So how do we reverse the growing carbon, land, and toxic chemical footprint of contemporary AVCs; expand the nutrient-rich food supply; and induce more equitable, inclusive, healthier food environments—and thus consumption patterns—so as to navigate from today’s unsustainable and precarious AVCs to a warmer, more urban, more African, and shock-prone world in which wealthier consumers place an ever-growing premium on the non-nutritive attributes of the foods they buy? Given the climate change, population and income growth, and urbanization baked into AFSs already, beneficial innovation is the only feasible pathway. And because innovation takes time, typically measurable in decades, we urgently need to accelerate innovative activity.

But no one-size-fits-all innovations exist. Many candidate socio-technical bundles are available, but those that can work in one system may be ill-suited for others. Appropriate paths from today to tomorrow necessarily differ by context.

The demographic, epidemiological, and nutritional transitions underway vary markedly across distinct AFSs and societies as the food environments in which people make dietary choices evolve differentially. Much of this evolution is influenced by the nutrition transition, the changes in dietary and physical activity patterns of populations primarily driven by a set of factors including increased and accelerated urbanization, globalization, and economic development in countries (Popkin et al. 2012). These changing dietary and physical activity patterns are correlated with a rise in the prevalence of overweight, obesity, and noncommunicable diseases in tandem with stymied undernutrition in LMICs (Popkin et al. 2020). Figure 1 shows how the double burden of malnutrition changes among each AFS typology (from rural and traditional to industrialized and consolidated). One clearly sees the sharp decline in child stunting prevalence as AFSs develop—as well as the continued existence of stunting even in the most advanced systems—and the corresponding rise in the prevalence of obesity among adults.

Fig. 1
figure 1

(Data source Marshall et al. 2021)

Stunting and obesity by system type

Drewnowski and Popkin (1997) earmarked distinct patterns that cut across the nutrition transition (Fig. 2).Footnote 1 Consistent with our rural and traditional AFS typology, people in Drewnowski and Popkin’s Patterns 1–3 have access to seasonally-dependent local foods, with much of their diet coming from staple grains and roots/tubers. Animal source foods are less available and affordable, and highly processed, packaged foods are sold in lower volumes, although that is changing (Baker et al. 2020). These populations are vulnerable to higher incidences of childhood wasting or stunting, high maternal and child mortality rates—often due to communicable diseases—and other factors that contribute to a shorter life expectancy (Frassetto et al. 2009; IFPRI 2015).

Fig. 2
figure 2

(Adapted from Drewnowski and Popkin 1997)

The nutrition transition in five patterns

As economies and AFSs transition due to economic growth and urbanization, countries in Pattern 4 shift more towards those classical patterns of industrialized and consolidated AFS types. Food supply chains, markets, and environments become more varied and diverse (Barrett et al. in press). Urbanization drives demographic and technological changes so that more women enter the labor force (Seto and Ramankutty 2016). In this Pattern 4, and with transitioning and emerging AFSs, there is access to more processed and convenient foods, street food, and fast food, and more and more people consume food away from home. This is reflected partly in the strong shift towards purchasing food for home consumption in modern retail outlets, as shown in Fig. 3. Physical activity often decreases due to changes in employment type and transportation (Kearney 2010). These changes in diets and activity have important implications for the onset of overweight, obesity, and non-communicable diseases (Popkin et al. 2020). Many countries categorized as emerging and transitioning AFS types are now reeling from a double burden of malnutrition among their population (Gómez et al. 2013).

Fig. 3
figure 3

(Data source Nielsen 2015)

Share of food purchases by type of vendor. Modern retail includes supermarkets, hypermarkets, hard discounters, and convenience stores

In modern or industrialized AFSs, behavioral change begins to reverse the negative tendencies of the preceding patterns, although currently this remains too rare, even in high-income settings. Figure 1 shows some suggestive evidence of modest improvements in adult obesity prevalence in industrialized and consolidated AFSs. Consumers with greater educational attainment, higher incomes, and better access to health care exhibit a higher level of concern about eating healthier and exhibit increased levels of purposeful physical activity (Popkin et al. 2012). Food acquisition also dramatically changes towards more personalized and digitized platforms. Globally, online grocery sales have grown rapidly, especially in China (Fig. 4), a trend that the COVID pandemic is expected to accelerate.

Fig. 4
figure 4

(Data source Euromonitor 2017 as cited in AAFC 2017)

Online grocery sales trends, 2012–2019 (left-hand panel) and share of global online grocery sales (right-hand panel). The data from 2012 to 2016 is historical, with 2017–2019 forecasted by Euromonitor

As we navigate change within any given AFS context, innovations do not automatically advance healthy diets, equitable and inclusive livelihoods, environmental and climate sustainability, or resilience, much less some combination of those objectives. We must not naïvely believe that profitable innovation is inevitably favorable in all aspects relevant to society, nor that societally desirable innovations offer an attractive return on private investment. Some scientific and social innovations may aggravate underlying dysfunction, reinforcing preexisting structures that cause, or at least aggravate, AFSs’ foundational weaknesses. The discovery and upscaling of low-cost high fructose corn syrup, for example, or of some toxic chemicals were impactful, but not in especially positive ways ultimately.

Nor do discoveries with great scientific promise necessarily translate into scalable impact. The institutional environment into which innovations get introduced matter enormously to whether the resulting path leads to impact. Consider the juxtaposition of two scientific breakthroughs in rice genetics: the IR8 and IR64 varieties originated in 1966 and 1985, respectively, by the International Rice Research Institute (IRRI), and the transgenic golden rice variety revealed in 2000 that biosynthesizes beta carotene, the precursor to vitamin A. Golden rice was arguably the more impressive scientific achievement and met a pressing societal need, as reflected in the US Patent and Trademark Office recognizing it with a Patent for Humanity Award in 2015. Yet 20 years after its discovery to great fanfare, golden rice has not yet received full approval for commercial cultivation, processing, and sale in any country. By contrast, the semi-dwarf IR8 was the first “miracle rice” and the third generation IR64 became purportedly the most diffused cereal seed variety in history. The different outcomes arose less from scientific differences than from social ones. In the face of broad popular distrust of genetic engineering, and faced with a dense thicket of patents to navigate, golden rice has failed to deliver on its fanfare, while the IRRI varieties developed using conventional plant breeding methods succeeded with publicly funded R&D and extension in an environment more trusting of science, and less reliant on private funding and intellectual property protections. The juxtaposition of these advances in rice genetics underscores how innovations that advance one or more productivity, health, environmental or other objective rarely emerge spontaneously, given the myriad obstacles to overcome. Navigating to beneficial innovation requires proactive efforts by key actors, as well as, perhaps, a bit of good fortune.

This requires paying close attention to five key considerations simultaneously, so as to avoid linear thinking about the future. Several considerations matter to selecting appropriate innovations to advance our four design objectives. Each of these comprises a spectrum that reflects trade-offs to be considered within each specific future systems context; there is no universal right answer. The design objectives are the following:

  • Spatial extent of supply chains: Short supply chains are often more transparent, more trusted, more valued socially, and have lower associated transport costs but may have limits on the diversity of crops available at any one time of year (Gómez and Ricketts 2013; Pradhan et al. 2020). Longer supply chains can be more efficient based on global comparative advantage—including with respect to environmental impacts (e.g., GHG emissions), given differences in transport modes—and are in some cases specific to the crop grown (e.g., coffee, cocoa, or tropical fruits that will only grow in certain regions). Localized AFSs may be more resilient to some disruptions (e.g., port and trade-related), and globalized AFSs to others (e.g., regional climate shocks). Localized AFSs may also benefit from local “ownership” (i.e., sovereignty) and thus have stronger concern for local environmental conservation, although potentially at the expense of less visible and more distant global environmental and climate objectives.

  • Scale of production: Highly concentrated systems can sometimes offer significant efficiencies due to economies of scale and/or scope, including the ability to mobilize financing to cover the considerable fixed costs of R&D. But more concentrated systems may also pose greater systemic risks in times of crisis (as COVID-19’s impact on highly concentrated meat supply chains illustrates) and be more prone to inefficient or exploitative market power. More distributed systems, on the other hand, tend to foster greater competition and perhaps also create more local ownership of problems and initiatives because AFS is integral to many communities.

  • Product diversity: Biodiverse AFSs are commonly more resilient to myriad shocks than are ones based on fewer species. Diverse diets are also typically healthier than ones based on fewer food types, given the varied and incomplete nutrients provided by individual foods. Diversity often comes at a cost when there exist economies of scope, however. Sometimes trade-offs arise as one seeks greater diversity within AVCs.

  • Functional redundancy: Redundancy typically increases average costs of production and distribution. Redundancy might create excess production, or wastage during storage, increasing pressure on land. But redundancy typically reduces vulnerability to systemic shocks, helps limit market power, and can promote greater diversity.

  • Internalization of externalities: Internalizing the environmental and health costs of food so that producers bear the full costs associated with environmental degradation (e.g., biodiversity loss; impacts on air, water, and soil quality; and climate change) and public health impacts (e.g., from toxic chemicals, hazardous additives, etc.) can reduce those damages by encouraging producers to find less harmful methods. But prices will almost surely increase, which can harm poor people’s access to affordable, healthy diets, unless subsidies shift to favor the affordability of nutrient-dense foodstuffs to grow and purchase.

Each of these five considerations impacts one or more of the four HERS design objectives for future AFSs: healthy diets, equitable and inclusive livelihoods, resilience, and sustainability. They help characterize the desired attributes of AFSs beyond simply minimizing the cost of calories, the primary design objective from a half century ago.