SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

!""#$#(*-./&0#

Supplementary Materials – Figures

"#

!""#$#%&'()*+,#

Figure S1. Extent of Global Agricultural Lands. This map illustrates the global extent of croplands (green) and pastures (brown), as estimated from satellite- and census-based data by
Ramankutty et al.1. According to U.N. FAO statistics, croplands currently extend over 1.53 billion hectares (~12% of the Earth’s land surface, not counting Greenland and Antarctica), while pastures cover another 3.38 billion hectares (~26% of global land). Altogether, agriculture occupies ~38% of the Earth’s terrestrial surface, emerging as the largest use, by far, of land on the planet1,2.

1 WWW.NATURE.COM/NATURE | 1

RESEARCH SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

Fig S2a

1.8

1.6

YIELD RATIO (1985=1.0)

GARLIC

NATURAL RUBBER

COFFEE
MAIZE

ALFALFA

RAPESEED

1.4
COTTON

CUCUMBERS

SOYBEAN
WHEAT

RICE
ONIONS

CASSAVA

OATS

1.2

TOMATOES
SUGARCANE

BARLEY
POTATOES

SUNFLOWER SEEDS

0.4

0.6

0.8

1.0
1.0

1.2

1.4

1.6

HARVESTED AREA RATIO (1985=1.0)
1.8

2.0

2.2

SORGHUM

0.8

CEREALS

PULSES

OIL CROPS

SUGAR CROPS

FIBER CROPS

FODDER/FORAGE

ROOTS/TUBERS

FRUITS VEGETABLES & MELONS

NUTS

OTHERS

Figure S2. Trends in Global Crop Production, 1985–2005. In Figure S2a, we illustrate recent changes in yield and harvested area for 174 crops. The vertical axis shows changes in yield, expressed as a ratio of yields reported in 2005 and 1985. The horizontal axis reports relative changes in harvested area between 1985 and 2005. The size of the circle is based on each crop’s harvested area in 2005, while the color corresponds to major crop groupings. We see that crops show changes in total production through changing harvested area (moving left or right), changing average yields per hectare (moving up or down), or both. The dotted curve divides the figure into two regions: Crops above the curve experienced increases in total production from 1985 to 2005 while production of crops below the curve declined.

2 WWW.NATURE.COM/NATURE | 2

RESEARCH SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

Fig S2b

Figure S2b shows a detailed map of yield trends (tonnes/ha/year2) for maize for 1985–2005.
The plot shows statistically significant (p < 0.1) trends based on a linear regression of estimated annual yield values between 1985 and 2005. The data used in this calculation are based on
Monfreda et al.3, extended with additional data to cover the entire period.

3 WWW.NATURE.COM/NATURE | 3

RESEARCH SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

Fig S3

Figure S3. Ratio of Current Agricultural Yields to the Historical Carbon Debt of the
World’s Croplands. Here we consider the trade-off between growing more food through agricultural expansion and the emissions of carbon dioxide into the atmosphere from clearing additional land for crops. West et al.4 reported that tropical lands typically provide average crop yields ~50% lower than those in temperate regions – with the notable exception of oil palm, sugarcane, and South American soybeans – yet release nearly two times more carbon for each unit of land cleared. The ratio of low yields to high carbon losses illustrates the difficult trade-offs of many tropical areas and highlights the environmental dangers of relying on tropical cropland expansion to meet future food demands.

4 WWW.NATURE.COM/NATURE | 4

RESEARCH SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

Fig S4a

Fig S4b

5 WWW.NATURE.COM/NATURE | 5

RESEARCH SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

Figure S4. Global Yield Gap Analysis. In Figure S4a, we show the global patterns of “yield attainment” for maize – the ratio of yields reported for any given location compared to the nearmaximum (95th percentile) yields reported for maize, controlling for global variations in climate and soil conditions (adapted from methods of Licker et al.5). For a given location, a ratio of 50% shows that crop yields are only reaching half of their potential compared with other regions with the same climatic conditions and soils.
Figure S4b shows which factors most limit maize production – nutrients, water, or crop yield ceilings associated with today’s genetics and seed quality. These limiting factors are quantified using simple relationships between agricultural inputs and yield (see Supplemental Information).
In much of the world, the lack of nutrients and water are key limiting factors, whereas in regions of high productivity yields are likely limited by crop genetics.

Fig S5

Figure S5. Irrigation Use Efficiency Across the Globe. Irrigation is one of our best tools for improving crop yields. However, the use and yield benefits of irrigation water are not distributed evenly across the globe. Here we show irrigation water required per kilocalorie of crop yield
(irrigation water requirements and yields of irrigated crops from Siebert and Döll6.
Use of irrigation water varies greatly across the world: the 16 staple crops analyzed here (barley, cassava, groundnut, maize, millet, potato, oil palm, rapeseed, rice, rye, sorghum, soybean, sugarbeet, sugarcane, sunflower, and wheat) require an average of ~0.3 liters of irrigation per kilocalorie of production. In this figure, we see that even higher water use (over 1 liter per kilocalorie) is required in northern India and portions of the Middle East.

6 WWW.NATURE.COM/NATURE | 6

RESEARCH SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

Fig S6a

Fig S6b

Fig S6c

Fig S6e

Fig S6f

Figure S6. Nutrient Applications, Nutrient Use Efficiency, and Excess Nutrients on the
Globe. Building on recent geospatial datasets and analyses of crop production and nutrient cycling (Monfreda et al.3; Potter et al.7; Liu et al.8; MacDonald et al.9) and utilising updated

7 WWW.NATURE.COM/NATURE | 7

RESEARCH SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

fertilizer and manure datasets we illustrate global patterns of nutrient inputs (Figure S6a,b), nutrient use efficiency (yield per unit nutrient input, Figure S6c,d), and estimated levels of excess nutrients (Figure S6e,f).
This analysis shows that there are “hot spots” of low nutrient use efficiency (Figure S6c,d) and large volumes of excess nutrients (Figure S6e,f). Nutrient excesses are especially large in
China, Northern India, USA, and Western Europe. Furthermore, 10% of the world’s croplands account for 32% of the global nitrogen surplus and 40% of the global phosphorus surplus.

8 WWW.NATURE.COM/NATURE | 8

RESEARCH SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

Fig S7a

Fig S7b

9 WWW.NATURE.COM/NATURE | 9

RESEARCH SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

Figure S7. Differences Between Intrinsic and Delivered Food Production. Here we compare global crop yields for 16 staple crops (barley, cassava, groundnut, maize, millet, potato, oil palm, rapeseed, rice, rye, sorghum, soybean, sugarbeet, sugarcane, sunflower, and wheat) in terms of their intrinsic food production (Figure S7a, calories that would be available if all crops were consumed by humans directly) and their delivered food production (Figure S7b, calories available based on today’s allocation of crops to food, animal feed, and other products, assuming standard conversion factors).

10 WWW.NATURE.COM/NATURE | 10

RESEARCH SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

Supplementary Materials – Methods
Geospatial Yield Data
National and sub-national cropland area, maize harvested area and production information was collected for the spatial units delineated by Monfreda et al.3 from crop census reports, agricultural yearbooks and FAOSTAT data2. We then combined these data with spatial maps of cropland areas from Monfreda et al.3, to put the estimates on a 5 minute latitude-longitude spatial grid (approximately 9 km by 9 km at the equator).
We then averaged the harvested area and production numbers for each 5 minute grid cells to generate 7-year averaged harvested area and production estimates for ~1985 to ~2005 in 5year time steps. Yield was estimated as the ratio of production and harvested area. Finally, we linearly regressed the yields from circa 1985 to 2005 to determine the trends of maize yields at
5 min spatial resolution.
Yield Gaps and Limiting Factors Calculations
To calculate yield gaps, we build on the work of Licker et al.5 and group yield variations from
Monfreda et al.3 into 100 equal-area “bins” of similar climate (annual precipitation and growing degree-day) characteristics. Crop-specific potential yields for the yield gap analyses are defined as the 95th percentile yield within a climate bin. Comparing observed yields to potential yields defines the yield gap or “potential yield attainment” of each grid cell.
Management practices that limit maize yield increases (Figure 6b) are calculated using simple climate-specific input-yield models. For each climate bin, we quantify the saturating relationship
(Mitscherlich-Baule functional form10) between yields and nitrogen fertilizer application, phosphate fertilizer application, potash fertilizer application (fertilizer data from Nathaniel D.
Mueller, personal communication July 6, 2011), and percent irrigated area (Portmann et al.11) using a nonlinear least-squares algorithm. Yield plateaus (Ymax) for the Mitscherlich-Baule response are defined as the 98th percentile yields in a bin. The y-intercepts for nutrient response
(defined by bN, bP, and bK) are tied to the 2nd percentile yields in a bin, while y-intercepts for irrigation are allowed to vary with rainfed yield potentials in each climate bin. Following von
Liebig’s “law of the minimum”10, yield can be limited by any one of the inputs (Eqn. S1).

(

)

Y =min Ymax (1" bNe(−cNN) ),Ymax (1" bPe(−cPP) ),Ymax (1" bKe(−cK K) ),Ymax (1" bIRRe(−cIRRIRR) )
!! mod

Eqn. S1

Using our empirically derived input-yield relationships, we model yields and assess what factors
€
– nutrients, nutrients and irrigation, irrigation, or yield ceiling (90% of bin-specific potential yields) – limit a 50% yield increase within each climate bin.
Nutrient Inputs and Nutrient Balance Calculations
For Figure 8, applied nitrogen and phosphorus fertilizer rates are expressed in terms of kg per hectare of land (cropland and non-agriculture land). Total nutrient consumption is calculated as the sum of crop-specific chemical-fertilizer application rates (Nathaniel D. Mueller, personal

11 WWW.NATURE.COM/NATURE | 11

RESEARCH SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

communication July 6, 2011) multiplied across crop areas from Monfreda et al.3. The sum of nutrient consumption across all crops is harmonized with FAO national-level nutrient consumption statistics.
Manure application rates are calculated from the manure production dataset of Potter et al.7. We assume stable-produced manure is available to be applied to croplands, and that stableproduced manure is produced in proportion to cultivated agricultural land in a grid cell (the ratio of cropland area / cropland and pasture area from Ramankutty et al.1). Available manure is then subject to cropland application rates of 66% in Western Europe and Canada, 87% in the U.S., and 90% elsewhere (following Liu et al.8). Manure nitrogen loss from volatilization is estimated as a constant 36% loss (following Bouwman et al.12).
Excess nutrients are calculated as a simple mass balance described in West et al., which is similar to recent efforts to estimate nutrient balances (Liu et al.8, MacDonald et al.9). Chemical fertilizer and manure data sets are inputs for both nitrogen and phosphorous models. The nitrogen has additional inputs from nitrogen deposition (Dentener et al.13) and nitrogen fixation by legumes. Nitrogen fixation is scaled as a function of yields using a range of Nfix values from the literature (Smil14) and yields (Monfreda et al.3). Nutrient removal from harvest is estimated as the product of yield (Monfreda et al.3), dry fraction (Monfreda et al.3), and nutrient density
(USDA15).
Diet Gap Calculations
The “diet gap” is the difference between calories produced and calories that become available for human consumption. We analyze sixteen staple crops: barley, cassava, groundnut, maize, millet, potato, oil palm, rapeseed, rice, rye, sorghum, soybean, sugarbeet, sugarcane, sunflower, and wheat. The proportion of crop production allocated to food, feed, and other products is determined using FAOSTAT data for crop production, use and trade2. To account for trade, crop production is separated into production that was consumed domestically, and production that was exported. Crop production that was consumed domestically is multiplied by country specific crop use proportions. Crop production that was exported is multiplied by global average crop use proportions.
Delivered food calories are the sum of the calories that directly go to the food system, as well as the calories that have been converted from animal feed to meat. The calories available from crop production directly allocated to food are the product of food production in tonnes and average calorie content of the given crop, as determined by FAOSTAT Food Balance Sheets2.
We use grain to edible meat conversions to convert feed to animal protein16. The feed to animal protein calorie conversion is dependent on the density of cattle, chicken and pig meat produced within a country2.

12 WWW.NATURE.COM/NATURE | 12

RESEARCH SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

Supplementary Materials – Tables
Table S1. Changes in Global Agricultural Land Between 1985 to 2005

Global
North America

Global agriculture changes 1985 to 2005
(Million hectares and % change)
Numbers may not add up exactly due to rounding
Cropland
Pasture
Agricultural
Ha
% change
Ha
% change
Ha
% change
35.89
2.41
117.78
3.61
153.67
3.23

Latin America
Europe-Central Asia
Africa
Oceania

-0.88
1.44
-3.04
3.08
-0.10

-3.30
23.13
19.80
19.93
-40.20

-0.10
0.71
0.61
0.61
-1.23

-16.42
44.54
-25.54
65.83
-41.67

-0.35
0.94
-0.54
1.38
-0.88

Asia

28.51

1.91

98.42

3.01

126.93

2.67

East Asia
Southeast Asia
South Asia

7.06
15.04
3.27

0.47
1.01
0.22

22.49
0.51
-9.07

0.69
0.02
-0.28

29.55
15.55
-5.80

0.62
0.33
-0.12

West Asia

-13.12
21.41
-45.34
45.90
-1.48

3.14

0.21

84.49

2.59

87.63

1.84

13 WWW.NATURE.COM/NATURE | 13

RESEARCH SUPPLEMENTARY INFORMATION

doi:10.1038/nature10452

References
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

Ramankutty, N., Evan, A. T., Monfreda, C. & Foley, J. A. Farming the planet: 1. Geographic distribution of global agricultural lands in the year 2000. Global Biogeochemical Cycles 22,
GB1003 (2008).
FAOSTAT. Available at http://faostat.fao.org/ (Accessed March, 2011).
Monfreda, C., Ramankutty, N. & Foley, J. A. Farming the planet: 2. Geographic distribution of crop areas, yields, physiological types, and net primary production in the year 2000. Global
Biogeochemical Cycles 22, 1-19 (2008).
West, P. C. et al. Trading carbon for food: Global comparison of carbon stocks vs. crop yields on agricultural land. Proceedings of the National Academy of Sciences 107, 19645 (2010).
Licker, R. et al. Mind the gap: how do climate and agricultural management explain the 'yield gap' of croplands around? Global Ecology and Biogeography (2010).
Siebert, S. & Döll, P. Quantifying blue and green virtual water contents in global crop production as well as potential production losses without irrigation. Journal of Hydrology 384, 198-217 (2010).
Potter, P., Ramankutty, N., Bennett, E. M. & Donner, S. D. Characterizing the spatial patterns of global fertilizer application and manure production. Earth Interactions 14, 1-22 (2010).
Liu, J. et al. A high-resolution assessment on global nitrogen flows in cropland. Proceedings of the National Academy of Sciences 107, 8035 (2010).
MacDonald, G. K., Bennett, E. M., Potter, P. A. & Ramankutty, N. Agronomic phosphorus imbalances across the world's croplands. Proceedings of the National Academy of Sciences 108,
3086 (2011).
Paris, Q. The von Liebig hypothesis. American Journal of Agricultural Economics 74, 1019-1028
(1992).
Portmann, F. T., Siebert, S. & Döll, P. MIRCA 2000: global monthly irrigated and rainfed crop areas around the year 2000: a new high-resolution data set for agricultural and hydrological modeling. Global Biogeochemical Cycles 24, GB1011 (2010).
Bouwman, A. et al. A global high-resolution emission inventory for ammonia. Global biogeochemical cycles 11, 561-587 (1997).
Dentener, F. et al. Nitrogen and sulfur deposition on regional and global scales: A multimodel evaluation. Global biogeochemical cycles 20, GB4003 (2006).
Smil, V. Nitrogen in crop production: An account of global flows. Global biogeochemical cycles 13,
647-662 (1999).
U.S.D.A. Agricultural Waste Field Management Handbook. Report nr 210-VI, NEH-651.
Smil, V. Nitrogen and food production: proteins for human diets. AMBIO: A Journal of the Human
Environment 31, 126-131 (2002).

14 WWW.NATURE.COM/NATURE | 14