Published October 15, 2014
Crop Ecology & Physiology
Yield and Yield Gaps in Central U.S. Corn Production Systems
D. B. Egli* and J. L. Hatfield
ABSTRACT
The magnitude of yield gaps (YGs) (potential yield – farmer yield) provides some indication of the prospects for increasing
crop yield. Quantile regression analysis was applied to county corn (Zea mays L.) yields (1972–2011) from Kentucky, Iowa, and
Nebraska (irrigated) (total of 115 counties) to estimate the attainable potential yield (APY) (yield in the most favorable environments in the 40-yr record) for each county. The YG for each year was the difference between the APY and the county yield.
Potential productivity (40-yr mean county yield) varied substantially within and among states (579–1001 g m–2). The mean
relative yield gap (RYG) varied from 9 to 24% of APY and decreased linearly (p < 0.0001) as the mean county yield increased for
each state. Large YGs were partially related to very low yields that occurred in some years and may reflect the ability of the soil
to supply water to the crop; other soil or management factors may also be involved since irrigation did not reduce the RYG in
Nebraska to zero. The highest mean county APY was 30% (Kentucky), 22% (Iowa) and 14% (Nebraska) higher than the lowest
county APY. This second YG occurred in favorable environments. Reducing the first YG in Kentucky and Iowa will probably
require irrigation. Reducing the second YG may be more difficult if it is related to intractable soil characteristics. These results
suggest that soil conditions may play an important role in determining the size of YGs in Midwestern corn production systems.
Yield gaps, the difference between potential yield and
farmer yield, provide an estimate of the yield improvement
possible without any change in potential yield (Cassman et al.,
2003; Lobell et al., 2009). Large YGs imply that there is more
potential for yield improvement. Evans (1993, p. 292) and
Evans and Fischer (1999) defined potential yield as “the yield of
a cultivar when grown in environments to which it is adapted;
with nutrients and water not limiting; and with pests, diseases, weeds, lodging, and other stresses effectively controlled”.
Potential yield represents the ability of a crop to convert solar
radiation to dry matter and economic yield in a specific environment in the absence of stress.
The practical definition of potential yield may not be as simple as implied by Evans (1993, p. 292) and Evans and Fischer
(1999). Potential yield is location specific making it necessary
to aggregate local estimates spatially and temporally to produce
average estimates for larger areas (van Ittersum et al., 2013).
Cultural practices (cultivar, planting date, plant population,
etc.) affect yield making if necessary to specify the use of best
management practices when defining potential yield (Loomis
D.B. Egli, Dep. of Plant and Soil Sciences, Univ. of Kentucky, Lexington
KY 40546-0312. J.L. Hatfield, USDA-ARS National Lab. for Agriculture
and the Environment, Ames, IA 50011. Published with the approval of the
Director of the Kentucky Agricultural Experiment Station as Paper 14-06029. Received 7 July 2014. *Corresponding author ([email protected]).
Published in Agron. J. 106:2248–2254 (2014)
doi:10.2134/agronj14.0348
Copyright © 2014 by the American Society of Agronomy, 5585 Guilford
Road, Madison, WI 53711. All rights reserved. No part of this periodical
may be reproduced or transmitted in any form or by any means, electronic or
mechanical, including photocopying, recording, or any information storage and
retrieval system, without permission in writing from the publisher.
2248
and Connor, 1992, p. 10; van Ittersum et al., 2013). It is not yet
clear if potential yield depends on soil type or if the elimination of water and nutrient stress decouples potential yield from
soil characteristics (van Ittersum et al., 2013; Egli and Hatfield,
2014). Loomis and Connor (1992, p. 10) and Lobell et al.
(2009) introduced the concept of water limited potential yield
(i.e., potential yield in rain-fed environments) which may be
more relevant when irrigation is not an option. Water limited
potential yield would clearly be affected by soil characteristics.
Estimates of potential yield can be obtained from record
farmer yields (i.e., winners of yield contests), maximum
farmer yields, yields from well-managed agricultural experiment station experiments, or estimates from well-validated
crop simulation models (Lobell et al., 2009). Global estimates
are sometimes based on maximum farmer yield in a specific
climate zone (Licker et al., 2010), but such estimates may not
account for the variation in climate conditions or cropping
systems at smaller scales (van Ittersum et al., 2013).
Yield gaps for corn, summarized by Lobell et al. (2009),
ranged from 44 (YG as a percent of potential yield) to 84%
with large YGs often associated with low farm yields. van
Wart’s et al. (2013) estimates of national corn yield gaps in the
United States (data from 1986–2008) ranged from 26% (rain
fed) to 22% (irrigated) of potential yield. Site-specific YGs for
individual corn fields varied substantially in western Kenya
(water limited potential yield), but were much less variable
among irrigated fields in Nebraska. The average YG in western
Abbreviations: APY, attainable potential yield; NCCPI, national
commodity crop productivity index; NCCPI-AG, national commodity crop
productivity index-agriculture; RYG, relative yield gap; YG, yield gap.
A g ro n o my J o u r n a l • Vo l u m e 10 6 , I s s u e 6 • 2 014
Kenya was 68% of the water limited potential yield compared
with only 11% of the potential yield in Nebraska (van Ittersum
et al., 2013).
Lobell et al. (2009) concluded that YGs for wheat (Triticum
spp.) and rice (Oryza sativa L.) from major cropping systems
of the world varied from 20 to 80% of the potential yield at
most locations. Larger YGs were often found in low technology
systems in comparison to high-technology irrigated systems.
Licker et al. (2010) reported similar variation on a global scale
and their estimates sometimes followed political boundaries,
apparently reflecting the effect of the political system on the
availability of technology.
The YG for a given location will depend on the farmer yield
(high or low-input production system) and the method used
to estimate potential yield (Liang et al., 2011; Meng et al.,
2013). Lobell et al. (2009) suggested that, for a specific farm
yield, the YG will probably be largest for modeled potential
yield, followed by experiment yields and finally by maximum
farmer yield. The plethora of potential yield estimates complicates evaluation of the variation in YGs and devising remedial
strategies. The larger the geographical scale of the estimates of
potential and farmer yield, the more difficult it is to evaluate
the possibilities of reducing the YG (van Ittersum et al., 2013).
Our previous evaluation of soybean [Glycine max (L.) Merr.]
YGs focused on production systems with variable farmer yields
(Kentucky, Iowa, and Nebraska–Irrigated), but only minor differences in the availability of technology, the economic climate
and skills of the producers (Egli and Hatfield, 2014). Our
objective in that study, which was based on county yield data
from 1972 to 2011, was to evaluate the magnitude, temporal
trends, and the fundamental basis of the difference between
the APY and county yield (i.e., YG). This manuscript reports
the extension of that work to corn, a somewhat dissimilar crop
that produces higher yield and generally requires more intense
management (Egli, 2008a), using the same counties and techniques used with soybean (Egli and Hatfield, 2014).
MATERIALS AND METHODS
County corn yield estimates from 1972 through 2011 were
obtained from the National Agricultural Statistics Service website (http//www.nass.usda.gov/, accessed 2 July 2014). Three
states–Kentucky, Iowa, and Nebraska were selected to provide
a range in productivity in close geographic proximity and with
similar access to production technology. Only irrigated production in Nebraska was included to provide a relatively low-stress
high-yield environment. Counties with less than roughly 4048
harvested hectares (10,000 acres) in 2011 were excluded from
the analysis to avoid introducing possible artifacts resulting
from small production areas. Thirty-two counties in Kentucky
and 36 in Nebraska met the criterion. All 99 counties in Iowa
exceeded the minimum harvested area, so 47 were chosen at
random to create a manageable data set. In Kentucky, 10 of
the 32 counties had less than 4048 harvested hectares in 1972,
but by 1976 only two counties were less than the minimum.
All 47 Iowa counties exceeded the minimum harvested area in
1972. Three counties in Nebraska had less than 4048 irrigated
ha in 1972, but by 1977 all counties exceeded the minimum
harvested area. The counties included in this analysis were
essentially the same counties used previously in an evaluation
Fig. 1. Estimates of the attainable potential yield (APY) for the highest
(Phelps County, Nebraska) and lowest (Marshall County, Kentucky)
yielding counties of the 115 counties included in this analysis. The
dotted line represents the attainable potential yield (APY), calculated
from the quantile regression at the 95% percentile, from 1972 to 2011.
of soybean yield trends (Egli, 2008b) and soybean YGs (Egli
and Hatfield, 2014).
A quantile regression analysis set at the 95th percentile
[PROC QUANTREG in SAS (SAS for Windows, v. 9.3, SAS
Institute, Cary, NC)] (Webb, 1972; Cade and Noon, 2003)
was used to estimate yield in the most favorable environments
that occurred in each county from 1972 to 2011. This estimate
represents the aggregate accomplishment of all of the farmers
in each county in favorable conditions so it is similar to the
maximum farmer yield estimate of potential yield described by
Lobell et al. (2009).
The YG (g m–2) was calculated by subtracting the county
yield for each year from the APY for that year calculated from
the quantile regression equation as described previously (Egli
and Hatfield, 2014). The RYG was calculated by dividing the
absolute YG by APY and multiplying by 100.
The national commodity crop productivity index (NCCPI)
for each county was obtained through use of gSSURGO data
and NCCPI values by soil map unit using the 2013 gSSURO
data for the conterminous United States which were resampled and mosaiced to create a national soils raster at 100
m resolution (Soil Survey Staff, 2012). The national soils raster
attribute table was joined with the NCCPI values by soil map
unit (Soil Survey Staff, 2013). Mean NCCPI values were calculated for all soils within a county using ArcGIS: ZonalStatisticsAsTable, for each of the selected counties in Kentucky, Iowa,
and Nebraska (ESRI 2012. ArcGIS Desktop: Release 10.1,
sp1. Redlands, CA:Esri.). To constrain the NCCPI values only
to agricultural land within each county, mean NCCPI values
were calculated using a mask of specified land use classes from
the 2009 NASS Cropland Data Layer (NASS, 2009) combined with ArcGIS: ZonalStatisticsAsTable for (agricultural
land/row crop agriculture) in each of the selected counties in
the three states (ESRI 2012. ArcGIS Desktop Release 10.1, sp1.
Redlands, CA:Esri.).
The statistical package in Sigma Plot (Sigma Plot 12.0, SSPS
Inc., Chicago, IL) was used for all regression analysis.
Agronomy Journal • Volume 106, Issue 6 • 2014
2249
Fig. 2. Relationship between the mean relative yield gap and mean
county yield (1972–2011) for each state. The regression equations
relating relative yield gap to county yield are: Kentucky Y = 37.501–
0.0276X, r 2 = 0.21** (dotted line); Iowa Y = 48.144 – 0.0385X, r 2 =
0.56*** (dashed line); Nebraska (Irrigated [Irr.]) Y = 43.904– 0.0348X,
r 2 = 0.42*** (solid line).
RESULTS
The results of the quantile regression analysis for the highest
and lowest yielding counties in the 115 counties evaluated are
illustrated in Fig. 1. The average county yield (1972–2011)
in Phelps County, Nebraska, was 1001 g m–2 compared with
579 g m–2 for Marshall County, Kentucky. The APY increased
with time in all 115 counties (data not shown), in step with
the general trend for increasing county yields. There was a
substantial range in the rate of increase of APY among counties
with the maximum rate (16.8 g m–2 yr–1, Cuming County,
Nebraska) 2.7 times the minimum rate (6.3 g m–2 yr–1, Calloway County, Kentucky), but only Iowa showed a significant
association between rate of growth in APY and average county
yield (r = 0.41*, n = 47).
Nebraska (Irrigated) had the highest mean county yield
and Kentucky the lowest (26% less than Nebraska), with
Iowa intermediate between the extremes (Table 1). There
was, however, substantial overlap among the three states.
There was nearly a twofold difference in average yield
(579–1001 g m–2) among the 115 counties. The mean APY in
Kentucky was 20% less than Nebraska (irrigated). The mean
APY was closely associated with mean county yields (r =
0.85**, 0.90** and 0.83** for Kentucky, Iowa, and Nebraska,
respectively) as expected given the method of estimating APY.
The mean RYG was smallest in Nebraska (9–16%) and there
was considerable overlap between Kentucky (14–24%) and
Iowa (11–26%) (Fig. 2). The mean RYG decreased significantly (p <
0.01 or 0.0001) within each state as the mean county yield increased
so the largest RYGs occurred in the less-productive counties.
Fig. 3. Mean relative yield gap as a function of time for the four highestand four lowest-yielding counties (1972–2011) in (A) Kentucky, (B)
Iowa, and (C) Nebraska (Irrigated, Irr.).
There was substantial variation of the mean RYG with time
for the four highest- and the four-lowest yielding counties in
each state (Fig. 3). The average RYG for the high-yield counties
(roughly 15% in Kentucky and Iowa and 9% in Nebraska) was
smaller than in the lowest-yielding counties (roughly 21% in
Kentucky and Iowa and 14% in Nebraska). The largest RYGs
were more than 50% in Kentucky and Iowa, but they never
exceeded 40% in Nebraska. Large and small RYGs usually
occurred in the same year in both high- and low-yield counties.
The year-to-year variation was substantially less in the lowstress irrigated environment in Nebraska (Fig. 3C).
There was no obvious temporal trend of RYG in Kentucky
where linear regression of RYG vs. time identified a significant
(p < 0.10) relationship in only 5 of 32 counties (two with a
positive slope and three with a negative slope). Linear regression was significant (p < 0.10) in 10 of 47 counties in Iowa (all
with a negative slope) and 16 of 36 counties in Nebraska (all
with a negative slope).
Table 1. Corn production statistics for Kentucky, Iowa, and Nebraska (irrigated). Mean yield (1972–2011).
State
Kentucky
Iowa
Nebraska–irrigated
2250
No. of counties
32
47
36
County yield
Attainable potential yield
Mean
Range
Mean
Range
—————————————————- g m–2 ————————-——————
680
837
786–579
946–727
810
968
887–662
1058–773
919
1040
1001–868
1121–986
Coefficient of variation
County
APY
————- % ————
7.3
6.5
8.6
6.0
3.6
2.8
Agronomy Journal • Volume 106, Issue 6 • 2014
Fig. 4. Relationship between the mean National Commodity Crop
Productivity Index- Agriculture (NCCPI-AG) for each county and the
mean county yield (1972–2011) for Kentucky and Iowa. The index does
not apply to irrigated crops, so data from Nebraska–Irrigated were not
included in the analysis.
Mean county yield in Kentucky and Iowa was associated with the national commodity crop productivity indexagriculture (NCCPI-AG) (linear regression was significant
at p < 0.0001, r 2 = 0.58) (Fig. 4). There was very little overlap
between the two states, but there was substantial variation in
mean yield that was not accounted for by the NCCPI-AG.
The NCCPI-AG was designed for non-irrigated production
(NRCS, 2008), so we did not relate it to yield in Nebraska.
DISCUSSION
The substantial range in productivity (579–1001 g m–2 ,
based on the 40-yr mean county yields, Table 1) in this study
provided the variation needed to evaluate the relationship
between productivity and YGs. This variation in county yield
could be related to environmental conditions, soil characteristics, availability of technology or the farmer’s ability and
willingness to deploy the available technology. Using the 40-yr
mean yield to characterize productivity should minimize
effects of random short-term variation in weather conditions.
There was some variation in mean summer (June, July, and
August) temperature among states (Kentucky was warmest,
followed by Nebraska and Iowa), but the differences were
only 1 to 2°C (Egli and Hatfield, 2014) and probably did not
contribute much to differences in yield. Kentucky had slightly
lower normal summer precipitation (10%) than Iowa (Egli and
Hatfield, 2014), which could have contributed to the generally
lower yield in Kentucky, but irrigation should have minimized
any effects of the lower precipitation in Nebraska. There was
not much variation in normal summer precipitation within
each state, but there were some north–south gradients in
temperature in Iowa and Nebraska (Egli and Hatfield, 2014).
It seems unlikely, however, that the substantial variation in the
40-yr mean yield within each state was related to variation in
the aboveground environment.
These three states are clustered in a highly developed, sophisticated, high-yielding production system in central United
States, making it unlikely that variation in the availability of
technology or political and economic conditions that could
influence the deployment of technology (Hazell and Wood,
2008; Keating et al., 2010; Licker et al., 2010; Laborte et al.,
2012) contributed significantly to the variation in productivity
within and among states. Improved technology is a primary
driver of yield improvement (see a recent review in Fischer et
al. (2014), p. 192–194), so the generally high yield and upward
trend during the study period (Egli, 2008a, 2008b; Hatfield,
2010) suggests that many farmers in each state were utilizing
the best technology available. It is not possible, however, to
eliminate the possibility that some farmers in low-yield counties found it un-economic to employ the same level of technology as in high-yield counties.
Our estimates of APY were based on county yields occurring
in the most favorable conditions in the 40-yr period, consequently they represent the collective effort of farmers producing from 3117 to 84,413 ha (7700–208,500 acres) in 2011 with
current technology. As such, APY represents the maximum
attainable yield estimated within the limits of the available
technology and the farmer’s skills (how well do they apply the
technology) (Connor et al., 2011, p. 11) when environmental
conditions were favorable.
Our APY is probably less than the true potential yield (Evans,
1993, p. 292; Evans and Fischer, 1999) or modeled potential
yield (Lobell et al., 2009; van Ittersum et al., 2013). Our APY
is no doubt limited by the failure of all farmers in a county to
deploy the “best” management practices or by the incomplete
elimination of all stress by the favorable environments (e.g., water
could still be limiting APY). The APY in 2011 for the county
with the highest average yield in Nebraska, however, was only
7% less than the average potential yield (1998–2008) of three
irrigated counties in central Nebraska (1520 g m–2) estimated
with a simulation model (Grassini et al., 2011). In the same
comparison in Iowa, our APY was 17% less than Fischer et al.’s
(2014) estimate of potential yield (1500 g m–2). The APY of the
lowest yielding county in 2011 was 20 and 30% less than the estimates of potential yield for Nebraska and Iowa, respectively. In
a very practical sense, APY describes what is possible by a group
of farmers in midwestern corn production systems in favorable
environmental conditions, so it is closer to the maximum achievable yield described by Licker et al. (2010) or estimates based on
maximum farmer yield (Lobell et al., 2009).
The difference between APY and county yield provides a
YG estimate (Lobell et al., 2009) that may be smaller than
estimates based on modeled potential yield (Lobell et al., 2009;
Liang et al., 2011; Laborte et al., 2012; Meng et al., 2013).
These YGs may be closer to the exploitable YG (i.e., the difference between the yield of the best farmers and country yield)
described by Connor et al. (2011, p. 11). Two advantages of our
approach are that it is based on yields that actually occurred,
not theoretical yields and it provides an estimate of APY for
each of the 40 yr in each county, making it possible to evaluate
spatial and temporal YG variation.
Although changes in production technology (improved
hybrids and management practices) were mostly responsible
for the upward trend in county yields (Egli, 2008a; Fischer et
al., 2014), it is unlikely that changes in technology deployment
were responsible for the peak yields that provided the basis for
the APY estimate. The peak yields were most likely a result of shortterm variation in the environment, which in Kentucky and Iowa was
probably a result of variation in rainfall (Hatfield et al., 2011).
Agronomy Journal • Volume 106, Issue 6 • 2014
2251
Fig. 5. The relationship between the mean relative yield gaps (RYGs)
for corn and soybean for Kentucky, Iowa, and Nebraska-Irrigated (Irr.).
Soybean data were taken from Egli and Hatfield (2014).
The mean RYG for the 115 counties varied between approximately 9 and 24% (Fig. 2) which is essentially the same range
reported for soybean in the same counties (Egli and Hatfield,
2014). The estimates of van Wart et al. (2013) using simulated
potential yield for irrigated and rain-fed corn in the United
States were near our maximum levels (23 and 27% of potential yield for rain-fed and irrigated production, respectively).
Grassini et al. (2011) reported an average YG of 21% for three
irrigated counties in central Nebraska. Fischer et al. (2014,
p. 195) concluded that the YG in Iowa was 36% of the 2010
potential yield. Other estimates are larger than those reported
here. For example, RYGs ranged from 44 to 84% of potential
yield in a summary from the United States, Latin America, and
Africa (Lobell et al., 2009). Liang et al. (2011) found that the
average total RYG in a wheat (Triticum aestivum L.)–maize
cropping system in China was 58% of potential yield. Meng et
al. (2013) used a crop simulation model to estimate potential
yield and reported RYGs ranging from 44 to 52% of potential
yield for maize in China. Kassie et al. (2014) reported average RYGs, based on simulated water limited yields, of 71% in the
central Rift Valley of Ethiopia. These large RYGs no doubt reflect
lower farmer yields and estimates of potential yield that are closer to
the true potential yield.
The inverse relationship between the mean RYG and mean
county yield was similar to the response reported for soybean
(Egli and Hatfield, 2014) and wheat (Lu and Fan, 2013). Large
mean YGs in some years were the result of very low yields,
but these occasional large YGs were generally less evident in
higher-yielding counties (Fig. 1). Low summer rainfall would
have a much greater negative effect on yield in counties where
soils with low water holding capacity were common. The
relationship between soil characteristics including water holding capacity is well documented (see Hatfield, 2012, and the
references there in) so it is possible that some of the variation
in YGs among counties in these states was related to the water
holding capacity of the soil as suggested by Egli and Hatfield
(2014) for soybean. The consistently small YGs in the irrigated
production in Nebraska is consistent with this argument as
soil water holding capacity would be less important when
water was not as limiting. The linear relationship between
2252
mean county yield (Fig. 5) and NCCPI-AG (which includes
soil characteristics) is also consistent with the argument that
variation in productivity is related to soil characteristics. Other
soil characteristics related to YGs include topographic position
(Boling et al., 2010) and soil nutrient status (Affholder et al.,
2013). It must be noted, however, that the RYG was not zero in
irrigated production in Nebraska and it varied among counties
(Fig. 2), which could simply be an indication that water was
not the only factor in the plant’s aboveground or belowground
environment that influenced the YG in this state and/or that it
is difficult (or not economically feasible) to manage irrigation
on substantial land areas to eliminate all water stress.
There were no consistent temporal trends in the RYG for the
four highest- and lowest-yielding counties in Kentucky. Linear
regression of the RYG vs. time was significant (p < 0.10) in
only 5 of 32 counties (two with a positive slope and three with a
negative slope). There were, however, significant negative trends
in 10 of 47 counties in Iowa and 16 of 36 counties in Nebraska.
There were no temporal trends for soybean in Kentucky in
agreement with the results for corn, but corn showed more
evidence for a decline than soybean in Iowa (Egli and Hatfield,
2014) while Nebraska had essentially the same trend for soybean as for corn (16 of 34 counties significant negative slopes).
Hatfield (2010), using a similar approach to the one used here,
found no temporal trends in YG for corn using average state
yields from Iowa, Illinois, Indiana, Kansas, South Dakota, and
Texas. An increase in environmental stress, as expected from climate change (Hatfield et al., 2011) could increase the RYG, but
the development of stress tolerant hybrids (Tollenaar and Lee,
2002; Campos et al., 2006; Fischer et al., 2014, p. 192–195) or
an increase in precipitation (Hatfield et al., 2011) could decrease
the RYG over time. These effects cannot be separated with this
data set, but it is possible that any increase in environmental
stress was cancelled by improved management practices or by
the development of hybrids better able to tolerate stress. Interestingly, the clearest trend for a decline in RYG with time of
both maize and soybean (Egli and Hatfield, 2014) occurred in
Nebraska, the least stressful environment evaluated as shown by
higher yields, smaller yield gaps, and less year-to-year variation.
Perhaps improved management practices (including irrigation
scheduling) made a larger contribution to reducing stress in this
relatively low stress environment. The large year-to-year variation in RYG, especially in Kentucky and Iowa (Fig. 3), however,
makes any evaluation of temporal trends difficult.
Yield gaps in many production systems can be reduced by
increasing inputs or by relatively simple changes in management practices (Tittonell et al., 2008; Liang et al., 2011;
Pasuquin et. al., 2014), but reducing YGs related to soil characteristics and soil water holding capacity may be more difficult.
Irrigation would no doubt reduce the YGs in Kentucky and
Iowa if a sustainable source of water is available and the practice
is economically feasible. Increasing soil water holding capacity
significantly, however, may be a more formidable challenge.
The yield increase, on average, if the YG (mean APY – mean
county yield, Table 1) was reduced to zero was 23% of the mean
county yield in Kentucky, 20% in Iowa, and 13% in Nebraska.
Irrigation would probably capture only a proportion of this
increase in Kentucky and Iowa, given that the YG was not zero
in irrigated production in Nebraska.
Agronomy Journal • Volume 106, Issue 6 • 2014
There was substantial variation in mean APY (1972–2011)
among counties within each state (range in APY, Table 1).
This variation occurred in the most favorable environments in
the 40-yr period. This variation among counties represents a
second YG (i.e., the difference between the maximum average
(1972–2011) county APY and the average APY of any other
county). The maximum county APY was 30% larger than the
minimum (calculated from the ranges in mean APY in Table
1) in Kentucky, 37% in Iowa (only 22% if the second lowest
APY was used), but only 14% in Nebraska. Since this second
YG occurred in favorable environments, rainfall and soil water
availability probably did not play a major role in determining
its magnitude, so this second YG may not respond to irrigation
in Kentucky and Iowa. Variation in the application of technology or soil characteristics not directly related to soil water
storage may be more important. Reducing this second YG may
be possible to the extent that management or amendable soil
limitations are involved, but, it will be difficult if the gap is
related to more intractable soil characteristics.
We previously applied these same techniques to soybean
using essentially the same counties and time span (1972–2011)
(Egli and Hatfield, 2014). These two crops have few characteristics in common. Soybean, a C3 legume, produces a seed
high in oil and protein compared with the high starch kernel
produced by C4 maize (Egli, 1998, p. 3). Soybean produces
lower yield (ratio of corn to soybean yield averages roughly 2.8
to 3.0, Egli, 2008a), usually has lower economic value per unit
harvested area and lower production costs per hectare (www.
extension.iastate.edu, accessed 2 July 2014), and requires less
management and fewer inputs than corn. In spite of these
considerable differences, the mean RYG increased in step with
corn (Fig. 5, r = 0.63**). The general association between them
suggests that there is some consistency in the response of the
two crops to the factors controlling the YG. Yield of the two
crops have been increasing at nearly the same rate in Iowa
(since 1972) and Kentucky (since 1982) (Egli, 2008a), which is
consistent with their similarities in RYGs. The scatter around
the one-to-one line suggests, however, that the response to the
variation among counties was not entirely the same for the two
crops. There was more scatter in the data from Iowa with some
counties exhibiting conspicuously lower RYGs for soybean
than for corn and others with much larger RYGs for soybean
than for corn. Apparently the characteristics that define the
differences between these two crops had some effect on their
response to stress in certain counties, but the basis of this differential response is not known.
CONCLUSIONS
Corn YG analysis of county yields from 1972 to 2011 in
three states with highly sophisticated production systems
that differed in productivity produced average RYGs (APY –
county yield) ranging from 9 to 24% of the APY. The largest
YGs occurred in the least productive counties (i.e., those with
the lowest average yields). The large average YGs were related to
very low yield in some years, which could be partially related to
soil characteristics. The variation in mean APY among counties
within a state defined a second YG (highest mean APY– APY
of any other county), a gap occurring in favorable environmental conditions. The maximum APY was 30% larger than
the minimum APY in Kentucky, 22% in Iowa, and 10% in
Nebraska. Eliminating both of these YGs in low-yield counties could increase mean yield by nearly 60% in Kentucky and
Iowa, but only 24% in irrigated production in Nebraska.
The magnitude of the first YG and possibly the second could
be partially related to variation in soil characteristics. The first
YG could probably be reduced by irrigation in Kentucky and
Iowa, but the second YG may be more intractable. The magnitude of the RYGs for corn was similar to those for soybean
(Egli and Hatfield, 2014) from the same counties.
ACKNOWLEDGMENTS
We thank Marcy Rucker, Research Analyst, Dep. of Plant and
Soil Sciences, University of Kentucky for acquiring and analyzing the county yield data; Kristen McQuerry, Research Assistant,
Department of Statistics, University of Kentucky for her help with
the quantile regression analysis; and David James, GIS Specialist, USDAARS, National Laboratory for Agriculture and the Environment, Ames,
IA, for developing the county level NCCPI values used in this manuscript.
REFERENCES
Affholder, F., C. Poeydebat, M. Corbeels, E. Scopel, and P. Tittonell. 2013.
The yield gap of major food crops in the tropics: Assessment and analysis through field surveys and modeling. Field Crops Res. 143:106–118.
doi:10.1016/j.fcr.2012.10.021
Boling, A.A., T.P. Tuong, H. van Keulen, B.A.M. Bouman, H. Suganda, and
J.H.J. Spiertz. 2010. Yield gap of rainfed rice in farmers fields in central
Java, Indonesia. Agric. Syst. 103:307–315. doi:10.1016/j.agsy.2010.02.003
Cade, B.S., and B.R. Noon. 2003. A gentle introduction to quantile regression for ecologists. Front. Ecol. Environ 1:412–420.
doi:10.1890/1540-9295(2003)001[0412:AGITQR]2.0.CO;2
Campos, H., M. Cooper, G.O. Edmeades, C. Loffler, J.R. Schussler, and M.
Ibanez. 2006. Changes in drought tolerance in maize associated with
fifty years of breeding for yield in the U.S. cornbelt. Maydica 51:369–381.
Cassman, K.G., A. Doberman, D.T. Walters, and H. Yang. 2003. Meeting
cereal demand while protecting natural resources and improving environmental quality. Ann. Rev. Environ. Resour. 28: 315–358.
Connor, D.J., R.L. Loomis, and K.G. Cassman. 2011. Crop ecology. Productivity and management in agricultural systems. 2nd ed. Cambridge Univ.
Press, Cambridge, UK.
Egli, D.B. 1998. Seed biology and the yield of grain crops. CAB Int., Wallingford, UK.
Egli, D.B. 2008a. Comparison of corn and soybean yields in the United States:
Historical trends and future prospects. Agron. J. 79:S79–S83.
Egli, D.B. 2008b. Soybean yield trends from 1972 to 2003 in mid-western
USA. Field Crops Res. 106:53–59. doi:10.1016/j.fcr.2007.10.014
Egli, D.B., and J.L. Hatfield. 2014. Yield gaps and yield relationships in central
U.S. soybean production systems. Agron. J. 106:560–566. doi:10.2134/
agronj2013.0364
Evans, L.T. 1993. Crop evolution, adaptation and yield. Cambridge Univ.
Press, Cambridge, UK.
Evans, L.T., and R.A. Fischer. 1999. Yield potential: Its definition, measurement, and significance. Crop Sci. 39:1544–1551. doi:10.2135/
cropsci1999.3961544x
Fischer, T., D. Byerlee, and G. Edmeades. 2014. Crop yields and global food
security. Will yield increases continue to feed the world? ACIAR
Monogr. 158. Australian Centre for Int. Agric. Res., Canberra.
Grassini, P., J. Thorbun, C. Burr, and K.G. Cassman. 2011. High-yield irrigated maize in the western U.S. cornbelt. I. On-farm yield, yield-potential and impact of agronomic practices. Field Crops Res. 120:142–150.
doi:10.1016/j.fcr.2010.09.012
Hatfield, J.L. 2010. Climate impacts on agriculture in the United States: The
value of past observations. In: D. Hillel and C. Rosenzweig, editors,
Handbook of climate change and agroecosystems: Impact, adaptation
and mitigation. Imperial College Press, London, UK. p. 239–253.
Agronomy Journal • Volume 106, Issue 6 • 2014
2253
Hatfield, J.L. 2012. Spatial patterns of water and nitrogen response within
corn production fields. In: G. Aflakpui, editor, Agricultural science.
Intech Publishers, Rijeka, Croatia. p. 73–96.
Hatfield, J.L., K.J. Boote, B.A. Kimball, L.H. Ziska, R.C. Izaurralde, D. Ort et
al. 2011. Climate impacts on agriculture: Implications for crop production. Agron. J. 103:351–370. doi:10.2134/agronj2010.0303
Hazell, P., and S. Wood. 2008. Drivers of change in global agriculture. Philos.
Tran. Royal Soc. B 363:495–515. doi:10.1098/rstb.2007.2166
Kassie, B.T., M.K. van Ittersum, H. Hengsdijk, S. Assang, J. Wolf, and R.P.
Rotter. 2014. Climate-induced yield variability and yield gaps of maize
(Zea mays L.) in the Central Rift Valley of Ethiopia. Field Crops Res.
160:41–53. doi:10.1016/j.fcr.2014.02.010
Keating, B.A., P. Carberry, P.S. Bindraban, S. Asseng, H. Meinke, and J. Dixon.
2010. Eco-efficient agriculture: Concepts, challenges, and opportunities.
Crop Sci. 50:S109–S119. doi:10.2135/cropsci2009.10.0594
Laborte, A.G., K.C.A.J.M. de Brie, E.M.A. Smaling, P.F. Moya, A.A. Boling, and M.K. van Ittersum. 2012. Rice yields and yield gaps in Southeast Asia: Past trends and future outlook. Eur. J. Agron. 36:9–20.
doi:10.1016/j.eja.2011.08.005
Liang, W.-L., P. Carberry, W. Gui-yan, L. Run-hai, L. Hong-zhan, and X. Ai-ping.
2011. Quantifying the yield gap in wheat-maize cropping systems of the Hebei
Plain, China. Field Crops Res. 124:180–185. doi:10.1016/j.fcr.2011.07.010
Licker, R., M. Johnson, J.A. Foley, C. Barford, C.J. Kuckarik, C. Monfreda, and
N. Ramankutty. 2010. Mind the gap: How do climate and agriculture
management explain the “yield gap” of croplands around the world. Glob.
Ecol. Biogeogr. 19:769–782. doi:10.1111/j.1466-8238.2010.00563.x
Lobell, D.B., K.G. Cassman, and C.B. Field. 2009. Crop yield gaps: Their
importance, magnitudes, and causes. Annu. Rev. Environ. Resour.
34:179–204. doi:10.1146/annurev.environ.041008.093740
Loomis, R.S., and D.J. Connor. 1992. Crop ecology: Productivity and management in agricultural systems. Cambridge Univ. Press, Cambridge.
Lu, C., and L. Fan. 2013. Winter wheat potentials and yield gaps in the North
China plain. Field Crops Res. 143:98–105. doi:10.1016/j.fcr.2012.09.015
2254
Meng, Q., P. Hou, X. Chen, Z. Chui, and F. Zhang. 2013. Understanding
production potentials and yield gaps in intensive maize production in
China. Field Crops Res. 143:91–97. doi:10.1016/j.fcr.2012.09.023
NASS. 2009. Published crop-specific data layer. USDA-Natl. Agric. Statistics
Serv. http://assgeodata.gmu.edu/CropScape/ (accessed 2 July 2014).
NRCS. 2008. User guide National Commodity Crop Productivity Index (NCCPI)
Version 1.0. USDA-Natl. Resources Conserv. Serv., Washington, DC.
Pasuquin, J.M., M.F. Pampolino, C. Witt, A. Dobermann, T. Oberthur, M.J.
Fischer, and K. Inubushi. 2014. Closing yield gaps in maize production in
Southeast Asia through site-specific nutrient management. Field Crops
Res. 156:219–230. doi:10.1016/j.fcr.2013.11.016
Soil Survey Staff. 2012. The gridded soil survey geographic (gSSURGO) database for Iowa, Nebraska, and Kentucky. USDA–NRCS, Washington, DC.
Soil Survey Staff. 2013. National value added look up (valu1) table for the gridded soil survey geographic (gSSURGO) database for the United States
and the Territories, Commonwealths, and the island Nations served
by USDA-NRCS. United States Department of Agriculture. Natl.
Resources Conserv. Serv., Washington, DC.
Tittonell, P., B. Vanlaume, M. Corbeels, and K.E. Giller. 2008. Yield gaps,
nutrient use efficiency and response to fertilisers by maize across heterogeneous small-holder farms in western Kenya. Plant Soil 313:19–37.
doi:10.1007/s11104-008-9676-3
Tollenaar, M., and E.A. Lee. 2002. Yield potential, yield stability and
stress tolerance in maize. Field Crops Res. 75:161–169. doi:10.1016/
S0378-4290(02)00024-2
van Ittersum, M.K., K.G. Cassman, P. Grassini, D. Wolf, P. Tittonell, and
Z. Hochman. 2013. Yield gap analysis with local to global relevance-A
review. Field Crops Res. 143:4–17. doi:10.1016/j.fcr.2012.09.009
van Wart, J., K.C. Kersebaum, S. Peng, M. Milner, and K.G. Cassman. 2013.
Estimating crop yield potential at regional to national levels. Field Crops
Res. 143:34–43. doi:10.1016/j.fcr.2012.11.018
Webb, R.A. 1972. The use of boundary lines in the analysis of biological data.
J. Hortic. Sci. 47:309–319.
Agronomy Journal • Volume 106, Issue 6 • 2014