Relocating croplands could dramatically reduce the environmental impact of global food production

We use the notation in Tab1.

Current crop production and areas P
_i(x), H
_i(x)

We used 5-arc minute maps of fresh-weight production P_i(x) (Mgyear¹) and cropping area H_i(x) (ha) of 25 major crops (Table2) in the year 2010³⁷. These are the most recent global data, both crop-specific and spatially explicit.⁷⁵. Separate maps were created for irrigated and rainfall-fed crops, which allowed us to estimate the worldwide percentage of irrigated land at 21%.

Agro-ecologically feasible yields ({widehat{Y}}_{i}(x))

5-arc-minute maps of the agroecologically feasible dry-weight yield (Mgha), were used. ¹Year¹) of the same 25 crops on worldwide potential growing areas (Supplementary Movie3) from the GAEZ v4 model, which incorporates thermal, moisture, agro-climatic, soil, and terrain conditions⁴². These yield estimates are based on rainfed water supply, which is without additional irrigation. They are available for current climatic conditions, and assume a CO₂Fertilisation effect, for four future climate scenarios (20712100 period), that correspond to representative concentration pathways. (RCPs)⁷⁶Simulation by HadGEM2-ES model⁷⁷. Potential rainfed yield estimates for current climate conditions were available for a lower- and higher-input crop level. This represented, respectively, subsistence based organic farming systems and advanced fully mechanised production using high yielding crop varieties and the best fertiliser and insecticide application⁴². We also considered potential yields that represent a medium input management scenario. This was determined by the median of the relevant low-and high-input yields. Future potential yields were only available for high-input management levels. We considered a total 175 (=253 future+254 present) potential yield maps. Potential dry-weight yields were converted into fresh-weight yields. ({widehat{Y}}_{i}(x))Using crop-specific conversion factors^42,78.

GAEZ v4’s current and future potential rainfed yields were simulated based upon daily weather data. They also account for short-term events, such as heat waves, frost, and dry spells.⁴². However, these estimates represent average annual yields for 30-year periods. Thus, although irrigation may not be required in certain dry years, it is possible to avoid the need for irrigation in cropping areas identified by our approach.⁷⁹Ad hoc irrigation is a practical option to preserve productivity in times of drought. Climate change is projected to affect different regions.^80,81.

Carbon impact C
_i(x)

An earlier approach was used⁸The carbon impact of crop production C_i(xIn a 5-arc minute grid cell, the difference in potential natural carbon stocks and cropland-specific stocks of carbon was calculated. Each stock was given by the sums of the relevant soil and vegetation-specific carbon. The difference in carbon stored in the potential vegetation is the measure of the change in vegetation carbon stock due to land conversion. This information is available as a 5-arc minute global map⁸(SupplementaryFig.1a) Carbon stored in the crops. We used available estimates^8,78. For soil, there are no spatially explicit global estimates for soil organic carbon (SOC), as a result of land cover changes. We chose to use a simple approach that is consistent with estimates across large spatial scalas rather than a complex spatially specific model which would have made it difficult to make robust predictions in light of the limited empirical data. An earlier approach was used.⁸… and supported by empirical meta analyses^{82,83,84,85,86}We assumed that a 25% decrease in the potential natural SOC would result from the conversion of natural habitats to cropland. We used a 5-arc-minute global map showing pre-agricultural SOC stock levels to determine the latter.⁷ (Supplementary Fig.1b). The total local carbon impact (Mgcha) is thus calculated.¹) of the production of crop iIn the grid cell xAs estimated

Where ({{C}}_{{{{{{rm{potential}}}}}},{{{{{rm{vegetation}}}}}}}(x))And ({C}_{{{{{{rm{potential}}}}}},{{{{{rm{SOC}}}}}}}(x))denotes the natural carbon stocks in the soil and plants x, respectively, and ({C}_{{{{{{rm{crop}}}}}}}(i))denotes the crop’s carbon stocks i(All in Mg Cha¹). The design of the approach allows us estimate the carbon impact of the conversion from natural habitat to cropland, regardless of whether an area has been cultivated.

We did not include greenhouse gases from other sources in our analysis, such as methane from rice paddies or nitrous emission from fertilised soils.⁸⁷. These ongoing emissions are not one-off, as they are created during the production process. This would mean that the amount of these emissions in a scenario where the total production of each crop is constant is approximately the same as that associated with the current distribution. We also didn’t consider emissions related to transport. However, they have been shown to be smaller than other food chain emissions.⁸⁸It is not well correlated with distance traveled by agricultural goods.⁸⁹.

Impact of biodiversity B
_i(x)

Analogous to the carbon approach, we calculated the biodiversity impact of crop cultivation. B_i(xIn a 5-arc minute grid cell, we define local biodiversity as the difference between the natural habitat and the cropland. Our main analysis consisted of quantifying local biodiversity in terms range rarity (given the sum of inverse species size; see below) of mammals, birds, amphibians. Range rarity is a biodiversity measure that is especially relevant for conservation planning.^{39,90,91,92,93}Protection of endemic species and in particular³⁹. We also considered biodiversity in terms species richness as part of a supplementary analysis.

5-arc-minute global maps were used to show the range rarity, species richness, and habitat cover of mammals, birds, amphibians, and cropland.⁹⁴. These data were generated using a methodology³⁸This combines species-specific extents (spatial envelopes for species beyond geographic limits)⁴⁰) and habitat preferences (lists of land cover categories in which species can live⁹⁵), both available for all mammals, birds, and amphibians^96,97This map shows the potential natural biomes worldwide.⁴⁴In order to determine which species would live in a grid cell that has natural habitat conditions, The habitat preferences also include information about species ability to survive in croplands. This allows us to identify species that can tolerate a local conversion from natural habitat to cropland. The species richness impact of cropping in a grid cell can then be calculated as the estimated number of species that will be lost locally if natural habitat is converted into cropland. Instead of weighing all species equally in a grid, the range rarity effect in a grid cells is calculated as a sum of the inverse natural range sizes for the species locally lost when natural environment is converted. Thus, increased weight can be attributed to range-restricted and endangered species.^40,41.

The approach, similar to carbon, allows us to estimate biodiversity impact of crop production in both cultivated and uncultivated areas.

Potentially available land for agriculture V(x)

We defined the area V(x) (ha) potentially available for crop production in a given grid cell x, the area not currently covered in water bodies⁴²Due to soil and terrain limitations, the land is unsuitable⁴²Built-up land (urban areas infrastructure, roads)¹, pasture lands¹We did not include these crops in our analysis.³⁷Protected areas⁴² (Supplementary Fig.1e). If crop production is partially relocated and croplands are not removed, the appropriate retained areas are subtracted from any potentially available area.

Transnational relocation: optimal

First, we will consider the scenario where all current croplands are moved across national borders based upon current climate (Fig.3a – dark blue line). For each crop iEach grid cell xWe then determined the local (i.e., grid cell-specific) area ({widehat{H}}_{i}(x))(ha) On which crop iCell culture is used to grow it xSo that each crop has a total production iThe current production is equal to the current production, and the environmental impact on the environment is minimal. Denoting by

The current crop production worldwide iAny solution. ({widehat{H}}_{i}(x))Must comply with the equality constraints

Requires that the total production of each crop after relocation must be equal to the current. Also, the inequality constraints must be met.

Ensure that the local sum of cropping area is not greater than the locally available area V(x) (see above). These constraints allow us to identify the global cropland configuration that minimizes the biodiversity or total carbon impact.

respectively. We can also reduce a combined biodiversity and carbon impact measure. Furthermore, we can examine trade-offs between minimising each one of these impacts by considering the weighted objective function.

What is the weighting parameter The range is between 0 to 1.