Published March, 2003
Prediction of Cultivar Performance Based on Single- versus Multiple-Year Tests
in Soybean
Weikai Yan and Istvan Rajcan*
ABSTRACT
Because of the omnipresent genotype ⫻ year or genotype ⫻
location ⫻ year interactions in crop performance trials, it is commonly
believed that multiple-year data should be used in selecting cultivars
for the next year. An implicated but rarely tested hypothesis is that
multiple-year data are more predictive than single-year data of cultivar
performance in the next year. Yield data of the 1991 to 2000 Ontario
Soybean Variety Trials in the 2800 Crop Heat Unit (CHU) area were
used to study the power of single-year, multiple-location trials in
predicting cultivar performances in the following year, and to see if
data from multiple-year trials are more predictive. Mixed models were
used to estimate best linear unbiased predictions (BLUP) of tested
genotypes on the basis of single- or multiple-year trials, and the t-sta-
tistic of BLUP (tBLUP) was used as a measure of cultivar perfor-
mance. Results indicated that a single-year, multiple-location trial had
sufficient power for identifying genotypes that would perform well
or poorly in the next year. Two to four years’ data gave only slightly
better predictions of next-year performances than single-year data
but allowed more genotypes to be evaluated conclusively. The tBLUP
of genotype effects based on 2 yr of multiple-location trials should
be used as a basis for soybean cultivar selection and recommendation
in the 2800 CHU area of Ontario.
R egional performance trials are conducted annu-
ally for all major crops throughout the world to
help growers select cultivars for the next year. Such a
decision would be straightforward were there no
genotype ⫻ environment (GE) interaction (Gauch and
Zobel, 1996). GE interaction is, however, almost omni-
present, which complicates the decision making.
GE interaction in multiple-location and multiple-year
trials can be dissected into genotype ⫻ location interac-
tion (GL), genotype ⫻ year interaction (GY), and geno-
type ⫻ location ⫻ year three-way interaction (GLY)
(Comstock and Moll, 1963; Annicchiarico and Perenzin,
1994). Presence of GL within a single year necessitates
multiple-location trials; presence of GY warrants multi-
ple-year trials; and presence of GLY requires both mul-
tiple-year and multiple-location trials. Since GE inter-
action is almost omnipresent, and as yearly variation
is typically the largest source of yield variation, it is
commonly believed that the greater the number of years
a genotype is tested, the more reliable its evaluation
will be. An extended belief is that results based on
more years of performance trials are more predictive
of cultivar performance in the next year, which becomes
a dogma for cultivar recommendation. Thus, it is recom-
mended in regional performance trial reports, almost
without exception, that cultivar selection should be
Dep. of Plant Agriculture, Crop Sci. Bldg., Univ. of Guelph, Guelph,
ON, N1G 2W1, Canada. Received 25 Feb. 2002. *Corresponding author
(irajcan@uoguelph.ca).
Published in Crop Sci. 43:549–555 (2003).
549
based on multiple-location trials across multiple years.
Surprisingly, the hypothesis that multiple-year data give
better prediction of the next year’s performance has
been tested by only a few researchers (Cross and Helm,
1986; Gellner, 1989; Bowman, 1998). On the other hand,
it is not feasible for researchers to make decisions on
the basis of many years of testing because few genotypes
(except the checks) are tested in many years and many
genotypes are withdrawn from the trials if they do not
perform well in the first year.
This project was set up to address the following ques-
tions: (i) what is the predictive power of single-year
multiple-location trials in cultivar selection; (ii) are data
from multiple-year trials more predictive than those
from a single-year trial; and (iii) what is the merit of
using multiple-year data?
MATERIALS AND METHODS
Data Source
Yield data from the 1991 to 2000 Ontario Soybean Variety
Test (OSVT) in the 2800 crop heat unit (CHU) (OMAFRA,
1993) area of Ontario, Canada, were used in this study. The
OSVT is an official annual test conducted by the Ontario Oil
and Protein Seed Crop Committee (OOPSCC) and supported
by Ontario Ministry of Agriculture and Food and the Ontario
Soybean Growers. The OSVT assumes the functions of both
soybean [Glycine max (L.) Merr.] registration trials and per-
formance trials and is conducted across Ontario covering culti-
vars from 2300 to 3400 CHU. The cultivars in the 2800 CHU
test belong approximately to relative maturity groups from
0.4 to 1.5. The 2800 CHU trials included four test locations,
namely, Exeter, St. Pauls, Woodstock, and Winchester.
Geographically, the first three locations are in southwestern
Ontario, whereas Winchester is in eastern Ontario. Although
there were strong crossover GL interactions each year, the
interaction pattern varied considerably across years (Yan and
Rajcan, 2002). Each year, 60 to 113 adapted cultivars or breed-
ing lines from public and/or private breeding programs were
tested. Many more entries were tested in recent years, as
compared with the earlier years (Table 1). Although the same
sets of genotypes were tested at all locations within a single
year, the genotypes varied greatly with the year. In general,
about 50% or more of the entries were removed each year;
and only one cultivar was tested in all 10 yr (Table 1). A total
of 526 genotypes were tested during the 10-yr period. Except
for the check cultivars, which were determined by the
OOPSCC, cultivar sponsors were solely responsible for the
entering of their cultivars into the tests. Also, in some years,
as a result of poor germination, data from only three locations
Abbreviations: BLUP, best linear unbiased prediction; CHU, crop
heat units; E, environment main effect; G, genotype main effect;
GE, genotype ⫻ environment interaction; GL, genotype ⫻ location
interaction; GLY, genotype ⫻ location ⫻ year interaction; GY,
genotype ⫻ year interaction; L, location main effect; tBLUP, t-statis-
tics of BLUP; Y, year main effect; OOPSCC, Ontario Oil and Protein
Seed Crop Committee; OSVT, Ontario Soybean Variety Test.
550 CROP SCIENCE, VOL. 43, MARCH–APRIL 2003

Table 1. Number of genotypes that are commonly tested in the The genotype ⫻ environment interaction (including GL, GY,
2800 Ontario Soybean Variety Test during the period from and GLY) was implicitly stated in the SAS statements and
1991 to 2000. confounded with the experimental error. When analysis was
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 based on single-year data, the model included a fixed effect
for the grand mean and random effects for genotype and
1991 60 34 22 16 9 7 6 3 2 1
1992 64 38 27 15 11 8 3 1 1 location. The genotype ⫻ location interaction was implicitly
1993 89 50 26 16 10 3 2 1 stated and confounded with the experimental error. The op-
1994 89 48 28 15 7 5 4 tion “DFM ⫽ SATTERTH” was used to estimate degrees of
1995 99 49 24 10 7 6 freedom for appropriate hypothesis testing.
1996 90 42 19 13 7
1997 104 43 29 15 Relevant outputs of the above SAS statements include two
1998 113 53 27 aspects: (i) the variance component for each random factor
1999 112 42 (genotype, location, etc.), which gives information on the rela-
2000 112 tive importance of the variation sources and (ii) the estimated
random effect (i.e., BLUP relative to the grand mean) for
were collected. Consequently, the dataset was highly unbal- each level of all random factors. Our interest was in the BLUP
anced, with 83.3% missing cells. of the genotype main effects. The estimated BLUP for a geno-
At each location, a lattice design (before 1998) or a nearest type, under the proposed model, is the predicted yield of the
neighbor design (1998 and later) with four replicates was used. genotype relative to the grand mean. Whether the BLUP of
The experiments were planted according to local practice with a genotype is significantly higher or lower than the grand
planting rate about 50 seeds m⫺2. The harvested plot size was mean is tested by the t-statistics of BLUP (tBLUP), which is
8.25 m2 (four 5.5-m rows with 37.5-cm spacing between rows). the ratio of BLUP to the associated prediction error, which
Mean grain yield was computed in accordance with experimen- is inversely related to the number of data points from which
tal design for each genotype at each location. The analysis the BLUP was estimated. Yan et al. (2002) indicated that
reported here is based on mean values. tBLUP is highly correlated with, and is more informative
than, BLUP per se as it provides an intuitive and convenient
measure of superiority. For two-tail test, a t ⱖ 2.0 indicates
Statistical Analysis superiority over the average, and a t ⱕ ⫺2 indicates inferiority
The following model was used in this analysis: to the average at the 0.05 probability level. For one-tail test,
the threshold |t| value can be reduced to 1.67 for the same
Yijk ⫽ ␮ ⫹ gi ⫹ yj ⫹ lk ⫹ (yl)jk ⫹ eijk level of significance. Based on this threshold level, the geno-
where Yijk is the mean yield of genotype i in year j at location types can be easily classified as superior (t ⱖ 1.67), inferior
k, ␮ is the grand mean, gi is the main effect of genotype i, yj (t ⱕ ⫺1.67), and intermediate (1.67 ⬍ t ⬎ ⫺1.67) groups.
is the main effect of year j, lk is the main effect of location k, The predictive power of single-year, multiple-location trials
and (yl)jk is the interaction between year j and location k. The was measured by the correlation coefficient between the
term eijk is the residual associated with genotype i, year j, and tBLUP estimated from one year and that from the following
location k and consists of genotype ⫻ environment interaction year across genotypes. Similarly, the predictive power of multi-
(including GL, GY, and GLY) confounded with experimental ple-year trials is measured by the correlation coefficient be-
error. All effects except the grand mean are assumed random, tween tBLUP estimated from data from a number of previous
with 0 mean and normally distributed variances. Restricted years and data from the current year across genotypes. For
maximum likelihood was used to estimate the variance compo- example, the predictive power of a 5-yr test for 2000 is mea-
nents of various factors and the random effects, i.e., the best sured by the correlation coefficient between tBLUP based on
linear unbiased predictors (BLUP) of the genotypes. Likeli- data from 1995 to 1999 and that based on data from 2000
hood ratio was used to test the significance of the estimated across the genotypes tested in 2000.
variance components (Littell et al., 1996). The SAS procedure
PROC MIXED (SAS Institute, 1996) was used to obtain the
variance components and the random effects. The SAS state- RESULTS
ments were: Variance Components
PROC MIXED; The location main effect (L) was a much greater
CLASS YEAR LOCATION GENOTYPE; source of variation than genotype main effect (G) and
MODEL YIELD ⫽/DFM ⫽ genotype ⫻ location interaction (GL) (confounded with
experimental error) for all years except 1995 and 1996,
SATTERTH SOLUTION; and GL was always greater than G (Table 2). The large
RANDOM GENOTYPE LOCATION; yearly GL interaction, which causes cultivar rank
changes at different locations (Yan and Rajcan, 2002),
YEAR/SOLUTION;
warrants multiple-location trials for cultivar evaluation.
Since the purpose was to predict the next-year performance When all 10 yr of data were analyzed together, the
of the genotypes, only genotypes tested in the year in which year ⫻ location interaction was found to be the most
the cultivar performance was to be predicted were included important source of variation for soybean yield, ac-
in the analysis. For example, if data from 1991 to 1995 were counting for 55% of the total variation. The second
jointly analyzed to generate predictions for cultivar perfor-
mance in 1996, only genotypes present in 1996 would be in- largest source of variation was the year main effect
cluded in the analysis. (15%), whereas the location effect was only 2% (Table
When analysis was based on multiple-year data, the model 3). The genotype main effect, G, explained 10% of the
included a fixed effect for the grand mean and random effects total yield variation, and GE, confounded with the ex-
for genotype, year, location, and year ⫻ location interaction. perimental error, explained 15% of the total variation.
 YAN & RAJCAN: PREDICTION OF SOYBEAN CULTIVAR PERFORMANCE 551

Table 2. Variance components of genotype (G), location (L), and Table 4. Correlation coefficients (r) between the tBLUP values
genotype ⫻ location (GL) interaction† in individual years. obtained from 1-yr, multiple-location trials and those from the
next year across genotypes in common.
Percentage of total variance
Previous vs. present year R
Year G L GL ⫹ Error
1991 vs. 1992 0.06
1991 20.2 56.9 22.9 1992 vs. 1993 0.47
1992 25.2 61.1 13.7 1993 vs. 1994 0.41
1993 5.6 86.6 7.8 1994 vs. 1995 0.71
1994 19.9 49.6 30.6 1995 vs. 1996 0.47
1995 32.8 26.0 41.2 1996 vs. 1997 0.53
1996 30.8 33.2 36.0 1997 vs. 1998 0.56
1997 15.4 64.4 20.1 1998 vs. 1999 0.61
1998 4.6 86.0 4.8 1999 vs. 2000 0.57
1999 15.9 63.2 20.9
2000 11.8 66.3 21.9
Mean 18.2 59.3 22.0
⫺1.67, scored “⫺1”), and intermediate (1.67 ⬍
† Genotype ⫻ location interaction is confounded with experimental errors. tBLUP ⬎ ⫺1.67, scored “0”) groups. Table 5 displays
only genotypes that were tested for three or more years.
However, the likelihood ratio test indicates that only the The scores of a genotype were largely consistent across
variance components of genotype and year ⫻ location years. There were changes between 1 and 0 or between
interaction were significant (Table 3). The variance 0 and –1, but a switch between 1 (superior) and ⫺1
components of year and location main effects were not (inferior) from one year to the next did not occur except
significant because of relatively small number of degrees for genotypes OAC Libra and RCAT 9002 between
of freedom and the large number of missing cells. 1991 and 1992 (Table 5), and this was because of the
Theoretically, GE should be partitioned into GL, GY, exceptionally low correlation between the two years
and GLY. This partition was not possible in our analysis, (Table 4). In general, if a cultivar performed well (scored
however, because of the large number of missing cells. ‘1’) in one year, it was likely to perform well in the
Therefore, the GE interactions were implicitly con- following year; its performance may be intermediate,
founded with the experimental error. This treatment but it will not be poor. Likewise, if a cultivar performs
was justified by a previous study, which revealed that poorly (scored ‘⫺1’) in one year, it is likely to perform
the yearly GL interaction was largely random and was poorly again in the next year; it may perform better but
not repeated across years (Yan and Rajcan, 2002). Con- not among the best. Therefore, although the correlation
sequently, the observed GE across years was likely a between consecutive years of performance was only
random GLY three-way interaction. Random GE can- modest (Table 4), 1-yr data appeared to have sufficient
not be exploited by dividing the target area into mega- power for selection of superior cultivars and for culling
environments; it must be avoided through testing in of inferior cultivars (Table 5).
multiple environments representative of the target envi-
ronment (Yan and Hunt, 1998; Yan and Rajcan, 2002). Are Multiple-Year Data Better Predictors
Large GLY three-way interaction calls for multiple-year of Cultivar Performance?
and multiple-location trials for reliable cultivar eval-
uation. If multiple-year data predict cultivar performance
better than single-year data, tBLUP values estimated
Predictive Power of a Single-year Trial from multiple-year data should have a larger correlation
coefficient to that estimated in the next year. The analy-
The predictive power of a single-year trial is measured sis did not generate strong support for this assumption,
by the correlation coefficient between cultivar perfor- however. The prediction of cultivar performance for
mance (i.e., tBLUP) in one year and that in the next 2000 by a single-year test (1999) was r ⫽ 0.57; the corre-
year across genotypes. Except for the correlation be- lation increased to 0.63 when it was based on two years’
tween 1991 and 1992, the correlation coefficients ranged data (1998 and 1999). The prediction for performance in
from 0.41 to 0.71 (Table 4), all being highly significant 2000 did not improve by including data from additional
(P ⬍ 0.01). years (Table 6). For four of the seven years (Table 6)
The predictive power of a single-year trial may be 2-yr data gave the best prediction. For predicting culti-
better appreciated when cultivars are divided into supe- var performance in 1995, data from 1994 alone were
rior (tBLUP ⬎ 1.67, scored “1”), inferior (tBLUP ⬍ better predictors than when data from earlier years were
included. In general, 2 yr of data seemed to predict
Table 3. Variance components based on joint analysis of multi- better than 1-yr data but inclusion of data from addi-
ple-year data.
tional years did not substantially improve the prediction.
Percentage
Source df Variance of total Chi-square P
Merit of Using Multiple-Year Data
Genotype 525 31 417 10.6 762 ⬍0.00001
Environment 39 213 857 72.4 The cultivars that scored “1” performed significantly
Location 3 5 968 2 1 0.18 above the grand mean, and those that scored “⫺1” per-
Year 9 44 383 15 2 0.08
Location ⫻ year 27 163 506 55.4 3 934 ⬍0.00001 formed significantly below the grand mean. The prog-
GE (residual) 2 947 46 063 15.6 ress of all other genotypes, with a score of “0,” however,
Total variance 3 511 505 195 100
was not conclusive. These genotypes may have truly
552 CROP SCIENCE, VOL. 43, MARCH–APRIL 2003

Table 5. Genotype scores† obtained for individual years from 1991 to 2000. Only genotypes that were tested in three or more years
are shown.
Genotypes 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
3131207 – – – – – – 0 0 0 –
877 0 0 ⫺1 – – – – – – –
9071 – – 0 0 0 0 0 – – –
9111 0 ⫺1 ⫺1 0 – – – – – –
9132 – – – – – 0 1 0 0 –
9173 – – – – – 0 0 0 – –
91B01 – – – – – – – 0 0 0
940900 – – – 0 1 1 0 – – –
961000 – – – – – – 0 0 0 –
A0868 – – – – – – – 0 0 0
A1553 – – – – – – 1 0 0 0
AMBERLY – – – – – – – 0 0 0
AP1347 – 1 0 0 0 0 0 – – –
ARKO – 1 0 0 – – – – – –
ARVA – – – – – – – 0 0 0
BECK – – – – 0 0 0 – – –
BICENTENNIAL ⫺1 0 0 ⫺1 – – – – – –
BOUNTY – – – – – 0 0 0 0 0
C9304 0 1 0 0 – – – – – –
CM 401028 – – – – – 1 1 1 1 –
CM 91133 – – 0 ⫺1 ⫺1 – – – – –
CM 98006 – – – – – – – 0 0 0
CM 98007 – – – – – – – 0 0 0
CRUSADER 1 0 0 0 – – – – – –
CX145 – – – – – 0 1 0 1 –
FIESTA – – – – – – 0 0 0 0
FS 238111 – – – – – 0 0 1 0 –
FS 271 – – – – 1 0 0 1 – –
FS 4011 – 0 0 0 – – – – – –
FS 415 – – – – 1 0 0 – – –
FS 4738 – – 0 0 – – – – – –
FS 481 – – – – 0 0 0 – – –
FS 484 – – – 0 1 0 – – – –
FS 523 – – – – 0 0 0 – – –
FS607 – – 0 0 0 0 – – – –
FX9643 – – – – – – 0 0 0 –
GL1593 0 1 1 – – – – – – –
HAROSON 0 0 0 0 – – – – – –
HISOY1452 – – – – 1 0 0 – – –
HS 2778 – – – – – – – ⫺1 ⫺1 ⫺1
HS 2888 – – – – – – – 0 0 0
HS 2988 – – – – – – – 0 0 0
J083 – – 0 0 ⫺1 0 – – – –
K10179 – 0 0 ⫺1 ⫺1 – – – – –
K9087 – – 0 0 0 – – – – –
KG60 – 1 0 0 0 ⫺1 – – – –
KG62 – 0 0 0 0 0 0 – – –
M080 – – – – 0 0 0 0 0 0
MAPLE DONOVAN 0 0 0 ⫺1 – – – – – –
MAPLE GLEN ⫺1 0 ⫺1 ⫺1 ⫺1 – – – – –
MARATHON 0 1 1 0 0 0 0 0 0 0
MS 0747 – – – – 0 0 0 0 0 0
OAC 9006 0 1 1 – – – – – – –
OAC 9007 0 1 0 – – – – – – –
OAC 9106 0 1 1 – – – – – – –
OAC 9208 – 1 1 1 – – – – – –
OAC 9306 – – 0 0 0 – – – – –
OAC 9408 – – – 1 ⫺1 0 – – – –
OAC 9706 – – – – – – 0 0 0 –
OAC AUBURN – – – 1 1 0 0 0 ⫺1 0
OAC BAYFIELD – – – 1 0 0 ⫺1 0 0 0
OAC DORADO 0 0 0 0 0 0 0 0 – –
OAC ECLIPSE 0 0 0 0 0 0 ⫺1 – – –
OAC EMBRO – – – – – 0 0 0 – –
OAC EXETER – – – 0 0 0 0 0 ⫺1 0
OAC LIBRA ⫺1 1 0 0 0 ⫺1 – – – –
OAC SHIRE 1 1 1 1 1 0 0 0 – –
OT913 0 0 0 0 – – – – – –
OT921 – – 0 0 – – – – – –
PR9230 – 0 0 0 ⫺1 – – – – –
PRO 280 – – – – – – 0 0 0 –
PS50 – – – 0 0 ⫺1 0 ⫺1 – –
PS63 – – – – – 0 0 0 ⫺1 –
PS73 – – – – – – – 0 1 1
RCAT 9002 1 ⫺1 0 – – – – – – –
RCAT 9201 – 1 0 0 – – – – – –
RCAT 9203 – 1 0 0 – – – – – –
Continued next page.
 YAN & RAJCAN: PREDICTION OF SOYBEAN CULTIVAR PERFORMANCE 553

Table 5. Continued.
Genotypes 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
RCAT 9301 – – – 1 1 0 – – – –
RCAT 9302 – – 0 1 1 0 – – – –
RCAT 9601 – – – – – 1 0 0 – –
RCAT 9702 – – – – – – 0 0 0 –
RCAT 9801 – – – – – – – 0 0 0
RCAT 9802 – – – – – – – 0 0 0
RCAT 9803 – – – – – – – 0 1 0
RCAT BOBCAT – – – – – – 0 0 0 0
RS1493 – 0 0 0 0 0 0 – – –
S 08-80 – – – – – – 1 0 0 1
S 12-49 – – – – 0 0 0 – – –
S 14-H4 – – – – – – 1 0 0 0
S 14-M7 – – – – – – 0 0 ⫺1 0
SECORD – 0 1 0 0 0 – – – –
T8508 1 – 1 0 0 0 1 – 1 –
T8902 0 1 0 0 0 0 0 – – –
T9212 – – 0 ⫺1 0 – – – – –
T9214 – – 0 0 0 – – – – –
T9217 – – 0 0 0 – – – – –
T9310 – – – 1 0 0 – – – –
T9311 – – – 0 0 0 – – – –
T9312 – – – 1 0 0 – – – –
T9313 – – – ⫺1 ⫺1 ⫺1 – – – –
TALON 0 0 ⫺1 0 0 – – – – –
† A score of 1 means significantly better than the grand mean; ⫺1 means significantly poorer than the grand mean, and 0 means not significantly different
from the grand mean.

yielded similar to the grand mean, or they may be inade-
quately evaluated (too few data points), which causes
the BLUP to shrink toward the grand mean. One merit
of using data from more than one year is that more
cultivars can be decisively evaluated. For example, when
the multiple-location trial of 2000 alone was used, 30
out of 112 cultivars were judged as superior or inferior.
When two-year data (1999 and 2000) were used, 35
cultivars were judged as superior or inferior. Forty-four
cultivars were decisively evaluated when 4 yr of data
(from 1997–2000) were used (Table 7). In general, the
larger the number of years the dataset contains, the
larger the number of cultivars that can be conclusively
evaluated (Table 7). The increase in the number of
genotypes that can be decisively evaluated, however,
becomes trivial beyond 3 yr.
DISCUSSION
One conclusion from this study is that data from sin-
gle-year, multiple-location soybean trials had sufficient
power to identify superior and inferior genotypes, and
2 yr of data gave slightly better prediction of next year’s
performance. This conclusion is consistent with three
previous studies that investigated the predictive power
of single vs. multiple year trials for next-year cultivar
performance. Recognizing that growers are interested
in finding a selection strategy that will maximize their
production the next year, Cross and Helm (1986) com-
pared strategies of corn [Zea mays L.] hybrid selection
based on yield data from one, two, or three previous
years’ tests at single or multiple locations. The criterion
was yield advantage of selected hybrids relative to the
mean yield of all entries in the next year. Prediction
based on data from one or two previous years was better
than that based on data from the previous three years.
Using performance trial data of spring wheat [Triticum
aestivum L] and oats [Avena sativa L.], Gellner (1989)
compared eight predictors of cultivar performance for
their ability to identify the five top-yielding cultivars in
the subsequent year. Three of these predictors were
mean yields of respective cultivars obtained from one,
two, or three previous years of multiple-location trials.
He concluded that prediction of next-year performance
on the basis of data from the previous year alone was
equally good as that based on data from three previ-
ous years.
Using a balanced subset of data from North Carolina
performance trials, Bowman (1998) compared probabil-
ities of predicting the top varieties in a subsequent year
based on single and 2-yr data for barley [Hordeum vul-
gare L.], corn, cotton [Gossypium hirsutum L.], oats,
soybean, and wheat. He concluded that it was appro-

Table 6. Correlation coefficients between tBLUP based on one to nine previous years and that based on the present year.
Number of previous years
The present year 1 2 3 4 5 6 7 8 9
2000 0.57 0.63 0.61 0.61 0.61 0.61 0.61 0.61 0.61
1999 0.57 0.57 0.57 0.66 0.65 0.65 0.65 0.65
1998 0.56 0.68 0.68 0.68 0.67 0.68 0.67
1997 0.51 0.56 0.52 0.52 0.53 0.54
1996 0.47 0.51 0.51 0.51 0.51
1995 0.71 0.65 0.65 0.68
1994 0.32 0.53 0.55
Mean 0.53 0.59 0.58 0.61 0.59 0.62 0.64 0.63 0.61
554 CROP SCIENCE, VOL. 43, MARCH–APRIL 2003
Table 7. Number of genotypes decisively evaluated on the basis data from 1 to 10 yr of multiple-location trials.
Number of
The present year genotypes 1 2
2000 112 30 35
1999 112 44 50
1998 112 9 22
1997 104 33 39
1996 90 18 21
1995 99 40 46
Mean 29.0 35.5
† No BLUP output was available due to zero genotypic variance.
priate to use one-year multiple-location trial data to
select midseason corn hybrids and 2-yr, multiple-loca-
tion trial data in selecting the other crops. The finding
that 1-yr performance trial data had sufficient power in
identifying superior and inferiors genotypes justifies the
common practice by breeders and cultivar sponsors of
withdrawing their cultivars from the test based on a
single-year trial.
Bowman (1998) was not able to compare predictions
of cultivar performance by data from more than two
years since very few cultivars were tested in more than
three consecutive years. For the same reason, Cross
and Helm (1986) and Gellner (1989) only examined
predictions using data from up to 3 yr. We were able
to investigate predictions based on data from up to 9
yr because of the use of mixed models. Mixed models
allow the use of unbalanced data and give BLUP of
cultivars. This allowed us to examine the merits of using
multiple-year data and conclude that although a single-
year, multiple-location test has enough power to identify
the best cultivars, use of multiple-year data allow more
genotypes to be evaluated conclusively.
With fewer data points, BLUP has the property of
shrinking toward the grand mean, which makes it partic-
ularly suitable for cultivar evaluation based on unbal-
anced data (Piepho, 1994; DeLacy et al., 1996). Panter
and Allen (1995) also concluded that BLUP was supe-
rior to least square means in predicting soybean culti-
var performance. Instead of using BLUP, we used the
t-statistic of BLUP, i.e., tBLUP, as a measure of cultivar
performance. The two measures are highly correlated;
tBLUP has an advantage over BLUP per se, however,
in that it is also an intuitive measure of reliability of
cultivar performance relative to the grand mean, as
t-statistics are directly related to the probability of being
different from the grand mean. Thus, tBLUP provides
a convenient means for classifying tested genotypes into
superior, inferior, and intermediate groups (Yan et al.,
2002).
Alternatively, three different yield-stability statistics
have been used to recommend cultivars for production
in three maturity zones in Minnesota (Pazdernik et al.,
1997). The stability characteristics studied consisted of
two nonparametric statistics based on cultivar means
and Kang’s yield stability statistic, YSi (Kang, 1993). It
was concluded that the stability statistics, in addition to
cultivar means, could be used by consultants, cultivar
testing personnel, and breeders to recommend the most
Number of years
3 4 5 6 7 8 9 10
38 44 45 45 46 46 46 46
55 54 55 54 54 54 54
30 † 31 32 † 35
38 39 39 39 39
25 26 28 27
46 46 48
38.7 41.8 41.0 39.4 46.3 45.0 50.0 46.0
appropriate cultivars for maxiumum yield and protein
concentrations (Pazdernik et al., 1997). Our study, how-
ever, was aimed at using single vs. multiple-year, multi-
ple-location yield trial data to select cultivars in the
following year, reflecting a practice that is widely used
by breeders and extension specialists.
The results of our study are in full agreement with a
previous study of historical data from the Ontario Win-
ter Wheat performance trials (Yan, unpublished). Win-
ter wheat performance trials in Ontario have been con-
ducted at eight to 10 locations across the province, a
much wider region than the 2800 CHU area reported
here for soybean. Nevertheless, similar conclusions were
reached with regard to the predictive power of single-
versus multiple-year trials. In conclusion, when multi-
ple-year performance trial data are available, we recom-
mend the use of tBLUP based on data from 2 yr of
performance as a basis for cultivar evaluation and rec-
ommendation. Estimates of tBLUP based on single-
year performance are also sufficient to select superior
cultivars and to discard inferior genotypes.
REFERENCES
Annicchiarico, P., and M. Perenzin. 1994. Adaptation patterns and
definition of macro-environments for selection and recommenda-
tion of common wheat genotypes in Italy. Plant Breed. 113:197–205.
Bowman, D.T. 1998. Using crop performance data to select hybrids
and varieties. J. Prod. Agric. 11:256–259.
Comstock, R.E., and R.H. Moll. 1963. Genotype-environment interac-
tions. p. 164–196. In W.D. Hanson and H.F. Robinson (ed.) Statisti-
cal genetics and plant breeding. National Academy of Sciences and
National Research Council, Washington DC.
Cross, H.Z., and J.L. Helm. 1986. Hybrid maize selection strategies
based on state yield trials. J. Agron. Educ. 15:110–113.
DeLacy, I.H., K.E. Basford, M. Cooper, J.K. Bull, and C.G. McLaren.
1996. Analysis of multi-environment trials—a historical perspec-
tive. p. 39–124. In M. Cooper and G.L. Hammer (ed.) Plant adapta-
tion and crop improvement. CAB Int., Wallingford, Oxon, UK.
Gauch, H.G., and R.W. Zobel. 1996. AMMI analysis of yield trials.
p. 85–122. In M.S. Kang and H.G. Gauch, Jr. (ed.) Genotype-by-
environment interaction. CRC Press, Boca Raton, FL.
Gellner, J.L. 1989. Predicting superior yielding spring wheat an oat
cultivars using past yield data. Agron. J. 81:194–197.
Kang., M.S. 1993. Simultaneous selection for high yield and stability
in crop performance trials: Consequences for growers. Agron. J.
85: 754–757.
Littell, R.C., G.A. Milliken, W.W. Stroup, and R.D. Wolfinger. 1996.
SAS system for mixed models. SAS Institute Inc., Cary, NC.
Ontario Ministry of Agriculture. Food and Rural Affairs (OMAFRA).
1993. Crop Heat Units for Corn and Other Warm-Season Crops
in Ontario. [Online], available at: http://www.gov.on.ca/OMAFRA/
 YAN & RAJCAN: PREDICTION OF SOYBEAN CULTIVAR PERFORMANCE
english/crops/facts/93-119.htm. [modified 3 July 2001; verified 13
September 2002]. Government of Ontario, ON, Canada.
Panter, D.M., and F.L. Allen. 1995. Using best linear unbiased predic-
tions to enhance breeding for yield in soybean. II Selection of
superior crosses from a limited number of yield trials. Crop Sci.
35:405–410.
Pazdernik, D.L., L.L. Hardman, and J.H. Orf. 1997. Agronomic per-
formance and stability of soybean vareieties grown in three matu-
rity zones of Minnesota. J. Prod. Agric. 10:425–430.
Piepho, H.P. 1994. Best linear unbiased prediction (BLUP) for regional
555
yield trials: A comparison to additive main effects and multiplicative
interaction (AMMI) analysis. Theor. Appl. Genet. 89:647–654.
SAS Institute. 1996. SAS/STAT user’s guide. SAS Institute Inc.,
Cary, NC.
Yan, W., and L.A. Hunt. 1998. Genotype-by-environment interaction
and crop yield. Plant Breed. Rev. 16:135–178.
Yan, W., L.A. Hunt, P. Johnson, G. Stewart, and X. Lu. 2002. On-
farm strip trials versus replicated performance trials for cultivar
evaluation. Crop Sci. 42:385–392.
Yan, W., and I. Rajcan. 2002. Biplot analysis of test sites and trait
relations of soybean in Ontario. Crop Sci. 42:11–20.