NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Committee on a Framework for Assessing the Health, Environmental, and Social Effects of the Food System; Food and Nutrition Board; Board on Agriculture and Natural Resources; Institute of Medicine; National Research Council; Nesheim MC, Oria M, Yih PT, editors. A Framework for Assessing Effects of the Food System. Washington (DC): National Academies Press (US); 2015 Jun 17.

Cover of A Framework for Assessing Effects of the Food System

A Framework for Assessing Effects of the Food System.

Show details

6The U.S. Food and Agriculture System as a Complex Adaptive System

The U.S. food system has many features characteristic of a complex adaptive system, both in its structure (see Chapter 2) and in its effects (see Chapters 3-5). The complex systems perspective can offer important insights for understanding the dynamics of both the current configuration of the food system and the potential alternative configurations of the food system. This chapter begins by describing the properties of a complex adaptive system (CAS), illustrated with examples specific to the food system and with references to other chapters in the report as appropriate. The chapter then reviews the implications of these properties for the development of a sufficiently rich and comprehensive framework, including consideration of how specific factors shape the complex dynamics of the food system with regard to health, environmental, social, and economic outcomes.


A CAS is a system composed of many heterogeneous pieces whose interactions drive system behavior in ways that cannot easily be understood from considering the components separately. Such systems, whether they are social, physical, or biological, tend to share a set of specific properties (Hammond, 2009; Holland, 1992; Miller and Page, 2007). Consideration of these properties and their implications from many scientific and policy perspectives have yielded important insights into system behavior. These perspectives include social science (Axelrod, 1997; Axtell et al., 2002; Epstein, 2002, 2007; Schelling, 1978; Tesfatsion and Judd, 2006), public health (Auchincloss and Roux, 2008; Diez Roux, 2007; Epstein, 2009; Eubank et al., 2004; Homer and Hirsch, 2006; Huang and Glass, 2008; IOM, 2012; Longini et al., 2005; Luke and Stamatakis, 2012; Mabry et al., 2008, 2010), biology (Axelrod et al., 2006; Segovia-Juarez et al., 2004), business (Sterman, 2000), and land/ecosystem management (Parker et al., 2003; Schluter and Pahl-Wostl, 2007). Each of the following sections describes one important general property of CAS and then illustrates its applicability to the U.S. food and agriculture system with specific examples.

Individuality and Adaptation

Complex systems generally contain a variety of autonomous actors. These may vary considerably in local context, motivation, exposure to information or environmental signals, or level of scale. The decentralized interaction of actors is often a key driver of system behavior. At the same time, the actors themselves often adapt through time in response to other actors or to changes in the system state. Adaptation can occur at different speeds and take different forms across individuals. A variety of actors and processes of adaptation can be found within the U.S. food and agriculture system. Human actors in the system include consumers, farmers, laborers, food processors and manufacturers, distributors, food service providers, and researchers. At a higher level of aggregation, multinational firms, governments, regulatory agencies, and universities may act as unified actors that play important roles. At lower levels of scale, pathogenic bacteria, agricultural pests, and even genetic material (e.g., in the resistome1) represent distinct actors.

In modern industrialized societies, a vast array of human actors and aggregate institutional actors play important roles in shaping the structure and dynamics of the food system. Individual decisions that shape food system outcomes are made daily by farmers, crop field workers, bankers, crop consultants, grain elevator operators, meat packers, corporate product developers, advertisers, grocery store managers, truck drivers, chefs, waiters, home food gatekeepers, nutritionists, garbage collectors, antihunger and environmental activists, state and federal legislators, government employees, researchers, and physicians (to name a few). Consumer decisions on what, where, when, and how to buy and eat are fundamental drivers of the food supply chain in most countries. These decisions likewise drive ancillary outcomes for health, environmental, social, and economic effects of the food system because they shape what foods are produced, how they are produced, how they are made available, and how our bodies respond to what we eat (or do not eat). Individuals make decisions within organizational and institutional contexts that shape their choice sets and alter the costs and benefits of different options. Leaders of large agricultural input companies, food processing and distribution firms, retail grocery and restaurant chains, and institutional food buyers (like schools and hospitals) are themselves actors—whose business decisions affect the choices of individuals who work for or buy from these firms. Market research guides advertising to influence consumer choices in ways that benefit the marketers. Politicians and public agency leaders develop tax, regulatory, trade, and research policies to respond to shifts in societal values and political power, which in turn constrain the behaviors of economic firms and individual actors.

Processes of adaptation by individual actors in the food system are varied, ranging from changing consumer preferences to changing farming practices to evolution of drug resistance. Changes to the food system thus have impacts across the component subsystems of the food supply chain, and also across space, that go beyond simply “ripples”—because interventions can trigger adaptive responses. Not all actors will adapt to any specific system change, and not all adaptations have “beneficial” (or discernible) effects. Considering the full set of adaptive responses (by multiple types of actors) that is triggered by any change can be important for sufficient understanding of likely system effects. For example, the introduction of herbicide tolerant crops (e.g., Roundup Ready™ soybeans) not only reduced tillage and soil erosion but also reduced labor and energy use per acre, induced land conversion to crop use, and fostered the evolution of herbicide-resistant weeds (Barrows et al., 2014).

Feedback and Interdependence

Just as complex systems usually contain a variety of distinct (but interacting) actors, they tend also to contain several distinct (but potentially linked) mechanisms or pathways. These may cross multiple levels of the system (e.g., the hedonic reward pathway driving some eating behavior, which involves micro-level biological processes within a human, the physical environments surrounding them, the social or market-level processes connecting them), and they often interact with each other, creating interdependence of factors in the system. Obesity is a classic example of a phenomenon driven by multiple interdependent factors (see Chapter 3). A central hallmark of complex systems is the presence of feedback between actors or factors in the system. Feedback describes a dynamic process in which change in one part of a system affects another component, which, in turn, affects the original component again (often with a time lag). Within a complex system, feedback may cross different levels of scale (e.g., within an organism and in the environment surrounding it), sectors (e.g., economic, health, and social), or spatial boundaries (e.g., U.S. consumers and South American agriculture). Feedback can be positive (reinforcing) or negative (balancing).

Numerous examples of feedback and interdependence can be found in the U.S. food and agriculture system. As illustrated in Figure 6-1, the food system can be conceptualized as a transformation process that both depends on and creates important feedbacks for natural resources and human society. Natural resources like air, soil, water, and biota (pollinators, natural enemies of food pests) are essential for agricultural production, as well as the manufacture of many foods like bread, cheese, and wine. Yet depletion and effluents from the food system influence the future status of natural resources. In Figure 6-1, these changes occur from time 0 to time 1. Likewise, the food system depends on a host of human systems that govern our health, markets, policy, and general well-being. These human systems provide the labor, entrepreneurship, capital, and technology needed to produce and distribute food. Once again, the food system generates feedbacks that influence human systems at a future period.

FIGURE 6-1. Food system as a dynamic process transforming the state of natural resources and human systems from one period to the next.


Food system as a dynamic process transforming the state of natural resources and human systems from one period to the next.

Another prominent feedback example of widespread concern is the evolution of pesticide and antibiotic resistance by insects, pests, weeds, and plant and animal pathogens, which now incurs multibillion-dollar costs each year. These costs reflect the prevalence of inadequate and ineffective strategies for limiting the strength of the selection pressures for resistance created by chemical controls that initially are efficacious. Given the limited availability of new chemistries for controlling pests and pathogens and the ability of resistant organisms to move and transmit genetic material, this form of feedback and interdependence may greatly affect future management options in food, agriculture, and health systems.

Certain grazing practices also can shift rangeland systems to a less productive regime by reducing vegetation cover, setting in motion a feedback relationship that decreases nutrient and water accumulation (Gordon et al., 2008). Similarly, policy efforts to increase animal welfare by promoting free-range housing for hens have in some cases adversely affected the health of the animals by increasing exposure to pathogens through the soil and cannibalistic pecking (see Chapter 7, Annex 5).

Complex feedbacks occur also in the socioeconomic aspects of the food system. Market supply and demand relationships shape prices that act as incentives on the behavior of producers and consumers. Many food grains have well-developed futures markets as well as current markets, allowing prices to adjust based on expectations of future supply and demand. At times markets can reveal surprising indirect effects, as when a U.S. biofuel mandate contributed to higher global corn prices that, in turn, shifted more land into agricultural use (Hayes et al., 2009; Searchinger et al., 2008). Indeed, this example highlights the fact that market effects are not limited to price feedbacks that communicate incentives to buyers and sellers. Markets also create repercussions for the availability of goods and services that lack clear property rights. Climate stability, a global ecosystem service, is a clear example where a lack of property rights makes markets fail to regulate greenhouse gas emissions (like those from indirect land use), so policy interventions are needed. The political feedback mechanisms that shape policy design represent another layer of complexity in the food system.


Actors and processes in a CAS often exhibit substantial heterogeneity—they differ from each other in ways that can strongly shape local dynamics in parts of the system. For example, actors within the system may have different goals, different decision-making procedures, different information, different local environmental exposures, or different constraints on their actions. These differences can shape divergent adaptation or responses to changes in the system. Heterogeneity very often occurs across types of actors (as described above). For example, multinational corporations are likely to have very different information and constraints than those faced by individual consumers, and crop insect pests have different behavioral repertoires than pathogenic bacteria do. There also may be substantial heterogeneity within a particular type of actor. For example, consumers may vary in income, health status, or preferences; food service operators may face different regulatory regimes in different places; and farms certainly vary in the composition of their soil and in size and sales volume (see Chapter 5).

A good case study example of heterogeneity in types of distinct actors in the food system can be found in fruit and vegetable intake (see Chapter 7, Annex 3). Changing the intake levels of fruits and vegetables is likely to involve farmers, farm workers, food manufacturers, retailers, marketers, restaurants, school food service workers, and household food gatekeepers, each with different incentives and facing different information sets, which must be considered in assessing the likely impact of an intervention in this area.

Socioeconomic, spatial, and cultural heterogeneity also can lead the impacts of food system changes to differ significantly for different subgroups (see Chapter 5). This is an important consideration in the case of cage-free eggs. Because cage-free chickens are more expensive to maintain than those that are confined, switching to such methods could involve a substantial increase in market price. Demand for eggs is relatively inelastic, so most of the impact of that price increase would fall on lower-income families that rely on eggs as an inexpensive source of protein. Ignoring differences among consumers would mask the distributional consequences of such a shift.

Population heterogeneity also is a major consideration for the health effects of the food system (see Chapter 3), where risk factors, exposures, and disease outcomes may all differ substantially.

Spatial Complexity

Complex systems often contain spatial organization that strongly shapes dynamics within them. These spatial properties can govern the interaction of actors, existence and speed of feedback, and heterogeneity across the system. Physical geography (whether naturally occurring or built) and networks (whether representing contacts, flows of materials or information, or relationships among groups such as species) are examples of spatial organization. Within the food and agriculture system, elements of spatial organization include supply chains, market segmentation, the patchwork of geographically specific regulations across states and counties, international borders, and ecosystems and food webs. Spatial structure can matter by directly shaping the local context experienced by actors, but it also can shape impacts at a distance, govern changes in environment over time (e.g., spatial displacement as in environmental effects like pollution; see Chapter 4), and create indirect and possibly unintended effects (e.g., resurgence of target pests or antibiotic resistance through the resistome; see Chapter 7, Box 7-7).

Because of the broad spatial extent of arable cropland, pastures, and rangelands in many regions of the United States, agricultural production systems can have marked effects on water quality and quantity and on wildlife habitat and population densities. A key factor determining the impacts of agricultural production systems on water, wildlife, and other natural resources is the spatial organization of system components. For example, connectivity of strips of non-crop vegetation across a landscape dominated by crops can foster migration corridors for birds of conservation concern. Strips of trees, shrubs, and grasses can dramatically reduce the quantity of soil sediment moving from croplands to adjacent streams. Spatial concentration of livestock production, meanwhile, can magnify environmental effects (see Chapter 4). Spatial structure also is an important driver of consumer behavior (see Chapter 5) and health effects (see Chapter 3). For example, obesity outcomes can be strongly shaped by geography (e.g., the availability and convenience of food or the presence of advertising) as well as by “social” spatial structures (e.g., peer networks) (see Chapter 3). The importance of spatial structure in chronic disease is easily observed in the spatial patterns of incidence that emerge (see Chapter 3).

Dynamic Complexity

The presence of feedback, interdependence, and adaptation in a complex system can produce dynamics with characteristic properties. These often include substantial nonlinearity or “tipping points,” path dependence, and system behaviors that appear to be “emergent,” that is, system-level behaviors that differ from what might be expected from the sum of behaviors of individual components of the system. Nonlinearity can yield large effects from relatively small changes in system configuration. Examples in the food system include the relationship between arable cropland conversion to conservation buffer strips composed of reconstructed prairie and the consequent reduction in the export of soil sediments from watersheds (see Chapter 4), or the metabolic changes that result from weight gain and loss (see Chapter 3). The coupling of social and ecological systems (each with their own nonlinear processes) within the food system can lead to even stronger nonlinearities in the response of the overall system to changes (see Chapter 5).

Path dependence refers to phenomena whose later dynamics are strongly shaped by the sequence of early events. Examples in the food system include the relative importance of early life nutrition experience in shaping later habits, behaviors, and chronic disease risk (see Chapter 3).

Management of fish stocks is an important (and canonical) example of dynamic complexity at work (see Chapter 7, Annex 1). Overfishing often results in sudden and dramatic collapses in fish stocks if not carefully monitored and managed. This type of phase transition can occur because overfishing both depletes the existing stock of fish and reduces the rate at which fish populations are replenished through breeding. Globally, 90 percent of fisheries are considered fully exploited or overly so. Increasing demand for fish and the effects of climate change threaten to tip many fisheries toward collapse. In many cases, transitioning to aquaculture does not relieve the pressure on natural fisheries because wild stocks of herring, anchovies, and sardines are still sometimes used as feed sources for aquaculture production.

Given the importance of feedbacks in a complex system, another dynamic system characteristic of special interest (as noted in the discussion of environmental effects in Chapter 4) is the degree of resilience the system manifests when stressed by physical and biotic factors. Resilience also is relevant in the context of social and economic stress factors. For all types of stressors, resilience can be viewed as an ability to bounce back from sudden shocks and long-term stressors. For agricultural systems, temperature extremes, droughts, floods, and pests are recurrent, though unpredictable, biophysical stresses. Similarly, rapid increases in input costs, sharp declines in market values of crops and livestock, and regulations form part of the matrix of socioeconomic stress factors acting on agricultural systems. Often, farmers can take actions that minimize risks and susceptibilities to stress factors (e.g., adding irrigation systems to make up for precipitation deficits, purchasing crop insurance to cover lost revenue), but these risk reduction measures can incur significant costs. Other approaches, such as diversifying cropping systems to include crops with different planting and harvest dates, and contrasting vulnerabilities to pests, may incur little or no additional cost. In some cases, as in the case of federally subsidized crop insurance, costs for increasing resilience may be distributed to society at large.


The U.S. food and agriculture system has many of the characteristics of a CAS. It has diverse and adaptive individual actors, with substantial feedback and interdependence among them, and it includes both spatial and temporal heterogeneity as well as an adaptive change dynamic. Recognition of the food system as a CAS has important implications for efforts to assess its effects, and thus for the framework presented in the next chapter (Chapter 7). The complex systems perspective highlights key systemic features that a framework should address and argues for consideration of approaches and methodologies that can appropriately capture these features. Although no one method or approach is likely able to capture all elements of the system at once, the discussion of key aspects of complexity above is intended to guide consideration of what to include in (and what may be left out of) any analysis. In Chapter 7, the committee lays out a framework designed to inform assessments of the food system with a complex system perspective in mind, considering complexity in four distinct ways across six distinct steps. Chapter 7 also discusses specific methods that are well suited to capturing key aspects of complex dynamics, recognizing, however, that not all analyses can (or should) address all of the elements of the complex food system.


  • Auchincloss AH, Roux AVD. A new tool for epidemiology: The usefulness of dynamic-agent models in understanding place effects on health. American Journal of Epidemiology. 2008;168(1):1–9. [PubMed: 18480064]
  • Axelrod R. The complexity of cooperation: Agent-based models of competition and collaboration. Princeton, NJ: Princeton University Press; 1997.
  • Axelrod R, Axelrod D, Pienta KJ. Evolution of cooperation among tumor cells. Proceedings of the National Academy of Sciences of the United States of America. 2006;103(36):13474–13479. [PMC free article: PMC1557388] [PubMed: 16938860]
  • Axtell RL, Epstein JM, Dean JS, Gumerman GJ, Swedlund AC, Harburger J, Chakravarty S, Hammond R, Parker J, Parker M. Population growth and collapse in a multiagent model of the Kayenta Anasazi in Long House Valley. Proceedings of the National Academy of Sciences of the United States of America. 2002;99(Suppl 3):7275–7279. [PMC free article: PMC128597] [PubMed: 12011406]
  • Barrows G, Sexton S, Zilberman D. Agricultural biotechnology: The promise and prospects of genetically modified crops. Journal of Economic Perspectives. 2014;28(1):99–120.
  • Diez Roux AV. Integrating social and biologic factors in health research: A systems view. Annals of Epidemiology. 2007;17(7):569–574. [PubMed: 17553703]
  • Epstein J. Modeling civil violence: An agent-based computational approach. Proceedings of the National Academy of Sciences of the United States of America. 2002;99(3):7243–7250. [PMC free article: PMC128592] [PubMed: 11997450]
  • Epstein JM. Generative social science. Princeton, NJ: Princeton University Press; 2007.
  • Epstein J. Modeling to contain pandemics. Nature. 2009;460:687. [PMC free article: PMC3785367] [PubMed: 19661897]
  • Eubank S, Guclu H, Kumar VS, Marathe MV, Srinivasan A, Toroczkai Z, Wang N. Modelling disease outbreaks in realistic urban social networks. Nature. 2004;429(6988):180–184. [PubMed: 15141212]
  • Gordon LJ, Peterson GD, Bennett EM. Agricultural modifications of hydrological flows create ecological surprises. Trends in Ecology & Evolution. 2008;23(4):211–219. [PubMed: 18308425]
  • Hammond R. Complex systems modeling for obesity research. Preventing Chronic Disease. 2009;6:1–10. [PMC free article: PMC2722404] [PubMed: 19527598]
  • Hayes D, Babcock B, Fabiosa J, Tokgoz S, Elobeid A, Yu T-H, Dong F, Hart C, Chavez E, Pan S, Carriquiry M, Dumortier J. Biofuels: Potential production capacity, effects on grain and livestock sectors, and implications for food prices and consumers. Journal of Agricultural and Applied Economics. 2009;41(2):465–491.
  • Holland JH. Adaptation in natural and artificial systems. Cambridge, MA: MIT Press; 1992.
  • Homer JB, Hirsch G. System dynamics modeling for public health: Background and opportunities. American Journal of Public Health. 2006;96:452–458. [PMC free article: PMC1470525] [PubMed: 16449591]
  • Huang TT, Glass TA. Transforming research strategies for understanding and preventing obesity. Journal of the American Medical Association. 2008;300(15):1811–1813. [PubMed: 18854544]
  • IOM (Institute of Medicine). Accelerating progress in obesity prevention: Solving the weight of the nation. Washington, DC: The National Academies Press; 2012. [PMC free article: PMC3648752] [PubMed: 22983849]
  • Longini IM Jr., Nizam A, Xu S, Ungchusak K, Hanshaoworakul W, Cummings DA, Halloran ME. Containing pandemic influenza at the source. Science. 2005;309(5737):1083–1087. [PubMed: 16079251]
  • Luke DA, Stamatakis KA. Systems science methods in public health: Dynamics, networks, and agents. Annual Review of Public Health. 2012;33:357–376. [PMC free article: PMC3644212] [PubMed: 22224885]
  • Mabry PL, Olster DH, Morgan GD, Abrams DB. Interdisciplinarity and systems science to improve population health: A view from the NIH Office of Behavioral and Social Sciences Research. American Journal of Preventive Medicine. 2008;35(2 Suppl):S211–S224. [PMC free article: PMC2587290] [PubMed: 18619402]
  • Mabry PL, Marcus SE, Clark PI, Leischow SJ, Mendez D. Systems science: A revolution in public health policy research. American Journal of Public Health. 2010;100(7):1161–1163. [PMC free article: PMC2882409] [PubMed: 20530757]
  • Miller JH, Page SE. Complex adaptive systems: An introduction to computational models of social life. Princeton, NJ: Princeton University Press; 2007.
  • Parker DC, Manson SM, Janssen MA, Hoffmann MJ, Deadman P. Multiagent systems for the simulation of land-use and land-cover change: A review. Annals of the Association of American Geographers. 2003;93(2):314–337.
  • Schelling TC. Micromotives and macrobehavior. New York: W.W. Norton and Co.; 1978.
  • Schluter M, Pahl-Wostl C. Mechanisms of resilience in common-pool resource management systems: An agent-based model of water use in a river basin. Ecology and Society. 2007;12(2):4.
  • Searchinger T, Heimlich R, Houghton RA, Dong FX, Elobeid A, Fabiosa J, Tokgoz S, Hayes D, Yu TH. Use of U.S. croplands for biofuels increases greenhouse gases through emissions from land-use change. Science. 2008;319(5867):1238–1240. [PubMed: 18258860]
  • Segovia-Juarez JL, Ganguli S, Kirschner D. Identifying control mechanisms of granuloma formation during m. Tuberculosis infection using an agent-based model. Journal of Theoretical Biology. 2004;231(3):357–376. [PubMed: 15501468]
  • Sterman JD. Business dynamics: Systems thinking and modeling for a complex world. Boston, MA: Irwin McGraw-Hill; 2000.
  • Tesfatsion L, Judd KJ. Handbook of computational economics, Vol. 2: Agent-based computational economics. Amsterdam, The Netherlands: North-Holland; 2006.



Resistome refers to the collection of antibiotic resistance genes and their precursors in both pathogenic and nonpathogenic bacteria.

Copyright 2015 by the National Academy of Sciences. All rights reserved.
Bookshelf ID: NBK305170


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (10M)

Related information

  • PMC
    PubMed Central citations
  • PubMed
    Links to PubMed

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...