A nutrition-driven perspective

Gussow and Clancy⁽ Reference Gussow and Clancy ^{11 )} were the first to suggest the term ‘sustainable diet’ to describe a diet ‘composed of foods chosen for their contribution not only to health but also to the sustainability (the capability of maintenance into the foreseeable future) of the (…) agricultural system’⁽ Reference Herrin and Gussow ^{12 )}. Literally, the concept of diet in nutrition refers to the sum of foods consumed by a person. Whole diet, or dietary pattern, analysis has emerged as an alternative and complementary approach to the study of individual nutrients or foods, highlighting the dynamic and multiple factors involved in eating practices⁽ Reference Hu ^{13 )}. It helped better communicate healthy eating messages that emphasise a balance of food and beverages within energy needs⁽ Reference Freeland-Graves and Nitzke ^{14 )}. More fundamentally, adopting a whole-diet approach is now seen as necessary to examine the relationships between nutrition and health⁽ Reference Popkin ^{15 )}. It reflects the increasing recognition of the multidimensional nature of diets and diet-related diseases, from nutrient intakes and metabolism to food consumption behaviours and attitudes⁽ Reference Kant, Leitzmann and Park ^{16 )}.

Multidimensionality is further enhanced as the impacts of diets not only on health, but also on the environment or the economy, are considered to assess the sustainability of food choices. Participants at the 2010 International Conference jointly organised by the Food and Agriculture Organization and Bioversity International agreed on a common definition of sustainable diets as ‘those diets with low environmental impacts which contribute to food and nutrition security and to healthy life for present and future generations. Sustainable diets are protective and respectful of biodiversity and ecosystems, culturally acceptable, accessible, economically fair and affordable; nutritionally adequate, safe and healthy; while optimising natural and human resources’^{( 17 )}.

The sustainable diet concept advocates for a more consumer-driven thinking on the sustainability of agriculture, promoting a research and policy agenda that introduce nutrition as one of its core dimensions. It claims that understanding the determinants of consumer choices can improve agricultural and food systems, the environment and the health. More fundamentally, it emphasises the health and food security purpose of food systems, and highlights the need for quality, not just quantity or access. Advocates promote economically, socially and environmentally sustainable food systems that concurrently ensure ‘physical and economic access to sufficient, safe and nutritious food to meet their dietary needs and food preferences for an active and healthy life’^{( 18 )}. This reminds us that food, or more precisely feeding people, is agriculture and food systems’ main reason for being⁽ Reference Haddad ^{19 )}. As such, the concept of sustainable diets provides a food and nutrition security-orientated perspective on the question of the sustainability of food systems.

Food security and sustainable development

Food and nutrition security is a major concern today with still 842 million people undernourished^{( 1 )}. Resulting undernutrition is affecting millions of people, in particular children under 5 years with about 165 million stunted children in developing countries^{( 20 )}. Sub-Saharan Africa and South Asia, with wide sub-regional variation, are the most affected regions by stunting. Undernutrition is accompanied by, in some cases linked to, micronutrient malnutrition. Malnutrition involves privation in essential micronutrients with low food diversity. Deficiencies in essential micronutrients have detrimental effects on health; vitamin A is required for multiple physiological processes, ranging from vision to embryonic development; iron is an important component of haemoglobin, the oxygen-carrying component of blood; iron also plays an important role in brain development and iron-deficiency anaemia can impair the cognitive development of children; iodine is essential for healthy thyroid function and growth, etc. Simultaneously, increased prevalences of overweight and obesity are reported in both low- and high-income countries and represent the major health threats. Excessive fat accumulation, measured by the BMI, is acknowledged to be a risk factor for various non-communicable diseases and health problems, including CVD, diabetes, cancers and osteoarthritis^{( 21 )}.

Simultaneously, climate change and environmental degradation are major challenges to sustainable development. The global climate and other life-supporting environmental systems are seriously perturbed and depleted⁽ Reference Stocker, Qin, Plattner, Tignor, Allen, Boschung, Nauels, Xia, Bex and Midgley ^{4 )}. These changes include higher temperatures, drought-prone and long-term drying conditions in some sub-tropical regions, rising sea levels, acidification of oceans, declining water quality, depleting fish stocks, increasing frequency and severity of floods and other climate-related natural disasters. Biodiversity is also at risk, with 20 930 species and ecological communities known to be threatened^{( 22 )}. Biodiversity, the basis of ecosystem health and future food security, has been more seriously harmed by human activities in the past 50 years than at any other time in human history^{( 5 )}. Agriculture and the food sector have historically been major contributors to environmental degradation. For instance, irrigated agriculture globally accounts for 70 % of the consumption of freshwater resources^{( 23 )}. However, there is a bi-directional relationship between environmental degradation and food system activities. People, particularly those living in developing countries, are vulnerable to environmental changes that result in reduced quantity, quality and affordability of food. Similarly, nutrition transition and food system transformation go together. The current global health crisis of malnutrition, both in developed and developing countries, and the contemporaneous urging environmental degradation present new challenges for food systems and calls for changes. Improved food systems could be a major partner in the environmental solution.

Diets as system outputs

A crucial question is then: can optimal diets be derived that concurrently meet dietary requirements while reducing detrimental environmental impacts? Contrary to conventional wisdom, recent evidence suggests that high nutritional quality products might not be necessarily more environment-friendly. Although plant-based foods have lower greenhouse gas emissions (GHGE) per unit weight, better quality diets were found associated with significantly higher GHGE after adjustment for energy intake⁽ Reference Vieux, Darmon and Touazi ^{24 ,} Reference Vieux, Soler and Touazi ^{25 )}. So, can consumers lower, for example, their carbon footprint through making changes in the kind of food they buy, and still meet nutritional adequacy recommendations? MacDiarmid et al. ⁽ Reference Macdiarmid, Kyle and Horgan ^{26 )} derived what would look like such an optimal diet for a representative UK consumer for different GHGE reduction targets. Using mathematical linear programming, they conclude that such a diet can be achieved for the GHGE objectives set for 2020 (−25 %); however, meeting the targets for 2050 (−70 %) and dietary recommendations will require a ‘radical shift in food consumed’. Ad hoc constraints were added to the model to maintain simulated diets within consumers’ acceptability limits. These results clearly show that demand-side approaches to the problem of environmental unsustainability are desirable and likely to contribute to improvements. However, as highlighted by MacDiarmid et al. ⁽ Reference Macdiarmid, Kyle and Horgan ^{26 )}, GHGE reductions should be made to both the demand and supply sides within the food chain, in particular to attain longer-term objectives. If other environmental, economic, social and ethical aspects of sustainability were to be included in the optimisation model, while strengthening the acceptability constraints, one can wonder if feasible solutions can be derived by changes at the sole food basket level. Linear optimisation theory tells us that if there are x decision variables, then a set of x equality constraints needs to be specified for one unique optimal solution and vice versa. Any extra equality constraints will overspecify the problem. If feasible solutions are to be identified, extra decision variables need to be considered. In other words, other levers need to be operated jointly with actions encouraging behaviour changes.

The processes underlying food insecurity and diet-related environmental, economic and social unsustainability derive from a shared food system. For instance, GHGE are not food attributes, but outputs of different activities along the value chain. Food consumption is a heavy contributor to ‘embodied’ or indirect emissions in products that result from activities prior to purchase⁽ Reference Kim and Neff ^{27 )}. In practice, these indirect emissions are very hard to be accurately estimated and attributed to a good or an individual. Modelling exercises of optimal diets have had to use so far averages coming from life-cycle assessment studies on, sometimes, rather aggregated food groups. There might thus be a high degree of variation around these average estimates. For instance, Lindenthal et al. ⁽ Reference Lindenthal, Markut and Hörtenhuber ^{28 )} report substantial differences in terms of GHGE between organic production methods as compared with conventional farming in Austria (10–21 % lower CO₂-eq/kg product for organic dairy, 25 % for organic wheat bread and 10–35 % for organic vegetable.). Similar studies elsewhere have reported the same results⁽ Reference Küstermann, Kainz and Hülsbergen ^{29 ,} Reference Petersen, Regina and Pöllinger ^{30 )}.

Consumers stand at the top of the food system and diets are outputs of longer and more complex food chains encompassing several activities. Technologies and policies affect the overall environmental performances, food security and health outcomes⁽ Reference Ingram ^{31 )}. To derive optimal sustainable diets, we need to look at all the variables that influence the flow of activities along the food system. These are the levers to act upon. To assess and enhance food sustainability, focus needs to move beyond the food basket while ultimately bearing in mind that diets and nutrition are the final reason for being of the food system. Burchi et al. ⁽ Reference Burchi, Fanzo and Frison ^{32 )} define a system as ‘a set of elements that function together as collective units which have properties greater than the sum of their component parts’. The food system concept describes the required inputs, processes and generated outputs involved in the provision of food and nutrients for sustenance and health, including growing, harvesting, processing, packaging, transporting, marketing, consuming and disposing of food⁽ Reference Ericksen, Ingram and Liverman ^{8 ,} Reference Rastoin and Ghersi ^{33 ,} Reference Rutten, Yaroch and Story ^{34 )}. The current joint crisis of malnutrition and unsustainability has roots in agricultural and food systems that do not deliver enough essential nutrients to meet dietary requirements for all⁽ Reference Graham, Welch and Saunders ^{35 ,} Reference DeClerck, Fanzo and Palm ^{36 )}. The solution to sustainable diets lies both in sustainability-orientated food choices and in changes in the food systems. And modern societies depend on complex social–ecological systems to provide food⁽ Reference Ericksen, Ingram and Liverman ^{8 ,} Reference Fraser, Mabee and Figge ^{37 –} Reference Ostrom ^{39 )}.

Food and nutrition as ecosystem services

Agriculture and the food sector at large have a first-hand touchpoint with nature: crops need soil, water, insects for pollination, etc. The analysis of the relationships between resource acquisition and living organisms, at the heart of the sustainable diet concept in the case of human organisms, is also an ecological question, and can surely benefit from insights from ecology. Ecosystems consist of a community of species, or biodiversity, interacting with each other and with their environment. The product of these interactions, which include competition, predation, reproduction and cooperation, is essential to human wellbeing. Human subjects depend on goods provided by natural and managed ecosystems. These goods and other benefits provided by ecosystems to mankind are collectively referred to as ecosystem services (Ecosystem services were defined in the Millennium Ecosystem Assessment^{( 7 )} as ‘the benefits people obtain from ecosystems’, both natural and managed. These services may be categorised as provisional (fibre, food, water and fuel), regulative (climate and disease regulation, water purification), cultural (aesthetics, heritage, education, recreation and spiritual) or supporting services (nutrient cycling, primary production and soil formation)). All are processes through which ecosystem sustain human livelihoods. Food production is an ecosystem service central to human welfare⁽ Reference Costanza, d'Arge and De Groot ^{40 )}. The capacity of ecosystems to provide us with the energy and nutrition for our daily life fully depends on the foods that agriculture and food systems provide us. Clear from this process-based interpretation, human nutrition should be considered one of the most fundamental ecosystem services, or alternatively as dependent on several ecosystem services, including provisioning, regulating, supporting and cultural services⁽ Reference DeClerck, Ingram and Rumbaitis del Rio ^{41 )}.

Agricultural biodiversity, or agrobiodiversity, is the sub-component of biodiversity that refers to the biological variety and variability of living organisms that are involved in food and agriculture. It can be considered at three main levels: ecosystem diversity, species diversity and genetic diversity⁽ Reference Heywood, Fanzo, Hunter, Fanzo, Hunter, Borelli and Mattei ^{42 , 43 )}. It includes habitats and species outside of farming systems that benefit agriculture and enhance ecosystem functions such as pollination, soil dynamics and control of GHGE. Agrobiodiversity comprises the constituents of biological diversity important to food and agriculture as well as for the agroecosystem⁽ Reference Frison, Cherfas and Hodgkin ^{6 ,} Reference Brussaard, Caron and Campbell ^{44 )}. Furthermore, it is the result of the deliberate interaction between human subjects and natural ecosystems. Subsequent agroecosystems are thus the product of not just physical elements of the environment and biological resources, but vary according to cultural and management systems⁽ Reference Heywood, Fanzo, Hunter, Fanzo, Hunter, Borelli and Mattei ^{42 )}. Agrobiodiversity includes a series of social, cultural and ethical variables.

Reduction in agrobiodiversity and simplification of diets

Modern agriculture and food systems are contributing to the simplification of the structure of the environment, replacing nature's biodiversity with a small number of domesticated plant species and animal breeds⁽ Reference Altieri ^{45 )}. This process has been one of the main factors that allowed much of the human population to enjoy unprecedented levels of development and improved health. However, as efforts have been directed at maximising production and productivity, uniformity has replaced diversity within cultivated systems⁽ Reference Sage ^{46 )}. Agricultural intensification, which implies specialisation and genetic standardisation, reduction of utilised species, conversion of forests and wild land to anthropogenic habitats, homogenisation of soils through amendments, is certainly the first human-related cause of biodiversity loss⁽ Reference Frison, Cherfas and Hodgkin ^{6 ,} Reference Rosen ^{47 ,} Reference Tilman, Cassman and Matson ^{48 )}. The increase in food supply has thus come with important trade-offs that include soil degradation and loss of many regulatory and supporting ecosystem services. These trade-offs can impair the ability of the ecosystems to deliver the essential nutrients for human diets⁽ Reference Palm, Sanchez and Ahamed ^{49 )}.

This increased reliance on domesticated species and selected crop varieties can be linked to a significant reduction in dietary diversity. Modern agriculture is genetically dependent on a handful of varieties for its major crops⁽ Reference Kahane, Hodgkin and Jaenicke ^{50 )}. The world's agricultural landscapes are planted mostly with some twelve species of grain crops, twenty-three vegetable crop species and about thirty-five fruit and nut crop species⁽ Reference Fowler and Mooney ^{51 )} (as a comparison, one single hectare of tropical rain forest contains on average over 100 species of trees ⁽ Reference Perry ^{52 )}; cited in⁽ Reference Altieri ^{53 )}). This process of simplification of agriculture generated a model where only a small number of crop species dominate our energy and nutritional intakes. Three crops alone (rice, wheat and maize) account for more than 55 % of human energy intake⁽ Reference Stamp, Messmer and Walter ^{54 )}.

Although varying in nutrient content, no single crop species is capable of providing all essential nutrients. Nutritional diversity is now widely recognised to be a key factor for adequate diets likely to satisfy the complex human nutritional needs⁽ Reference Arimond, Wiesmann and Becquey ^{55 –} Reference Torheim, Ouattara and Diarra ^{58 )}. Evidence of the valuable outcomes of diversity in decreasing malnutrition, morbidity and mortality⁽ Reference Frison, Cherfas and Hodgkin ^{6 ,} Reference Tucker ^{59 )} is completed by indications of positive correlation with child growth and survival⁽ Reference Arimond and Ruel ^{60 ,} Reference Pelletier and Frongillo ^{61 )}. The importance of nutrient diversity for human wellbeing calls for dietary diversification. However, the quality of nutritional supply and human health is in danger because of a loss in biodiversity. A reduction in the consumption of varied, ‘nutritionally-rich’ and ‘functionally-healthy’ plant-based foods is reported in most developed and emerging countries⁽ Reference Johns and Eyzaguirre ^{62 )}. The pre-eminent simplification of human diets, associated with changing lifestyles, led to nutrient deficiencies and excess energy consumption. However, the elimination of most essential nutrient deficiencies (most important micronutrients usually reported are vitamin A, iodine and iron, zinc; Graham et al. ⁽ Reference Graham, Welch and Saunders ^{35 )} provide a list of fifty-one essential nutrients for sustaining human life) requires only small increases in the variety of food items an individual consumes⁽ Reference Ruel ^{63 )}. As a result, balanced nutrition in human diet can depend significantly on the diversity within crops⁽ Reference Mouillé, Charrondière and Burlingame ^{64 )}.

Ecological interdependences are key factors of the dietary content of most living species. Some lesser-known cultivars and wild varieties have been reported to be micronutrient superior over other more extensively utilised cultivars. For example, recent analyses have shown that provitamin-A carotenoid content of bananas differs by a factor of 8500 between different cultivars⁽ Reference Burlingame, Charrondiere and Mouille ^{65 )}. In Micronesia, the local ‘karat’ banana has been found to contain high levels of provitamin-A carotenoids, which contribute to protection against vitamin A deficiency and chronic diseases, including certain cancers, heart disease and diabetes⁽ Reference Engelberger, Aalsbersberg and Ravi ^{66 )} (cited in⁽ Reference Sajise ^{67 )}). In this regard, the term ‘neglected and underutilised species’ or ‘development opportunity crop’ refers to those species whose potential to improve people's livelihood is not being fully exploited (given the current lack of detailed and comprehensive nutritional information about diversity within crops at the cultivar level and the role it plays in nutrition, the Food and Agriculture Organization has launched the INFOODS initiative⁽ Reference Mouillé, Charrondière and Burlingame ^{64 )}).⁽ Reference Padulosi, Heywood, Hunter, Yadav, Redden, Hatfield, Lotze-Campen and Hall ^{68 )}. For instance, a local fruit, Berchemia discolor, was found to contribute in a low-cost manner to closing nutrient gaps in Kenya⁽ Reference Termote, Cogill and Deptford ^{69 )}. Peach palm (Bactris gasipaes) provides, under low soil fertility and extreme rainfall conditions, starchy fruits with high protein density, rich in monounsaturated oleic acids, carotenoids, vitamin E and potassium⁽ Reference Graefe, Dufour and van Zonneveld ^{70 )}. Amaranth, as a leafy vegetable, is nutritionally comparable with spinach while showing strong photosynthetic activity and water use efficiency⁽ Reference Wang, Ebert, Holmer, Linwattana, Nath and Keatinge ^{71 ,} Reference Ebert ^{72 )}. The drumstick tree (Moringa oleifera) combines the traits of high yield and high nutrient density in essential micronutrients, vitamins, antioxidants and bioavailable iron, making it a good supplement for children and pregnant and lactating women⁽ Reference Ebert ^{72 )}. All these examples demonstrate how intraspecific biodiversity and the consumption of neglected species and varieties can be essential to nutrition security.

Increasing the number of crops available to local communities, in particular in developing countries, increases the likelihood of obtaining the nutrients needed for healthy and productive lives⁽ Reference DeClerck, Fanzo and Palm ^{36 )}. Deckelbaum et al. ⁽ Reference Deckelbaum, Palm and Mutuo ^{73 )} showed that biodiversity and hunger hotspots geographically correspond, reminding us of the link that Jared Diamond unravelled about the spatial relationship between biodiversity availability and society development⁽ Reference Diamond ^{74 )}. This evidence, demonstrating the correlation between hunger and biodiversity-losing areas, confirms the need for local biodiverse agricultural systems. DeClerck et al. ⁽ Reference DeClerck, Fanzo and Palm ^{36 )} further observed that improving functional agrobiodiversity in Kenya reduces anaemia incidence, and that interventions supporting environmental sustainability, through biodiversity, can have multiple direct and indirect outcomes on human health and nutritional wellbeing. Similarly, in rice-based aquatic production systems, Halwart⁽ Reference Halwart ^{75 )} found that vegetal agrobiodiversity allowed improved biological diversity and diverse nutritional sources for human subjects (calcium, iron, zinc, vitamin A, some fatty acids and limiting amino acids). Moreover, through fish biodiversity, rice yields increase and the presence of several aquatic organisms in rice ecosystems allows a better biological control of vectors and pests. Animal and vegetal agrobiodiversity in rice-based ecosystems increases income through yield growth and lower costs for pesticides through biological control⁽ Reference Diamond ^{74 )}. These issues suggest, for tackling malnutrition, but also other aspects of food insecurity, the need to link ecology and agriculture to human nutrition and health.

Agrobiodiversity and resilience for food security

On top of nutritional issues, agricultural biodiversity is an essential component in the sustainable delivery of a more secure food supply. Agrobiodiversity is the outcome of thousands of years of efforts by farmers, selecting, breeding and developing appropriate production systems and methods. It plays a crucial role in productivity and livelihood of farmers, by providing the wide range of resources they need to increase productivity in favourable settings or to adapt to variable conditions. Biodiversity simplification resulted in an artificial ecosystem that requires constant human intervention, whereas plant biodiversity allows internal regulation of essential functions in natural ecosystems⁽ Reference Altieri ^{53 )}. Several nature- and human-related drivers of change threaten the ability of social–ecological systems to maintain vital functions and processes: climate change, natural resources exploitation, habitat depletion, pollution, etc.

Understanding how agrobiodiversity is likely to impact agricultural and ecosystems is key. Climate change is a potent risk to the world's food supply in coming decades, likely to undermine production and driving up prices⁽ Reference Godfray, Beddington and Crute ^{76 ,} Reference Ingram, Ericksen and Liverman ^{77 )}. Agricultural biodiversity will be absolutely essential to cope with the predicted impacts of climate change. Crop genetic diversity provides partial resistance to diseases, and enables farmers to exploit different soil types and microclimates for a variety of nutritional and other uses⁽ Reference Altieri ^{53 )}. Improved resilience, to climatic shocks among others, is observed in highly biodiverse ecosystems⁽ Reference Swift, Izac and Van Noordwijk ^{78 ,} Reference Rees, Condit and Crawley ^{79 )}. In Malawi, Mozambique and Zambia, between 26 and 50 % of rural households relied on indigenous fruits as a coping strategy during critical seasonal hunger periods⁽ Reference Akinnifesi, Kwesiga and Mhango ^{80 ,} Reference Powell, Lckowitz and McMullin ^{81 )}. Furthermore, biodiversity in agroecosystems accomplishes multiple ecological services beyond the production of food such as: nutrients recycling, hydrological regulation, purification of toxic chemical compounds, etc. For instance, improvement in agroforestry biodiversity reduces nutrient leaching and soil erosion and refurbishes key nutrients from the lower soil layers^{( 82 )}. To assess the role of agricultural biodiversity in sustainable and secure food production, cross-sectoral approaches are necessary as potential benefits can be manifested at different ecological and human scales⁽ Reference Frison, Cherfas and Hodgkin ^{6 )}. Farmers also conserve, and modify their use of, agrobiodiversity to better adapt to different environmental conditions, but also to changing market conditions⁽ Reference Pascual, Narloch and Nordhagen ^{83 )}. In Indonesia, the conservation of high levels of biodiversity in rubber agro-forests helped secure population livelihood during the 2008 fall of rubber prices by providing an alternative source of income from secondary products⁽ Reference Powell, Lckowitz and McMullin ^{81 ,} Reference Feintrenie and Levang ^{84 )}. Agrobiodiversity can thus be seen as a crucial asset to keep multiple options open. As a general rule, increasing the number of species in a community will enhance the number of functions provided by that community, and will reinforce the stability of the provision of those functions⁽ Reference DeClerck, Fanzo and Palm ^{36 )}.

Modelling activities, capturing diversities

In recent years, there has been a significant development of bio-economic models, enhanced by the recognition of the multifunctionality of agriculture and the multiplicity of objectives assigned to the agricultural policies⁽ Reference Janssen and van Ittersum ^{85 )}. The subsequent increasing demand for integrated assessment called also for more dialogue and co-operation between scientists from various disciplines, and bio-economic models have been advocated as an adequate tool for such a purpose⁽ Reference Kragt, Seppelt, Voinov, Lange and Bankamp ^{86 )}. Bio-economic models refer to models that couple both an economic and a biophysical component. Brown⁽ Reference Brown ^{87 )} more precisely identifies models primarily concerned with ‘biological process (…) to which an economic analysis component has been added’. Another kind of model consists of ‘economic optimisation models which include various biophysical components as activities among the various choices for optimization’. In between, he suggests a third category that integrates in an interactive manner the biophysical and the economic modules. This last category genuinely deserves to be called ‘bio-economic’.

At the heart of most bio-economic models lies the paradigm that, for analysing the relationships and trade-offs between socio-economic systems and biophysical and ecological processes, and to help evaluate how management actions affect different policy objectives, it is necessary to model activities⁽ Reference Flichman, Louhichi, Boisson and Flichman ^{88 )}. What produces biodiversity depletion or soil erosion is not wheat or maize production per se, but the way it is produced. And there are several ways of producing the same product. The degree of pressure on the environment will depend on the crop selected and its combination with other crops, the tillage technique, the type of soil, the production system, the period of harvest, the seasonality and many other technical issues. It is therefore not adequate to associate a final product with a single simple production function. The relationships between a final product and the inputs associated with its production, highly non-linear because of the large set of possible combinations, might be better captured by considering the variety of activities or production processes.

Land, water, seeds (of different species and varieties), labour, energy, machinery, fertilizer, etc. are taken into account as inputs to the agricultural production. Food are outputs, as well as pollution, changes in landscape, depletion of natural resources, soil erosion, loss of underground water, habitat destruction, biodiversity losses, etc. There are numerous possible combinations of inputs to produce several outputs. Using an example from agricultural production, wheat systems do not only produce grain, but also straw and different types of pollution. They are ‘joint products’⁽ Reference Baumgärtner, Dyckhoff and Faber ^{89 ,} Reference Pasinetti ^{90 )}. Thus each activity can produce several products (e.g. grain, straw and pollution), and in turn each product can be produced by several activities (e.g. several ways of producing grain). As a consequence, modelling the relationships between a final product and the ‘externalities’ become even more challenging to synthesize.

Bio-economic models represent production activities in an explicit manner. A production activity describes a specific production process. Usually called an engineering production function, it describes explicitly the relationships between factors of production and products expressed in physical quantities (e.g., kg fertilizer/ha, m³ water for irrigation, etc.). In agriculture, an activity is defined by the technical coefficients that represent the use of inputs needed to produce different outputs⁽ Reference Flichman, Louhichi, Boisson and Flichman ^{88 )}. These engineering production functions, which use primal variables (physical quantities), constitute the essential link between the biophysical and economic processes. Models based on cost functions, which use dual variables (prices), can hardly analyse the relationships between inputs and outputs in a straightforward and proper manner. The fact that one product is obtained through several production activities, explains in part the complex and non-linear relationships between inputs and outputs observed per product, which are difficult to capture mathematically. On the contrary, the average cost can more realistically be assumed equal to the marginal cost when considered per activity. Relationships between inputs and joint products by activity are thus linear functions of Leontief type. The use of engineering production functions creates a strong information demand, requiring data framed in terms of physical input–output matrices. However, thanks to this representation, positive and negative jointness can be simultaneously taken into account. This more direct approach can help assess the joint interactions between biodiversity and nutrition.

Biodiversity in bio-economic modelling

There are basically four approaches to introduce biodiversity or agrobiodiversity in bio-economic optimisation models. In a normative approach, it is possible to include biodiversity conservation targets directly in a multi-objective function. Multi-objective optimisation models are goal-oriented models, where optimal decisions need to be taken in the presence of trade-offs between two or more conflicting objectives. Holzkämper and Seppelt⁽ Reference Holzkämper and Seppelt ^{91 )} developed a spatially explicit optimisation model with respect to ecological and economic goals, namely habitat suitability for three target species and profit losses from different land-use options. Results show that optimum agricultural land-use patterns differ between species, as well as between study sites. Groot et al. ⁽ Reference Groot, Rossing and Jellema ^{92 )} explore the synergies and trade-offs between financial returns, landscape quality, nature conservation and environmental quality in a spatially explicit land-use allocation model, which combines agronomic, economic and environmental indicators with biodiversity and landscape quality indicators. More specifically, their Landscape IMAGES model couples an agroecological model to a multi-objective optimisation algorithm that generates a set of alternative landscape configurations. An agroecological engineering approach was used to design production activities.

Alternatively, impacts on biodiversity of different land-use management options and policy scenario can be assessed through optimisation models. Schönhart et al. ⁽ Reference Schönhart, Schauppenlehner and Schmid ^{93 )} address the effects of land use intensity and landscape development on biodiversity at farm and landscape levels. Their integrated land use model combines a crop rotation model with a biophysical process model (erosion–productivity impact calculator) and a spatially explicit farm optimisation model. Field- and farm-specific crop yields, crop rotations and environmental outcomes of the biophysical model are inputs to the farm optimisation model, which maximises total farm gross margin subject to resource endowments and several balance equations. Decisions in integrated land use model are assumed to reflect actual producers’ choices postulating efficient farm resource utilisation. This structure allows introducing landscape metrics, such as the Shannon's diversity index, to quantify the spatial biodiversity impacts of landscape development scenarios. Scenario analysis is used to assess the cost-effectiveness of different agro-environmental measures to achieve biodiversity targets. One asset of the integrated land use models is the inclusion of spatial modelling of landscape elements. Similarly, Mouysset et al. ⁽ Reference Mouysset, Doyen and Jiguet ^{94 )} adopt a multi-criteria approach to assess jointly the impacts of public policy options on conservation of biodiversity and farming production. Assuming income-maximising farmers under technical constraints, the authors test different taxation scenarios on economic performances and farmland bird abundance.

As argued by Schönhart et al. ⁽ Reference Schönhart, Schauppenlehner and Schmid ^{93 )}, biodiversity conservation targets can also be introduced in the model as constraints. Van Wenum et al. ⁽ Reference Van Wenum, Wossink and Renkema ^{95 )} study optimal wildlife management on crop farms using integer programming. They compute a wildlife-cost frontier at the farm level evaluating the optimal trade-off between species richness and total gross margins. Their model derives sets of management activities that maximise farm income under incrementally varying wildlife conservation requirements. Results provide the extent to which stepwise increases in species richness objectives impact negatively farm profits.

A last approach consists of integrating agrobiodiversity at the core of the model in the definition of the agricultural activities. All the earlier examples of models that consider impacts on biodiversity are based, at least partially, on engineering production functions. It implies defining activities such as obtaining constant marginal costs. Combinations of crops and rotation schemes, in interaction with the environment and agronomic technique, on the farm at the field level have to be considered to specify activities. For example, maize, beans and squash (the indigenous ‘American three-sisters’) planted simultaneously would be modelled as a specific activity, different from an activity involving only one of the three crops or any other combination. The planting techniques used, either just in the same field or in the same hole, would also be distinguished. Conceptually, this approach by activity would allow setting agrobiodiversity at the core of the model, and better match an understanding of the environmental and nutritional outcomes of diets as system outputs. Indeed, specific environment and nutrition impacts can thus be specified by activity and not by product. However, one strong limitation regards data requirements. Given the wide array of possible combinations, a large number of technical coefficients, which enter the model as external variables, need to be available and properly estimated to result in real improvement to existing modelling exercise.

Joint assessment of nutrition and biodiversity

The increasing demand for integrated assessment, including nutrition⁽ Reference Hammond and Dubé ^{10 ,} Reference Frongillo, Tofail and Hamadani ^{96 –} Reference Remans, Smukler, Fanzo, Hunter, Borelli and Mattei ^{98 )} calls also for bio-economic models integrating consumer choices and dietary patterns, and subsequent sets of food consumption and nutrition indicators. A nutrition-driven food system, which also ensures that environmental integrity, economic self-reliance and social well-being are maintained and enhanced, places people, as consumers, as one of its central focus⁽ Reference Burchi, Fanzo and Frison ^{32 )}. Not only should we be able to determine food and nutrient availability at the farm or food system level, resulting from the use of biodiversity for instance, but we also need to understand and consider how it translates into actual consumption at the household and individual level. To achieve this, models of food consumption patterns and behaviours need to be integrated into the bio-economic models. This type of tool will allow a proper nutritional analysis and evaluation of required changes in the food systems to reach sustainable diets. In the context of developing countries, farm-household models offer the conceptual background to expand existing bio-economic farm models, to capture the interactions between ecological dimensions and agronomic decisions with consumers’ choices (and acceptability of simulated options in terms of consumers’ preferences) and nutritional outcomes. Small-holder farmers are vital for developing countries’ economies, supporting today one-third of humanity^{( 99 )}. Farm households, while increasingly selling and relying on markets, represent an ‘easier to control for’ food system at the smallest scale.

In the case of small-holder farmers in developing countries, the deciding entity is both a producer and a consumer. In the existence of market failures, non-separability regarding production and consumption decisions has to be assumed, and a farm-household approach becomes necessary⁽ Reference Sadoulet and De Janvry ^{100 )}. Several attempts have been made to couple bio-economic and farm-household models⁽ Reference Janssen and van Ittersum ^{85 ,} Reference Holden, Shiferaw and Pender ^{101 –} Reference Shiferaw, Holden, Aune, Heerink, Keulen and Kuiper ^{104 )}. In particular, the Joint Research Centre, with the CIHEAM-IAMM and other partners, further developed the FSSIM model of the European Commission for application in developing countries. The FSSIM-Dev (Farm System Simulator for Developing Countries) model is a bio-economic farm-household optimisation model, with a first application to Sierra Leone⁽ Reference Louhichi, Gomez y Paloma and Belhouchette ^{105 )}. A household module has been added to the modular structure of FSSIM. Production, and related environmental outcomes, as well as food consumption are outputs of the model. Conditionally on the quality of the data about the environmental impacts associated with each production activity, and about food consumption and associated nutritional intakes entered and generated out of the model, such a model could assess the farming practices best suited to improve different sets of nutrition and/or biodiversity indicators and the associated trade-offs. In a normative approach, this approach could help define optimal combinations of activities and resulting diets. In a more positive approach, it could identify through simulation analysis the factors more likely to help attain some of these optimal combinations.