Yield Gap Analysis:Implications for Research and Policy

Kenneth G. CassmanHeuermann Professor of Agronomy, andDirector, NE Center for Energy Sciences

University of Nebraska—Lincolnwww.ncesr.unl.edu

Drier savanna

Moist savanna

Humid forest

Midaltitude savanna

What do these have in common?

• Ensuring global food security in face of climate change

• Land use and indirect land use change– Impact of climate change on agriculture and

vica versa• Competition between food and biofuel• Conservation of natural ecosystems and

biodiversity

EACH DEPENDS ON RATE OF GAIN IN CROP YIELDS ON EXISTING FARM LAND

• Will business as usual meet projected global food demand in 2050?

Global area for cereal crops declining

48% increase in cereal production needed by 2050 (40 yr) = 1.2% yr-1 of current average yield, or…..

1% annual exponential growth rate

• If business as usual won’t do it, what is needed to change course?

Year1960 1970 1980 1990 2000 2010

Gra

in h

arve

sted

are

a (M

ha)

400

500

600

700

800

Global Cereal Area Trends, 1966-2006

Total cereal area(maize, rice, wheat, sorghum, millet, barley, oats)

19661966--19801980+4.6 +4.6 MhaMha yy--11

19811981--20062006--1.7 1.7 MhaMha yy--11

maize + rice + wheat area

19661966--19801980+3.9 +3.9 MhaMha yy--11

19811981--20062006+0.5 +0.5 MhaMha yy--11

Urban-industrial expansion onto prime farmland at the periphery of Kunming (+6 million), the capital of Yunnan Province, China,

Year1966 1976 1986 1996 2006

Gra

in Y

ield

(kg

ha-1

)

1000

2000

3000

4000

5000

Maize Yieldy = 2260 + 62.5xr2 = 0.94

Rice Yieldy = 2097 + 53.5xr2 = 0.98

Wheat Yieldy = 1373 + 40.1xr2 = 0.97

Global Cereal Yield Trends, 1966-2006

THESE RATES OF INCREASE ARE NOT FAST ENOUGH TO MEET EXPECTED DEMAND ON EXISTING FARM LAND! source: FAOSTAT

+2.8%

+1.3%

2010 2020 2030 2040 2050

Glo

bal A

vera

ge M

aize

Yie

ld (k

g ha

-1)

4500

5000

5500

6000

6500

7000

7500

Rel

ativ

e R

ate

of G

ain

(% a

vera

ge y

ield

)

0.8

0.9

1.0

1.1

1.2

1.3

1.4Global Maize Grain Yield kg ha-1 Relative Rate of Gain

Tyranny of constant rate of yield gain:Decreasing relative rate of gain

100

150

200

250

300

1965 1975 1985 1995 2005

YEAR

IRR

IGA

TED

AR

EA (M

ha)

10.0

12.5

15.0

17.5

20.0

Irrigated Area

% of total cultivated area

Global Irrigated Area and as a % of Total Cultivated Land Area, 1966-2004

Irrigated systems occupied 18% of cultivated land area but produced 40% of human food supply

Decreasing water supply in all major irrigated areas

Yet, irrigated agriculture produces 40% of global food supply on just 18% of the cropped area.

Year

1960 1970 1980 1990 2000 2010

Cor

n gr

ain

yiel

d (b

u ac

-1)

60

80

100

120

140

160

180

linear rate of gain = 1.86 bu ac-1 yr-1

double- to single-crosshybrids

Expansion of irrigated area

Conservation tilliage and soil testing

Increased N fertilizer rates

Improved balance in N,P,K fertilization

Precision planters

Transgenic (Bt)insect resistance

Electronicauto-steer

Multi-location hybrid testing in1000s of on-farm strip trials

Integrated pest management

USA Corn Yield Trends, 1966-2009(powerful support from science and technology)

From NCGA website, 8 May 2008: http://www.ncga.com/PDFs/NCGA%20Presentation%20on%20Food%20and%20Fuel%205-7-08.pdf

Bottom Line on Yield Trends• Expansion of crop area limited by lack of good

quality arable soils and concerns about loss of wildlife habitat and biodiversity– USA conservation reserve land– Rainforests and wetlands in Latin America, SE Asia, SSA

• Current rates of gain in crop yields not adequate to meet expected demand for food, feed, fiber, and fuel on existing crop land

• Little scope for increasing irrigated crop area due to competition for water with other sectors

• Little increase in yield potential of maize or rice for the last 30-40 years; yield stagnation in some areas

• Little scope for quantum leap in yields from biotechnology?

• Need for ecological intensification (global sense)

Clearing virgin rain forest in Brazil

Ecological Intensification• How close can average farm yields

come to the yield potential ceiling using crop and soil management practices that conserve natural resources, protect environmental quality, give acceptable rate of economic return?

• Necessary, but not sufficient…..– food affordability and access, markets,

good governance, policies, etc.

•Based on data from 123 fields farmer’s irrigated fields in 2005-2007 seasons, Nebraska, USA.Grassini et al., submitted, Field Crops Res.

High yields and high nitrogen use efficiency (NUE) are possible*

140

160

180

200

220

Continuous maize Soybean-maize

N fe

rtiliz

er ra

te (k

g N

ha-1

)

50

60

70

80

90

Continuous maize Soybean-maize

NU

E (k

g gr

ain

kg-1

N fe

rtiliz

er)

• Higher N rates but also higher NUE compared to U.S. averages, especially under soybean-corn rotation due to higher yields and lower N rate than continuous corn.

• No difference in N fertilizer rate under continuous corn with NT or CT; under soybean-corn rotation, N fertilizer tended to be higher under NT than CT.

• NUE tended to be higher under conventional tillage due to (i) higher yields at the same N rate under continuous corn and (ii) same yield with lower N rate under soybean-corn rotation.

Conventional (CT): disk

Conservation (NT): strip-, ridge-, and no-till.

Tillage system:12.7 13.3 kg ha-1

13.5 13.4

U.S. maize averages

Treatment Yield(t/ha)

Irrigation(cm)

Irrigation Efficiency (t/cm)

100% ET 14.9 22.617.011.4

0

0.6675% ET* 14.7 0.8650% ET 12.8 1.12

Rainfed 5.8 --

More crop per drop

High yields and high water use efficiency are possible: maize in south-central NE, 2-y means 2005-06

*75% ET replacement except 10 d before and 7 d after silking

Ecological Intensification• How close can average farm yields

come to the yield potential ceiling using crop and soil management practices that conserve natural resources, protect environmental quality, give acceptable rate of economic return?

• Necessary, but not sufficient…..– food affordability and access, markets,

good governance, policies, etc.

Maize

China

Brazil

USA - rainfed

USA - irrigated

0

2

4

6

8

10

12

1960 1970 1980 1990 2000 2010Year

Wheat

China

India

Northwest Europe

0

1

2

3

4

5

6

7

8

1960 1970 1980 1990 2000 2010Year

Rice

India

R. KoreaChina

Indonesia

0

1

2

3

4

5

6

7

1960 1970 1980 1990 2000 2010Year

Yiel

d (M

g ha

-1)

Yiel

d (t

ha-1

)

Evidence that average national yields begin to plateau when they reach 70-80% of yield potential

Note yield plateaus in Korea and China for rice, wheat in northwest Europe and India, and maize in China and……..perhaps for irrigated maize in the USA

Cassman, 1999. PNAS, 96: 5952-5959 ?

Cassman et al, 2003, ARER 28: 315-358

Crop Sci. 50:1882–1890 (2010)Genetic Improvement in Winter Wheat Yields in the Great Plains of North

America, 1959–2008Robert A. Graybosch* and C. James Peterson

Abstract……Linear regressions of relative grain yields vs. year over the time period 1984 to 2008, however, showed no statistically significant trend in the SRPN. For the same time period in the NRPN, a statistically significant positive slope of 0.83 was observed, though the coefficient of determination (R2) was only 0.28. …….. and further improvement in the genetic potential for grain yield awaits some new technological or biological advance.

Field Crops Research 119 (2010) 201–212Why are wheat yields stagnating in Europe? A comprehensive data

analysis for FranceNadine Brissona,∗, Philippe Gateb, David Gouacheb, Gilles Charmetc, Francois-

Xavier Ouryc, Frédéric Huarda

AbstractThe last two decades are witnessing a decline in the growth trend of cereal yields in many European countries. The present study analyses yield trends in France using various sources of data: national and regional statistics, scattered trials, results of agroclimatic models using climatic data.

What are reasons for yield stagnation?

• Secret plot among farmers to limit production?

• Climate change?• Unidentified constraints and/or lack of

appropriate crop and soil management technologies to take full advantage of solar radiation and water resources?

• Average farm yields are bumping up against the yield potential (Yp) ceiling (75-85% of Yp)– Size of exploitable yield gap is too small

Yield potential and yield gaps

Yiel

dGap 1

Gap 2

Yield potential Water-limitedyield potential

Actual yield

Other limiting factors:

NutrientsWeedsPestsOthers

CO2Solar radiationTemperature

Growth periodPlant density

Attainable yieldwith availablewater supply:

SoilRainfall

Irrigation

Modified from: van Ittersum and Rabbinge, 1997

When producing crops near the yield potential ceiling, there is a razor-thin margin for error in managing inputs (too much, too little, too early, too late), and also in managing pests because of dynamic interactions

between nutrient status and pest incidence and severity

Interactions between plant nutrient status and disease become more prominent at high yield levels:

For many crops, there is often a positive relationship between leaf N concentration and the prevalence and severity of several foliar diseases. In this case, sheath blight on rice in Vietnam. Hence, a critical need to precisely balance N supply with crop demand without excess or deficiency.

Yield potential and yield gaps

Yiel

dGap 1

Gap 2

Yield potential Water-limitedyield potential

Actual yield

Other limiting factors:

NutrientsWeedsPestsOthers

CO2Solar radiationTemperature

Growth periodPlant density

Attainable yieldwith availablewater supply:

SoilRainfall

Irrigation

75-85% Yp

70-80% Yp-r

Developing a transparent, robust, reproducible protocol to estimate crop yield potential (Yp) in a given

field, region, state, or nation

PhD. Project---Justin van Wart

Rainfed Maize area and NOAA weather stations with +20 yr data (daily Tmax/Tmin, rainfall, dewpoint, wind speed)

Rank of NOAA weather stations by density of harvested rainfed maize area

Select reference weather stations (RWS) with greatest maize area density within 100-km buffers, with minimum overlap, until >50% of U.S maize area is covered. Weight Yp estimates for each RWS buffer by maize area density to estimate national Yp.

Total area within buffers account for 52% of USA RF maize production

Appropriate crop simulation model(validated against yields that approach Yp)

Complicated systems: Maize, rice, wheat in China

Source: Monfreda et al, 2008

NOAA weather stations with +20 yrs data: daily Tmax, Tmin, precip, dewpoint, wind speed and harvested rice area density in China

Intensive lowland rice systems in AsiaCentral LuzonPhilippines Rice-Rice

Mekong DeltaVietnam

Rice-Rice-Rice

West JavaIndonesia Rice-Rice

Red River D.Vietnam

Rice-Rice-Maize

SouthernChina

Rice-Rice(hybrid)

J F M A M J J A S O N DMonth

DS rice WS rice

DS rice WS rice

WS rice DS rice

Spring-rice Summer-rice Maize

Early-rice Late-rice

Cauvery DeltaSouth India

Rice-Rice-Pulses

Pulses WS rice DS rice

Indo-GangeticPlain & SC China

Rice-Wheat WS riceWheat

WS rice

Central JavaIndonesia

Rice-Rice-Maize Rice Maize Rice

Province Cropping season Crop establish Seeding rate* Hills per m2 Seeding date Transplanting date Flowering date Maturity date Area (ha) Variety matchHubei Early season Transplanting 25 266,667 IR72

Direct seeding 45 (H), 90 (I) 266,667 IR72 Middle season Transplanting 21 533,333 2You725

Direct seeding 45 (H), 90 (I) 800,000 2You725 Late season Transplanting 22.5 200,000 IR72

Hunan Early season Transplanting 25 1,133,333 IR72 Direct seeding 30 (H), 60 (I) 466,667 IR72

Middle season Transplanting 22.5 500,000 2You725 Single late season Transplanting 18.4 366,667 2You725 Double late season Transplanting 18.4 1,600,000 IR72

Anhui Early season Transplanting 37.5 80,000 IR72 Direct seeding 75 (I) 186,667 IR72

Middle season Transplanting 22.5 1,466,667 2You501 Direct seeding 45 (I) 200,000 2You501

Late season Transplanting 37.5 300,000 IR72 Jiangxi Early season Transplanting 30 1,333,333 IR72

Direct seeding 37.5 (H), 60 (I) 133,333 IR72 Middle season Transplanting 27 333,333 2You501 Late season Transplanting 27 1,400,000 IR72

Guangdong Early season Transplanting 22.5 1,000,000 IR72 Late season Transplanting 22.5 1,000,000 IR72

Sichuan Single season Transplanting 16 2,666,667 2You501 Guangxi Early season Transplanting 30 1,000,000 IR72

Middle season Transplanting 28.5 133,333 2You501 Late season Transplanting 27 1,000,000 IR72

Fujian Early season Transplanting 22.5 233,333 IR72 Middle season Transplanting 18 533,333 IR72 Late season Transplanting 19.5 233,333 IR72

Jiangsu Single season Transplanting 25.5 1,200,000 Wuxiangjing 9 Direct seeding 60 (I) 800,000 Wuxiangjing 9

Jilin Single season Transplanting 22.5 733,333 Jin Dao 305 Heilongjiang Single season Transplanting 24 2,600,000 Jin Dao 305 Liaoning Single season Transplanting 22.5 666,667 Jin Dao 305 Henan Single season Transplanting 20.5 633,333 XD90247 Zhejiang Early season Transplanting 30 93,333 IR72

Direct seeding 75 (I) 40,000 IR72 Middle season Transplanting 20 333,333 Wuxiangjing 9

Direct seeding 52.5 (I) 333,333 Wuxiangjing 9 Late season Transplanting 22.5 200,000 IR72

Yunan Single season (japonica) Transplanting 55.5 600,000 Jin Dao 305 Single season (indica) Transplanting 55.5 333,333 2You501 Upland rice Direct seeding 75 (I) 100,000 HD297

Guizhou Single season Transplanting 15 733,333 2You725 Hainan Early season Transplanting 33 166,667 IR72

Late season Transplanting 33 213,333 IR72 *kg/ha, H=hybrid and I=inbred 29,146,667

Major rice cropping systems in China: 17 provinces, 44 systems (94% of total rice area)

Simulat

e rice

Yp for each

of these s

ystems

Selected reference weather stations (RWS) and 100-km buffer zones over geospatial distribution of harvested rice area. Selected buffer zones account for 50% of total rice area in China. Simulation with ORYZA2000,

Once selected, each RWS database undergoes rigorous Q/A vetting and gap filling for missing data, then used in simulation model to estimate Yp

Collaboration with: Kurt Christian KersebaumInst. of Landscape Systems Analysis, Muncheberg, Germany

Wheat area density (green) and reference weather stations with 100 km buffers in Germany covering 74% of total harvested area

Country Crop Total harvested area, Mha

% of total area in Ypestimate

Ya†(year)

Yp

50%6.6

(2008)

10.1(2009)

12.4(2009)

8.1 (2008)

52%

8.8

13.2

15.160%

74% 9.5

Ya/Yp(%)

China Irrigated rice 29.5 75%

77%

82%

85%

USA Rainfedmaize 28.7

USA Irrigated maize 3.6

Germany Rainfedwheat 3.2

Production-weighted yield potential (Yp) estimates based on current crop management compared to average national yield (Ya), and coverage of crop area included in Yp estimates.

†US data from USDA NASS, 2009; China rice and German wheat data from FAO, 2008.

How high can average farm yields go?

• Best current estimate:– 75-85% for irrigated crops– 70-80% for rainfed crops

• Actual threshold depends on:– Relative prices for outputs versus inputs– Crop sensitivity to lodging, diseases and

insect pests, other stresses at high yield levels, and probability of intense storms

• Rice more sensitive than maize (lower ceiling)• Wheat varies: southern (India) > northern

(northern Europe)

Conclusions

• Crop yield potential (Yp) can be estimated using methods that are transparent, robust, and reproducible based on weather data, geospatial distribution of crop area, current crop management (plant date and maturity, plant population) and appropriate crop models

• Average farm yields begin to stagnate when they reach 70-85% of Yp

• Yield gap analysis depends on accurate estimate of Yp

Need for Yield Potential--Yield Gap Atlas

• Interpret historical yield trends and yield plateaus (predict yield plateaus?)

• Prioritize research to ensure global food security in the face of climate change– Identify areas with largest unexploited yield gaps,

identify constraints, close yield gaps through ecological intensification

• Estimate agricultural land requirements and indirect land use change (and associated GHG emissions) as influenced by policies– e.g. competition between food and biofuel