Different assumptions were made to project the key variables for the global agricultural and food systems. Here we provide an overview of the choices underlying the projections of diets, crop yields and cultivable land availabilities. More details are given in the technical report of the study, where the assumptions relating to the other variables are described (feed efficiencies, biofuel feedstocks, cropping intensity).

Two alternative assumptions for diets

The demand for agricultural commodities for human consumption is driven by diets and population dynamics. UN population projections are used to estimate the population of the 21 regions in 2050. For the diets, the simulations were produced using two contrasting hypotheses. The first envisages so-called “trend-based” diets. It extends past trends, with a continuation of current diets in developed countries and a continued nutritional transition in developing countries. According to this hypothesis, daily caloric needs would not be met worldwide, with sub-Saharan Africa in particular remaining well below the nutritional recommendations in 2050. The second assumption, reflecting a “healthy” diet, projects a shift in consumption towards diets that more closely align with WHO’s nutritional recommendations. In this case, the diets adopted in the different regions of the world tend to converge while retaining regional specificities.

Two sets of projections for yields

In this study, the focus was on characterising the uncertainties related to crop yields in 2050, under the combined impacts of technical evolutions (plant breeding, technological advances, agricultural practices, etc.) and climate change.

It is difficult to predict the dynamics of technical evolutions by 2050 (plant breeding, technological advances, agricultural practices, etc.) and the extent of their effects on yield growth. Assuming moderate technical developments, yields could increase by at least 20% to 40% depending on the region, with increases of up to 80% to 90% in sub-Saharan Africa if more sustained technical developments are achieved.

Moreover, the increase in atmospheric CO2 concentration associated with climate change is favourable to photosynthesis, and therefore to plant yields, provided that plants’ water and nitrogen needs are met. This so-called “CO2 effect” could offset the deleterious effects of the rise in average temperatures and the decrease in rainfall affecting certain regions. It would thus lead to increases in average yields of between 1% and 8% depending on the region. Conversely, if the CO2 effect does not occur in the field, climate change could depress yields.

To reflect these uncertainties regarding the dynamics of technical developments and the CO2 effect, the simulations use two sets of projections named “high ” and “moderate” yield growth which define a range of possible variations of yields by 2050.

Different options to define the land constraint

The land constraint corresponds to the maximum cultivable area in each region. The definition chosen for cultivable areas is important in the GlobAgri-AE2050 model because they are the limiting factor for potential expansion of cultivated areas. The cultivable land availabilities were projected based on the Global Agroecological Zones procedure implemented by IIASA and the FAO, factoring in the impacts of climate change on the soils’ agro-climatic potential. The definition used in the reference scenarios assumes that an area is cultivable if its agro-climatic potential can accommodate a crop (annual or perennial) no matter how it is currently utilised. In these assumptions, global cultivable acreage would remain relatively stable by 2050 compared with “2010”, hovering at around 5 billion ha, as losses in the two Latin American regions, the three African regions, and Oceania would be offset by gains, mostly in the former USSR, USA-Canada and, to a much lesser extent, China and the “rest of Europe” region.

As an alternative to this fairly flexible definition, a more restrictive acceptation of the maximum cultivable area considers that the areas wooded in 2010 cannot be used for crops in 2050. This second assumption also excludes areas likely to be urbanized by 2050.

Finally, an even more restrictive definition of land constraint is to prohibit the expansion of cultivated areas beyond their “2010” levels.

9 scenarios have been produced in addition to the 2010 reference year.

The adopted geographic breakdown attempts to balance parsimony by distinguishing a limited number of regions corresponding to one or more countries, with relevance respecting to the objectives assigned to the study. This leads to the distinction of eight representative European regions that reflect the diversity of agricultural production conditions. By adopting such a geographic breakdown, this work differs from many other studies, which consider Europe as a single block. Europe is taken in the sense of the European Union extended to Switzerland, Norway, Serbia, and the Western Balkan countries. France, Germany, Poland, and the United Kingdom are distinguished due to their size and agricultural and geopolitical specificities. In order to avoid multiplying the number of regions, Spain and Italy, two other major agricultural nations, are included in Southern Europe, a region that also includes Portugal and the Balkan countries except for Bulgaria and Serbia. Romania constitutes the basis of the Eastern Europe region, which also includes Bulgaria, Hungary, and Serbia. Central Europe includes Austria, the Czech Republic, and Slovakia, to which Switzerland has been added. Finally, the rest of Europe is mainly composed of the countries of Northern Europe supplemented by Ireland, Belgium, the Netherlands, and Luxembourg.

The model establishes a balance (in tonnes) for 31 agri-food products and 5 forage products. Considered products are reported below.