Access to this full-text is provided by MDPI.
Content available from Agronomy
This content is subject to copyright.
Citation: Zori´c, M.; Gunjaˇca, J.; Gali´c,
V.; Juki´c, G.; Varnica, I.; Šimi´c, D. Best
Linear Unbiased Predictions of
Environmental Effects on Grain Yield
in Maize Variety Trials of Different
Maturity Groups. Agronomy 2022,12,
922. https://doi.org/10.3390/
agronomy12040922
Academic Editors:
Gniewko Niedbała,
Magdalena Piekutowska,
Tomasz Wojciechowski and
Mohsen Niazian
Received: 15 March 2022
Accepted: 11 April 2022
Published: 12 April 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
agronomy
Article
Best Linear Unbiased Predictions of Environmental Effects on
Grain Yield in Maize Variety Trials of Different
Maturity Groups
Marina Zori´c 1, Jerko Gunjaˇca 2,3 , Vlatko Gali´c 4, Goran Juki´c 1, Ivan Varnica 1and Domagoj Šimi´c 3,4,*
1Croatian Agency for Agriculture and Food, Vinkovacka cesta 63c, 31000 Osijek, Croatia;
marina.zoric@hapih.hr (M.Z.); goran.jukic@hapih.hr (G.J.); ivan.varnica@hapih.hr (I.V.)
2
Faculty of Agriculture, University of Zagreb, Svetosimunska cesta 25, 10000 Zagreb, Croatia; jgunjaca@agr.hr
3
Centre of Excellence for Biodiversity and Molecular Plant Breeding, Svetosimunska 25, 10000 Zagreb, Croatia
4Agricultural Institute Osijek, Juzno predgradje 17, 31000 Osijek, Croatia; vlatko.galic@poljinos.hr
*Correspondence: domagoj.simic@poljinos.hr
Abstract:
Development of new cultivars and agronomic improvements are key factors of increasing
in future grain yield in maize grown in environments affected by climate change. Assessment of
value for cultivation and use (VCU) reflects the results of latest breeding efforts showing yield trends,
whereby external environmental covariates were rarely used. This study aimed to analyze several
environmental effects including stress degree days (SDD) on grain yields in Croatian VCU trials
in three maturity groups using linear mixed model for the estimation of fixed and random effects.
Best linear unbiased predictions (BLUPs) of location-year interaction showed no pattern among
maturity groups. SDD showed mostly non-significant coefficients of regression on location BLUPs
for yield. Analyzing location BLUPs, it was shown that the effect became consistently stronger with
later maturity, either positive or negative. The effects of management might play more critical role
in maize phenology and yield formation compared with climate change, at least in suboptimum
growing conditions often found in Southeast Europe. To facilitate more robust predictions of the
crop improvement, the traditional forked approach dealing with G
×
E by breeders and E
×
M by
agronomists should be integrated to G
×
E
×
M framework, to assess the full gradient of combinations
forming the adaptation landscape.
Keywords: maize; grain yield; heat stress; maturity groups; BLUPs; VCU trials
1. Introduction
Long-term trials for assessment of value for cultivation and use (VCU) [
1
,
2
] applying
during maize registration i.e., maize variety trials are valuable source of information on
general maize yield trends [
3
,
4
] and across maturity groups [
5
]. However, the genetic
values of genotypes in multi-environment VCU trials is hidden by variation caused by
complex genotype by environment interaction effects [
6
]. In order to disentangle this
complexity in typically unbalanced VCU trials, mixed-effect models were used, which
allow attributing yield variability to randomly distributed independent effects. In this
context, the means can be analyzed adjusting by best linear unbiased predictions (BLUPs)
and best linear unbiased estimators (BLUEs) [
7
]. BLUP is a method with favorable feature
of shrinkage of estimators towards the mean, reducing its variance, but increasing its
predictive accuracy [
8
]. The method was first developed in animal breeding for predicting
the animal breeding values [
9
], receiving afterwards considerable attention in plant multi
environment trial analysis [
10
–
12
]. In contrast to the BLUE methodology calculated using
fixed-effect linear models, mixed models with fixed and random effects allow calculation
of the random effect matrix needed to extract BLUPs [13].
Agronomy 2022,12, 922. https://doi.org/10.3390/agronomy12040922 https://www.mdpi.com/journal/agronomy
Agronomy 2022,12, 922 2 of 13
Yield information on varieties and environments in VCU trials can be analyzed apply-
ing different models aiming at describing as much of the yield variation between varieties
and their interactions with the environments as possible using external environment-
specific covariates which can detect (a) biotic stresses. Several environmental (climatic)
indices were proposed incorporating weather variables to enable reliable within-season
predictions in new environments [
6
,
14
]. In Southeast Europe, an aridity index [
15
] and an
improved Palmer Drought Severity Index [
16
] were used among others for quantifying
drought stress in maize. It was demonstrated though, that extreme heat as a stressor could
have more critical role for maize production than drought in the US [
17
] corroborating
previous statistical studies of rainfed maize yields showing a strong negative yield response
to accumulation of extreme temperatures (>30
◦
C) and relative weak response to seasonal
rainfall. The concept of Stress Degree Days (SDD), ref. [
18
] utilized for decades to measure
heat stress, is also appearing in recent studies e.g., [
19
,
20
]. Buhiniˇcek et al. [
20
] used a
long-term pre-registration yield trials in Croatia and official weather data to show that
irregular climatic conditions during maize growing season are becoming more prominent
in the last three decades. In that study based on experimental and simulation data, it was
demonstrated that choosing right maturity group should play even more important role in
the future. However, genotype by environment interaction effects were not captured.
Globally, hybrid relative maturity in maize, i.e., plant cycle duration is becoming
more important in the context of climate change to maximize yield, whereby farmers
should continuously adapt maize cycle duration and planting dates to the diversity of
environmental conditions [
21
,
22
]. Several systems for rating of hybrid maize maturity are
employed around the world. Various thermal units are used in North America including
relative maturity (RM) and growing degree days (GDD) [
23
], whereas the uniform method
recommended by Food and Agriculture Organization of the United Nations (FAO) [
24
] is
still used in Europe [
25
]. Similar to RM, FAO maturity groups represent the length of time
necessary for a hybrid to reach harvest-ready moisture determined relatively to a standard
hybrid. This maturity classification is an important component for maize VCU evaluation
as well as for recommended or national lists of varieties released for commercial use [24].
The objectives of this study were to determine environmental effects on grain yield
using BLUPs and BLUEs across three maturity groups in Croatian VCU maize trials over
the last two decades, and to evaluate the use of SDD as a climatic covariate to determine
the impact of climate change on grain yield in maize.
2. Materials and Methods
Yield data from official Croatian variety trials assessing VCU of maize in three maturity
groups (FAO 300, FAO 400 and FAO 500) for the period 2001–2019 were used in this study.
Data were collected from five locations in northern (continental) part of Croatia: Zagreb
(Zg: N 45.7; E 16.3), Kutjevo (Ku: N 45.3; E 17.9), Osijek (Os: N 45.5; E 18.5), Beli Manastir
(Bm: N 45.7; E 18.6) and Vukovar (Vu: N 45.3; E 19.1). The adjacent yield trials of the three
maturity groups were set on marsh gley partly hydroameliorated (Zg), acid brown soil
over clastite and metamorphite (Ku), semigley on leached loess (Os) and chernozem on
loess (Bm and Vu) [
26
]. In total, there were 37–49 trials per maturity group in the 19-year
period (Supplementary Table S1). The number of genotypes included all entries in all
trials: control cultivars, the genotypes entering the first trial year, subsequent withdrawn
genotypes by breeders, as well the genotypes (cultivars) finally released. All trials were set
as randomized complete block designs with four replicates and were machine-planted on
11.2 m
2
sized plots. Grain yield data were calculated on the 14% moisture basis. Standard
cultural practices (fertilization, weed, pest, and disease control) for maize high-yielding
production were used before and during crop growing. The data sets were unbalanced due
to a number of genotypes leaving or entering the trials from one year to another.
Agronomy 2022,12, 922 3 of 13
Variance components were assessed within three maturity groups, FAO 300, FAO 400
and FAO 500. The general three-way mixed model used for modelling of grain yield
variance was set as:
yijk =µ+ (Gi)+(Lj)+(Yk)+(LYjk)+(GLij)+(GYik)+(GLYijk) + εijk
in which yield y
ijk
was observed for igenotypes in jlocations over kconsecutive years. Main
effect of year Y
k
was treated as fixed, whereas the effects of genotypes (G
i
) locations (L
j
)
and interaction terms of genotype-by-location (GL
ij
), genotype-by-year (GY
ik
), genotype-
by-location-by-year (GLY
ijk
) and the error
εijk
were assumed to be independent and have
constant variances over levels of effects. Locations were assumed to be crossed with years
(LY
jk
) as most of the locations were used in most of the years. Interactions of genotypes
by years and locations were calculated on the basis that most of the genotypes were
screened for at least two years, whereas some genotypes (checks) are present in the dataset
over more than seven years. All models were set using the restricted maximum likelihood
(REML) in R [
27
] library lme4 [
28
]. The intraclass correlation coefficients (ICC) and Marginal
R
2
/Conditional R
2
values of the models were calculated in the sjPlot library [
29
] based on
the methodology proposed by Nakagawa et al. [
30
,
31
]. The ICC quantifies the proportion of
variance explained by random factors in multi-level or hierarchical datasets. The best linear
unbiased predictions (BLUP) [
12
] of random effects and best linear unbiased estimates
(BLUE) [
32
] of year effects (Figure S2) were extracted using the R’s coef () function. In
our study, the focus was on environmental random effects and their BLUPs denoted as
L_BLUPs and LY_BLUPs for L and LY effects, respectively.
Daily weather data for precipitation, air temperature (minimum and maximum) and
solar radiation was obtained from the AGRI4CAST database [
33
] using information for
the grids from west to east: 79,129 Zg (N 45.68793; E 16.30936); 78,134 Ku (N 45.35677;
E 17.87238); 79,136 Os (N 45.53097; E 18.54105); 80,136 Bm (N 45.75453; E18.57620) and
78,138 Vu (N 45.25416, E 19.13910) for the period from April to October of each year.
The impact of heat stress on grain yield was estimated using stress degree days (SDD)
concept [18–20] calculated as
SDD ∞
30 =∑N
t=1DDt,DDt =0, when Ta <30
Ta −30, when Ta ≥30
where trepresents the daily time step, Nis the total number of days in each growing
period, DD is degree days, and Ta is air temperature. The SDD index was chosen as an
environmental covariate in this study due to highest correlations with grain yield (data
not shown). Correlations between some other climate covariates (growing degree days,
precipitation, vapor pressure deficit) and grain yield were similar to those presented by
Buhiniˇcek et al. [
20
]: they were mostly weak and non-significant. A simple linear regression
model was used for fitting the data for grain yield on SDD values across the five locations.
Simulations were performed using the APSIM platform [
34
] for the same period
in three geographically distant locations included in the VCU trials (Zg, Ku, Os) select-
ing default variety options. Details were presented by Stepinac et al. [
35
]. Briefly, the
maize genotypes representing three FAO maturity groups were Pioneer
®
cultivars P38H20
(FAO 300), P34K77 (FAO 400) and P33M54 (FAO 500) (Pioneer, Johnston, IA, USA). An
uncalibrated model of APSIM evaluations was used assuming unchanged (invariant) geno-
types with optimum planting density of 7.1 plants m
−2
(FAO 300), 6.4 plants m
−2
(FAO 400)
and 6.0 plants m
−2
(FAO 500) similar to the VCU trials. The intention of APSIM simula-
tions was to assess relationships of observed/simulated average grain yields and BLUPs
with SDD values across FAO groups in the locations where regression coefficients were
not significant.
Agronomy 2022,12, 922 4 of 13
3. Results
The magnitudes of main variance component of genotype (G) were similar for all
three maturity groups, whereby the G component for late maturity group of FAO 500 was
somewhat smaller. The main variance component of location (L) was the largest for FAO 500
maturity group having generally high standard errors. The largest variance components
in all three maturity groups were estimated for the environmental
location ×year
(LY)
interaction followed by the main G and L effects in FAO 300 and FAO 400 (Table 1). In
FAO 500, the second greatest variance component was from the L effect. On the other hand,
the GL effect was negligible in all maturity groups. The proportion of variance explained
by the G effect was 8.82% averaged over the FAO groups.
Table 1.
Estimated variance components and associated standard errors (s.e.) for yield of maize from
official Croatian variety trials in the period 2001–2019 along with intraclass correlation coefficients
(ICC), values of Marginal R2/Conditional R2(MargR2/CondR2) and heritability estimates.
Effect
/Parameter
FAO 300 FAO 400 FAO 500
Variance s.e. Variance s.e. Variance s.e.
Gi0.418 0.042 0.409 0.039 0.328 0.041
Lj0.275 0.335 0.388 0.331 0.759 0.233
LYjk 2.251 0.154 1.763 0.138 2.369 0.133
GLij 0.057 0.039 0.027 0.043 0.000 0.060
GYik 0.091 0.032 0.146 0.028 0.539 0.026
GLYijk 0.211 0.026 0.289 0.019 0.363 0.026
εijk 1.034 0.009 0.883 0.007 0.811 0.009
ICC 0.76 0.77 0.84
MargR2/CondR20.29/0.83 0.28/0.84 0.19/0.87
Heritability 0.48 0.53 0.40
ICC values between 0.76 and 0.84 indicate good repeatability, whereas heritability
estimates were notably lower, between 0.40 and 0.53 due to relatively large size of inter-
action effects compared with the genotypic (G) effects. According to the ratio Marginal
R
2
/Conditional R
2
, proportion of variance explained by the fixed factor Y ranged from 0.19
to 0.29. Overall, proportion of variance explained by both the fixed and random factors
is >0.8.
Unadjusted mean values for grain yield over the 19-year period grouped in three
FAO maturity groups were given in Supplementary Figure S1. Generally, mean values
ranged broadly from 2.94 to 15.33 t/ha with similar results across the maturity groups.
There were no significant yield trends over the period. Caterpillar dot plots of random LY
effects showed the most positive values for the Os location in 2018 and the most negative
values for Vu location in 2012 in all three FAO groups (Figure 1). When compared with
other locations, Os and Vu locations varied the most in yield over the years having similar
pattern across the maturity groups.
SDD values fluctuated considerably during the two-decade period ranging from
around 30 in 2014 to 250 in 2012 for the Vu location (Figure 2). Generally, SDD values
were alike at all five locations in a particular year, except for the period 2007–2009 and
in 2012 when SDD values diverged to some extent. However, the observed SDD values
were consistently higher in the locations of Eastern Croatia than those in west: Bm and Os
in 2003 and Vu in very hot years of 2012, 2015 and 2017.
Regression analysis revealed mostly non-significant slope coefficients of the L_BLUPs
on SDD values (Table 2). The exceptions were the coefficients in Bm and Vu locations at
most instances: they were significantly positive in Bm and significantly negative in Vu.
Thus, higher values of SDD at these two locations had an impact on grain yield particularly
in Vu where the significance level of regression coefficient for all maturity groups was
p≤0.056
. On the other hand, no discernible differences were found among the estimates
across the maturity groups.
Agronomy 2022,12, 922 5 of 13
Agronomy 2021, 11, x FOR PEER REVIEW 5 of 13
Figure 1. Dot plots of relative best linear unbiased predictions of random LY effects (Year:Location)
(LY_BLUPs) in VCU trials in maize genotypes belonging to three maturity groups (a) FAO 300; (b)
FAO 400; and (c) FAO 500.
SDD values fluctuated considerably during the two-decade period ranging from
around 30 in 2014 to 250 in 2012 for the Vu location (Figure 2). Generally, SDD values were
alike at all five locations in a particular year, except for the period 2007–2009 and in 2012
when SDD values diverged to some extent. However, the observed SDD values were
consistently higher in the locations of Eastern Croatia than those in west: Bm and Os in
2003 and Vu in very hot years of 2012, 2015 and 2017.
Figure 1.
Dot plots of relative best linear unbiased predictions of random LY effects (Year:Location)
(LY_BLUPs) in VCU trials in maize genotypes belonging to three maturity groups (
a
) FAO 300;
(b) FAO 400; and (c) FAO 500.
Agronomy 2021, 11, x FOR PEER REVIEW 5 of 13
Figure 1. Dot plots of relative best linear unbiased predictions of random LY effects (Year:Location)
(LY_BLUPs) in VCU trials in maize genotypes belonging to three maturity groups (a) FAO 300; (b)
FAO 400; and (c) FAO 500.
SDD values fluctuated considerably during the two-decade period ranging from
around 30 in 2014 to 250 in 2012 for the Vu location (Figure 2). Generally, SDD values were
alike at all five locations in a particular year, except for the period 2007–2009 and in 2012
when SDD values diverged to some extent. However, the observed SDD values were
consistently higher in the locations of Eastern Croatia than those in west: Bm and Os in
2003 and Vu in very hot years of 2012, 2015 and 2017.
Figure 2.
Stress Degree Days (SDD) for five locations in the continental part of Croatia in the
two-decade period from 1999 to 2019.
Average grain yield was the highest in Zg and Bm locations for all three maturity
groups (Figure 3). In both locations, the highest yielding FAO group was the latest one
(FAO 500) gradually followed by two earlier ones (FAO 400 and 300, respectively). No
such pattern was observed for the FAO groups in other three lower yielding locations (Ku,
Os and Vu). In the high-yielding location Bm, unadjusted means for grain yield between
the maturity groups FAO 300 and FAO 500 were significantly different. There were no
significant differences among maturity groups in all other locations.
Agronomy 2022,12, 922 6 of 13
Table 2.
Coefficients of regression of the L_BLUPs on SDD values estimated by a simple linear model.
SE, t, and pdesignate the standard error, the t statistic and p-value of the fitted parameter, respectively.
p-values lower than 0.05 are shown in bold.
Location Maturity group Term Estimate SE t p
Zg FAO300 Intercept 10.703 0.703 15.218 0.000
Slope 0.011 0.007 1.556 0.138
FAO400 Intercept 11.317 0.563 20.113 0.000
Slope 0.008 0.006 1.385 0.184
FAO500 Intercept 11.334 0.667 17.003 0.000
Slope 0.008 0.007 1.162 0.261
Ku FAO300 Intercept 11.602 0.619 18.732 0.000
Slope −0.003 0.006 −0.446 0.663
FAO400 Intercept 11.722 0.519 22.601 0.000
Slope 0.001 0.005 0.097 0.924
FAO500 Intercept 12.172 0.626 19.453 0.000
Slope −0.004 0.006 −0.687 0.504
Os FAO300 Intercept 11.791 0.645 18.288 0.000
Slope −0.006 0.005 −1.116 0.281
FAO400 Intercept 12.038 0.652 18.451 0.000
Slope −0.004 0.005 −0.682 0.505
FAO500 Intercept 11.403 0.751 15.184 0.000
Slope 0.003 0.006 0.434 0.670
Bm FAO300 Intercept 10.407 0.366 28.441 0.000
Slope 0.011 0.004 3.039 0.008
FAO400 Intercept 11.157 0.433 25.765 0.000
Slope 0.008 0.004 1.885 0.079
FAO500 Intercept 10.940 0.403 27.181 0.000
Slope 0.012 0.004 2.872 0.012
Vu FAO300 Intercept 12.385 0.649 19.074 0.000
Slope −0.010 0.005 −2.056 0.056
FAO400 Intercept 12.933 0.541 23.925 0.000
Slope −0.011 0.004 −2.564 0.020
FAO500 Intercept 13.454 0.583 23.077 0.000
Slope −0.015 0.005 −3.316 0.004
By analyzing best linear unbiased predictions, it was revealed that L_BLUPs are
somewhat different compared with unadjusted means over the locations. Markedly, the
L_BLUPs became stronger with later maturity, either positive or negative (Figure 4). It
was consistent across all five locations and three maturity groups. There is no geographic
pattern though, indicating that the effect of location is discrete. Due to large standard
errors of all L_BLUPs, there were no significant differences among maturity groups within
a location.
Predicted grain yield according to APSIM simulated data averaged over the period
2001–2019 (Figure 5) was considerably lower in Zg and Ku locations than observed un-
adjusted means obtained at the same locations (Figure 3). Although non-significant, the
highest simulated grain yields were in genotypes of early FAO 300 group in these loca-
tions. Observed unadjusted and simulated means for Os location were similar in all three
maturity groups together with larger standard errors than those in Zg and Ku.
Correlations between grain yield and SDD were consistently negative across the
three locations and the three maturity groups (Figure 6). Significant negative correlation
coefficients were estimated in genotypes of all three FAO groups for average grain yield
obtained by observed unadjusted means and APSIM simulated means, but not obtained
by L_BLUPs. Notably, the respective correlations with grain yield estimated by observed
unadjusted means and APSIM simulated means were similar. Correlations between grain
yield obtained by L_BLUPs and SDD were mostly weak.
Agronomy 2022,12, 922 7 of 13
Agronomy 2021, 11, x FOR PEER REVIEW 7 of 13
the maturity groups FAO 300 and FAO 500 were significantly different. There were no
significant differences among maturity groups in all other locations.
Figure 3. Unadjusted means for grain yield with corresponding standard errors in VCU trials across
five locations in maize genotypes belonging to three maturity groups FAO 300, FAO 400, and FAO
500.
By analyzing best linear unbiased predictions, it was revealed that L_BLUPs are
somewhat different compared with unadjusted means over the locations. Markedly, the
L_BLUPs became stronger with later maturity, either positive or negative (Figure 4). It
was consistent across all five locations and three maturity groups. There is no geographic
pattern though, indicating that the effect of location is discrete. Due to large standard
errors of all L_BLUPs, there were no significant differences among maturity groups within
a location.
Figure 4. The best linear unbiased predictions of random Location effects (L_BLUPs) with
corresponding standard errors in VCU trials in maize genotypes belonging to three maturity groups
FAO 300, FAO 400, and FAO 500.
Figure 3.
Unadjusted means for grain yield with corresponding standard errors in VCU trials across
five locations in maize genotypes belonging to three maturity groups FAO 300, FAO 400, and
FAO 500
.
Agronomy 2021, 11, x FOR PEER REVIEW 7 of 13
the maturity groups FAO 300 and FAO 500 were significantly different. There were no
significant differences among maturity groups in all other locations.
Figure 3. Unadjusted means for grain yield with corresponding standard errors in VCU trials across
five locations in maize genotypes belonging to three maturity groups FAO 300, FAO 400, and FAO
500.
By analyzing best linear unbiased predictions, it was revealed that L_BLUPs are
somewhat different compared with unadjusted means over the locations. Markedly, the
L_BLUPs became stronger with later maturity, either positive or negative (Figure 4). It
was consistent across all five locations and three maturity groups. There is no geographic
pattern though, indicating that the effect of location is discrete. Due to large standard
errors of all L_BLUPs, there were no significant differences among maturity groups within
a location.
Figure 4. The best linear unbiased predictions of random Location effects (L_BLUPs) with
corresponding standard errors in VCU trials in maize genotypes belonging to three maturity groups
FAO 300, FAO 400, and FAO 500.
Figure 4.
The best linear unbiased predictions of random Location effects (L_BLUPs) with corre-
sponding standard errors in VCU trials in maize genotypes belonging to three maturity groups
FAO 300, FAO 400, and FAO 500.
Agronomy 2021, 11, x FOR PEER REVIEW 8 of 13
Predicted grain yield according to APSIM simulated data averaged over the period
2001–2019 (Figure 5) was considerably lower in Zg and Ku locations than observed
unadjusted means obtained at the same locations (Figure 3). Although non-significant, the
highest simulated grain yields were in genotypes of early FAO 300 group in these
locations. Observed unadjusted and simulated means for Os location were similar in all
three maturity groups together with larger standard errors than those in Zg and Ku.
Figure 5. Predicted grain yield according to APSIM simulated data averaged over the period 2001–
2019 with corresponding standard errors in maize genotypes belonging to three maturity groups
FAO 300, FAO 400, and FAO 500 at three geographically distinct locations in Croatia.
Correlations between grain yield and SDD were consistently negative across the
three locations and the three maturity groups (Figure 6). Significant negative correlation
coefficients were estimated in genotypes of all three FAO groups for average grain yield
obtained by observed unadjusted means and APSIM simulated means, but not obtained
by L_BLUPs. Notably, the respective correlations with grain yield estimated by observed
unadjusted means and APSIM simulated means were similar. Correlations between grain
yield obtained by L_BLUPs and SDD were mostly weak.
Figure 5.
Predicted grain yield according to APSIM simulated data averaged over the period
2001–2019
with corresponding standard errors in maize genotypes belonging to three maturity
groups FAO 300, FAO 400, and FAO 500 at three geographically distinct locations in Croatia.
Agronomy 2022,12, 922 8 of 13
Agronomy 2021, 11, x FOR PEER REVIEW 8 of 13
Predicted grain yield according to APSIM simulated data averaged over the period
2001–2019 (Figure 5) was considerably lower in Zg and Ku locations than observed
unadjusted means obtained at the same locations (Figure 3). Although non-significant, the
highest simulated grain yields were in genotypes of early FAO 300 group in these
locations. Observed unadjusted and simulated means for Os location were similar in all
three maturity groups together with larger standard errors than those in Zg and Ku.
Figure 5. Predicted grain yield according to APSIM simulated data averaged over the period 2001–
2019 with corresponding standard errors in maize genotypes belonging to three maturity groups
FAO 300, FAO 400, and FAO 500 at three geographically distinct locations in Croatia.
Correlations between grain yield and SDD were consistently negative across the
three locations and the three maturity groups (Figure 6). Significant negative correlation
coefficients were estimated in genotypes of all three FAO groups for average grain yield
obtained by observed unadjusted means and APSIM simulated means, but not obtained
by L_BLUPs. Notably, the respective correlations with grain yield estimated by observed
unadjusted means and APSIM simulated means were similar. Correlations between grain
yield obtained by L_BLUPs and SDD were mostly weak.
Figure 6.
Correlation coefficients between grain yield estimated by observed unadjusted means,
APSIM simulated means, L_BLUPs for maize genotypes of three maturity groups and SDD values
across three geographically distinct locations in Croatia in the two-decade period. The dashed vertical
line denotes significance level at p< 0.05.
4. Discussion
By estimating the components of variance and analyzing their magnitude, it is possible
to decouple the complexity of genotype by environment interactions. This is important for
planning VCU experiments to determine optimum resource allocation [
36
,
37
] particularly
in large areas where diverse climatic and geographical conditions with different cycle
duration take place. In India [
38
], variance components for all effects used also in our study
were quite close for all investigated maize maturity groups (extra early, early, medium and
late). In our VCU trials where geographic conditions were not notably different, the greatest
estimated variance component was the LY interaction for all three maturity groups. In the
FAO 300 and 400 maturity groups, it is followed by a considerably lower proportion of
the G effect, whereas in the FAO 500 group the next component of variance by magnitude
is the L effect. In the German VCU maize trials [
37
], the prevailing variance for grain
yield was from the L effect, followed by LY and G effects in all three (early, medium, late)
maturity groups. Comparable results were presented for maize yield trials conducted
in African [
39
,
40
] and Asian environments [
41
]. In yield trials of late maturing groups
(FAO 500, 600 and 700) in different environments in Greece [
42
] the highest percentage
of yield variation was explained by the main environmental effect (68.89%) when used a
simple two-way genotype by environment model with no decoupling of L and Y effects,
whereby the proportion of G effect was 3.95%. In our study, the proportion of variance
explained by genotype was in average 8.82% compared with 11.56% in our previous
research [5] and 6.65% obtained from the German VCU trials [37].
The means in our study were analyzed in the context of BLUPs and not of BLUEs as
recommended by Robinson [
8
] giving several arguments for using it: (1) smaller expected
mean square errors, (2) it is suitable in variety trials when the aim is to predict the future
variety performance, and (3) it is appropriate for small-area estimation. As a rule of thumb,
negative L_BLUPs may be declared as sites with low soil yield potential, whereas sites with
positive L_BLUPs are sites with high soil yield potential. To our knowledge, our study
is the first attempt to apply BLUPs of environmental effects to evaluate appropriateness
of individual maturity groups of maize for particular growing area. However, soil yield
potential can be evaluated precisely when a set of soil properties are known, such as rooting
depth, topsoil structure or soil compaction [
43
]. The same is true for APSIM simulations
which should be calibrated to generate results according to local soil data [
44
]. This is
Agronomy 2022,12, 922 9 of 13
particularly important when soil type (chernozem) is the same such as in Bm and Vu, but
soil yield potential seems to be different.
Over longer periods in VCU trials, yield increases are expected (as well as increases
in other important economic traits) due to the trend of improvement of new cultivars
through breeding programs [
4
,
5
]. However, in our study no considerable yield increase
was observed for any of the maturity groups (Supplementary Figure S1). This general trend
may be more or less noticeable due to the action of different climatic and agronomic factors,
especially in the context of the growing impact of climate change on all major agricultural
plant species [
45
]. Bönecke et al. [
43
] determined the effects of several agrometeorological
factors on the development of German winter wheat yield in the period between 1958 and
2015 using 298 nitrogen (N) fertilization trials. For this purpose, they separated climatic
from genetic and agronomic yield effects using several linear mixed effect models and
estimating climatic impact based on the coefficient of determination for those models.
The results indicate a general and strong effect of climatic changes on yield development,
especially due to the increase in mean temperatures and heat stress during the grain-filling
period. Except on days of heat stress with more than 31
◦
C, yields in locations with
higher yield potential were less prone to adverse weather conditions than in places with
lower yields.
In maize, it was observed that high temperatures above 30
◦
C play a more critical role
than drought due to increased air moisture deficit and increased
evapotranspiration [17,20,46]
.
The increase in air temperature had a negative effect on yield even under irrigated con-
ditions [
46
]. Also, high temperatures can limit yields by reducing cycle duration [
47
]. In
general, global maize yields are declining with climate change due to increasing in air
temperatures. Zhu et al. [
19
] combined crop models, satellite observations, and field data
to investigate how heat stress affects maize yield in the U.S. Midwest. When the effects of
warming were decomposed into direct heat stress and indirect water stress, observational
data suggest that the yield was reduced more by direct heat stress than by indirect water
stress. They suggest that adaptation strategies should focus on heat stress during grain
formation as it poses a marked threat to cause the decline in maize yields and that changes
in management should be designed to adequately assess the effects of heat stress during
different developmental stages.
Our results corroborate previous findings that SDD values had similar amplitudes
for locations in Croatia for a given year [
20
]. The exceptions were the years 2007, 2008,
2009 and 2012 when SDDs differed slightly among locations. This could be relevant for
interpreting genotype by environment in VCU yield trials. However, regression analysis
showed mostly non-significant regression coefficient for yield on SDD value for all three
maturity groups due to shrinkage property of BLUPs. On the other hand, correlation
coefficients between SDD values and observed unadjusted/simulated means for grain
yield demonstrated that relationship between the environmental covariate of SDD and
grain yield did exist. Nevertheless, in two locations with the highest SDD values in east the
regression coefficient was mostly significant.
Climate change had the effect on maize cropping system in Europe allowing earlier
planting and/or growing early maturity cultivars/hybrids. These avoiding strategies
are commonly applied in maize where stress can be circumvented by earlier planting
dates or planting earlier hybrids to avoid assumed adverse weather conditions mostly
during flowering. However, the global trends in temperature and precipitation indicate
that extreme weather events may occur at any time throughout the growing season. Thus,
the negative effect of temperature during the growing season of maize can be mitigated
by selecting later hybrid maturities that require a longer thermal time period to complete
development [
46
]. Generally, a full-season hybrid can take more advantage of available
heat units and performs better when everything else is balanced. However, recent climate
change considerably impacts maize production causing various (a) biotic stresses [
48
] and
multiple interactions among them, both documented or new [
49
]. Moreover, our study
Agronomy 2022,12, 922 10 of 13
indicates that late hybrids seem to be more prone to adverse climatic conditions in the sites
with low soil yield potential.
Nevertheless, investigations in France [
50
], India [
38
] and China [
51
,
52
] showed that
one of the ways to mitigate the effect of heat stress is the detection and selection of the
optimal maturity group for each breeding area that can prevent a reduction in maize yield.
However, Abendroth et al. [
22
] stated that factors other than thermal availability are more
important in the US Midwest when choosing maize hybrids, such as drying costs, field
operability, labor constraints or crop genetics availability. Moreover, it seems that the
contribution of crop management to maize phenology, i.e., life cycle is larger than climate
change [
53
]. Ultimately, the increase in maize yield is equally influenced by improved
management and the development of new genotypes [54].
The negative effect of the location can be mitigated by applying different agro-technical
measures [
55
,
56
]. However, in Southeast Europe, there are inadequate (suboptimum) cul-
tural practice and crop management due to economic reasons [
5
,
57
,
58
] resulting in complex
and mostly unknown multiple stress growing conditions in field environments. Altogether,
investigating simultaneous genotype
×
environment
×
management (G
×
E
×
M) inter-
actions should be employed [
59
,
60
] to overcome traditional crop improvement approach
seeking firstly genotypes adapted broadly under a standard management regime, and
then manipulation of management regionally in response to average local environmental
conditions. Traditional discipline-centered approaches that dealt with separate compo-
nents of G
×
E
×
M, usually as G
×
E by breeders and E
×
M by agronomists should be
integrated in order to search the full spectrum of G
×
E
×
M combinations forming the
adaptation landscape.
Supplementary Materials:
The following are available online at https://www.mdpi.com/article/
10.3390/agronomy12040922/s1, Table S1: Overview of Croatian VCU maize trials grouped in three
FAO maturity groups evaluated in the period 2001–2019, Figure S1: Mean values for grain yield from
Croatian VCU maize trials grouped in three FAO maturity groups evaluated in the period 2001–2019
in several locations (grey dots) and their average for the particular year (red dots), Figure S2. Best
linear unbiased estimators (BLUEs) of the year effect with corresponding standard errors for grain
yield from Croatian VCU maize trials grouped in three FAO maturity groups evaluated in the period
2001–2019 in several locations (Table S1).
Author Contributions:
Conceptualization, J.G. and D.Š.; methodology, J.G., V.G. and D.Š.; software,
V.G.; formal analysis, V.G. and D.Š.; investigation, M.Z. and I.V.; resources, G.J.; data curation, G.J.and
I.V.; writing—original draft preparation, M.Z.; writing—review and editing, J.G., V.G. and D.Š.;
funding acquisition, G.J. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement:
1. Publicly available weather datasets were analyzed in this study.
This data can be found at https://agri4cast.jrc.ec.europa.eu/DataPortal/Index.aspx (accessed on
4 October 2021
), 2. The detailed grain yield data are available upon request from Croatian Agency
for Agriculture and Food, Vinkovacka cesta 63c, 31000 Osijek, Croatia; goran.jukic@hapih.hr (G.J.).
Conflicts of Interest: The authors declare no conflict of interest.
References
1.
Waes, J.V. Harmonization of VCU Testing Methods for Maize Varieties in a European Context. Acta Agron. Hung.
2006
,54,
365–377. [CrossRef]
2.
Zori´c, M.; Varnica, I.; Juki´c, G.; Juri´c, R. Varieties Registration in the Republic of Croatia. Sjemenarstvo
2020
,31, 41–52. [CrossRef]
3.
Schils, R.L.M.; Van den Berg, W.; Van der Schoot, J.R.; Groten, J.A.M.; Rijk, B.; Van de Ven, G.W.J.; Van Middelkoop, J.C.; Holshof,
G.; Van Ittersum, M.K. Disentangling Genetic and Non-Genetic Components of Yield Trends of Dutch Forage Crops in the
Netherlands. Field Crops Res. 2020,249, 107755. [CrossRef]
Agronomy 2022,12, 922 11 of 13
4.
Laidig, F.; Piepho, H.-P.; Drobek, T.; Meyer, U. Genetic and Non-Genetic Long-Term Trends of 12 Different Crops in German
Official Variety Performance Trials and on-Farm Yield Trends. Theor. Appl. Genet. 2014,127, 2599–2617. [CrossRef] [PubMed]
5.
Zori´c, M.; Gunjaˇca, J.; Šimi´c, D. Genotypic and Environmental Variability of Yield from Seven Different Crops in Croatian Official
Variety Trials and Comparison with On-Farm Trends. J. Agric. Sci. 2016,155, 804–811. [CrossRef]
6.
Kang, Y.; Ozdogan, M.; Zhu, X.; Ye, Z.; Hain, C.; Anderson, M. Comparative Assessment of Environmental Variables and Machine
Learning Algorithms for Maize Yield Prediction in the US Midwest. Environ. Res. Lett. 2020,15, 064005. [CrossRef]
7.
Henderson, C.R. Best Linear Unbiased Estimation and Prediction under a Selection Model. Biometrics
1975
,31, 423–447. [CrossRef]
8. Robinson, G.K. That BLUP Is a Good Thing: The Estimation of Random Effects. Stat. Sci. 1991,6, 15–32. [CrossRef]
9.
Patterson, H.D.; Thompson, R. Recovery of Inter-Block Information When Block Sizes Are Unequal. Biometrika
1971
,58, 545–554.
[CrossRef]
10.
Piepho, H.-P. Best Linear Unbiased Prediction (BLUP) for Regional Yield Trials: A Comparison to Additive Main Effects and
Multiplicative Interaction (AMMI) Analysis. Theor. Appl. Genet. 1994,89, 647–654. [CrossRef]
11.
Kelly, A.M.; Smith, A.B.; Eccleston, J.A.; Cullis, B.R. The Accuracy of Varietal Selection Using Factor Analytic Models for
Multi-Environment Plant Breeding Trials. Crop Sci. 2007,47, 1063–1070. [CrossRef]
12.
Piepho, H.P.; Möhring, J.; Melchinger, A.E.; Büchse, A. BLUP for Phenotypic Selection in Plant Breeding and Variety Testing.
Euphytica 2008,161, 209–228. [CrossRef]
13.
Searle, S.R.; Casella, G.; McCulloch, C.E. Variance Components, 2nd ed.; Wiley: New York, NY, USA, 2006; ISBN 978-0-470-00959-8.
14.
Wang, R.; Cherkauer, K.; Bowling, L. Corn Response to Climate Stress Detected with Satellite-Based NDVI Time Series. Remote
Sens. 2016,8, 269. [CrossRef]
15.
Kovacevic, V.; Šimi´c, D.; Sootaric, J.; Josipovi´c, M. Precipitation and Temperature Regime Impacts on Maize Yields in Eastern
Croatia. Maydica 2007,52, 301–305.
16.
Pandži´c, K.; Likso, T.; Peji´c, I.; Šarˇcevi´c, H.; Pecina, M.; Šestak, I.; Tomši´c, D.; Mahovi´c, N.S. Application of the Self-Calibrated
Palmer Drought Severity Index for Estimation of Drought Impact on Maize Grain Yield in Pannonian Part of Croatia. Meteorology
2021, 1–23. [CrossRef]
17.
Lobell, D.B.; Hammer, G.L.; McLean, G.; Messina, C.; Roberts, M.J.; Schlenker, W. The Critical Role of Extreme Heat for Maize
Production in the United States. Nat. Clim. Change 2013,3, 497–501. [CrossRef]
18.
Idso, S.B.; Jackson, R.D.; Pinter, P.J.; Reginato, R.J.; Hatfield, J.L. Normalizing the Stress-Degree-Day Parameter for Environmental
Variability. Agric. Meteorol. 1981,24, 45–55. [CrossRef]
19.
Zhu, P.; Zhuang, Q.; Archontoulis, S.V.; Bernacchi, C.; Müller, C. Dissecting the Nonlinear Response of Maize Yield to High
Temperature Stress with Model-Data Integration. Glob. Change Biol. 2019,25, 2470–2484. [CrossRef]
20.
Buhiniˇcek, I.; Kauˇci´c, D.; Kozi´c, Z.; Juki´c, M.; Gunjaˇca, J.; Šarˇcevi´c, H.; Stepinac, D.; Šimi´c, D. Trends in Maize Grain Yields across
Five Maturity Groups in a Long-Term Experiment with Changing Genotypes. Agriculture 2021,11, 887. [CrossRef]
21.
Parent, B.; Leclere, M.; Lacube, S.; Semenov, M.A.; Welcker, C.; Martre, P.; Tardieu, F. Maize Yields over Europe May Increase in
Spite of Climate Change, with an Appropriate Use of the Genetic Variability of Flowering Time. Proc. Natl. Acad. Sci. USA
2018
,
115, 10642–10647. [CrossRef]
22.
Abendroth, L.J.; Miguez, F.E.; Castellano, M.J.; Carter, P.R.; Messina, C.D.; Dixon, P.M.; Hatfield, J.L. Lengthening of Maize
Maturity Time Is Not a Widespread Climate Change Adaptation Strategy in the US Midwest. Glob. Change Biol.
2021
,27,
2426–2440. [CrossRef]
23.
Dwyer, L.M.; Stewart, D.W.; Carrigan, L.; Ma, B.L.; Neave, P.; Balchin, D. Guidelines for Comparisons among Different Maize
Maturity Rating Systems. Agron. J. 1999,91, 946–949. [CrossRef]
24. Jugenheimer, R.W. Corn: Improvement, Seed Production, and Uses; Wiley: New York, NY, USA, 1976.
25.
Publications Office of European Union. Common Catalogue of Varieties of Agricultural Plant Species—Supplement 2020/2.
Available online: http://op.europa.eu/en/publication-detail/-/publication/55e07b9c-6297-11ea-b735-01aa75ed71a1/language-
en (accessed on 20 October 2021).
26.
World Reference Base for Soil Resources. Available online: https://www.fao.org/3/w8594e/w8594e00.htm (accessed on
10 February 2022).
27.
R Core Team. R: A Language and Environment for Statistical Computing. 2020. Available online: https://www.eea.
europa.eu/data-and-maps/indicators/oxygen-consuming-substances-in-rivers/r-development-core-team-2006 (accessed on
20 October 2021).
28.
Bates, D.; Mächler, M.; Bolker, B.; Walker, S. Fitting Linear Mixed-Effects Models Using Lme4. J. Stat. Softw.
2015
,67, 1–48.
[CrossRef]
29.
Lüdecke, D.; Bartel, A.; Schwemmer, C.; Powell, C.; Djalovski, A.; Titz, J. SjPlot: Data Visualization for Statistics in Social Science.
R Package Version 2.8.10. 2021. Available online: https://cran.r-project.org/package=sjPlot (accessed on 21 October 2021).
30.
Nakagawa, S.; Schielzeth, H. A General and Simple Method for Obtaining R2 from Generalized Linear Mixed-Effects Models.
Methods Ecol. Evol. 2013,4, 133–142. [CrossRef]
31.
Nakagawa, S.; Johnson, P.C.D.; Schielzeth, H. The Coefficient of Determination R2 and Intra-Class Correlation Coefficient from
Generalized Linear Mixed-Effects Models Revisited and Expanded. J. R. Soc. Interface 2017,14, 20170213. [CrossRef]
32.
Bernardo, R. Reinventing Quantitative Genetics for Plant Breeding: Something Old, Something New, Something Borrowed,
Something BLUE. Heredity 2020,125, 375–385. [CrossRef]
Agronomy 2022,12, 922 12 of 13
33.
AGRI4CAST Resources Portal. Available online: https://ec.europa.eu/jrc/en/scientific-tool/agri4cast-resources-portal (accessed
on 20 October 2021).
34.
Holzworth, D.P.; Huth, N.I.; Devoil, P.G.; Zurcher, E.J.; Herrmann, N.I.; Mclean, G.; Chenu, K.; Van Oosterom, E.J.; Snow, V.;
Murphy, C.; et al. APSIM—Evolution towards a New Generation of Agricultural Systems Simulation. Environ. Model. Softw.
2014
,
62, 327–350. [CrossRef]
35.
Stepinac, D.; Peji´c, I.; Šimi´c, D. Modeling the Long-Term Response of Yield to Heat Stress for Maize Genotypes of Different
Maturity. In Proceedings of the 56th Croatian and 16th International Symposium on Agriculture, Vodice, Croatia, 16–19 February
2021; Rozman, Z., Antunovi´c, Z., Eds.; Faculty of Agrobiotechnical Sciences: Osijek, Croatia, 2021; pp. 368–372.
36. Moro, J.; Alonso, R.; Rodriguez, A. Variety Trials in Spain. Biuletyn Oceny Odmian 1989,21, 88–104.
37.
Laidig, F.; Drobek, T.; Meyer, U. Genotypic and Environmental Variability of Yield for Cultivars from 30 Different Crops in
German Official Variety Trials. Plant Breed. 2008,127, 541–547. [CrossRef]
38.
Kleinknecht, K.; Möhring, J.; Singh, K.P.; Zaidi, P.H.; Atlin, G.N.; Piepho, H.P. Comparison of the Performance of Best Linear
Unbiased Estimation and Best Linear Unbiased Prediction of Genotype Effects from Zoned Indian Maize Data. Crop Sci.
2013
,53,
1384–1391. [CrossRef]
39.
Sallah, P.Y.K.; Abdula, M.S.; Obeng-Antwi, K. Genotype x Environment Interactions in Three Maturity Groups of Maize Cultivars.
Afr. Crop Sci. J. 2004,12, 95–104. [CrossRef]
40.
Haruna, A.; Adu, G.B.; Buah, S.S.; Kanton, R.A.L.; Kudzo, A.I.; Seidu, A.M.; Kwadwo, O.-A. Analysis of Genotype by Environment
Interaction for Grain Yield of Intermediate Maturing Drought Tolerant Top-Cross Maize Hybrids under Rain-Fed Conditions.
Cogent Food Agric. 2017,3, 1333243. [CrossRef]
41.
Shojaei, S.H.; Mostafavi, K.; Omrani, A.; Omrani, S.; Nasir Mousavi, S.M.; Illés, Á.; Bojtor, C.; Nagy, J. Yield Stability Analysis of
Maize (Zea mays L.) Hybrids Using Parametric and AMMI Methods. Scientifica 2021,2021, 5576691. [CrossRef] [PubMed]
42.
Katsenios, N.; Sparangis, P.; Chanioti, S.; Giannoglou, M.; Leonidakis, D.; Christopoulos, M.V.; Katsaros, G.; Efthimiadou, A.
Genotype
×
Environment Interaction of Yield and Grain Quality Traits of Maize Hybrids in Greece. Agronomy
2021
,11, 357.
[CrossRef]
43.
Bönecke, E.; Breitsameter, L.; Brüggemann, N.; Chen, T.-W.; Feike, T.; Kage, H.; Kersebaum, K.-C.; Piepho, H.-P.; Stützel, H.
Decoupling of Impact Factors Reveals the Response of German Winter Wheat Yields to Climatic Changes. Glob. Change Biol.
2020
,
26, 3601–3626. [CrossRef]
44.
Archontoulis, S.V.; Miguez, F.; Moore, K. Evaluating APSIM Maize, Soil Water, Soil Nitrogen, Manure, and Soil Temperature
Modules in the Midwestern United States. Agron. J. 2014,106, 1025–1040. [CrossRef]
45.
Lobell, D.B.; Gourdji, S.M. The Influence of Climate Change on Global Crop Productivity. Plant Physiol.
2012
,160, 1686–1697.
[CrossRef]
46.
Di Salvo, J.I.; Lee, C.; Salmerón, M. Regional Multi-Environment Analysis of Corn Productivity and Yield Stability as Impacted by
Hybrid Maturity. Field Crops Res. 2021,262, 108025. [CrossRef]
47.
Bonhomme, R.; Derieux, M.; Edmeades, G.O. Flowering of Diverse Maize Cultivars in Relation to Temperature and Photoperiod
in Multilocation Field Trials. Crop Sci. 1994,34, 156–164. [CrossRef]
48.
Field, C.B.; Barros, V.; Stocker, T.F.; Dahe, Q. (Eds.) Managing the Risks of Extreme Events and Disasters to Advance Climate Change
Adaptation—IPCC; Cambridge University Press: Cambridge, UK, 2012.
49.
Malenica, N.; Duni´c, J.A.; Vukadinovi´c, L.; Cesar, V.; Šimi´c, D. Genetic Approaches to Enhance Multiple Stress Tolerance in Maize.
Genes 2021,12, 1760. [CrossRef] [PubMed]
50.
Caubel, J.; de Cortazar-Atauri, I.G.; Vivant, A.C.; Launay, M.; de Noblet-Ducoudré, N. Assessing Future Meteorological Stresses
for Grain Maize in France. Agric. Syst. 2018,159, 237–247. [CrossRef]
51. Liu, Z.; Hubbard, K.G.; Lin, X.; Yang, X. Negative Effects of Climate Warming on Maize Yield Are Reversed by the Changing of
Sowing Date and Cultivar Selection in Northeast China. Glob. Change Biol. 2013,19, 3481–3492. [CrossRef]
52.
Liu, Z.; Yang, X.; Chen, F.; Wang, E. The Effects of Past Climate Change on the Northern Limits of Maize Planting in Northeast
China. Clim. Change 2013,117, 891–902. [CrossRef]
53.
Liu, Y.; Zhang, J.; Pan, T.; Ge, Q. Assessing the Adaptability of Maize Phenology to Climate Change: The Role of Anthropogenic-
Management Practices. J. Environ. Manag. 2021,293, 112874. [CrossRef]
54.
Duvick, D.N. The Contribution of Breeding to Yield Advances in Maize (Zea mays L.). In Advances in Agronomy; Academic Press:
Cambridge, MA, USA, 2005; Volume 86, pp. 83–145.
55.
Berzsenyi, Z.Z.; Dang, Q.L. Effect of Crop Production Factors on the Yield and Yield Stability of Maize (Zea mays L.) Hybrids.
Acta Agron. Hung. 2006,54, 413–424. [CrossRef]
56.
Pepo, P. Effect of Agrotechnical Factors on Soil Chemical Traits and Maize Yield on Chernozem in the Long-Term Experiment.
Plant Soil Environ. 2021,67, 453–459. [CrossRef]
57.
Sudari´c, A.; Šimi´c, D.; Vratari´c, M. Characterization of Genotype by Environment Interactions in Soybean Breeding Programmes
of Southeast Europe. Plant Breed. 2006,125, 191–194. [CrossRef]
58.
Hristov, N.; Mladenov, N.; Djuric, V.; Kondic-Spika, A.; Jeromela, A.; Šimi´c, D. Genotype by Environment Interactions in Wheat
Quality Breeding Programs in Southeast Europe. Euphytica 2010,174, 315–324. [CrossRef]
Agronomy 2022,12, 922 13 of 13
59.
Messina, C.; Hammer, G.; Dong, Z.; Podlich, D.; Cooper, M. Modelling Crop Improvement in a G
×
E
×
M Framework via
Gene–Trait–Phenotype Relationships. Crop Physiol. Appl. Genet. Improv. Agron. 2009,10, 235–265. [CrossRef]
60.
Cooper, M.; Voss-Fels, K.P.; Messina, C.D.; Tang, T.; Hammer, G.L. Tackling G
×
E
×
M Interactions to Close On-Farm Yield-Gaps:
Creating Novel Pathways for Crop Improvement by Predicting Contributions of Genetics and Management to Crop Productivity.
Theor. Appl. Genet. 2021,134, 1625–1644. [CrossRef]
Content uploaded by Domagoj Šimić
Author content
All content in this area was uploaded by Domagoj Šimić on Apr 15, 2022
Content may be subject to copyright.
