Historical perspective

The term ‘sustainable intensification’ originated from development efforts that aimed to increase the productivity of sub-Saharan agriculture in the 1990s (Pretty, Reference Pretty1997). It was used in the context of increasing production from existing agricultural land in ways that lower environmental impact and do not lead to further land conversion or loss of ‘undisturbed’ natural ecosystems (Campbell et al., Reference Campbell, Thornton, Zougmoré, van Asten and Lipper2014). In the SI framework, changes in land-use intensity (higher yields, multiple cropping seasons, higher livestock stocking density) are accompanied by changes in the levels of biophysical and socioeconomic inputs to the land (e.g., labour, feed resources or capital). There are a number of illustrations showing that SI can have huge effects on domestic food budgets, social infrastructure, business development and the well-being of both the rural and urban populations (Pretty et al. Reference Pretty, Toulmin and Williams2011). Proponents of SI originally emphasised the importance of using local knowledge and developing agricultural methods suited to local conditions. Participation of smallholder farmers was considered crucial for the development of more productive technologies (Pretty, Reference Pretty1997). A wide range of bottom-up, integrated technologies was therefore used to conserve water and soils and to manage nutrient flows and pests. However, the term remained loosely defined, so that SI was subsequently firmly embraced by the industry and by a number of international organisations (Food and Agriculture Organisation (FAO), 2010; Campbell et al., Reference Campbell, Thornton, Zougmoré, van Asten and Lipper2014) for whom an increase in food demand is an inevitable response to population growth and dietary shifts towards more animal proteins.

In a recent assessment of the concept of SI, many implementers of this framework indicated there is no clear difference between SI and ‘traditional’ intensification and modernisation practices (Petersen and Snapp, Reference Petersen and Snapp2015). In practice, SI has long been ‘narrowly’ focused on production and has been criticised for lacking engagement with the key social principles of sustainability (Loos et al., Reference Loos, Abson, Chappell, Hanspach, Mikulcak, Tichit and Fischer2014). According to these authors, focusing on the need to increase food production reflects a fundamental misunderstanding of the primary causes of food insecurity, that is, poverty and political and structural problems. Issues such as equity, access to food and food distribution would thus need to be prioritised before increases in production can improve food security. Recently, the SI framework has been more broadly conceived in a way that addresses dietary issues and improves equity of access to food. In this perspective, Garnett et al. (Reference Garnett, Appleby, Balmford, Bateman, Benton, Bloomer, Burlingame, Dawkins, Dolan, Fraser, Herrero, Hoffman, Smith, Thornton, Toulmin, Vermeulen and Godfray2013) do not consider SI as a ‘business-as-usual’ food production with ‘marginal’ improvements in sustainability, but rather as a rethinking of a food system that reduces its environmental footprint supports rural economies and enhances human nutrition and animal welfare. Improvements in animal diets could for instance benefit animal welfare. However, there are potentially negative effects related to (i) more confined environments that affect the ability of the animals to express their natural behaviour and (ii) the risk of more animal health problems due to breeding for higher yields and growth rates (Röös et al., Reference Röös, Bajželj, Smith, Patel, Little and Garnett2017; Huber, Reference Huber2018).

Sustainable intensification as an efficiency-oriented perspective

One priority of SI is to close yield gaps (i.e. the difference between the actual and attainable yield on the basis of the genetic potential and optimal production conditions without yield limiting and reducing factors) to produce more food while using less land. Rao et al. (Reference Rao, Peters, Castro, Schultze-Kraft, White, Fisher, Miles, Lascano, Blümmel, Bungenstab, Tapasco, Hyman, Bolliger, Paul, Van der Hoek, Maass, Tiemann, Cuchillo, Douxchamps, Villanueva, Rincón, Ayarza, Rosenstock, Subbarao, Arango, Cardoso, Worthington, Chirinda, Notenbaert, Jenet, Schmidt, Vivas, Lefroy, Fahrney, Guimarães, Tohme, Cook, Herrero, Chacón, Searchinger and Rudel2015) discussed that the SI of forage-based systems is based on several intensification processes, including genetic intensification, that is, the deployment of productive livestock breeds, and the development and use of grass and legume cultivars selected because of their higher biomass production, nutritive value and persistence relative to native grasses. One key to the successful intensification of tropical forage-based systems is the adequate selection of fodder species, for instance, leucaena (Leucaena leucocephala, Lam.), a nitrogen-fixing shrub that can serve as the backbone of the system. Sustainable intensification of tropical forage-based systems is likely to increase their productivity (zu Ermgassen et al., Reference zu Ermgassen, de Alcântara, Balmford, Barioni, Beduschi Neto, Bettarello, de Brito, Carrero, de AS Florence, Garcia, Gonçalves, da Luz, Mallman, Strassburg, Valentim and Latawiec2018) while providing regulating ecosystem services, including a contribution to soil fertility, limitation of erosion, and climate regulation via CO₂ sequestration. Integration of forage systems within cropping systems should enhance the coupling of carbon (C) and nitrogen (N) cycles in grasslands, cover-crops, and ley-farming systems, thus reducing the presence of reactive N in the soils and consequently, the risk of nitrate leaching and nitrous oxide emissions to the atmosphere (Lemaire et al., Reference Lemaire, Franzluebbers, de Faccio Carvalho and Dedieu2014). A large part of grass-based biomass and crop residues can be used for feeding ruminants. Genetic intensification of animals has frequently led to the use of high-yielding breeds. As an example, dairy producers in many places around the world view Holstein Friesian cattle as the ultimate cow for their farm to show their competence as dairy farmers. However, these animals are very dependent on controlled and optimal conditions; they can be vulnerable to climate change and are little adapted to feed on roughages (Phocas et al., Reference Phocas, Belloc, Bidanel, Delaby, Dourmad, Dumont, Ezanno, Fortun-Lamothe, Foucras, Frappat, González-García, Hazard, Larzul, Lubac, Mignon-Grasteau, Moreno, Tixier-Boichard and Brochard2016). As a consequence, production levels are often far below the attainable yield.

The search for increased efficiency requires precise adjustment of the diets of individual animals to their requirements. Precision livestock farming, by recording herbage biomass and quality and animal physiological and behavioural traits, offers new possibilities to efficiently monitor and manage highly productive livestock systems. According to Campbell et al. (Reference Campbell, Thornton, Zougmoré, van Asten and Lipper2014), SI and climate smart agriculture are highly complementary, with SI being an essential lever for decreasing GHG emissions per unit of product (Röös et al., Reference Röös, Bajželj, Smith, Patel, Little and Garnett2017). This efficiency-oriented perspective is in line with the logic of substituting cereals and soybean meal in ruminant diets with human-inedible resources. Crop residues such as culled Brussel sprouts, waste tomatoes and carrot pulp after juice extraction are used to supplement grazing animals or forages in areas where intensive vegetable production is common. Another innovative option consists of producing insects from food wastes and then using the insects as protein sources for livestock, including dairy cows (Stamer, Reference Stamer2015). Other human-inedible feed resources can be used, mainly for dry cows, growing heifers and small ruminants, including dried distillers grains with solubles, palm kernel cake or spineless cactus (Makkar, Reference Makkar2018).

Is sustainable intensification sustainable?

Within the SI framework, the focus on improving resource use efficiency may, however, not necessarily lead to a reduction in the environmental footprint. This was observed in the Brazilian Amazon region, where more efficient agricultural systems were seen as profitable to farmers, which resulted in the expansion of the cultivated area and further deforestation (Lambin and Meyfroidt, Reference Lambin and Meyfroidt2011). The most efficient systems are not necessarily the most sustainable. Rodríguez-Ortega et al. (Reference Rodríguez-Ortega, Bernués, Olaizola and Brown2017) compared three Mediterranean sheep farming systems by applying the emergy methodology that is based on the amount of direct and indirect materials and energy sources embodied in final products. Lamb meat production was 1.9 and 1.3 times more intensive and efficient, respectively, in a partially integrated mixed system than in a pasture-based sheep system but 5.1 times less sustainable as a result of lower self-sufficiency and renewability. One problem of crop-livestock systems that sell meat and cereal crops is that animals and crops frequently do not ecologically complement each other (Altieri et al., Reference Altieri, Nicholls and Montalba2017). As a consequence, farmers keep on buying external chemical inputs. Conversely, in a suckler-cow farm network of the Charolais-area, organic farmers grow crops on farm to feed the cows and efficiently exploit the diversity of feed resources. These farms had the lowest GHG emissions and non-renewable energy consumption per ha, the lowest operational costs and the highest net income per worker (Veysset et al., Reference Veysset, Lherm, Bébin and Roulenc2014).

The definition of sustainability within the SI framework largely addresses the reduction of harmful effects, while little attention has been paid to producing positive environmental outcomes such as increasing the supply of clean water and the stocks of natural resources including fertile soils. Moreover, the focus is essentially utilitarian, that is, reducing harmful effects is done to ensure continuing agricultural production while remaining within ‘planetary boundaries’. According to Struik and Kuyper (Reference Struik and Kuyper2017), the words ‘sustainable’ and ‘intensification’ are often not assigned equal weight. For instance, the Irish grassland-based dairy system developed according to the principles of SI causes biodiversity losses (Sullivan et al., Reference Sullivan, Skeffington, Gormally and Finn2010). The Food and Agriculture Organisation has advocated using an ecosystem services approach for SI (FAO, 2010). Given the classical trade-off between food production on the one side and many regulating and cultural services on the other, it is difficult to enhance all ecosystem services simultaneously. According to Smith (Reference Smith2013), SI should thus be regarded as a ‘guiding principle in decisions about land-use, rather than as an end-point’. By interpreting SI as an inevitable response to population growth and the search for global food security, it is also only after productivity concerns are covered that most environmental and social issues are addressed. This limits the emergence of converging and transformative solutions that jointly account for productive and environmental issues and that meet the expectations of all types of stakeholders (Howe et al., Reference Howe, Suich, Vira and Mace2014). Solutions are thus mostly sub-optimal for the environmental and social dimensions. For instance, Thorlakson et al. (Reference Thorlakson, de Zegher and Lambin2018) recently analysed how 449 companies in the food, textile and wood-products sectors have adopted voluntary practices to improve the environmental and/or social management of their suppliers’ activities. Half of these companies use some form of sustainable sourcing practices. However, these practices are limited in scope, with 71% covering only one or a few input materials, one quarter applying to only a single product line, and many primarily focusing on labour rights and compliance with national laws.

Historical perspective

The term ‘agroecology’ can be traced back to the 1930s and has been used to denote a scientific discipline, a set of agricultural practices, and a social movement that promotes culturally sensitive, socially fair and economically viable farming systems (Wezel et al., Reference Wezel, Bellon, Doré, Francis, Vallod and David2009). Until the 1960s, AE was only referred to as a scientific discipline. From the 1970s, AE gradually emerged as a movement in line with environmental movements that went against industrial agriculture creating greater specialisation and intensification. As a scientific discipline, AE applies ecological theory to the design and management of sustainable agroecosystems (Altieri, Reference Altieri2002; Wezel et al., Reference Wezel, Bellon, Doré, Francis, Vallod and David2009) or of the entire food system (Francis et al., Reference Francis, Lieblein, Gliessman, Breland, Creamer, Harwood, Salomonsson, Helenius, Rickerl, Salvador, Wiedenhoeft, Simmons, Allen, Altieri, Flora and Poincelot2003). It aims to stimulate natural processes to design agricultural systems that are weakly artificialised, productive, environmentally friendly and less dependent on chemical inputs. As a movement, AE counts thousands of researchers and practitioners, mainly in Latin America, where it does not work through any standard or certification system (Tittonell, Reference Tittonell2014). It promotes traditional farming systems that sustain year-round yields through the use of agrobiodiversity. Diverse plant species and genetic resources and optimal interactions between system components (including between crops and livestock) are used to enhance agroecosystem functions rather than to introduce chemical inputs. Agroecology also promotes food sovereignty, local autonomy, and community control of land, water and genetic resources (Altieri et al., Reference Altieri, Nicholls and Montalba2017). Farmer-to-farmer networks play a key role in the extension process and dissemination of knowledge (Rosset et al., Reference Rosset, Machín Sosa, Roque Jaime and Ávila Lozano2011).

Despite the recent surge in the academic literature on AE, livestock farming systems have scarcely been considered in most agroecological thinking until recently. Disconnectedness from the land is probably the main problem threatening the sustainability of livestock farming. On the basis of a study conducted by Altieri (Reference Altieri2002), who identified the key ecological processes to be optimised in agricultural systems, Dumont et al. (Reference Dumont, Fortun-Lamothe, Jouven, Thomas and Tichit2013) proposed five principles as a guideline to implement site-specific combinations of agroecological practices in livestock farming systems: (i) achieve integrated animal health management, (ii) decrease the external inputs needed for production, (iii) decrease pollution by optimising the metabolic functioning of farming systems, (iv) enhance functional diversity within livestock farming systems to strengthen their resilience, and (v) preserve biological diversity by adapting management at farm and landscape scales. The application of these principles was shown to generate environmental and economic benefits that were quantified across a broad range of ruminant, pig, poultry, aquaculture, and integrated crop-livestock systems. The extent to which the five principles were mobilised differed between production systems and their degree of intensification (Dumont et al., Reference Dumont, Fortun-Lamothe, Jouven, Thomas and Tichit2013; Reference Dumont, Jouven, Bonaudo, Botreau and Sabatier2017).

Agroecology promotes system redesign and can lead to win-wins between production and environmental goals

Agroecology promotes system redesign and questions production goals. Within the French CIVAM network (Centres d’Initiatives pour Valoriser l’Agriculture et le Milieu rural), the target goal was to create added value from dairy production without always maximising outputs per animal or per unit area. This led to an increase in the use of grazed herbage at the expense of maize silage, as well as a reduction in the use of concentrate feeds. Limiting pesticides and insecticides through more complex crop rotations and an increase in grassland area at the farm level had various direct and indirect benefits on soils, water and biodiversity. Grasslands comprised a large proportion of grass–legume mixtures, and the grazing season was extended into late autumn and winter. Herd management was tailored to adapt animal requirements to resources by grouping calving periods. Inputs, feed costs and mechanisation costs were lower than in conventional systems from the same area. In spite of slightly lower milk production (-13%), the gross margin was higher (+26%). Greenhouse gas emissions were similar between the two systems, that is, 1.1 kg CO₂-eq/l but accounting for carbon sequestration in permanent grasslands and hedgerows led to a 14% reduction of net GHG emissions in the CIVAM compared with conventional dairy farms of the same area (0.87 v. 1.02 kg CO₂-eq/l; Duru and Therond, Reference Duru and Therond2015).

One strength of agroecological systems lies in their self-sufficiency, which, through interacting with their environment and the recycling of on-farm wastes, can produce part of the resources needed for production. Self-sufficiency reduces dependency on erratic market prices but increases dependency on climatic conditions, for instance, summer droughts that can drastically reduce grassland biomass production. The potential of all-forage diets still needs to be demonstrated for productive dairy cattle breeds with high nutrient requirements. An example is the lower ovarian cyclicity during early lactation, which results in poor the reproductive performance of Holstein-Friesian and Montbéliarde cows in low-input systems, in spite of spring calving that aims to optimise grassland utilisation (Pires et al., Reference Pires, Chilliard, Delavaud, Rouel, Pomies and Blanc2015). Use of animal genotypes adapted to their environmental conditions, and access to pasture so that animals express their natural behaviour are essential principles of AE (Dumont et al., Reference Dumont, Fortun-Lamothe, Jouven, Thomas and Tichit2013; Reference Dumont, González-García, Thomas, Fortun-Lamothe, Ducrot, Dourmad and Tichit2014), and are assumed to guarantee animal welfare. Beef cattle, dairy cattle, sheep and dairy goat farmers in agroecological systems did not establish strong priorities among breeding goals (i.e., between feed efficiency, animal health, reproduction, docility, product quality, etc.), but rather searched for animals with ‘balanced’ characteristics that are classically referred to as being ‘robust’ (Phocas et al., Reference Phocas, Belloc, Bidanel, Delaby, Dourmad, Dumont, Ezanno, Fortun-Lamothe, Foucras, Frappat, González-García, Hazard, Larzul, Lubac, Mignon-Grasteau, Moreno, Tixier-Boichard and Brochard2016).

High productivity levels can be achieved in tropical and Mediterranean silvopastoral systems that are based on highly diverse feed resources. In a Colombian dairy system, a tree-rich matrix allowed an increased stocking rate and increased milk production by 130% (Murgueitio et al., Reference Murgueitio, Calle, Uribe, Calle and Solorio2011) while completely eliminating the use of chemical fertilisers and contributing to climate regulation via the maintenance of soil organic matter and complex soil food webs. In tropical silvopastoral systems, the presence of trees within pastures benefits cattle welfare through less exposure to stressful climatic conditions and might increase milk yield (Paciullo et al., Reference Paciullo, Pires, Aroeira, Morenz, Maurício, Gomide and Silveira2014). Observations in Latin America have emphasised that enhancing tree species diversity and structural complexity could increase system resilience after a hurricane (Altieri et al., Reference Altieri, Nicholls and Montalba2017). An increased number of bee species in silvopastoral systems increased coffee production by an average of 5% (Cardoso and Mendes, Reference Cardoso and Mendes2014). As trees mature, they also provide ecological corridors that connect wildlife-friendly habitats. In addition, trees provide timber and firewood materials to farmers, roots and bark for medicinal uses, green forage and pods for livestock, and fruits and honey for human consumption (Murgueitio et al., Reference Murgueitio, Calle, Uribe, Calle and Solorio2011). Overall, win-win solutions are the result of collective decisions rather than situations where only individual interests and power relations prevail; this has been shown in various temperate and tropical environments (Groot et al., Reference Groot, Jellema and Rossing2010; Carmona-Torres et al., Reference Carmona-Torres, Parra-López, Groot and Rossing2011; Howe et al., Reference Howe, Suich, Vira and Mace2014).

How much can agroecology contribute to feeding the world?

One of the main criticisms of AE is that the productivity of agroecological systems is assumed to be lower than that of sustainably or traditionally intensified systems. This is denied by AE advocates who stress the high long-term performance (Murgueitio et al., Reference Murgueitio, Calle, Uribe, Calle and Solorio2011) and higher resilience of tropical agroecological systems (Altieri et al., Reference Altieri, Nicholls and Montalba2017). In addition, an increase in the complexity of crop rotations with temporary fodders, catch and cover-crops and traditional mixtures of crops and legumes (such as ‘milpa’ in Mexico or ‘méteils’ in France), mostly to the detriment of cereals, will likely enhance production and ecosystem services in both tropical and temperate environments (Altieri et al., Reference Altieri, Nicholls and Montalba2017; Barbieri et al., Reference Barbieri, Pellerin and Nesme2017). There is indeed no a priori for system intensification in the agroecological framework, as previously illustrated by the example of French CIVAM farms, for which profitability was high for farmers (Dumont et al., Reference Dumont, Fortun-Lamothe, Jouven, Thomas and Tichit2013; Duru and Therond, Reference Duru and Therond2015). One characteristic of most agroecological systems is that more time is required to supervise and observe the system. Consequently, labour productivity and productivity per unit area can be lower than in classically or sustainably intensified systems. This can be buffered by creating added value on the farm from higher-value products due to their better sensory quality or image, associated certification, and sometimes on-farm processing. Intrinsically, functional diversity plays a key role in the design of agroecological ruminant systems so that they both benefit from and provide ecosystem services. Scaling-up agroecological livestock farming would thus benefit from payment for environmental services.

Functional diversity is assumed to increase system resilience, redundancy being the underlying property that supports mechanisms of stabilisation since the collapse of any species can be offset by another species with similar characteristics. The ‘portfolio effect’ states that communities with higher species richness are more likely to include the species that is best adapted to any given condition in fluctuating environments so that the number of species per se has a positive effect on system resilience (Dumont et al., Reference Dumont, Jouven, Bonaudo, Botreau and Sabatier2017). Within herds, diversity of individual trade-offs between functions that was quantified in dairy goats (Puillet et al., Reference Puillet, Martin, Sauvant and Tichit2010) and dairy cows (Ollion et al., Reference Ollion, Ingrand, Delaby, Trommenschlager, Colette-Leurent and Blanc2016) does not increase herd milk production but is assumed to stabilise it under fluctuating environmental conditions. Abiotic and biotic interactions, including biogeochemical cycles and trophic interactions, lend ruminant systems further resilience properties (Dumont et al., Reference Dumont, Jouven, Bonaudo, Botreau and Sabatier2017). The extent to which ecological processes benefit long-term system performances needs to be analysed for a fair evaluation of agroecological ruminant systems. So far, there has been a huge gap in terms of investments in research between traditionally or sustainably intensified systems and agroecological systems since the former have received the majority of governmental funding and almost the total investment in research by the private sector (Tittonell, Reference Tittonell2014). A fair answer to the question of how much can AE can contribute to feeding the world thus requires the opening of a vast research agenda for the animal science community (Dumont et al., Reference Dumont, González-García, Thomas, Fortun-Lamothe, Ducrot, Dourmad and Tichit2014). In the AE paradigm, researchers no longer produce generic solutions or tools but consider farmer skills and knowledge (Altieri et al., Reference Altieri, Nicholls and Montalba2017). This calls for a transdisciplinary approach of the research-development-innovation chain to bridge the gap between science and practice.

Convergence between sustainable intensification and agroecology requires moving towards a wider food system scale

Agroecology and SI historically did not pay the same attention to the social dimensions of sustainability (Wezel et al., Reference Wezel, Bellon, Doré, Francis, Vallod and David2009; Loos et al., Reference Loos, Abson, Chappell, Hanspach, Mikulcak, Tichit and Fischer2014). Agroecology and SI could now converge as some researchers investigating SI have proposed the adoption of a food-system transformation perspective (Garnett et al., Reference Garnett, Appleby, Balmford, Bateman, Benton, Bloomer, Burlingame, Dawkins, Dolan, Fraser, Herrero, Hoffman, Smith, Thornton, Toulmin, Vermeulen and Godfray2013). Central to this is the conviction that excessive meat consumption is a leading cause of the environmental crisis. For instance, Popp et al. (Reference Popp, Lotze-Campen and Bodirsky2010) examined non-CO₂ GHG emissions from agriculture under different assumptions of food demand and concluded that reduced meat consumption would be far more effective than any technical mitigation measure. This goes in line with outputs from a recent meta-analysis (Aleksandrowicz et al., Reference Aleksandrowicz, Green, Joy, Smith and Haines2016). Consistently, most ongoing scenarios simulate a decline in animal protein consumption and limit livestock production to pasture and co-products from human food (e.g., van Zanten et al., Reference van Zanten, Meerburg, Bikker, Herrero and de Boer2016). However, some proponents of SI, including actors and lobbies from food and farming industry, and researchers, still highlight the short-term negative effects such options could have on livestock business and economic growth in the agricultural sector (Röös et al., Reference Röös, Bajželj, Smith, Patel, Little and Garnett2017).

While both frameworks aim to optimise land-use at the global level, there are still different views on where food should be produced. Indeed, while enough food is produced globally there remains strong food security challenges. This mismatch emerges through the complex patterns of global trade in food products (Smith, Reference Smith2013; Gordon et al., Reference Gordon, Bignet, Crona, Henriksson, Van Holt, Jonell, Lindahl, Troell, Barthel, Deutsch, Folke, Haider, Rockström and Queiroz2017). Integrated global markets have often been advocated by SI proponents (Röös et al., Reference Röös, Bajželj, Smith, Patel, Little and Garnett2017). For instance, demographic shifts and economic development in emerging countries have stimulated EU exports, notably for the dairy and pork sectors. By contrast, AE calls for new forms of regionally embedded agri-food systems, which implies rethinking market mechanisms. It has given priority to local autonomy and community control of land since the very beginning (Altieri, Reference Altieri2002; Rosset et al., Reference Rosset, Machín Sosa, Roque Jaime and Ávila Lozano2011; Cardoso and Mendes, Reference Cardoso and Mendes2014). In addition, AE calls to strengthen links between producers and consumers. In Europe and North America, local production emerged to restore consumer confidence in food systems. In Europe, local production creates added value for high-quality products with a strong territorial identity. In developing countries, the local production aims to secure food sovereignty based on low-input production systems. As no space is available for agricultural land expansion, for instance, in South Asia and East Asia (Smith, Reference Smith2013), local production will imply a search for methods to increase yield, an increase in cropping intensity (e.g., double or triple cropping within a year), and/or increased and more resilient production in complex silvopastoral systems. This can be achieved through the use of adapted crossbreeds or local breeds rather than the use of high-yielding livestock breeds that may be ill-adapted to climate change. Intensification efforts should also consider the social context and are made easier when innovations have been co-designed by farmers and NGO’s involved in rural development from the start of the process (e.g. Kosgey et al., Reference Kosgey, Baker, Udo and Van Arendonk2006 for small ruminant breeding programmes in the tropics).

Beyond closing yield gaps, meeting the projected demands of population growth calls for improving food distribution and access, as well as market infrastructure, which will require changing policies and global mechanisms that rule the food system (Foley et al., Reference Foley, Ramankutty, Brauman, Cassidy, Gerber, Johnston, Mueller, O’Connell, Ray, West, Balzer, Bennett, Carpenter, Hill, Monfreda, Polasky, Rockström, Sheehan, Siebert, Tilman and Zaks2011). As reviewed by De Schutter and Vanloqueren (Reference De Schutter and Vanloqueren2011), market and political obstacles could be overcome through six policy principles: (i) focusing efforts on the needs of smallholders; (ii) redistributing public goods as part of food security policies; (iii) gaining a richer understanding of innovation that includes traditional knowledge; (iv) involving meaningful participation of smallholders in local programmes and nationwide policies; (v) using public procurement to speed the transition towards sustainable agriculture; and (vi) redefining performance criteria used to monitor agricultural projects beyond classical measures such as productivity per unit of land or water. This confirms that SI and AE cannot be considered from their biotechnical definition only and should further move towards a wider food system scale, encompassing productive, environmental and social dimensions.

Use of new technologies is widespread in sustainable intensification and offers opportunities for scaling-up agroecology

The application of new technologies has a different share within the SI and AE frameworks. Within SI, new technologies provide a foundation for the transition towards more efficient and intensified systems, for instance, through the optimal timing and amount of feed offered to the animals, nutrient recycling, early detection of livestock diseases, and alternative reproduction techniques (Food and Agriculture Organisation, 2011; Dumont et al., Reference Dumont, Fortun-Lamothe, Jouven, Thomas and Tichit2013; Garnett et al., Reference Garnett, Appleby, Balmford, Bateman, Benton, Bloomer, Burlingame, Dawkins, Dolan, Fraser, Herrero, Hoffman, Smith, Thornton, Toulmin, Vermeulen and Godfray2013; Campbell et al., Reference Campbell, Thornton, Zougmoré, van Asten and Lipper2014). Makkar (Reference Makkar2018) advocated the need to develop a business model around these technologies and to bring private companies on board. Through the SI framework, biotechnology and the use of genomic information have also found a new avenue to promote themselves as a solution to world hunger (Tittonell, Reference Tittonell2014). The most recent example is the Bill and Melinda Gates Foundation funding a non-profit research alliance to select cows that are both more productive and resistant to heat stress, for instance by providing genomic tools for screening young animals that have the desired traits for a particular environment. Early detection of livestock diseases through precise animal monitoring would benefit animal welfare. More broadly, making appropriate use of digital sciences is likely to help in the gathering of individual-based data and the monitoring of every component of the system to increase its overall efficiency (Ingrand, Reference Ingrand2018). Livestock farmers are indeed becoming increasingly reliant on new digital technologies and communication tools that increase knowledge dissemination and connections among actors. For instance, pastoralist communities in East Africa and Mongolia use smartphones as an early warning system for predicting forage availability in arid rangelands, which increases system long-term efficiency. The key issue for making digital science part of the agroecological transition is that it remains at a low cost and is thus accessible to smallholder farmers.

Strict certifications could offer opportunities to boost sustainable intensification and agroecology

Consumers and civil society pressures have led to increased adoption of sustainable practices by food companies (Thorlakson et al., Reference Thorlakson, de Zegher and Lambin2018), which is likely to have some animal welfare and environmental benefits. Although varying across regions and socio-demographic characteristics of consumers, the willingness of consumers to pay for farm animal welfare is higher for beef and dairy cows than for pigs, poultry and aquaculture (Clark et al., Reference Clark, Stewart, Panzone, Kyriazakis and Frewer2017). This reveals the potential of labelling ruminant farming systems through strict welfare certifications so that citizens can develop their own system of consumption ethics (Dumont et al., Reference Dumont, González-García, Thomas, Fortun-Lamothe, Ducrot, Dourmad and Tichit2014). Regarding environmental benefits, compliance with the Brazilian Forest Code was achieved thanks to increased transparency in the Brazilian cattle supply chain (Guéneau, Reference Guéneau2018; zu Ermgassen et al., Reference zu Ermgassen, de Alcântara, Balmford, Barioni, Beduschi Neto, Bettarello, de Brito, Carrero, de AS Florence, Garcia, Gonçalves, da Luz, Mallman, Strassburg, Valentim and Latawiec2018). This increased compliance did not affect the productivity of beef and dairy farms. Success depended on the independent audit of slaughterhouses, to ensure they only purchased livestock from farms/municipalities in compliance with the Forest Code (zu Ermgassen et al., Reference zu Ermgassen, de Alcântara, Balmford, Barioni, Beduschi Neto, Bettarello, de Brito, Carrero, de AS Florence, Garcia, Gonçalves, da Luz, Mallman, Strassburg, Valentim and Latawiec2018). In Latin America, AE, however, does not operate through any standard or certification system (Tittonell, Reference Tittonell2014). Botreau et al. (Reference Botreau, Farruggia, Martin, Pomiès and Dumont2014) showed that it is possible to build a multicriteria evaluation tool that qualifies the compliance of mountain dairy systems with agroecological principles. By crossing agroecological principles with categories of state variables on which the fulfilment of these principles should have a positive effect, it becomes possible to identify a complete range of practice-based criteria accounting for ecological processes and interactions within a system. The criteria should be understood and acceptable by farmers and technical advisors and are specific to the mountain context (Botreau et al., Reference Botreau, Farruggia, Martin, Pomiès and Dumont2014), as agroecological practices must be adapted to local conditions (Altieri, Reference Altieri2002). This methodology allows the comparison of mountain dairy farms and can thus lead to a strict and verifiable certification of agroecological systems.