Sustainable agriculture seeks to create an economically viable, socially supportive, and environmentally sound farming system. However, there is limited empirical research on measuring agricultural sustainability, its components, and spatial distribution. Drawing data from the 2012 Census of Agriculture, a confirmatory factor analysis helps identify a three-factor structure of sustainable farming across counties in the United States: the environmental sustainability component refers to practices that reduce environmental degradation; the economic dimension highlights the efficiency of agricultural production; and the social component identifies practices which tend to promote economic viability and social support for communities. A cluster analysis on factor score indexes further indicates that the typology of sustainable agriculture includes: a majority of counties in the high environment & low economy and moderate intensity categories are widespread in this country; a low environment & high economy group is concentrated in the Northern Great Plains and in Iowa and Illinois; a few counties in the extremely low environment & extremely high economy cluster are located along the Lower Mississippi Valley, and dispersed throughout the Midwest, California, and Florida; and a small number of counties in the high social group are primarily concentrated in the New England, along the Pacific Coast, and around the Northern Great Lakes. In general, an agroecosystem framework that focuses on the interplay between social and ecological process of the agricultural system provides insights at the systems level and explicitly interprets the core attributes of sustainable agriculture, both in definition and measurement.
Key Words: environmental sustainability, economic sustainability, social sustainability, measurement, U.S. county agricultural statistics
Sustainable agriculture has emerged as a new paradigm for farming practices that differ from those associated with conventional agriculture. A sustainable mode of agricultural production, according to Ikerd, is "capable of maintaining its productivity and usefulness to society indefinitely. Such an agriculture must use farming systems that conserve resources, protect the environment, produce efficiently, compete commercially, and enhance the quality of life for farmers and society overall (1993: 151)." In other words, sustainable agriculture seeks to create "integrated, resource-conserving, and equitable farming systems", and sustainable production practices also promote economically viable, socially supportive, commercially competitive, and environmentally sound farming (Allen, Van Dusen, Lundy, & Gliessman, 1991; Francis & Youngberg, 1990).
To investigate the features of sustainable agriculture, previous studies have developed concepts and models to illustrate sustainable modes of farming, while paying little attention to its empirical components, typologies, and national spatial distribution. Drawing data from the 2012 Census of Agriculture, this study examines the structure of sustainable agriculture in the United States to more accurately delineate sustainability as a multifaceted concept, and to classify different sustainable farming segments. Such analyses advance studies on sustainable agriculture in three ways. Firstly, this study assesses the components of agricultural sustainability from an empirical perspective to encompass the well-defined multidimensional concept. Secondly, this study helps explain how sustainable farming is structured nationally. While previous findings have provided useful definitions for identifying key features of sustainability, the current study tests the measure of agricultural sustainability across meso-scale geographies (Lobao, 1996) in the United States. Finally, the empirical analyses help classify different sustainable farming segments and their spatial distribution. I plan to identify the composition of sustainable farming by means of measurement modeling, which may pinpoint the need for further exploring the comparability of measures in order to examine the disparity of sustainability across nationwide geographies. In sum, the current study aims to advance conceptual and methodological contributions to the existing literature amid the recent surge of more empirical evidence in the United States.
II. Dimensions of Agricultural Sustainability
Advocates of sustainable agriculture generally suggest that sustainable farming provides a holistic paradigm presenting production units as organisms whose component parts interact in a state of equilibrium, produce sufficient yields, and to react as a whole to external physical, biological, and socioeconomic stimuli and limits (Francis & Madden, 1993; Ikerd, 1993). In other words, agricultural sustainability is a multi-dimensional concept which includes environmental, economic, and social components. In a more sociological context, sustainable agriculture is often considered as an activity which maintains a particular social system. Agriculture not only produces food but also sustains rural landscapes, protects biodiversity, generates employment, and contributes to the viability of rural areas. Sustainable agriculture in particular has the potential to foster rural cohesion, amplifying the many social, environmental, economic, and rural community benefits. A body of literature, therefore, is surfacing to tackle the growing links between the environment, on-farm sustainability, and social justice, reflecting an emerging systemic understanding of agriculture as an ecological and social activity (Fernandez, Goodall, Olson, & Mendez, 2013).
As a whole, sustainable agriculture focuses on optimizing the whole production system. Emerging as a response to the negative externalities of conventional agriculture, agroecology has contributed to the development of the concept of sustainability in agriculture (Altieri, 1987). In its early stage, agroecology has been variously defined as the ecology of agriculture, being used to study, diagnose, and propose alternative low-input management of agroecosystems. Its whole-systems approach and knowledge of dynamic equilibrium has provided a theoretical basis for sustainability. Although agroecology initially dealt primarily with crop production and farm ecology, in recent decades it was enriched with integrative concepts and methods from the social sciences (Hecht, 1995; Mendez, Bacon, & Cohen, 2013; Wezel et al., 2009). The field of agroecology then evolved toward a system-based, transdisciplinary, and multidimensional approach to help us better understand the complexity of agriculture that emerges from unique social and cultural contexts.
A primary foundation of agroecology is the concept of ecosystem, defined as a functional system of complementary relations between living organisms and their environment, which in space and time appears to maintain a steady yet dynamic equilibrium. In other words, an ecosystem has structural components with particular relationships that together take part in dynamic processes. In an agricultural setting, an agroecosystem is created when human manipulation and alternation of ecosystems take place for the purpose of establishing agricultural production (Gliessman, 2014). When we consider farm systems as agroecosystems, this concept then provides a foundation for understanding the interactions and relationships among the diverse components of the agrifood production system.
To conceptualize the interactions among ecological and social factors of the agroecosystem, a number of researchers have developed models to illustrate these complex relationships. Gliessman (2014), for example, has suggested that a sustainable agroecosystem develops when components from a broad social and ecological foundation are combined into a system with a structure, function, and coevolution that reflects the interaction of human knowledge and preferences with the ecological components of the agroecosystem (Figure 1). In general, the agroecosystem concept involving the interplay between social, economic, and environmental process of the agricultural system serves the core of sustainable agriculture, both in definition and measurement.
Empirical research on sustainability and sustainable agriculture is already quite extensive but nonetheless growing rapidly. While a complex concept and difficult to implement, previous studies have proposed various parameters and indicators to evaluate agricultural sustainability. In general, scholars have proposed that agricultural sustainability consists of three key dimensions: ecological, economic, and social components. In terms of the ecological aspect, the core concerns are to enhance nutrients, organic materials, and local resource conservation, and reduction of negative environmental externalities. The indicators of this dimension may include usage of chemicals, inorganic fertilizers, and synthetic pesticides, of animal/organic manures, crop agrodiversity, cropping patterns, cover crops, energy, and water resource management (Neher, 1992).
The economic dimension of agricultural sustainability highlights systems undergirding farmers' livelihoods and measures of production such as yield, productivity, and competition. Appropriate indicators for measuring economic criteria may include farm labor, average of crop production, monetary income from, and outside of, the farm, and market availability (Hayati, Ranjbar, & Karami, 2011). The social dimension is usually associated with food security, farmer participation, promotion of local institutions, culture and farming communities, and contribution to employment generation. The appropriate socioeconomic indicators may include land tenure, nutritional and health status of family members, farmer's degree of satisfaction, the quality of rural life, and social equity (Flores & Sarandon, 2004; Senanayake, 1991). Overall, as Francis and Youngberg (1990) have suggested, sustainable agriculture is based on human goals and on understanding the long-term impact of our activities on the environment and other socioeconomic species. Typically, agricultural sustainability is defined primarily in terms of resource conservation and profitability, and of social aspects of the food and agriculture system (Allen et al., 1991; Novak & Goodman, 1995).
The complexity of agricultural sustainability comes from its multifaceted components and various methodological approaches and techniques for measuring and assessing the sustainability of agriculture (Zinck et al., 2004). Generally, these assessment approaches involve different scales of analysis, such as field or land parcel, farm, community, and nation. While studies on of the measurement of agricultural sustainability at the field-level are more likely to focus on the ecological and/or economic aspect(s) (Pimentel, Hepperly, Hanson, Douds, & Seidel, 2005; Reganold, Glover, Andrews, & Hinman, 2001), research on monitoring sustainable agriculture at the national-level tend to integrate the ecological, economic, and social dimensions of sustainability in order to evaluate certain agricultural policy impacts and to ensure successful implementation of sustainable systems for a country (Lehtonen, Aakkula, & Rikkonen, 2005). Given the site-specific nature of sustainability management, numerous studies on measuring agricultural sustainability at the farm-level have used a variety of indicators, directly measured or ranked in their models, to reflect different aspects of sustainability (Frater & Franks, 2013; Nambiar, Gupta, Fu, & Li, 2001). Nevertheless, tools currently available to examine the development of sustainable agriculture at the mesoscale, such as communities or counties, are rarely available.
Quantitative analyses of sustainable agriculture in previous research are usually based on primary or secondary data sources. Primary data sources rely on the perceptions of participants, such as farmers and stakeholders, to determine the composite variables of sustainability measures (Biggelaar & Suvedi, 2000; Pilgeram, 2011; Taylor, Mohamed, Shamsudin, Mohayidin, & Chiew, 1993). Hence, such approaches typically provide relative measures of sustainability with scores assigned to management practices by comparing differences in the perceptions of the individuals concerned. Related indexes such as the agro-ecosystem performance index (API) (Bonny, Prasad, & Paulose, 2010), the composite sustainability index (CSI) (Frater & Franks, 2013), and the global sustainability index (Sg) (Castoldi & Bechini, 2010) are based on such participatory methodological paradigms. The other approach is to employ a variety of secondary data sources on selected variables to construct composite indexes through different models or simulation scenarios. Related indexes, such as the agricultural sustainability index (ASI) in the Zhejiang coastal zone of China (Nambiar et al., 2001), the composite indicators of agricultural sustainability (CIAS) in the Duero basin of Spain (Gomez-Limon & Riesgo, 2009), and the environmental sustainability index (ESI) in southeastern Colorado (Sands & Podmore, 2000), are proposed to access the characteristics of agricultural sustainability in different regions. However, little has been revealed about the features of sustainable agriculture in the United States by employing such quantitative measures.
III. Data and Methods
The present research uses data taken from the 2012 Census of Agriculture, which is the latest wave of census conducted by the U.S. Department of Agriculture's National Agricultural Statistics Service (NASS, USDA). The Census of Agriculture has been recognized as the leading and official source of statistical information regarding agricultural production in the United States. The units of analysis in this study are counties in the 48 contiguous states, excluding Alaska and Hawaii due to their unique economies and geographies. The valid sample consists of a total of 3,067 counties, excluding San Juan in Colorado and Hudson in New Jersey because there are no farms in these counties.
To better understand the structural components of sustainable agriculture in the United States, I first use an exploratory factor analysis (EFA) to identify latent variables from interrelations among observed variables. With a small number of factors, the EFA can help uncover succinct latent dimensions of sustainable farming, although a prior model design based on relevant theories is not specified. The observed items measure various farm production expenses, farming practices, and the demographics and financial wellbeing of producers.
Observed items include the following: chemicals purchased per farm ($1,000), gasoline, fuels, and oils purchased per farm ($1,000), farms practicing rotational or management-intensive grazing (%), manure used (%), organic farms (%), farms marketing products through community supported agriculture (CSA) (%), farms selling products directly to individuals for human consumption (%), family or individual farms (%), farms whose principle operators are tenants (%), operators whose primary occupation is farming (%), hired farm labor per farm (number), market value of agricultural products sold per farm ($1,000), and net farm income of the operations per farm ($1,000). These individual values are converted into corresponding standard scores, which are standard deviations from the mean, to remove scale differences.
The EFA, based on the Kaiser-Guttman criterion and other factor determination methods such as scree plots of eigenvalues, help identify three distinct factors, accounting for 65% of the variance explained in the data. The first factor grouped such items as chemicals purchased, fuels purchased, rotational grazing, operators whose primary occupation is farming, hired labor, value of products, and net income. The second factor combined such farming and marketing practices as organic farms, CSA farms, and farms selling to individuals. The third component grouped items including manure used, family or individual farms, and tenant farming. As the third latent variable may not represent a distinct dimension exclusive to a common frame, the second-stage proceeds using the ten observed variables outlined above, excluding the three items extracted from the third factor, in the subsequent analyses.
To confirm a more rigid structural model, what follows is a confirmatory factor analysis (CFA) from structural equation modeling (SEM) to test a hypothesized model of latent constructs and examin whether the data fit these latent constructs. As an extension of measurement models, the CFA accounts for both random and nonrandom measurement errors in observed items to generate consistent and stable estimates of the latent traits in structural models (Bollen, 1989). According to the agroecosystem framework and the latent structure derived from the EFA, it is possible to specify how each latent construct consists of different indicators (see Figure 2).
Based on a review on relevant literature, a theoretical covariance among the three latent constructs is assumed: environmental, economic, and social sustainability. To better test errors and improve model fit in the measurement, three further correlated error terms are added between observed indicators: chemicals purchased and net income; fuels purchased and value of products; and value of products and net income. This confirmatory factor analysis meets such assumptions as having a general normal distribution and an adequate sample size, without perfect multicollinearity. Furthermore, a range of fit statistics, such as incremental fit index and root mean square of approximation, enable an overall assessment of model fit to examine whether this model adequately reproduces the observed relations across individual counties.
In addition, to characterize the typology of sustainable agriculture, a hierarchical agglomerative cluster analysis is employed to identify and describe the classification of sustainable farming at the county-level. Agglomerative cluster analysis is a multivariate statistical procedure for attempting to reorganize a sample of entities into relatively homogenous groups (Everitt, Landau, Leese, & Stahl, 2011). In this analysis, squared Euclidean distance is used to measure distances between clusters and counties based on environmental, economic, and social sustainability index scores derived from the CFA. The formula for squared Euclidean distance is based on the Minkowski generalized distance metric and is given in Equation 1, where [d.sub.ij] is the distance between counties i and j, [x.sub.ik] is the value of the [k.sup.th] index score of sustainability measurement for the [i.sup.th] county, and [x.sub.jk] is the value of the [k.sup.th] index score of sustainability measurement for the [j.sup.th] county (Everitt et al. 2011). Clusters and counties are joined together using Ward's minimum variance method, which minimizing the within-cluster sum of squares by merging two clusters from the previous generation, E, until all counties are grouped into one cluster. The formula for Ward's method is presented in Equation 2, where [[bar.x].sub.m,k] is the mean of the [m.sup.t]h cluster for the [k.sup.th] index score; and [x.sub.ml,k] is the score on the [k.sup.th] standardized value (k=1 ... p) for the [l.sup.th] county (l=1 ... [n.sub.m]) in the [m.sup.th] cluster (m=1 ... g) (Everitt et al. 2011). As the Ward's method is sensitive to outliers, standard scores larger (lower) than 4 (-4) are winsorized in order to maintain high (low) sustainable counties while alleviating the effects of skewness due to extreme outliers.
[mathematical expression not reproducible] (1)
[mathematical expression not reproducible] (2)
To clarify the latent relationships among measures of sustainability, a CFA is conducted to generate consistent estimates in structural models, as shown in Figure 2. The CFA results are indicative of a moderate fit of the model to the data. While the significance X2 tests rejected the null hypothesis that the implied model covariance adequately reproduced the observed covariance, the excessive power that came from a large sample size made the [x.sup.2] tests restrictive. The incremental fit index (IFI) and the Tucker-Lewis index (TLI), which measure the proportional reduction in [x.sup.2] values from the baseline to the hypothesized model, while adjusting for degrees of freedom, are alternative indices commonly used to estimate goodness of fit. For both fit statistics, which range from 0 to 1, values above .90 indicate a good fit. The IFI and TLI in this model turned out to be .929 and .878 respectively. Furthermore, the root mean squared error of approximation (RMSEA) measures the degree of mismatch between the observed and implied covariance matrix. Typically, values between .05 and .10 indicate moderate fit. Overall, these results indicate that such structural models represent a decent latent construct of sustainability measures.
The component fit further reveals the model parameter estimates (Figure 2). The standardized factor loadings represent how the observed variables change, in standard deviation units, for each standard deviation unit change in the latent factor. The significance of factor loadings was tested using a t-distribution. All factor loadings for sustainability latent constructs were significant, thus confirming that the observed variables in the hypothesized model were valid measures of the three latent constructs (environmental, economic, and social sustainability).
Moreover, the positive covariance between environmental and social sustainability indicated that these two dimensions of agricultural sustainability are positively correlated. The connection between social and economic sustainability, however, turns out to be negative, although the association was minor. Notably, counties of higher economic sustainability tend to be environmentally unsustainable as the results indicate a negative covariance between these two latent constructs and the magnitude of such correlation is quite high.
To better examine the spatial distribution of agricultural sustainability, Figure 3-5 maps the location of the sustainability index scores derived from the CFA. Counties with higher environmental sustainability (Figure 3) are primarily located in the Western Mountains, the New England, the Northern Great Lakes, the interior of the South Atlantic, around the Missouri-Arkansas-Oklahoma border, along the northern Pacific Coast of Washington and Oregon, and in states such as Kentucky, Tennessee, and Texas. The Moran's I further indicates a high degree of positive spatial correlation regarding the intensity of sustainability (I=0.694), which means that counties of high (low) environmental sustainability tend to be located near other high (low) environmentally sustainable counties.
The spatial distribution of economic sustainability (Figure 4), however, looks much different from that of environmental sustainability. High intensity areas tend to be concentrated in the Midwest, specifically in the Dakotas, Nebraska, Iowa, Illinois, and Kansas. Some sporadic areas are located along the coast of the South Atlantic, in Georgia, Florida, Washington, and California. [Text incomplete in original source.] the Lower Mississippi Valley. The spatial statistics indicate that economic sustainability is spatially dependent (Moran's I= 0.680), rather than randomly distributed geographically.
In terms of social sustainability (Figure 5), counties with higher scores are concentrated along both coasts of the United States. These counties are primarily located in the Northeast, the Pacific coasts of California, Washington, and Oregon, and around the Northern Great Lakes, while others are dispersed in the Western Mountains and the South Atlantic. The social dimension of agricultural sustainability is also found to be spatially clustered in certain areas (Moran's I= 0.709). In sum, these findings revealed that sustainable farming is unevenly distributed in the United States. Regional variations are markedly found in each of the three dimensions of farming practices across meso-scale geographies.
To characterize the typology of sustainable agriculture, a cluster analysis is employed to classify the sustainable farming at the county-level. Based on the environmental, economic and social sustainability index scores, the results indicate that the 3,067 counties can be divided into five groups, which account for 83 percent of the original variance in the data. The presence of five clusters is based on loss of information diagnostics, as shown in Table 1. The distance statistics show a loss of information at State 4, suggesting a five cluster solution. Breaks in Mojena value at Stage 2 and 4 indicated a three or five cluster solution. Relatively high values of pseudo F indicated the presence of five or seven clusters. To validate the solution, cluster analysis using average linkage method is run which generally produces similar cluster groupings. All the assumptions of cluster analysis are met, although there are some moderate outliers in the data.
The means of the environmental, economic, and social sustainability index across these clusters reveal five distinct groups (Table 2). The majority of counties in the United States (1,682 or 54.8%) fell in the high environment & low economy category. While their social sustainability index is around average, the environmental index was 0.51 standard deviations above average and the economic index was 0.41 standard deviations below average. On the contrary, the low environment & high economy group consisted of 481 counties (15.7% of total) representing low rates of environmental and social sustainability practices, being 0.93 and 0.28 standard deviations below average, respectively. However, they displayed high intensity in the economic dimension, being 0.79 standard deviations above average.
While the findings reveal 651 counties (21.2% of total) expressing moderate intensity sustainable farming, the high social category of 131 counties (4.3%) represents considerably high rates of social sustainability. Notably, the extremely low environment & extremely high economy group includes 122 counties (4.0% of total) with very low rates of environmental sustainability, and very high rates of sustainable practices in the economic dimension. In sum, these distinct groups reveal substantial variations across counties in different measures of agricultural sustainability.
To identify their spatial distribution, Figure 6 maps these sustainable agriculture clusters. In terms of geography, the high social category was primarily located in New England, along the Pacific coast of California, Washington, and Oregon, and around the Northern Great Lakes. Counties in the extremely low environment & extremely high economy group are particularly concentrated in the Lower Mississippi Valley. The others in this group are dispersed in states in the Midwest and in California, North Carolina, and Florida, for example. While both of the high environment & low economy and moderate intensity groups tend to be widespread in the United States, they seem to be absent in the Midwest. Counties in the low environment & high economy group, however, are primarily located in the Northern Great Plains, Iowa, and Illinois.
V. Discussion and Conclusion
Although much of the literature on sustainable agriculture seeks to define the concept and suggests performance indicators to promote sustainability, current tools that monitor the sustainability of agriculture at the county-level across the United States are not readily available. Conceptually, attempting to narrow down a precise definition of sustainable agriculture is problematic because there are conflicts among different parties and farmers over how to achieve sustainability, due to different regional and socioeconomic backgrounds (National Research Council, 2010; Rigby & Caceres, 2001).
Based on the structure and function of agroecosystems, studies in general agree that sustainable agriculture has several basic characteristics and key latent components. One interpretation of sustainability which has received considerable attention is resource conservation and profitability. Such efforts focus on indicators such as preserving soil productivity, reducing the use of fertilizers, pesticides, and cultivation, and increasing farming profits (Novak & Goodman, 1995). Moreover, an expanding concept of agricultural sustainability is defined in terms of pressing social problems in food and agricultural systems. The key ideals of sustainability, therefore, are related to food security, social justice, and intergenerational equity (Allen et al., 1991; Smit & Smithers, 1993). Taken together, an inclusive concept of agricultural sustainability is one which is environmentally sound, economically viable, socially supportive, and which serves as the foundation for fostering future generations.
Understanding the features of sustainable agriculture and its geographical variation is crucial to furthering our knowledge of how U.S. farming is organized. To quantify the main attributes of sustainable agriculture from an agroecosystem approach, one must have a selection of indicators which are comparable over time or space to examine the latent relationships among measures of agricultural sustainability. Using data taken from the 2012 Census of Agriculture, a confirmatory factor analysis is conducted to generate consistent estimates in structural models at the county-level. The results identified three distinct latent constructs: environmental, economic, and social sustainability.
Among these three latent constructs, the first factor grouped variables which quantify the expense of chemicals including insecticides and herbicides, all purchases of gasoline, fuels and oils, and farms practicing rotational grazing. The benefits associated with popular rotational grazing systems usually include reducing field erosion, improving soil structure, enhancing organic matter, and ultimately improving animal health and increasing pasture productivity (Finch, Samuel, & Lane, 2014). The environmental sustainability component, therefore, refers to farming practices that reduce environmental degradation and promote resource conservation. The economic sustainability component, which combines indicators including hired farm labor, value of products, net income, and operators whose primary occupation is farming, generally highlights farmers' livelihood and the efficiency of agricultural production. The third factor, which combines organic farming, CSA farming, and farms selling to individuals, tends to signify higher community engagement and sustainable livelihood security through local market linkages (DeLind, 2002; Getz, 2008). Thus, the social sustainability component helped identify practices likely to promote economic viability and social support for rural communities. The core idea in the latent variable approach is that for the measurements to be consistent and stable the data needs to fit the hypothesized latent constructs as a whole.
In addition, the significant covariance between these three latent dimensions of sustainable agriculture demonstrates that the structural components of agricultural sustainability are correlated and that the components may interact and together take part in a dynamic process. Although a sustainable mode of agriculture system focuses on the interactions among the broader environment and the social community, the findings indicate that the association between environmental and the economic sustainability are negatively correlated. The results somewhat support what the literature has shown regarding the connection between industrial farming and environmental pollution characteristics (Harun & Ogneva-Himmelberger, 2013).
To identify the classification of sustainable agriculture, one needs multiple measures that examine the disparity of sustainability across nationwide geographies. Based on a cluster analysis on factor score indexes derived from the latent constructs, results indicate that the typology of sustainable agriculture include a small number of counties in the extremely low environment & extremely high economy and the high social groups, a moderate intensity cluster with average sustainability rates, a low environment & high economy group saw above average scores on the economic dimension but below average environmental and social scores, and a majority of high environment & low economy places where their environmental sustainability scores were above average while economic scores were below average.
In terms of geography, the regional distribution of socially and environmentally sustainable farming is fairly concentrated in New England, the western mountain regions, around the Great Lakes, the interior of the southern Atlantic coast, and along the northern Pacific coast. On the contrary, there is an intense concentration of economically sustainable farms in the Lower Mississippi Valley and the Midwest, especially in the Northern Great Plains, Iowa, and Illinois. These places are typically dominated by conventional agriculture and may have received substantial government support to produce commodity crops. Since the findings indicate that economic sustainability is at odds with environmental sustainability, conservation assistance and subsidies targeting these areas should compensate farmers for addressing the social and environmental pillars of sustainability.
Overall, there are several key findings in this research that suggest opportunities and directions for future research on sustainable agriculture in the United States. First, sustainable farming practices may contain three distinct dimensions: environmental, economic, and social. The regional disparities of sustainability between these dimensions of farming practices in the United States provide evidence that agricultural sustainability is conceptually and empirically multi-dimensional. The agroecosystem framework, which focuses on the interplay between social and ecological process of the agricultural system, provides insights at the systems level and explicitly interprets the core attributes of sustainable agriculture, both in definition and measurement.
While agricultural sustainability usually represents a transformative approach combining production goals with broader social objectives, the findings reveal that the adoption of sustainable agriculture was uneven across regions in the United States. To better understand the geography of agricultural sustainability, this research suggests multiple measures of farming practices which should be site-specific and comparable across meso-scale geographies. However, this current study is still limited by parameters to evaluate agricultural sustainability. In terms of the social dimension of sustainability, for example, more indicators quantifying the quality of rural life and social equity would enrich our knowledge about social aspects of agricultural sustainability. This current study is also limited by using single year measures that cannot show increases or decreases in sustainable farming on the national-scale.
Moreover, the findings reveal that sustainable agriculture seems to be spatially dependent, which means that high (low) intensity of sustainable farming areas tend to be located near other high (low) intensity areas. For future studies to investigate the spatial patterns of sustainable agriculture and the factors that drive such geographical variations, the analyses should be aware of the "neighborhood effects" on sustainable farming conversion. While this current study employs a whole systems approach to empirically test the agricultural sustainability measures, there is a need for future studies to explain the geographic disparities of sustainability across the United States.
Allen, P., van Dusen, D., Lundy, J., & Gliessman, S. (1991). Integrating social, environmental, and economic issues in sustainable agriculture. American Journal of Alternative Agriculture, 6, 1: 34-39.
Altieri, M. A. (1987). Agroecology: The scientific basis of alternative agriculture. Boulder, CO: Westview Press.
Biggelaar, C., & Suvedi, M. (2000). Farmers' definitions, goals, and bottlenecks of sustainable agriculture in the North-Central Region. Agriculture and Human Values, 17, 4: 347-358.
Bollen, K. A. (1989). Structural equations with latent variables. New York? Wiley.
Bonny, B. P., Prasad, R. M., & Paulose, S. (2010). Agro- ecosystem Performance Index (API)--A quantitative approach to evaluate the sustainability of rice production systems. Journal of Sustainable Agriculture, 34, 7: 758-777.
Castoldi, N., & Bechini, L. (2010). Integrated sustainability assessment of cropping systems with agro-ecological and economic indicators in northern Italy. European Journal of Agronomy, 32, 1: 59-72.
DeLind, L. B. (2002). Place, work, and civic agriculture: Common fields for cultivation. Agriculture and Human Values, 19, 3: 217-224.
Everitt, B. S., Landau, S., Leese, M., & Stahl, D. (2011). Cluster analysis. West Sussex, United Kingdom: John Wiley & Sons.
Fernandez, M., Goodall, K., Olson, M., & Mendez, V. E. (2013). Agroecology and alternative agri-food movements in the United States: Toward a sustainable agri-food system. Agroecology and Sustainable Food Systems, 37, 1: 115-126.
Finch, H. J. S., Samuel, A. M., & Lane, G. P. F. (2014). Grazing management. In S. Finch, A. M. Samuel, & G. P. Lane (Eds.), Lockhart & Wiseman's crop husbandry including grassland (9th ed., pp. 499-512). Cambridge, UK? Woodhead.
Flores, C. C., & Sarandon, S. J. (2004). Limitations of neoclassical economics for evaluating sustainability of agricultural systems? Comparing organic and conventional systems. Journal of Sustainable Agriculture, 24, 2: 77-91.
Francis, C. A., & Madden, J. P. (1993). Designing the future: Sustainable agriculture in the US. Agriculture, Ecosystems & Environment, 46, 1-4: 123-134.
Francis, C. A., & Youngberg, G. (1990). Sustainable agriculture: An overview. In C. A. Francis, C. B. Flora, & L. D. King (Eds.), Sustainable agriculture in temperate zones (pp. 1-23). New York: Wiley.
Frater, P., & Franks, J. (2013). Measuring agricultural sustainability at the farm-level: A pragmatic approach. International Journal of Agricultural Management, 2, 4: 207-225.
Getz, C. (2008). Social capital, organic agriculture, and sustainable livelihood security: Rethinking agrarian change in Mexico. Rural Sociology, 73, 4: 555-579.
Gliessman, S. R. (2014). Agroecology: The ecology of sustainable food systems (3rd ed.). Boca Raton, FL: CRC Press.
Gomez-Limon, J. A., & Riesgo, L. (2009). Alternative approaches to the construction of a composite indicator of agricultural sustainability: An application to irrigated agriculture in the Duero basin in Spain. Journal of Environmental Management, 90, 11: 3345-3362.
Hayati, D., Ranjbar, Z., & Karami, E. (2011). Measuring agricultural sustainability. In E. Lichtfouse (Ed.), Biodiversity, biofuels, agroforestry and conservation agriculture (pp. 73-100). Dordrecht, the Netherlands: Springer.
Harun, S. M. R., & Ogneva-Himmelberger, Y. (2013). Distribution of industrial farms in the United States and socioeconomic, health, and environmental characterristics of counties. Geography Journal, 2013: 1-12.
Hecht, S. B. (1995). The evolution of agroecological thought. In M. A. Altieri (Ed.), Agroecology: The science of sustainable agriculture (pp. 1-20). Boulder, CO: Westview.
Ikerd, J. E. (1993). The need for a system approach to sustainable agriculture. Agriculture, Ecosystems & Environment, 46: 147-160.
Lehtonen, H., Aakkula, J., & Rikkonen, P. (2005). Alternative agricultural policy scenarios, sector modelling and indicators: A sustainability assessment. Journal of Sustainable Agriculture, 26, 4: 63-93.
Lobao, L. (1996). A sociology of the periphery versus a peripheral sociology: Rural sociology and the dimension of space. Rural Sociology, 61, 1: 77-102.
Mendez, V. E., Bacon, C. M., & Cohen, R. (2013). Agroecology as a transdisciplinary, participatory and action-oriented approach. Agroecology and Sustainable Food Systems, 37, 1: 3-18.
Nambiar, K. K. M., Gupta, A. P., Fu, Q., & Li, S. (2001). Biophysical, chemical and socio-economic indicators for assessing agricultural sustainability in the Chinese coastal zone. Agriculture, Ecosystems & Environment, 87, 2: 209-214.
National Research Council. (2010). Toward sustainable agricultural systems in the 21st century. Washington, DC: The National Academies Press.
Neher, D. (1992). Ecological sustainability in agricultural systems. Journal of Sustainable Agriculture, 2, 3: 51-61.
Novak, J. L., & Goodman, R. W. (1995). Profitable and environmentally sound agriculture: A sustainable approach to the future. Journal of Agricultural & Food Information, 2, 4: 43-62.
Pilgeram, R. (2011). "The only thing that isn't sustainable ... is the farmer"? Social sustainability and the politics of class among pacific northwest farmers engaged in sustainable farming. Rural Sociology, 76, 3: 375-393.
Pimentel, D., Hepperly, P., Hanson, J., Douds, D., & Seidel, R. (2005). Environmental, energetic, and economic comparisons of organic and conventional farming systems. BioScience, 55, 7: 573-582.
Reganold, J. P., Glover, J. D., Andrews, P. K., & Hinman, H. R. (2001). Sustainability of three apple production systems. Nature, 410, 6831: 926-930.
Rigby, D., & Caceres, D. (2001). Organic farming and the sustainability of agricultural systems. Agricultural Systems, 68, 1: 21-40.
Sands, G. R., & Podmore, T. H. (2000). A generalized environmental sustainability index for agricultural systems. Agriculture, Ecosystems & Environment, 79, 1: 29-41.
Senanayake, R. (1991). Sustainable agriculture. Journal of Sustainable Agriculture, 1, 4: 7-28.
Smit, B., & Smithers, J. (1993). Sustainable agriculture: Interpretations, analyses and prospects. Canadian Journal of Regional Science, 16, 3: 499-524.
Taylor, D. C., Mohamed, Z. A., Shamsudin, M. N., Mohayidin, M. G., & Chiew, E. F. C. (1993). Creating a farmer sustainability index: A Malaysian case study. American Journal of Alternative Agriculture, 8, 4: 175-184.
Wezel, A., Bellon, S., Dore, T., Francis, C., Vallod, D., & David, C. (2009). Agroecology as a science, a movement and a practice. A review. Agronomy for Sustainable Development, 29, 4: 503-515.
Zinck, J. A., Berroteran, J. L., Farshad, A., Moameni, A., Wokabi, S., & Ranst, E. V. (2004). Approaches to assessing sustainable agriculture. Journal of Sustainable Agriculture, 23, 4: 87-109.
Institute of Sociology, Academia Sinica
No. 128, Sec. 2, Academia Rd., Taipei 11529, Taiwan
Received October 16, 2017; accepted April 20, 2018; last revised April 9, 2018 Proofreaders: Min-Fang Tsai, Alex C. Chang, Tsai-Ying Lu
* The previous version of this manuscript was presented at the 2017 annual meeting of the Rural Sociological Society. I would like to thank Dr. David Peters from Iowa State University for his insightful suggestions, and thank the anonymous reviewers for their careful reading and helpful comments.
[Please note: Some non-Latin characters were omitted from this article]
Caption: Figure 1 The Interaction Among the Social and Ecological Components of Sustainable Agroecosystems (Adopted from Gliessman, 2014)
Caption: Figure 2 A Measurement Model of Sustainable Agriculture. Factor Loadings Present Unstandardized Values (Standardized Values Are in Parentheses). [chi square]: 1328.141, p<.001; IFI: .929, TLI: .878, RMSEA: .115
Caption: Figure 3 Intensity of Environmental Sustainability for N = 3,067
Caption: Figure 4 Intensity of Economic Sustainability for N = 3,067 Counties in the United States
Caption: Figure 5 Intensity of Social Sustainability for N = 3,067 Counties in the United States
Caption: Figure 6 Sustainable Agriculture Clusters for N = 3,067 Counties in the United States
Table 1 Hierarchical Agglomerative Cluster Analysis of Sustainable Agriculture for N = 3,067 Counties in the United States Cluster Distance Slope [R.sup.2] (E) [R.sup.2] (p)F 15 0.00314 0.222 0.949 0.944 4064 14 0.00391 0.243 0.945 0.941 4050 13 0.00397 0.0157 0.941 0.937 4075 12 0.00468 0.180 0.937 0.932 4098 11 0.00720 0.536 0.929 0.927 4019 10 0.00755 0.0490 0.922 0.920 4003 9 0.0104 0.384 0.911 0.912 3929 8 0.0110 0.0574 0.900 0.902 3946 7 0.0167 0.510 0.884 0.889 3873 6 0.0260 0.557 0.858 0.871 3689 5 0.0296 0.140 0.828 0.847 3687 4 0.0681 1.30 0.760 0.810 3233 3 0.0880 0.293 0.672 0.749 3138 2 0.137 0.562 0.535 0.624 3519 1 0.535 2.89 0.000 0.000 -- Cluster (p)[t.sup.2] 15 87.7 14 476 13 166 12 194 11 169 10 1121 9 712 8 409 7 595 6 293 5 1448 4 1715 3 1250 2 1238 1 3519 Table 2 Intensity of Sustainable Agriculture by Clusters for N = 3,067 Counties in the United States (A) Extremely low environment extremely high economy Number of Counties 122 Sustainable Agriculture Measures Environmental sustainability (index) [-2.853.sup.BCDE] Economic sustainability (index) [2.364.sup.BCDE] Social sustainability (index) [-0.319.sup.CDE] (B) Low environment high economy Number of Counties 481 Sustainable Agriculture Measures Environmental sustainability (index) [-0.929.sup.ACDE] Economic sustainability (index) [0.793.sup.ACDE] Social sustainability (index) [-0.280.sup.CDE] (C) Moderate Number of Counties 651 Sustainable Agriculture Measures Environmental sustainability (index) [-0.108.sup.ABDE] Economic sustainability (index) [0.094.sup.ABDE] Social sustainability (index) [-0.084.sup.ABDE] (D) High environment low economy Number of Counties 1682 Sustainable Agriculture Measures Environmental sustainability (index) [0.505.sup.ABCE] Economic sustainability (index) [-0.407.sup.ABCE] Social sustainability (index) [0.001.sup.ABCE] (E) High social Number of Counties 131 Sustainable Agriculture Measures Environmental sustainability (index) [0.541.sup.ABCD] Economic sustainability (index) [-0.476.sup.ABCD] Social sustainability (index) [1.713.sup.ABCD] Note: Scheffe's test indicates significant differences at p<0.05 between groups.