃciency the establishment of ponds, landscape levelling, and the is, however, not supported by our ﬁndings. 702 P. Tipraqsa et al. / Agricultural Systems 94 (2007) 694–703 For farmers to switch from commercial to integrated Galpin, M., Dorward, P.T., Shepherd, D.D., 2000. Participatory Farm farming, the high start-up costs might be a constraint while Management (PFM) Methods for Agricultural Research and Exten- sion: A Training Manual. The University of Reading, UK. increasing labour scarcity could put pressure on integrated Garcia, S., 1997. Indicators for sustainable development of ﬁsheries. In: farms to maintain their diversiﬁed resource base. Further Benites, J.R., Tschirley, J.B. (Eds.), Land Quality Indicators and Their research could focus on identifying the farms for which Used in Sustainable Agriculture and Rural Development. FAO, integrated farming is a viable option, and identifying ways Rome, Italy, pp. 131–154. in which farmers who are relatively poor in resources can Gomiero, T., Giampietro, M., Bukkens, S.G.F., Paoletti, M.G., 1999. Environmental and socioeconomic constraints to the development of be made to beneﬁt from resource integration. freshwater ﬁsh aquaculture in China. Crit. Rev. Plant Sci. 18 (3), 359– 371. Acknowledgements Halwart, M., Bartley, D., Burlingame, B., Funge-Smith, S., James, D., 2006. FAO regional technical expert workshop on aquatic biodiversity, This article draws on the ﬁrst authors PhD research at its nutritional composition, and human consumption in rice-based systems. J. Food Compos. Anal. 19 (6–7), 752–755. the Centre for Development Research (ZEF-Bonn). The re- Hussain, M.J., Doane, D.L., 1995. Socio-ecological determinants of land search was made possible by a ﬁnancial grant of the Ger- degradation and rural poverty in northeast Thailand. Environ. man Academic Exchange Service (DAAD) and the Conserv. 22 (1), 44–50. International Water Management Institute (IWMI). The Jamu, D.M., Piedrahita, R.H., 2002. An organic matter and nitrogen article greatly beneﬁted from the comments of the journal dynamics model for the ecological analysis of integrated aquaculture/ agriculture systems: II, model evaluation and application. Environ. editor and two anonymous reviewers. Model. Softw. 17 (6), 583–592. Johnson, R.B., 1997. Examining the validity structure of qualitative References research. Education 118 (2), 282–292. Konboon, Y., Wijnhoud, J.D., Lefroy, R.D.B., 2001. Nutrient manage- ADB, 1998. Regional Technical Assistance for Third Agriculture and ment in rainfed lowland rice-based systems in northeast Thailand. In: Natural Resources Research at CGIAR Centers – Improving Man- ´ Kam, S.P., Hoanh, C.T., Trebuil, G., Hardy, B. (Eds.), Natural agement of Natural Resources for Sustainable Rainfed Agriculture. Resource Management Issues in the Korat Basin of Northeast Asian Development Bank, Manila, The Philippines. Thailand. IRRI, Manila, Philippines, pp. 137–146. Alsagoﬀ, S.A.K., Clonts, H.A., Jolly, C.M., 1990. An integrated poultry, Lamprecht, H., Pancel, L., 1993. Silviculture in the tropical natural multi-species aquaculture for malaysian rice farmers: a mixed integer forests. In: Pancel, L. (Ed.), Tropical Forestry Hand Book. Springer, programming approach. Agr. Syst. 32 (3), 207–231. Heidelberg, Berlin, pp. 727–810. Berg, H., 2002. Rice monoculture and integrated rice-ﬁsh farming in the Lightfoot, C., Pullin, R.S.V., 1994. Why an integrated resource manage- Mekong Delta, Vietnam – economic and ecological considerations. ment approach? In: Burmmett, R.E. (Ed.), Aquaculture Policy Options Ecol. Econ. 41 (1), 95–107. for Integrated Resource Management in Sub-Saharan Africa. ICL- Bonham, C.D., 1989. Measurements for Terrestrial Vegetation. Wiley, ARM Conference Proceeding, pp. 7–9. New York. Lightfoot, C., Bimbao, M.P., Dalsgaard, J.P.T., Pullin, R.S.V., 1993. Chalamwong, Y., 2001. Recent trends in migration ﬂows and policies in Aquaculture and sustainability through integrated resources manage- Thailand. TDRI Quart. Rev. 16 (2), 10–18. ment. Outlook Agricul. 22 (3), 143–150. Conway, G.R., Barbier, E.W., 1991. After the Green Revolution: Lightfoot, P., Hull, U.K., Fuller, T., 1983. Circular rural–urban move- Sustainable Agriculture for Development. Earthscan Publications ment and development planning in northeast Thailand. Geoforum. 14 Limited, London. (3), 277–287. Coxhead, I., Plangpraphan, J., 1999. Economic boom, ﬁnancial bust, and Naegel, L.C.A., 1994. Development of an integrated aquafarming system. the decline of Thai agriculture: was growth in the 1990s too fast? Aquacult. Int. 2 (1), 1–9. Chulalongkorn J. Econ. 11 (1). Nelson, D.W., Sommers, L.E., 1996. Total carbon, organic carbon, and Craswell, T.E., 2002. Maintaining Soil and Water Quality: The Founda- organic matter. In: Sparks, D.L. (Ed.) Methods of Soil Analysis: Part tion of Sustainable Land Management. The International Commission 3–Chemical Methods. pp. 961–1010. for Earth Sciences in Africa (ICESA). Noble, A.D., Gillman, G.P., Ruaysoongnern, S., 2000. A cation exchange Dalsgaard, J.P.T., Oﬁcial, R.T., 1997. A quantitative approach for index for assessing degradation of acid soil by further acidiﬁcation assessing the productive performance and ecological contributions of under permanent agriculture in the tropics. Eur. J. Soil Sci. 51, 233– smallholder farms. Agr. Syst. 55 (4), 503–533. 243. Daly, B., 1992. Recommended Methods for Soil and Plant Analysis. The OAE, 2003. Agricultural Statistics of Thailand Crop Year 2002/2003. International Board for Soil Research and Management, Bangkok, Oﬃce of Agricultural Economics, Centre for Agricultural Information, Thailand. Ministry of Agriculture and Cooperatives, Bangkok, Thailand. Devendra, C., Thomas, D., 2002a. Smallholder farming systems in Asia. OECD, 2001. Multifunctionality: Towards and Analytical Framework. Agr. Syst. 71 (1–2), 17–25. Organization for Economic Cooperation and Development, Paris Devendra, C., Thomas, D., 2002b. Crop–animal interactions in mixed Cedex, France. farming systems in Asia. Agr. Syst. 71 (1–2), 27–40. Pant, J., Demaine, H., Edwards, P., 2005. Bio-resource ﬂow in integrated Dey, M.M., Kambewa, P., Prein, M., Jamu, D., Paraguas, F.J., Pemsl, agriculture-aquaculture systems in a tropical monsoonal climate: a D.E., Briones, R.M., 2006. Impact of development and dissemination case study in northeast Thailand. Agr. Syst. 83 (2), 203–219. of integrated aquaculture-agriculture (IAA) technologies in Malawi. Paris, T.R., 2002. Crop–animal systems in Asia: socio-economic beneﬁts NAGA, WorldFish Center Quarterly 29 (1–2). and impacts on rural livelihoods. Agr. Syst. 71 (1–2), 147–168. Dillon, J., Hardaker, J.B., 1993. Farm Management Research for Small Pitcher, T.J., Preikshot, D., 2001. Rapﬁsh: a rapid appraisal technique Farmer Development. FAO, Rome, Italy. to evaluate the sustainability status of ﬁsheries. Fish Res. 49, 255– Dillon, J., McConnell, D.J., 1997. Farm Management for Asia: A Systems 270. Approach. FAO, Rome, Italy. Potter, C., Burney, J., 2002. Agricultural multifunctionality in the WTO– Frei, M., Becker, K., 2005. Integrated rice-ﬁsh culture: coupled produc- legitimate non-trade concern or disguised protectionism? J. Rural. tion saves resources. Nat. Resour. Forum. 29 (2), 135–143. Stud. 18 (1), 35–47. P. Tipraqsa et al. / Agricultural Systems 94 (2007) 694–703 703 Prein, M., 2002. Integration of aquaculture into crop-animal systems in Tipraqsa, P., 2005. The opportunities and constrains of integrated farming Asia. Agr. Syst. 71 (1–2), 127–146. system in Northeast Thailand. In: A Case Study of Huai Nong Ian Ruaysoongnern, S., Penning de Vries, F., 2005. Learning alliances Catchment, Khon Kaen Province. Cuvillier Verlag, Goettingen, development for scaling up of multi-purpose farm ponds in a semi- Germany. arid region of the Mekong Basin. In: Smits, S., Fonseca, C., Pels, J. Vereijken, P., 1997. A methodical way of prototyping integrated and (Eds.), Proceedings of the Symposium on Learning Alliances for ecological arable farming systems (I/EAFS) in interaction with pilot Scaling up Innovative Approaches in the Water and Sanitation sector farms. Eur. J. Agron. 7 (1), 235–250. held in Delft, The Netherlands. Vereijken, P., 2003. Transition to multifunctional land use and agriculture. Smitinand, T., Kanjunt, C., Oberhauser, U., Ramsiri, B., 1994. Study on NJAS-Wagen. J. Life Sc. 50, 171–180. Forest Structure of Diﬀerent Stages of Succession on Old Farm Plots Viyakorn, P., 2001. The role of trees in countering land degradation in of Mixed Deciduous and Hill Evergreen Forest. Forest Research cultivated ﬁelds in Northeast Thailand. Southeast Asia Studies 39 (3), Oﬃce, Royal Forest Department, Bangkok, Thailand. 398–416. Smyth, A.J., Dumanski, J., 1993. FESLM: An International Framework Wijnhoud, J.D., Konboon, Y., Lefroy, R.D.B., 2001. Accounting for for Evaluating Sustainable Land Management. Report No. 73. FAO, environmental and socioeconomic sustainability in Northeast Thai- Rome, Italy. land: towards decision support for farmers and extension workers. In: Suwanraks, R., 2000. Suﬃciency economy. TDRI Quart. Rev. 15 (1), 6– Simmons, R.W., Noble, A.D., Lefroy, R.D.B. (Eds.), Nutrient Balance 17. for Sustainable Agricultural Production And Natural Resource Talpaz, H., Tsur, Y., 1982. Optimising aquaculture management of a Management in Southeast Asia. International Water Management single-species ﬁsh population. Agr. Syst. 9 (2), 127–142. Institute.