Available via license: CC BY
Content may be subject to copyright.
RESEARCH ARTICLE
Towards increased shading capacity: A combined phenotypic
and genetic analysis of rice shoot architecture
Martina Huber
1
| Magdalena M. Julkowska
2
| L. Basten Snoek
3
|
Hans van Veen
1,4
| Justine Toulotte
1,4
| Virender Kumar
5
|
Kaisa Kajala
1
| Rashmi Sasidharan
1,4
| Ronald Pierik
1
1
Plant-Environment Signaling, Utrecht
University, Utrecht, The Netherlands
2
Boyce Thompson Institute, Ithaca, USA
3
Theoretical Biology and Bioinformatics,
Utrecht University, Utrecht, The Netherlands
4
Plant Stress Resilience, Utrecht University,
Utrecht, The Netherlands
5
Sustainable Impact Platform, International
Rice Research Institute, Los Baños, Philippines
Correspondence
Ronald Pierik, Plant-Environment Signaling,
Utrecht University, Padualaan 8, 3584 CH
Utrecht, The Netherlands.
Email: r.pierik@uu.nl
Funding information
This research was funded by the Netherlands
Organisation for Scientific Research (NWO)
(Project Number 14700.RS) in collaboration
with The International Rice Research Institute.
Societal Impact Statement
Rice farming is transitioning from transplanting rice seedlings towards the less
labour-intensive and less water-demanding method of directly seeding rice. This,
however, is accompanied by increased weed proliferation. To tackle this issue, this
study seeks to identify how the crop itself can better suppress weeds, with a focus
on light competition via shading. Using a rice diversity panel, traits were identified
that contribute to enhanced shading capacity, and these traits were encapsulated
into a single shading capacity metric. This was followed by the identification of the
genetic loci underpinning variation in the core traits. The identified haplotypes can
be used in breeding programmes to improve weed suppression by rice, thus contrib-
uting to sustainable agriculture.
Summary
•In modern rice farming, one of the major constraints is weed proliferation and the
entailed ecological impact of herbicide application. This requires increased weed
competitiveness in current rice varieties, achieved via enhanced shade casting to
limit the growth of shade-sensitive weeds.
•To identify traits that increase rice shading capacity, we exhaustively phenotyped
a rice diversity panel of 344 varieties at an early vegetative stage. A genome-wide
association study (GWAS) revealed genetic loci underlying variation in canopy
architecture traits linked with shading capacity.
•The screen shows considerable natural variation in shoot architecture for 13 exam-
ined traits, of which shading potential is mostly determined by projected shoot
area, number of leaves, culm height and canopy solidity. The shading rank, a metric
based on these core traits, identifies varieties with the highest shading potential.
Five genetic loci were found to be associated with canopy architecture, shading
potential and early vigour.
•Identification of traits contributing to shading capacity and underlying allelic varia-
tion will serve future genomic-assisted breeding programmes. Implementing the
presented genetic resources for increased shading and weed competitiveness in
Received: 13 December 2022 Revised: 25 May 2023 Accepted: 26 May 2023
DOI: 10.1002/ppp3.10419
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2023 The Authors. Plants, People, Planet published by John Wiley & Sons Ltd on behalf of New Phytologist Foundation.
Plants People Planet. 2023;1–20. wileyonlinelibrary.com/journal/ppp3 1
rice breeding will make its farming less dependent on herbicides and contribute
towards more environmentally sustainable agriculture.
KEYWORDS
allelic variation, genome-wide association study (GWAS), growth vigour, haplotype analysis,
plant competition, rice diversity panel, shading capacity, shoot architecture
1|INTRODUCTION
Rice feeds more than half of the world's population as a staple food
(Kennedy & Burlingame, 2003; Wing et al., 2018). In traditional rice
farming, seedlings are transplanted into flooded paddy fields. This
works as a natural way to prevent weed infestation since it gives rice
seedlings a size advantage in addition to flood-suppressed germina-
tion and the growth of weeds. This practice is increasingly problem-
atic, both because of the high manual labour input (Chakraborty
et al., 2017; Kumar & Ladha, 2011) and because global climate change
is reducing the availability of fresh water not only for rice farmers but
for the global agricultural sector (FAO, 2019; Oliver et al., 2019). The
traditional rice farming system is transitioning towards direct-seeded
rice, where rice seeds are directly sown into the fields. This practice
drastically reduces the water requirement and labour input (Chauhan
et al., 2017; Farooq et al., 2011; Kumar & Ladha, 2011). Besides all of
its advantages, the major constraint for direct-seeded rice is the abun-
dant proliferation of weeds (Rao et al., 2007; Xu et al., 2019). In
direct-seeded rice (DSR) practice, rice seedlings are directly competing
with weeds as they lose their seedling size advantage. Waterlogging
cannot be applied to suppress emerging weeds, as most modern rice
cultivars do not germinate under water (Chauhan, 2012; Ghosal
et al., 2019; Kretzschmar et al., 2015). Currently, weeds are sup-
pressed with herbicides, leading to the evolution of herbicide-resistant
weeds and groundwater pollution (Heap, 2014; Kraehmer et al.,
2016). This creates a pressing need for the deployment of sustainable
weed management options (Chauhan, 2012; Chauhan & Yadav, 2013;
Mennan et al., 2012; Zhao et al., 2006a). One possible solution to this
problem is to increase the weed competitiveness of the rice seedling
(Chauhan, 2013; Dass et al., 2017; Dimaano et al., 2017; Johnson
et al., 1998; Rao et al., 2007; Sakamoto et al., 2006; Zhao et al., 2007).
Just like their wild ancestors, shade-casting crop varieties com-
pete with invading weeds by reducing the weed's access to full sun-
light, thereby impeding their growth. However, the traits contributing
to shading potential were neglected or even selected against in breed-
ing efforts since tall plants and droopy leaves are generally considered
undesirable because they make harvesting more difficult. Here we
propose to develop weed-competitive rice varieties by selecting for
an ideotype with faster growth and high shade-casting potential on
proximate weeds. A large projected shoot area and therefore ground
cover are associated with weed competitiveness (Caton et al., 2003;
Dingkuhn et al., 2001; Haefele et al., 2004; Mennan et al., 2012;
Namuco et al., 2009; Rao et al., 2007; Zhao et al., 2006b,2007). In
addition, a high number of leaves and tillering capacity, as well as
plant biomass and early vigour are advantageous for competition
against weeds (Haefele et al., 2004; Mahajan & Chauhan, 2013;
Namuco et al., 2009; Zhao et al., 2006a), but these are not specific
architecture traits.
Shoot architecture traits that help plants gain an advantage over
their neighbours through light competition include increased leaf area,
increased planar angle of leaves and tillers and leaf droopiness
(Andrew et al., 2015; Brainard et al., 2005; Dingkuhn et al., 1999;
Mahajan & Chauhan, 2013; Seavers & Wright, 1999; Worthington &
Reberg-Horton, 2013). Accelerated vertical growth might provide an
additional advantage for outcompeting neighbours, yet plant height
has been strongly selected against during the green revolution of most
cereals, including rice. Indeed, increased shading potential of the crop
has a clear potential for improvement towards sustainable weed sup-
pression (Pantazopoulou et al., 2021; Peerzada et al., 2017; Seavers &
Wright, 1999), which also applies for cereal canopies, as has been
shown for wheat and other cereals where a rapidly closing crop can-
opy achieved through higher planting density and/or uniform planting
pattern depleted weeds from access to light (Chauhan & Abugho,
2013; Chauhan et al., 2011; Marín & Weiner, 2014; Ottis & Talbert,
2007; Park et al., 2003; Weiner et al., 2010; Wolfe et al., 2008;Wu
et al., 2021). The critical period of weed competition in rice is from
the moment of sowing up to at least six weeks after sowing in a DSR
system (Abdullah Al Mamun, 2014; Azmi et al., 2007; Chauhan &
Johnson, 2011; Mennan et al., 2012; Raj & Syriac, 2017). Especially in
the context of DSR, shading by the crop canopy would have to occur
early in the season, and especially seedling vigour would substantially
reduce weed growth (Mahajan & Chauhan, 2013; Subedi et al., 2019;
Zhao et al., 2006a). A consequence of such early weed suppression, in
addition to less need for herbicides, would be an increased yield at
the harvest stage of the crop (de Vida et al., 2006; Laca et al., 2006;
Mahender et al., 2015; Namuco et al., 2009; Subedi et al., 2019; Zhao
et al., 2006a).
Building on the idea to increase shading for improved weed com-
petitiveness, here (1) we phenotyped a rice diversity panel of
344 globally distributed varieties where we recorded 13 quantitative
traits. Based on these, (2) we determined key architectural character-
istics of shading potential in the early growth phase. (3) We combined
these core traits into one parameter to develop the shading rank,
where the rice varieties were ranked for their shading potential. (4) A
genome-wide association study (GWAS) revealed association with
eight genetic loci for traits contributing to shade potential. The results
of this study form a primer for the identification of alleles contributing
to increased shading and early plant vigour.
2HUBER ET AL.
2|MATERIALS AND METHODS
2.1 |Plant material
Three hundred forty-four Asian rice (Oryza sativa) varieties were used
out of the rice diversity panel 1 (RDP1) (Eizenga et al., 2014) and one
Oryza glaberrima variety (TOG7192) was included. The RDP1 is a col-
lection of homozygous varieties from 82 countries. The panel includes
landraces and elite rice cultivars from five subpopulations: indica and
aus (of the Indica group), tropical japonica, temperate japonica and aro-
matic (comprising the Japonica group) and the admixture group (Liakat
Ali et al., 2011; Zhao et al., 2011). Detailed information on the full
panel is provided in Table S1.
2.2 |Growth conditions
Rice plants were grown in the screenhouse facilities of the Interna-
tional Rice Research Institute (IRRI) in the Philippines from October
2017 to April 2018. Temperatures ranged from 37C during the day
to 27C at night, with a relative humidity of 75% and 80%, respec-
tively, and a photoperiod of 11 to 12 h. The experiment followed a
randomised block design, with the four replicate blocks separated in
time. Each block contained three individual plants for each of all the
investigated varieties, of which two were measured and harvested,
and the third served as a backup in case of any potential failure. Plants
were grown in single pots at a 30 cm 30 cm distance. In the first
experiment, seeds received from the IRRI gene bank were exposed to
40C for 5 days to break dormancy, followed by 24 h at 21C. For ger-
mination, seeds were put in Petri dishes on wet filter paper and incu-
bated at 32C for 24 h. Seeds were planted directly on the soil: four
seeds were placed per pot (diameter of 16 cm and 13 cm high, with-
out drainage holes), filled with sterilised clay-loam field soil mixed with
NPK fertiliser (with 46/18/60 g per kg soil) and covered with a thin
layer of soil. From planting onwards, the soil was kept moist. Seven
days after sowing (DAS), surplus seedlings were removed, retaining
one seedling per pot. At 14 DAS, fertiliser with 50% N of the concen-
tration of the first application was added. From 15 DAS onwards, the
layer of water was maintained for water-logged conditions.
2.3 |Phenotyping
At 28 DAS, the following traits were measured manually: number of
leaves and tillers, plant height, culm height and length of longest leaf.
Plants were photographed from the top and side using two digital
cameras in a fixed imaging set-up. The shoot was harvested, and dry
weight was recorded after 48 h of drying at 70C (IRRI, 2013). In
Table 1, detailed evaluation methods for each trait are given. Dataset
S1 contains the raw data for each replicate, and Table S2 provides the
results of statistical analysis on phenotypic trait values. Figure S1
depicts scatter plots for all pair-wise phenotype comparisons.
2.4 |Data processing and statistical analysis
For the analysis of RGB images, an automatised image analysis pipe-
line was established using the open-source, Python-based PlantCV
software (PlantCV version 3.7) (Fahlgren et al., 2015; Gehan
et al., 2017). The script was optimised for monocots to extract values
for shoot area, hull area and perimeter. The original Python script can
be accessed at https://plantcv.readthedocs.io/en/stable/, and the
customised Jupiter notebook used in this study is given in Dataset S2.
Tiller angle, leaf angle and leaf erectness were measured using the
free ImageJ software (https://imagej.nih.gov/ij/). Tiller angles were
taken between the two outermost tillers and the culm, respectively.
Similarly, leaf angles were recorded between the second and third
youngest leaves and the culm. Leaf droopiness was measured on the
TABLE 1 Description of 13 investigated shoot architecture and growth traits in rice, with unit of measure and method of measurement.
Trait Unit Description
Number of leaves Number of all visible green leaf blades
Number of tillers Number of side branches classified as tillers as soon as it splits off the culm, having two leaves
Total plant height cm Height from soil to the straightened topmost leaf tip
Culm height cm Mother stem—from soil to highest node, where youngest leaf blade bends off
Leaf length cm Length of longest leaf blade
Projected shoot area cm
2
All green leaf area projected from top view
Convex hull area cm
2
Smallest area enclosing outermost leaf tips
Shoot perimeter cm Outline of the projected shoot area
Leaf initiation angle Angle between culm and leaf blade initiation measured for second and third leaf
Tiller angle Angle between the culm and tillers, measured for the left and right outermost tillers
Leaf droopiness Interception angle of two tangents aligned to initiation and tip of leaf blade measured for second and third leaves
Dry weight shoot g Dry matter of shoot biomass after drying in oven at 70 C for 48 h
Solidity Ratio of projected shoot area divided by convex hull area
HUBER ET AL.3
same leaves, defined as the interception angle of two tangents aligned
to the initiation and the tip of the leaf blade.
In the first block, 62 varieties were excluded as their position
within the greenhouse received shading. In the later three blocks, we
did not use this partially shaded area of the greenhouse anymore to
ensure equal light conditions for all the studied plants. Prior to statisti-
cal analysis, the raw data were curated for outliers (using 1.5*IQR
from the mean), and the mean was calculated out of the eight repli-
cates (four blocks with two plants each). Statistical analysis such as
ANOVA, Pearson correlation and hierarchical clustering were
performed using R (R Version: 3.6.1-1bionic; R Core Team, 2020) and
the online tool MVapp https://mvapp.kaust.edu.sa (Julkowska
et al., 2019). The Pearson correlation coefficients between traits were
calculated using raw data. Trait values were normalised per trait using
the Fisher z-transformation and scaled prior to clustering by applying
Ward D2 hierarchical clustering. For traits that had a correlation >.5,
one core trait was chosen: shoot area, leaf number, solidity, culm
height, leaf angle, tiller angle and leaf droopiness. The shading rank
was calculated as follows:
First, trait values were normalised:
tn
variety ¼
tvariety min tvariety
max tvariety
min tvariety
x100,
where tvariety is the value of a certain trait measured for a certain plant
in the investigated population, and min and max are the minimum and
maximum values of the measured trait in this population, with the
normalised values ranging from 0 to 100.
Next, for each variety, a shading score was calculated:
SSvariety ¼Pcoretraitstn
variety where Pis the sum of the normalised
values of the core traits. From this, we derive the shading rank (SR),
which is the rank given to each variety according to its SS, ordering
the varieties from 1 (lowest) to 344 (highest). The list of 344 varieties
with their normalised core trait values, the sum of normalised core
trait values, and their shading rank can be found in Table S3.
2.5 |GWAS
2.5.1 | Phenotype data
Mean values of all phenotypes were included, only O. glaberrima
TOG7192 was excluded. We tested for normal distribution across the
recorded traits prior to running the GWAS. Table S4 includes the list
for all 344 varieties with 13 shoot trait values (as the mean value out
of eight replicates) used for GWAS, that were cross-correlated in
Figure S1.
2.5.2 | Genotype data
We used the imputed HDRA with 4.8 M SNPs from 3010 O. sativa
varieties, including RDP1, RDP2 and NIAS (McCouch et al., 2016), and
the 3000 Rice Genomes data set (D. R. Wang, Agosto-Pérez,
et al., 2018), available at http://ricediversity.org/data/index.cfm
tools/. The data was curated by filtering for unique SNPs with a 90%
minimum count and a minor allele frequency ≥5%. For the 344 pheno-
typed varieties, this resulted in a total of 1.7 M SNPs used for the
GWAS. As an average genome-wide linkage disequilibrium (LD) decay
in rice, we used previously calculated values (Huang et al., 2010; Zhao
et al., 2011).
We used two different software packages for GWAS just to con-
firm the robustness of any identified associations, and only loci that
were identified by both methods were followed up on afterwards.
The first is an R package (R version 3.6.1) of Genomic Association
and Prediction Integrated Tool (GAPIT) (Li et al., 2014; Tang
et al., 2016; Wang, Tian, et al., 2018). We employed a mixed linear
model (MLM) (Yu et al., 2006) with the optimal number of principal
components based on the calculated Bayesian information criterion
(BIC) for each trait, including as coefficients a kinship matrix
(K-matrix) based on clustering analysis to account for genetic rela-
tionships between individuals, together with the population structure
(Q-matrix). The second software package is lme4QTL. GWAS was
performed as described in Ziyatdinov et al. (2018), taking population
structure into account by using a kinship matrix, and calculated with
the cov() function in R 3.6 (Figure S2). The decomposition matrix to
correct for population structure was made by following the lme4QTL
protocol and using the relmatLmer(), varcov() and decompose_varcov
() functions in order. The obtained decomposition matrix, together
with the traits and binary SNP matrix, is then used in the matlm()
function to calculate the significance and effect per SNP. The full list
of detected significant SNP associations can be found in Dataset S3.
As a confirmation for the reliability of SNP trait associations, we com-
pared the results of the two methods applied here. The Manhattan
plots for all investigated traits using lme4QTL can be found in
Dataset S4 and for GAPIT in Dataset S5). An exact overlap was not
expected, as there are minor differences in the calculation of the
SNP-trait associations: GAPIT uses MLM to create the SNP-trait
association and a principal component analysis that we set to “opti-
mal”to calculate the optimal number of PCs needed to describe the
variation in population structure. The lme4QTL is a mixed model
solver (Linear Mixed Effects 4 QTL) that can use a kinship matrix as a
covariance to control for population structure. The narrow-sense
heritability (h
2
) of the analysed traits was calculated with GAPIT
(Table S5). To set the significance threshold, the rather conservative
Bonferroni correction was applied, calculated by the –log
10
(p-value
of .05/ΣSNPs), which corresponds to log10(.05/1.700.000)
=7.53 for the imputed HDRA data set. To examine the GWAS
model performance and estimate possible model overfitting, QQ plots
were generated (Dataset S6).
2.6 |Post-GWAS analysis
For all follow-up analysis, the output of the GWAS using the lme4QTL
method on the raw, untransformed phenotype data was used.
4HUBER ET AL.
2.6.1 | Locus definition
Loci that were detected with both methods (lme4QTL and GAPIT)
were considered most robust and subsequently used as a basis for the
selection of loci of interest and follow-up analysis. A locus of interest
was determined if several significantly associated SNPs were found in
close proximity. Single SNPs passing the threshold were neglected
since whole-genome sequencing data provides sufficient markers in
each LD block. Rice has a low rate of LD decay, which makes the iden-
tification of causal genes more difficult (Wang et al., 2020). Therefore,
the local LD analysis was used to define LD clumps surrounding the
index SNPs using LD clumping in PLINK (http://zzz.bwh.harvard.edu/
plink/). We set three selection criteria: strong LD between SNPs, a p-
value threshold of .01 and a physical distance of 250 kb, given with
the R
2
value. We considered SNPs with –log
10
(p-value) > 5 as index
SNPs to perform the analysis and clumped SNPs with a p-value > 4.
For the determination of loci of interest, we focused on the core traits
culm height, shoot area, solidity and number of leaves. For culm
height and number of leaves, single significant SNPs were not found
to be surrounded by other significant SNPs within LD and therefore
did not meet our selection criteria. Since dry weight was highly corre-
lated with the traits of branchiness, the loci found for dry weight were
included as representatives for branchiness, and similarly, the loci for
plant height were included as representatives of height-related traits.
2.6.2 | Gene models
Genetic regions covered by significant SNPs were searched for candi-
date genes using two different gene annotation models, which were
then merged: the Michigan State University (MSU; October
31, 2011—Release 7; http://rice.plantbiology.msu.edu/) and the Rice
Annotation Project Database (RAP-DB; March 24, 2020; https://
rapdb.dna.affrc.go.jp/). Other data resources used were the gene ID
converter (https://rapdb.dna.affrc.go.jp/tools/converter), GALAXY—
rice genome browser (http://13.250.174.27:8080/?tool_id=
getgenes&version=1.0.0&__identifer=pxuu9t4bnk) and SNP seek
(http://snp-seek.irri.org/).
2.7 |Haplotype analysis
To facilitate the identification of candidate genes within the detected
loci related to canopy architecture, we performed haplotype analysis
spanning the coding sequence regions of the genes within each locus.
Since haplotype analysis can only be performed for annotated genes,
SNPs that are not in a coding sequence cannot be included in haplo-
type analysis. For each locus, we used the combined gene model
annotation (MSU and RAP-DB) to identify the coding sequences of
the individual genes (Table S6). Subsequently, all SNPs within the cod-
ing sequence region were compiled into one haplotype, and all varie-
ties were grouped based on their haplotype sequence, with
haplotypes represented by two or less varieties being excluded from
the analysis. If significant SNPs were too far away to be within LD,
they were not defined as loci and were not included in further haplo-
type analysis. Based on the haplotype grouping for each coding
sequence, we performed an ANOVA followed by a post-hoc test for
significant differences between the haplotypes for each measured
trait. The individual haplotypes are represented by A/T, where A
stands for a variant of the reference accession and T for any alterna-
tive variant. Dataset S7 shows haplotype groups for all determined
loci of interest with their phenotype effects for 13 investigated shoot
traits. For more information on all investigated loci, consult Dataset
S8, which contains the full list of haplotype sequences within the
defined loci of interest.
2.8 |Canopy shading experiment
Rice plants were grown in the greenhouse facilities of Utrecht
University, The Netherlands, in February 2021. Temperatures were
set to 29C during the day and 25C during the night, with a
photoperiod from 8 a.m. to 8 p.m., with a minimal light intensity of
400 μmol m
2
s
1
and artificial light (Valoya, Model Rx400 500 mA
5730, Spectrum AP673L) switching on if sunlight dropped below
400 μmol m
2
s
1
. Automatic watering kept the soil water-saturated.
The selected O. sativa varieties were Shim Balte, Mudgo, Della and
Luk Takhar, with shading ranks of 344, 330, 49 and 1, respectively.
The germination protocol was followed as described above. Four
plants were grown per pot in each of the corners of a square pot
(10 10 11 cm) in a substrate mix of black soil, vermiculite and
sand in a ratio of 5:3:2 together with 6 g of Osmocote slow-release
fertiliser and 1 L of Yoshida nutrient solution per kg of substrate. Pots
were arranged at a distance of 10 cm in mixed plots. The experiment
units (the eight plants that were measured) were surrounded by bor-
dering plants to avoid border effects on the experimental units. Light
intensity (photosynthetic active radiation [PAR] of the 400–700 nm
waveband) was measured at ground level between rice plants (with
six measurements in each of the three replicates) and above the plants
for reference at the same time to calculate light extinction. PAR values
can be found in Table S7.
3|RESULTS
3.1 |Shoot architectural variation between rice
varieties
In order to evaluate the variation in shading potential within the rice
diversity panel (Table S1), we measured 13 traits on 4-week-old seed-
lings in the screenhouse (Figure 1, Table 1, Table S2).
Substantial variation was observed for all measured traits among
the varieties belonging to different subpopulations (Figure 1;
Table S2). The indica subpopulation showed the highest dry weight,
number of leaves and number of tillers, followed by the aus subpopu-
lation, and aromatic,tropical and temperate japonica ranked lowest for
HUBER ET AL.5
these parameters (Table S2). Shoot and hull area were also observed
to be higher in indica and aus subpopulations, intermediate in aromatic
subpopulation, and lowest in japonicas and admixture subpopulations.
Indica and aus on average develop the most compact shoots (highest
solidity), contrasting with the low solidity of japonicas and admixed.In
culm height, indica and aus were shortest, and temperate japonica and
admixed subpopulation were tallest. When taking the entire diversity
panel of 344 varieties, five traits (shoot area, hull area, solidity, plant
height and dry weight) already showed a significant difference
between the individual varieties at 4 weeks after sowing (Table S2).
When grouped together in subpopulations, all traits showed signifi-
cant differences between subpopulations (Table S2). Overall, it
appears that relatively large variation between subpopulations was
observed for traits related to area and branchiness, whereas traits
related to height showed only little variation between subpopulations.
These differences are clearly determined by differences in genetic
background since the growth conditions were constant. The high vari-
ation observed for traits related to shading potential suggests that the
investigated rice diversity panel offers the genetic variation needed to
inspire improvement of shading potential in elite-breeding varieties.
3.2 |Correlation of shoot architectural traits
To explore the relationship between individually measured traits and
determine which traits are independent of each other, we performed a
FIGURE 1 Shoot traits in rice differ between subpopulations. Distribution of investigated shoot traits in the screened diversity panel. The
plots represent the trait value (y-axis) observed for varieties grouped according to different subpopulations on x-axis. (a) Shoot area [cm
2
], (b) hull
area [cm
2
], (c) perimeter [cm], (d) solidity, (e) dry weight [g], (f) number of leaves/plant, (g) number of tillers/plant, (h) plant height [cm], (i) leaf
length [cm], (j) culm height [cm], (k) leaf angle [], (l) tiller angle [] and (m) droopiness []. Each data point represents the mean out of eight
replicates for each of the 344 varieties. The colours represent different groups of subpopulations: ind,indica;aus;adm,admixed; aro, aromatic;trj,
tropical japonica and tej,temperate japonica. The letters in the graphs represent the significantly different groups, as determined with Tukey's HSD
with p-value <.05, error bars indicate standard deviation, horizontal bars indicate the median. Mean values for all 13 traits and the sum of the
normalised traits including results for Tukey's pairwise post hoc test can be found in Table S2.
6HUBER ET AL.
Pearson correlation analysis (Figure 2a, Figure S1). Shoot area and hull
area showed a strong positive correlation with shoot dry weight. Leaf
and tiller numbers were highly correlated with shoot area and dry
weight. Height-associated traits, such as plant height, culm height and
leaf length, were positively correlated with each other. On the other
hand, a negative correlation was found between culm height and the
number of leaves and tillers. Solidity, leaf angle, tiller angle and droopi-
ness did not display strong correlations with other measured traits.
To examine the types of canopy architecture exhibited within the
rice diversity panel, we performed hierarchical clustering (Figure 2b),
resulting in seven trait clusters. The clustering shows how traits are
grouped together according to the patterns observed across all rice
varieties. Taking the correlation and clustering analyses together, we
can classify core groups of traits: area-related (shoot area, hull area,
perimeter), branchiness (number of leaves and tillers and dry weight),
height-related (plant and culm height and leaf length), solidity, leaf
angle, tiller angle and droopiness (Table 2).
3.3 |Defining ‘shading potential’from shoot
architecture traits
The shading potential of a plant expresses the effectiveness with
which it can cover ground area. The traits chosen to be included in a
rank that stands for shading potential were included based on litera-
ture research (Andrew et al., 2015; Caton et al., 2003; Dingkuhn
et al., 2001; Haefele et al., 2004; Mahajan & Chauhan, 2013; Mennan
et al., 2012; Namuco et al., 2009; Rao et al., 2007; Worthington &
Reberg-Horton, 2013; Zhao et al., 2006b,2007) together with a gen-
eral understanding of plant architecture and light extinction. From the
large set of shoot architectural traits, we defined core traits based on
clustering and correlation analysis (Figure 2a), revealing which traits
are closely linked and would therefore be overrepresented if all taken
into account with the same weight for ranking the varieties. Core
traits are chosen as representative of a correlating group of traits. To
quantify shading potential, we ranked varieties for the sum of the core
traits contributing to shading potential (projected shoot area, number
of leaves, solidity, culm height, leaf angle, tiller angle and leaf
FIGURE 2 Correlation and clustering of 13 shoot growth and architecture traits in rice highlights trait groups that define the core trait
groups. (a) Pearson correlation coefficients between traits. The colour and size of the circles reflect the strength of the correlation (see colour
scale below diagram). (b) Hierarchical cluster analysis. Traits are clustered using ward. D2 method. Rows represent 13 studied shoot traits, the
length of branches is proportional to the distance in clustering. The values of individual samples are normalised per trait using Z-Fisher
transformation scaled prior to clustering. Based on a cut off at seven clusters and together with the correlation coefficients, we grouped together
the traits into defined core groups.
TABLE 2 Core groups of shoot traits in rice. Measured shoot
architecture traits were grouped based on their correlation and
clustering. For core groups with multiple traits, we have selected a
representative trait as the core trait, indicated in bold.
Core groups Measured shoot architectural traits
Area Projected shoot area, convex hull area, perimeter
Branchiness Number of leaves, number of tillers, dry weight
Height Culm height, leaf length, plant height
Solidity Solidity
Leaf angle Leaf angle
Tiller angle Tiller angle
Droopiness Droopiness
HUBER ET AL.7
droopiness, bold in Table 2). Each of the traits was then normalised so
as to become comparable across traits that have different magnitudes
and measured units by rescaling the values to a range from 0 to
100 while keeping the relative differences of trait values between dif-
ferent varieties unchanged, and these relative differences are also
reflected in the sum of the normalised trait values. Varieties were then
ranked according to their sum of normalised trait values, from
344 (highest) to 1 (lowest), resulting in the shading rank (for detailed
information, see Section 2.4). The resulting shading ranks within this
diversity panel are shown in Table S3. Since the diversity panel was
evaluated 28 days after sowing, a large shoot size in high-ranking vari-
eties also indicates rapid growth and seedling vigour. From the
25 highest-ranking varieties, 14 belong to the indica subpopulation
and eight to aus. Low-ranking varieties in terms of shading potential
include widely grown varieties such as IR 64 and Nipponbare
(Acevedo-Siaca et al., 2020), ranking 74th and 73rd, respectively
(Table 3). This suggests that some of the current elite rice varieties
could have a rather poor shading potential, and through breeding with
varieties from indica and aus subpopulations, the shading potential
and weed competitiveness can possibly be increased.
The distribution of the different varieties with respect to the core
trait groups area, branchiness, height and solidity is shown in Figure 3,
together with top images of representative varieties. None of the top-
ranking varieties showed the highest values for all core shading traits
(Figure 3), hinting at trade-offs between shading traits. For example,
Sze Guen Zim ranks highest for shoot area and number of leaves but
is one of the lower-ranking varieties for culm height. The variety with
the highest shading rank (344), Shim Balte, has a very high number of
leaves and solidity but has a close to average culm height. Mudgo
reaches a rank of 340 despite its relatively low number of leaves and
TABLE 3 Shading rank for 10 highest and 10 lowest ranking rice varieties, and for varieties of special interest (Mudgo, IR 64-21, Nipponbare
and Della) with normalised core trait values (between 0 as lowest and 100 highest) and the sum of the core traits. Varieties in bold are visualised
in Figure 3. The shading rank ranges from 344 as the highest and 1 as the lowest shading. The list of shading ranks for the entire panel can be
found in Table S3. Subpopulations: ind,indica;aus;adm,admixed; aro, aromatic;trj,tropical japonica and tej,temperate japonica; norm, normalised
trait value without unit; SUM_norm_traits, sum of normalised trait values.
Variety Subpopulation
Shoot
area
norm
Number
of leaves
norm
Solidity
norm
Culm
height
norm
Leaf
angle
norm
Tiller
angle
norm
Droopiness
norm
SUM_
norm_
traits
Shading
rank
SHIM BALTE aus 78 85 73 86 94 65 79 561 344
SZE GUEN ZIM ind 100 100 95 38 15 55 67 470 343
PARAIBA CHINES
NOVA
ind 77 55 51 64 25 100 90 462 342
P 737 aus 91 56 69 84 42 49 68 458 341
SHIRKATI aus 93 61 68 51 8 85 80 446 340
SABHARAJ ind 94 78 63 54 23 57 73 443 339
PAUNG MALAUNG aus 89 56 97 52 16 45 85 440 338
NIRA ind 80 64 56 47 32 70 82 431 337
SATHI aus 67 59 66 73 22 52 81 420 336
MTU9 ind 86 46 57 79 19 48 82 417 335
MUDGO ind 73 30 57 79 20 53 95 407 330
IR 64-21 ind 16 59 41 13 16 32 78 254 74
NIPPONBARE tej 19 25 52 25 13 42 77 253 73
DELLA trj 11 6 12 38 66 46 56 234 49
COCODRIE trj 10 11 22 39 23 26 38 168 10
L 202 trj 1 10 9 27 14 44 61 166 9
TRIOMPHE DU
MAROC
tej 2 10 51 52 22 25 2 165 8
S 4542 A 3-49B-
2-12
trj 4 8 7 48 5 43 43 159 7
TAINAN IKU 487 tej 5 24 38 36 12 19 19 154 6
PI 298967-1 adm 5 11 1 42 17 34 34 143 5
SHIROGANE tej 4 17 14 19 12 34 43 142 4
BUL ZO tej 10 8 20 45 22 21 11 137 3
GUINEANDAO adm 10 14 9 38 8 40 9 127 2
LUK TAKHAR tej 3 8 26 17 5 44 0 103 1
8HUBER ET AL.
solidity. At the other end of the spectrum, Della ranks 49 and is low
for all traits except for culm height, whereas Luk Takhar (rank 1)
shows low values for all core traits.
3.4 |Predicted competitive varieties are casting
more shade
To validate our shading rank and assess functional shading capacity,
we grew varieties with varying shading rank and evaluated them for
canopy shading. We selected two of the predicted competitive (Shim
Balte with a Shading Rank of 344 and Mudgo ranking 330) and two
predicted non-competitive rice varieties (Luk Takhar ranking 1 and
Della ranking 49) (Figure 4, Table 3). By measuring the light quantity
under the canopies of selected varieties (Table S4), we indeed
observed strong shading by varieties with a high shading rank (Shim
Balte and Mudgo) and less severe shading by varieties with a low
shading rank (Luk Takhar and Della). This result validates our shading
rank, at least for the varieties tested and the selection of shoot archi-
tecture traits to effectively predict shade casting.
3.5 |SNPs associated with seedling establishment
and shoot architectural traits
The high phenotypic variability found in the studied diversity panel
(Table S4), together with the high genetic variation (Wang,
Longkumer, et al., 2018), provides a strong basis for a GWAS. We
FIGURE 3 Shoot architecture traits associated with shading capacity and projected shading in the investigated rice diversity panel. (a–d)
Scatter plots showing the distribution of 344 rice varieties in pair-wise combination of four core traits of the shading rank, shoot area, number of
leaves, solidity and culm height. Representative high (344, 343 and 330) and low (49 and 1) ranking varieties together with Nipponbare (73) and
IR 64-21 (74) are highlighted in colours. (e) Top view images of representative varieties, with colour coded frames. Numbers are respective
shading ranks as found in Table 3.
HUBER ET AL.9
observed high narrow-sense heritability for all measured traits
(Table S5). We investigated the genomic trait associations on the lat-
est available SNP set at the time of analysis, with 4.8 M SNPs from
3010 O. sativa varieties (McCouch et al., 2016, D. R. Wang, Agosto-
Pérez, et al., 2018), with two different software packages (lme4QTL
(Ziyatdinov et al., 2018) and Genomic Association and Prediction Inte-
grated Tool (GAPIT) (Tang et al., 2016; Wang, Tian, et al., 2018); see
methods for a detailed description). The total list of p-values for SNPs
association across all measured traits can be found in Dataset S3,
resulting from the lme4QTL package.
Despite solidity being a very complex and likely a polygenic trait,
the analysis revealed a strong association with 14 SNPs in the locus
on chromosome 3 (Figure 5). Three genomic regions were associated
with plant height, located on chromosomes 3, 5 and 6 (Figure 5). The
peak on chromosome 3 was also detected for other height-related
traits: culm height and leaf length (Dataset S4). Overall, the associa-
tions with culm height showed lower LOD scores (Dataset S4), and
thus we followed up the loci in plant height. The results for droopi-
ness reveal strong associations with SNPs on chromosomes 1 and
10, sharing the association on chromosome 1 with tiller angle
(Figure 5, Dataset S4). Leaf angle could be associated with a highly
significant peak of SNPs on chromosome 12. The associations
between leaf or tiller number found for SNPs on chromosomes
11 and 12 were shared between these two traits (Dataset S4). These
two loci were also found for dry weight. This suggests that the genetic
components underlying the formation of new leaves and tillers might
have a common genetic constituent, consistent with the high correla-
tion in their phenotypes (Figure 2). The analysis for dry weight
revealed significant associations on chromosomes 3, 7 and 12, overlap-
ping with the associations found for shoot area (Figure 5). The strong
accumulation of significantly associated SNPs on chromosome 1 were
also found to be associated with solidity, shoot area and dry weight,
representing three of the core traits. When taking shading potential
together as the sum of all core traits, a GWAS on this composite trait
yielded a rather random pattern of SNP associations (Figure S4). This
further highlights our earlier findings (Table 3), that shading can be
achieved through various strategies, and shading potential, as such, is
genetically a highly complex trait. Therefore, genetic mapping of shoot
architecture components that contribute to shading capacity is an
effective approach to identifying genetic components that contribute
to shading and potential weed competitiveness.
3.6 |Identification of alleles associated with
increased shading potential
The genomic regions that consisted of multiple SNPs above the
Bonferroni threshold within the calculated local average LD (Table 4)
were investigated in more detail. Because the traits related to canopy
shading potential are the primary focus of this work, we prioritised
the loci associated with culm height, shoot area, solidity and dry
weight (Figure 5). In total, we determined six loci to be followed up
with a haplotype analysis to identify specific alleles that could con-
tribute to traits determining shading potential. By grouping varieties
according to SNPs within one coding region and examining the phe-
notypic differences between identified haplotypes, we identified alle-
lic variation associated with high shading potential (Figure 6). On
chromosome 1, we found two loci, the first one for droopiness
(Figure 6a) in the coding sequence for a GTP-binding protein
(Os01g0225200), where one haplotype (hap03) had a significantly
lower droopiness compared with all others. The second locus on
chromosome 1 was found to overlap between shoot area, solidity
and dry weight (Figure 6d–f) in a sequence encoding a protein with
protein phosphorylation function (Os01g0810800). The haplotypes
FIGURE 4 Canopy shading capacity is
consistent with shading rank of high and low
ranking rice varieties. Significant difference in light
extinction between canopies of different rice
varieties at 5 weeks after sowing. The plot shows
the reduction in light intensity (%PAR) measured
at the ground level under the rice canopy
compared to above the canopy, for different rice
varieties on x-axis, where Della and Luk Takhar
were classified as non-competitive (blue) with
shading ranks of 49 and 1, respectively, and
Mudgo and Shim Balte as competitive (green) with
shading ranks of 330 and 344, respectively.
Letters indicate significance (ANOVA with Tukey's
pairwise comparison post hoc test p< .05).
Measured PAR values (photosynthetic active
radiation of 400–700 nm waveband) can be found
in Table S7.
10 HUBER ET AL.
FIGURE 5 Genome-wide association study (GWAS) identifies the genetic regions underlying shoot architectural traits and seedling vigour in
4-week-old rice seedlings, reflecting the early vegetative growth stage. We used single-trait GWAS with a mixed linear model (MLM) for solidity,
plant height, shoot area, dry weight, droopiness and leaf angle. The Manhattan plots depict the single nucleotide polymorphisms (SNPs) with
minor allele frequencies (MAFs) >.05. Negative logarithmic p-values on the y-axis, for 1.7 M SNPs across the 12 rice chromosomes along the
x-axis. Genomic regions highlighted in green are loci of interest (numbered L1.1-L12). These results were generated by using the software
package lme4QTL.
HUBER ET AL.11
of three coding regions in locus 3 (Figure 6g–i), associated with
solidity, were observed to have significantly lower solidity than
the most abundant haplotype. These are annotated as a FAR1
domain-containing protein (Os03g0843700), Pirin-like protein
(Os03g0845000) and a RPB17 fragment (Os03g0845700). Locus
5 (Figure 6j–k), associated with plant height, includes Cytochrome C
(Os05g0420600) and a conserved hypothetical protein
(Os05g0420900), where for both, the most abundant haplotype was
linked to the shortest plants. In locus 7, associated with shoot area
and dry weight, we found that only one gene (Os07g0623200, anno-
tated as ATPase and heavy metal transporter protein) showed clear
separation across the haplotypes, where all the non-reference haplo-
types showed higher shading potential, indicated by higher shoot area
and dry weight (Figure 6l,m). Within locus 11, associated with solidity
and dry weight (Figure 6b,c), there is only one gene located, encoding
a pyruvate kinase family protein (Os11g0216000). We found that the
second most abundant haplotype was associated with increased
shading due to the higher dry weight of varieties that shared this spe-
cific combination of SNPs.
We then summarised these main haplotypes into a table where
we express their contribution (positive/neutral/negative) to the
expression of the core traits that we used to compile the shading rank.
This facilitates a direct impression of how specific alleles, detected as
haplotypes as mentioned above, affect all the core traits that together
determine the shading rank (Figure 7). As such, this would facilitate
decisions on which alleles would be expected to be a potential source
for improving shading potential in elite varieties. From this integrative
table, it becomes clear that a number of haplotypes are highly
desirable. Hap2 from locus L-1.1 (Os01g0225200), hap2 from locus
L-1.2 (Os01g0810800), hap3, hap4 and hap5 from locus L-3
(Os03g0843700) and hap3 from another gene in locus L-3
(Os03g0845000) are all alleles that positively affect values of multiple
core traits that contribute to shading potential without negatively
affecting any of the other core traits. Since these alleles are not highly
abundant in the studied population (Figure 6), it is highly likely that
they are not represented in the current elite varieties. From this table,
it also becomes clear that allelic effects on trait expression are quite
consistent between shoot area, dry weight and leaf number. Alleles
that are positive for these traits, however, are often negative or neu-
tral for culm height. Furthermore, all alleles that seem to stimulate
culm height reduce expression of (multiple) other core traits, indicat-
ing that within this population there is little potential for improvement
TABLE 4 Summary of determined loci and genes of interest of core rice trait groups for shading potential with index SNPs (significant SNPs
with LOD >5) and locus span in kb (clumped SNPs with LOD >4 in local LD up- and downstream), with the locus ID and gene annotation. Genes
represented in Figure 6are highlighted in bold. Full list of SNP positions in loci of interest with gene annotation and gene ontology categories can
be found in Table S6. These results were generated by using the software package lme4QTL. Chr, chromosome; SNP, single nucleotide
polymorphism; index SNP_ID, ID of identified index SNP within the span of the locus; Locus_ID, annotated for Oryza sativa.
Trait Locus Chr Index SNP_ID
Locus
span [kb] Locus_ID Gene annotation
Droopiness L-1.1 1 1.01395336 49 Os01g0225200 Predicted protein; BP: GTP binding
Shoot area L-1.2 1 1.07664139 6 Os01g0810800 Hypothetical conserved gene; BP: protein
phosphorylation
Dry weight
Solidity
Solidity L-3 3 3.35500735 404 Os03g0841800 GSK3/SHAGGY-like kinase
Os03g0841850 Hypothetical protein.
Os03g0843700 FAR1 domain containing protein
Os03g0845000 Similar to Pirin-like protein
Os03g0845700 Similar to RPB17 (fragment)
Os03g0845800 Conserved hypothetical protein
Os03g0848700 Coiled-coil, nucleotide-binding, and leucine-rich
repeat protein
Plant height L-5 5 5.20612311 59 Os05g0420500 Conserved hypothetical protein
Os05g0420600 Cytochrome c
Os05g0420900 Conserved hypothetical protein
Plant height L6 6 6.13994152 240 Os06g0269300 TolB-like domain containing protein
Os06g0346300 Acyl-CoA oxidase/oxidoreductase
Shoot area L-7 7 7.25787749 146 Os07g0623200 Heavy metal transporter protein; ATPase, P-type
Dry weight 7 Os07g0623501 Hypothetical gene
7 Os07g0623600 Similar to mRNA, clone: RTFL01-43-H20
Dry weight L-11 11 11.6059294 23 Os11g0216000 Pyruvate kinase family protein
Leaf angle L-12 12 12.085063092 1 Os12g0557800 Pentatricopeptide repeat domain containing protein
12 HUBER ET AL.
FIGURE 6 Haplotype groups and the associated effect on expression of phenotypes of core traits for shading capacity in rice. L-1.1 was
detected for (a) droopiness with haplotypes in the coding sequence of the gene Os01g0225200 for a GTP binding protein. L11 was detected for
(b) solidity and (c) dry weight with haplotypes in the coding sequence of the gene Os11g0216000 for a pyruvate kinase family protein. L1.2 was
detected for (d) shoot area, (e) solidity and (f) dry weight with haplotypes in the coding sequence of the gene Os01g0810800 for a protein with
the function of protein phosphorylation. L3 was detected for solidity, encoding the genes (g) Os03g0843700 for a FAR1 domain containing
protein, (h) Os03g0845000 for a Pirin-like protein and (i) Os03g0845700 for an RPB17 fragment. L5 was found for plant height, encoding
(j) Os05g0420600 for cytochrome C and (k) Os05g0420900 a conserved protein. L7 was found for (l) shoot area and (m) dry weight encoding
only one gene Os07g0623200 for a heavy metal transporter protein. Dot plots with letters representing the significantly different groups, as
determined with Tukey's HSD with p-value <.05. Y-axis trait value, x-axis groups of haplotypes, the most abundant haplotype is highlighted in
blue. Additional information about the detected genes can be found in Table 4and dot plots for haplotypes for all 13 traits found in loci of
interest are shown in Dataset S7.
HUBER ET AL.13
of plant height without undesirable trade-offs with other traits.
Although only a few of the detected alleles affect leaf angles, the ones
that do have the opposite effect on tiller angles.
4|DISCUSSION
We studied phenotypic and genetic variation in rice shoot architec-
ture to identify traits and their underlying genetic loci that contribute
to canopy shading. We investigated variability across a natural rice
diversity panel in shoot architecture at the early vegetative stage. The
traits investigated here encompass both early vigour and shade cast-
ing through shoot architecture, which are linked to weed suppression
in rice fields (Andrew et al., 2015; Brainard et al., 2005; Mahajan &
Chauhan, 2013; Seavers & Wright, 1999; Worthington & Reberg-
Horton, 2013). It is well documented that weeds are substantially
affected in their growth by shading and effectively suppressed by rice
canopy shading (Chauhan, 2012,2013; Koarai & Morita, 2003). It has
also been shown that early weed suppression correlates with
enhanced yield (Mahender et al., 2015; Namuco et al., 2009; Subedi
et al., 2019; Zhao et al., 2006a). Traits related to shoot architecture,
such as leaf angle or droopiness, are of special interest as they do not
require substantial resource investment while creating more optimal
3D canopy distribution of the shoot biomass for increased shading
potential. Other traits, such as leaf area, number of leaves or shoot
biomass, likely require considerable resource investments and are
typically associated with growth vigour, that is, rapid seedling
establishment.
4.1 |Shoot architecture traits for shading
Shading potential can be defined in two-dimensional measures, such
as ground cover or projected shoot area, including a third dimension
where plant height is considered as space resource utilisation (Zhang
et al., 2019). We identified how all measured traits are related to one
another and identified core traits that capture the observed variance
(Figure 2). We identified groups of traits related to branchiness (num-
ber of leaves and tillers) and height (plant height, culm height and leaf
length) and added solidity, leaf and tiller angles and droopiness to cal-
culate a shading rank. The varieties with the highest shading rank
belong to the indica and aus subpopulations, which have also been
found in earlier studies to have higher yields and less weed biomass in
weedy fields compared with japonicas (Zhao et al., 2006b). We found
admixed, tropical japonica and aus subpopulations to typically range
between temperate japonica and indica. This pattern could be found in
the majority of the measured traits and is in line with the phylogenetic
relatedness of the different subpopulations (Eizenga et al., 2014;
Liakat Ali et al., 2011; McCouch et al., 2016; Zhao et al., 2011). Early
seedling vigour is particularly important for weed competition during
FIGURE 7 Haplotype effects on all core trait phenotypes in rice. The heatmap shows for each specific haplotype within each gene of
interest, an increased trait value in yellow, no significant change in grey and a decreased value in blue. The most abundant haplotype within the
studied population and the haplotype comprising the reference genome are indicated. BP: biological process. Additional information about the
detected genes in loci of interest can be found in Table 4.
14 HUBER ET AL.
the critical period of weed control, and some of the high-ranking vari-
eties, such as Shim Balte, Paung Malaung and Sabharaj, are also
known by breeders for their early vigour. Increased shading ability is
intrinsic to early vigour because it follows to some extent from large
size. However, the shading rank proposed here is more comprehen-
sive to additional traits such as solidity and plant architecture that
may involve less resource investment than vigour traits. With this
improved way of ranking a plant's shading capacity, our study exem-
plifies a new method of selection for high-shading varieties and
genetic loci associated with traits that contribute to high-shade can-
opy architecture. It also helps to narrow down any selection to a mod-
est number of core traits, making phenotyping more efficient. We
propose that varieties that have a high shading rank, are potentially
more weed-competitive varieties, whereas those that rank low are
likely to be weaker competitors. Indeed, our experiment proved that
canopies of high-ranking varieties allow significantly less light penetra-
tion than low ranking ones (Figure 4).
The correlations between traits encapsulated within each of the
trait groups that were used for the shading rank often underlines nat-
ural growth patterns; the more tillers a plant has, the more leaves it
bears since each tiller has the potential to develop a certain number
of leaves. A strong correlation was previously observed between tiller
formation and relative growth rate (Dingkuhn et al., 2001). Likewise,
in our study, the number of leaves and leaf area were positively corre-
lated with shoot dry weight (Figure 2, Figure S1). This well-established
relationship (Caton et al., 2003; Dingkuhn et al., 2001; Poorter
et al., 2012) probably follows from a larger shoot area providing a
higher capacity for photosynthesis and thereby leading to a higher
overall growth rate (Caton et al., 2003). However, not all traits
showed expected correlations; while solidity is the ratio of shoot area
and hull area, it is only weakly correlated with shoot area (Figure 2,
Figure S1). This suggests that shoot solidity is independent of how
large its total shoot area, leaf number or angles are. Since solidity indi-
cates the uniformity of the plant's ability to shade its circumference, it
is a valuable trait for shading capacity analysis; a large projected shoot
area with low solidity would still leave many open spaces within a sin-
gle plant's sphere for light penetration where weeds can proliferate.
Inverse correlations were found between branchiness (number of
leaves and tillers) and height traits. This trade-off between height and
branching is well documented as apical dominance, where the height
growth of the main shoot is promoted at the expense of branching
(Roig-Villanova & Martínez-García, 2016; Teichmann & Muhr, 2015).
Summarising, the trends observed within this study are in line with
earlier observations, whereas we identify new, informative trait
groups that allow interpretations at the canopy level and that contrib-
ute independently to the shading potential of rice plants.
4.2 |Elucidating the genetic components of
shading potential
We screened a large diversity panel representing different subpopula-
tions, which adds new information to several available studies on
specific subpopulations or recombinant inbred lines (Cordero-Lara
et al., 2016; Hoang et al., 2019; To et al., 2019; Wang et al., 2011).
4.2.1 | Architecture
The SNP dataset from the rice diversity panel (Eizenga et al., 2014)
was combined with the observed phenotypic variation to identify
putative genetic loci underlying high shading potential. This varia-
tion (Figure 1, Table S4), together with a high trait heritability
(Table S5), provides a strong basis for GWAS. Plant height and leaf
length were associated with loci on chromosomes 5 and 6. The
locus on chromosome 5 harbours two genes encoding cytochrome
C and a conserved hypothetical protein. The haplotype analysis
revealed one allele for both genes that was associated with a highly
significant increase in plant height (Figure 6). The locus on chromo-
some 6 encodes the heading date (Hd1) locus that was also previ-
ously associated with plant height in vegetative rice plants (Yang
et al., 2014; Zhang et al., 2012). Subedi et al. (2019) performed a
GWAS on plant height at plant maturity and found peaks on chro-
mosomes 1 and 11. This discrepancy could indicate that at differ-
ent developmental stages, plant height is determined by different
genomic regions, but since Subedi et al. (2019) used a specifically
constructed genetic population stemming from six parents, the
genetic starting material was also fundamentally different from the
population used here. Interestingly, haplotypes associated with high
culm height exhibit low plant height and vice versa (Dataset S7).
Haplotypes associated with high plant height typically show longer
leaf length (Dataset S7). While all the height-related traits were
highly correlated at the phenotypic level (Figure 2), the lack of
common loci for all the traits (Dataset S4) and opposite trends
within the haplotype groups (Dataset S7) suggest that the three
components of plant height are regulated independently at the
genetic level.
Although we consider solidity a composite trait, we revealed only
one strong locus with several significant SNP associations on chromo-
some 3 (Figure 5). When we grouped varieties into haplotype groups
for two coding regions (Os03g0845000 and Os03g0845700,
Figure 6a,b), encoding a Pirin-like protein and a RPB17 fragment
within this locus, the phenotypes of the haplotype groups appeared
to differ not just in solidity but also in shoot area, dry weight and leaf
number (Figure 6h,i, Dataset S7). This indicates that genetic regulation
of solidity could still be associated with traits of plant vigour.
4.2.2 | Vigour
Vigour-related traits (i.e., dry weight, shoot area and number of leaves)
are all strongly correlated and share associated loci on chromosomes
1, 7, 11 and 12 (Figure 5, Dataset S4). The locus on chromosome 11
was also reported by Yang et al. (2014) for dry weight and fresh
weight at the late tillering stage, which is comparable to the develop-
mental stage studied here. A closer look at this locus revealed that
HUBER ET AL.15
only one gene is located within the linkage disequilibrium of associ-
ated SNPs. Interestingly, the haplotype analysis for SNPs within this
gene, encoding a pyruvate kinase family protein, revealed a significant
difference in dry weight between the two haplotype groups
(Figure 6g). Significant differences were also observed for shoot area
and the number of leaves and tillers for the same two haplotype
groups. As only one gene was located within this locus and one spe-
cific haplotype was related to high biomass, this locus is a promising
candidate for follow-up studies and breeding programmes. The locus
on chromosome 7 associated with shoot area and dry weight
(Figure 6e–f) harbours two genes, and we found that the haplotypes
were associated with an increased shoot area and dry weight but also
an increased number of leaves and tillers. These loci for plant vigour
complement those found in a QTL study for height at 7 and 14 days
after sowing and fresh weight, using exclusively temperate japonica
genotypes (Cordero-Lara et al., 2016), thus having an intrinsically dif-
ferent pool of biological variation that can provide different
genomic leads.
4.3 |Improving shading potential and weed
suppression
The large phenotypic variation, high abundance of haplotypes that do
not positively contribute to shading potential, and low shading rank of
several commercially important varieties in this studied diversity panel
together indicate a strong potential for improvement of shading
capacity in such varieties. For example, IR 64 and Nipponbare are
widely grown rice varieties that have a very low shading rank,
ranking in the lowest quartile of our population (Table S3). We
identified a suite of alleles of the Os01g0225200, Os01g0810800,
Os03g0843700 and Os03g0845000 genes that contribute positively
to shading potential (Figure 6). The IR 64 variety, which gave rise to
many of the current widely grown rice varieties (Acevedo-Siaca et al.,
2020; Mackill & Kush, 2018), is typically not carrying the favourable
alleles for these genes, and this is true for Nipponbare as well.
Remarkably, the most abundant haplotype, tends to be the most infe-
rior one for the target traits of high shade casting, in the diversity
FIGURE 8 Schematic of rice plant architecture ideotypes (a,d) of high shading capacity, compared to suboptimal (b,c,e,f) shoot architecture
for shading. The ideotype for a high shade casting rice plant at early growth stages, would combine strong growth vigour (many leaves and tillers
and increased height) with architectural aspects, such as wide tiller and leaf angles as well as droopy leaves. This is opposed to shoots that, even if
they have many leaves, have a very erect stature with narrow angles and straight up leaves (b) or a plant with long but few leaves (c). The
ideotype (d) would result in a large and dense ground cover (a big shoot area, together with high solidity). A plant with very long leaves and wide
inclination angles but low in vigour, would range far but only cover a small percentage of ground within this circle (low solidity) (f). The opposite, a
shoot with very narrow inclination angles and upright leaves would have a high solidity but still not reach a large ground cover. Side views shown
in panels a–c, top views shown in panels d–f.
16 HUBER ET AL.
panel screened here. Based on the insights from this study, we can
now guide improvements for shading potential in these varieties
through conventional breeding, where we provide information for
optimal alleles. That a certain set of shoot architectural traits, defined
here as core traits, indeed contribute to increased shading, i.e. an
increased light extinction below the canopy, was confirmed in the
greenhouse experiment with varieties contrasting in their Shading
Rank, with high ranking varieties significantly shading more than low
ranging varieties. The single metric of the Shading Rank, collapses the
multidimensional description of our phenotypic screen, with the pur-
pose of making it a more graspable and practically, easy-to-use metric
for future genetic screening by scientists and breeders alike. The
Shading Rank was not developed as a score for shoot architecture of
any rice variety, but as a simple metric for high or low shading poten-
tial within this diversity panel. Shoot architecture traits that have been
previously shown to be relevant for weed-competitiveness through
light competition, are rapid leaf formation and tillering, together with
wide leaf and tiller angles and leaf droopiness, leading to increased
leaf area, as well as plant height (Andrew et al., 2015; Brainard
et al., 2005; Dass et al., 2017; Mahajan & Chauhan, 2013; Seavers &
Wright, 1999; Worthington & Reberg-Horton, 2013). An ideotype for
strong shade casting ability would combine traits related to growth
and vigour, together with an optimal 3D arrangement of leaves and
tillers to maximise ground cover (Figure 8), leading not only to maxi-
mum shading over competitive plants but also optimising light extinc-
tion for photosynthetic activity, which in turn promotes plant growth.
Future studies could resolve if such improved varieties would
indeed have superior weed-suppressive properties in field trials, as
predicted from our analyses. Such tests are especially relevant
because rice is a highly plastic species known for its strong ability to
fill up empty spaces with tillers (Bahuguna et al., 2021). Since we have
performed our experiments under stable conditions in a controlled
environment, it will be relevant to perform field trials when testing
improved varieties since rice varieties may differentially adapt some
of the observed architecture variables under different planting densi-
ties and the associated changes in light composition and availability.
Furthermore, plant architecture is also plastic throughout develop-
ment, and although relatively horizontal angles would improve weed-
shading early on, more erect leaves would prevent crop shading in
later stages (Mantilla-Perez et al., 2020; Murchie & Burgess, 2022;
Natukunda et al., 2022). Another aspect of weed competitiveness that
was not covered in our study would be the root systems, for which
the rapidly evolving high-throughput phenotyping methods are a
major opportunity to resolve comparable questions as done here for
shoot architecture. We conclude that breeding for specific vigour
traits will likely have additional beneficial effects, as indicated by the
haplotype studies. Vigour from root growth can then be an added
layer at a later step towards field-grown, weed competitive varieties
that can be farmed in a sustainable manner. Having worked from a
broad diversity panel rather than a focused or limited population and
including traits such as angles, droopiness and solidity has enabled us
to identify alleles in existing varieties that can now be used in rice
improvement programmes for sustainable weed competitiveness.
AUTHOR CONTRIBUTIONS
Martina Huber, Ronald Pierik and Rashmi Sasidharan designed the
experiments, with additional input from Kaisa Kajala, Justine
Toulotte and Hans van Veen. Martina Huber performed all the
experiments, analysed the data and wrote the article with the
contributions of all authors. Magdalena M. Julkowska carried out
the haplotype analysis and assisted with statistical data analysis
and data visualisation. L. Basten Snoek provided technical assis-
tance for the genome-wide association studies (GWAS) and per-
formed part of the analysis. Hans van Veen provided assistance for
statistical analysis. Justine Toulotte performed part of the experi-
ment and measurements. Virender Kumar contributed to the
research plan and experiment support at IRRI. Ronald Pierik serves
as the author responsible for contact and ensures communication,
supervised all experiments, revises the manuscript draft and,
together with Rashmi Sasidharan, conceives the research plan and
project design.
ACKNOWLEDGEMENTS
We thank Ricardo Eugenio and James Edgane for their substantial
assistance in the phenotyping at the International Rice Research Insti-
tute. Special thanks goes to Haley Schuhl and the PlanctCV team for
their support and effort in tailoring the image analysis software to our
data set. We thank Roel van Bezouw and Tom Theeuwen for helpful
discussions about GWAS and Rens Voesenek, Evelyn Aparicio,
Jochem Evers and Jonne Rodenburg for useful discussions on this
research project.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflict of interest for this research.
DATA AVAILABILITY STATEMENT
All custom R scripts and supplementary data are available at https://
zenodo.org/record/6793832 or upon request to the author.
ORCID
Martina Huber https://orcid.org/0000-0002-9670-0460
Magdalena M. Julkowska https://orcid.org/0000-0002-4259-8296
L. Basten Snoek https://orcid.org/0000-0001-5321-2996
Hans van Veen https://orcid.org/0000-0001-9772-5929
Justine Toulotte https://orcid.org/0000-0003-2017-9645
Virender Kumar https://orcid.org/0000-0002-2521-7578
Kaisa Kajala https://orcid.org/0000-0001-6483-7473
Rashmi Sasidharan https://orcid.org/0000-0002-6940-0657
Ronald Pierik https://orcid.org/0000-0002-5320-6817
REFERENCES
Abdullah Al Mamun, M. (2014). Modelling rice-weed competition in direct-
seeded rice cultivation. Agricultural Research,3, 346–352. https://doi.
org/10.1007/s40003-014-0138-2
Acevedo-Siaca, L. G., Coe, R., Wang, Y., Kromdijk, J., Quick, W. P., &
Long, S. P. (2020). Variation in photosynthetic induction between rice
accessions and its potential for improving productivity. The New
Phytologist,227, 1097–1108. https://doi.org/10.1111/nph.16454
HUBER ET AL.17
Andrew, I. K. S., Storkey, J., & Sparkes, D. L. (2015). A review of the poten-
tial for competitive cereal cultivars as a tool in integrated weed man-
agement. Weed Research,55, 239–248. https://doi.org/10.1111/wre.
12137
Azmi, M., Shukor, J. A., & Najib, M. Y. M. (2007). Critical period for weedy
rice control in direct-seeded rice. Journal of Tropical Agriculture and
Food Science,35(2), 333–339.
Bahuguna, R. N., Chaturvedi, A. K., Pal, M., Viswanathan, C.,
Jagadish, S. V. K., & Pareek, A. (2021). Carbon dioxide responsiveness
mitigates rice yield loss under high night temperature. Plant Physiology,
188(1), 285–300. https://doi.org/10.1093/plphys/kiab470
Brainard, D. C., Bellinder, R. R., & DiTommaso, A. (2005). Effects of canopy
shade on the morphology, phenology, and seed characteristics of
Powell amaranth (Amaranthus powellii). Weed Science,53, 175–186.
https://doi.org/10.1614/WS-04-067R1
Caton, B. P., Cope, A. E., & Mortimer, M. (2003). Growth traits of diverse
rice cultivars under severe competition: Implications for screening for
competitiveness. Field Crops Research,83, 157–172. https://doi.org/
10.1016/S0378-4290(03)00072-8
Chakraborty, D., Ladha, J. K., Rana, D. S., Jat, M. L., Gathala, M. K.,
Yadav, S., Rao, A. N., Ramesha, M. S., & Raman, A. (2017). A global
analysis of alternative tillage and crop establishment practices for eco-
nomically and environmentally efficient rice production. Scientific
Reports,7, 9342. https://doi.org/10.1038/s41598-017-09742-9
Chauhan, B. S. (2012). Weed ecology and Weed Management strategies
for dry-seeded Rice in Asia. Weed Technology,26,1–13. https://doi.
org/10.1614/WT-D-11-00105.1
Chauhan, B. S. (2013). Shade reduces growth and seed production
of Echinochloa colona, Echinochloa crus-galli, and Echinochloa
glabrescens. Crop Protection,43, 241–245. https://doi.org/10.1016/j.
cropro.2012.10.009
Chauhan, B. S., & Abugho, S. B. (2013). Effects of water regime, nitrogen
fertilization, and rice plant density on growth and reproduction of low-
land weed Echinochloa crus-galli. Crop Protection,54, 142–147.
https://doi.org/10.1016/j.cropro.2013.08.005
Chauhan, B. S., Jabran, K., & Mahajan, G. (2017). Rice production worldwide.
(Springer Nature). https://doi.org/10.1007/978-3-319-47516-5
Chauhan, B. S., & Johnson, D. E. (2011). Ecological studies on Echinochloa
crus-galli and the implications for weed management in direct-seeded
rice. Crop Protection,30, 1385–1391. https://doi.org/10.1016/j.
cropro.2011.07.013
Chauhan, B. S., Singh, V. P., Kumar, A., & Johnson, D. E. (2011). Relations
of rice seeding rates to crop and weed growth in aerobic rice. Field
Crops Research,121, 105–115. https://doi.org/10.1016/j.fcr.2010.
11.019
Chauhan, B. S., & Yadav, A. (2013). Weed management approaches for
dry-seeded rice in India: A review. Indian Journal of Weed Science,45,
1–6.
Cordero-Lara, K. I., Kim, H., & Tai, T. H. (2016). Identification of seedling
vigor-associated quantitative trait loci in temperate japonica Rice.
Plant Breeding and Biotechnology,4, 426–440. https://doi.org/10.
9787/PBB.2016.4.4.426
Dass, A., Shekhawat, K., Choudhary, A. K., Sepat, S., Rathore, S. S.,
Chauhan, B. S., Mahajan, G., & Chauhan, B. S. (2017). Weed manage-
ment in rice using crop competition-a review. Crop Protection,95,
45–52. https://doi.org/10.1016/j.cropro.2016.08.005
de Vida, F. B. P., Laca, E. A., Mackill, D. J., Fernández, G. M., & Fischer, A. J.
(2006). Relating rice traits to weed competitiveness and yield: A path
analysis. Weed Science,54, 1122–1131. https://doi.org/10.1614/WS-
06-042R.1
Dimaano, N. G. B., Ali, J., Sta. Cruz, P. C., Baltazar, A. M., Diaz, M. G. Q.,
Acero, B. L. Jr., & Li, Z. (2017). Performance of newly developed
weed-competitive Rice cultivars under lowland and upland weedy
conditions. Weed Science,65, 798–817. https://doi.org/10.1017/wsc.
2017.57
Dingkuhn, M., Johnson, D. E., Sow, A., & Audebert, A. Y. (1999). Relation-
ships between upland rice canopy characteristics and weed competi-
tiveness. Field Crops Research,61,79–95. https://doi.org/10.1016/
S0378-4290(98)00152-X
Dingkuhn, M., Tivet, F., Siband, P., Asch, F., Audebert, A., & Sow, A. (2001).
Varietal differences in specific leaf area: a common physiological
determinant of tillering ability and early growth vigor? In Rice research
for food security and poverty alleviation. Proceedings of the international
Rice research conference, Los Baños, Philippines, 31 March-3 April, 2000
(pp. 95–108). International Rice Research Institute (IRRI).
Eizenga, G. C., Ali, M. L., Bryant, R. J., Yeater, K. M., McClung, A. M., &
McCouch, S. R. (2014). Registration of the Rice diversity panel 1 for
Genomewide association studies. Journal of Plant Registrations,8,
109–116. https://doi.org/10.3198/jpr2013.03.0013crmp
Fahlgren, N., Feldman, M., Gehan, M. A., Wilson, M. S., Shyu, C.,
Bryant, D. W., Hill, S. T., McEntee, C. J., Warnasooriya, S. N., Kumar, I.,
Ficor, T., Turnipseed, S., Gilbert, K. B., Brutnell, T. P., Carrington, J. C.,
Mockler, T. C., & Baxter, I. (2015). A versatile phenotyping system and
analytics platform reveals diverse temporal responses to water avail-
ability in Setaria. Molecular Plant,8, 1520–1535. https://doi.org/10.
1016/j.molp.2015.06.005
FAO. (2019). World Food and Agriculture—Statistical pocketbook 2019.
Farooq, M., Siddique, K. H. M. M., Rehman, H., Aziz, T., Lee, D.-J. J., &
Wahid, A. (2011). Rice direct seeding: Experiences, challenges and
opportunities. Soil and Tillage Research,111,87–98. https://doi.org/
10.1016/j.still.2010.10.008
Gehan, M. A., Fahlgren, N., Abbasi, A., Berry, J. C., Callen, S. T., Chavez, L.,
Doust, A. N., Feldman, M. J., Gilbert, K. B., Hodge, J. G., Hoyer, J. S.,
Lin, A., Liu, S., Lizárraga, C., Lorence, A., Miller, M., Platon, E.,
Tessman, M., & Sax, T. (2017). PlantCV v2: Image analysis software for
high-throughput plant phenotyping. PeerJ,5, e4088. https://doi.org/
10.7717/peerj.4088
Ghosal, S., Casal, C., Quilloy, F. A., Septiningsih, E. M., Mendioro, M. S., &
Dixit, S. (2019). Deciphering genetics underlying stable anaerobic ger-
mination in Rice: Phenotyping, QTL identification, and interaction
analysis. Rice,12, 50. https://doi.org/10.1186/s12284-019-0305-y
Haefele, S. M., Johnson, D. E., M'Bodj, D., Wopereis, M. C. C. S., &
Miezan, K. M. (2004). Field screening of diverse rice genotypes for
weed competitiveness in irrigated lowland ecosystems. Field Crops
Research,88,29–46. https://doi.org/10.1016/j.fcr.2003.11.010
Heap, I. (2014). Global perspective of herbicide-resistant weeds. Pest Man-
agement Science,70, 1306–1315. https://doi.org/10.1002/ps.3696
Hoang, G. T., Gantet, P., Nguyen, K. H., Phung, N. T. P., Ha, L. T.,
Nguyen, T. T., Lebrun, M., Courtois, B., & Pham, X. H. (2019). Genome-
wide association mapping of leaf mass traits in a Vietnamese rice land-
race panel. PLoS ONE,14, e0219274. https://doi.org/10.1371/journal.
pone.0219274
Huang, X., Wei, X., Sang, T., Zhao, Q., Feng, Q., Zhao, Y., Li, C., Zhu, C.,
Lu, T., Zhang, Z., Li, M., Fan, D., Guo, Y., Wang, A., Wang, L., Deng, L.,
Li, W., Lu, Y., Weng, Q., …Han, B. (2010). Genome-wide association
studies of 14 agronomic traits in rice landraces. Nature Genetics,42,
961–967. https://doi.org/10.1038/ng.695
International Rice Research Institute. (2013). SES (Standard Evaluation
System) for Rice (Manila, Philippines).
Johnson, D. E., Dingkuhn, M., Jones, M. P. P., & Mahamane, M. C. (1998).
The influence of rice plant type on the effect of weed competition on
Oryza sativa and Oryza glaberrima. Weed Research,38, 207–216.
https://doi.org/10.1046/j.1365-3180.1998.00092.x
Julkowska, M. M., Saade, S., Agarwal, G., Gao, G., Pailles, Y., Morton, M.,
Awlia, M., & Tester, M. (2019). MVApp—Multivariate analysis applica-
tion for streamlined data analysis and curation. Plant Physiology,180,
1261–1276. https://doi.org/10.1104/pp.19.00235
Kennedy, G., & Burlingame, B. (2003). Analysis of food composition data
on rice from a plant genetic resources perspective. Food Chemistry,80,
589–596. https://doi.org/10.1016/S0308-8146(02)00507-1
18 HUBER ET AL.
Koarai, A., & Morita, H. (2003). Evaluation of the suppression ability of
rice (Oryza sativa) on Monochoria vaginalis by measuring photosyn-
thetic photon flux density below rice canopy. Weed Biology and Man-
agement,3, 172–178. https://doi.org/10.1046/j.1445-6664.2003.
00104.x
Kraehmer, H., Jabran, K., Mennan, H., & Chauhan, B. S. (2016). Global dis-
tribution of rice weeds –A review. Crop Protection,80,73–86.
https://doi.org/10.1016/j.cropro.2015.10.027
Kretzschmar, T., Pelayo, M. A. F., Trijatmiko, K. R., Gabunada, L. F. M.,
Alam, R., Jimenez, R., Mendioro, M. S., Slamet-Loedin, I. H.,
Sreenivasulu, N., Bailey-Serres, J., Ismail, A. M., Mackill, D. J., &
Septiningsih, E. M. (2015). A trehalose-6-phosphate phosphatase
enhances anaerobic germination tolerance in rice. Nature Plants,1,
15124. https://doi.org/10.1038/nplants.2015.124
Kumar, V., & Ladha, J. K. (2011). Direct seeding of Rice. Recent developments
and future research needs. Academic Press. https://doi.org/10.1016/
B978-0-12-387689-8.00001-1
Laca, E. A., Mackill, D. J., & Fischer, J. (2006). Relating rice traits to weed
competitiveness and yield: A path analysis. Weed Science,54, 1122–
1131. https://doi.org/10.1614/WS-06-042R.1
Li, M., Liu, X., Bradbury, P., Yu, J., Zhang, Y. M., Todhunter, R. J.,
Buckler, E. S., & Zhang, Z. (2014). Enrichment of statistical power for
genome-wide association studies. BMC Biology,12, 73. https://doi.
org/10.1186/s12915-014-0073-5
Liakat Ali, M., McClung, A. M., Jia, M. H., Kimball, J. A., McCouch, S. R.,
Eizenga, G. C., McCouch, S. R., & Georgia, C. E. (2011). A Rice diversity
panel evaluated for genetic and agro-morphological diversity between
subpopulations and its geographic distribution. Crop Science,51,
2021–2035. https://doi.org/10.2135/cropsci2010.11.0641
Mackill, D. J., & Khush, G. S. (2018). IR64: A high-quality and high-yielding
mega variety. Rice,11, 18. https://doi.org/10.1186/s12284-018-
0208-3
Mahajan, G., & Chauhan, B. S. (2013). The role of cultivars in managing
weeds in dry-seeded rice production systems. Crop Protection,49,52–
57. https://doi.org/10.1016/j.cropro.2013.03.008
Mahender, A., Anandan, A., & Pradhan, S. K. (2015). Early seedling vigour,
an imperative trait for direct-seeded rice: An overview on physio-
morphological parameters and molecular markers. Planta,241, 1027–
1050. https://doi.org/10.1007/s00425-015-2273-9
Mantilla-Perez, M. B., Bao, Y., Tang, L., Schnable, P. S., & Salas-
Fernandez, M. G. (2020). Toward “smart canopy”sorghum: Discovery
of the genetic control of leaf angle across layers. Plant Physiology,
184(4), 1927–1940. https://doi.org/10.1104/pp.20.00632
Marín, C., & Weiner, J. (2014). Effects of density and sowing pattern on
weed suppression and grain yield in three varieties of maize under
high weed pressure. Weed Research,54, 467–474. https://doi.org/10.
1111/wre.12101
McCouch, S. R., Wright, M. H., Tung, C.-W., Maron, L. G., McNally, K. L.,
Fitzgerald, M., Singh, N., DeClerck, G., Agosto-Perez, F. J., Korniliev, P.,
Greenberg, A. J., Naredo, M. E. B., Mercado, S. M. Q., Harrington, S. E.,
Shi, Y., Branchini, D. A., Kuser-Falc˜ao, P. R., Leung, H., Ebana, K., …
Mezey, J. (2016). Open access resources for genome-wide association
mapping in rice. Nature Communications,7, 10532. https://doi.org/10.
1038/ncomms10532
Mennan, H., Ngouajio, M., Sahin, M., Isık, D., & Altop, K. (2012). Competi-
tiveness of rice (Oryza sativa L.) cultivars against Echinochloa crus-galli
(L.) Beauv. In water-seeded production systems. Crop Protection,41,
1–9. https://doi.org/10.1016/j.cropro.2012.04.027
Murchie, E. H., & Burgess, A. J. (2022). Casting light on the architecture of
crop yield. Crop and Environment,1(1), 74–85. https://doi.org/10.
1016/j.crope.2022.03.009
Namuco, O. S. S., Cairns, J. E. E., & Johnson, D. E. (2009). Investigating
early vigour in upland rice (Oryza sativa L.): Part I. seedling growth and
grain yield in competition with weeds. Field Crops Research,113, 197–
206. https://doi.org/10.1016/j.fcr.2009.05.008
Natukunda, M. I., Mantilla-Perez, M. B., Graham, M. A., Liu, P., & Salas-
Fernandez, M. G. (2022). Dissection of canopy layer-specific genetic
control of leaf angle in Sorghum bicolor by RNA sequencing. BMC
Genomics,23, 95. https://doi.org/10.1186/s12864-021-08251-4
Oliver, V., Cochrane, N., Magnusson, J., Brachi, E., Monaco, S., Volante, A.,
Courtois, B., Vale, G., Price, A., & Teh, Y. A. (2019). Effects of water
management and cultivar on carbon dynamics, plant productivity and
biomass allocation in European rice systems. Science of the Total Envi-
ronment,685, 1139–1151. https://doi.org/10.1016/j.scitotenv.2019.
06.110
Ottis, B. V., & Talbert, R. E. (2007). Barnyardgrass (Echinochloa crus-galli)
control and Rice density effects on Rice yield components. Weed Tech-
nology,21, 110–118. https://doi.org/10.1614/WT-06-018.1
Pantazopoulou, C. K., Bongers, F. J., & Pierik, R. (2021). Reducing shade
avoidance can improve Arabidopsis canopy performance against com-
petitors. Plant, Cell and Environment,44, 1130–1141. https://doi.org/
10.1111/pce.13905
Park, S. E., Benjamin, L. R., & Watkinson, A. R. (2003). The theory and
application of plant competition models: An agronomic perspective.
Annals of Botany,92, 741–748. https://doi.org/10.1093/aob/mcg204
Peerzada, A. M., Ali, H. H., Chauhan, B. S., Masood Peerzada, A.,
Ali, H. H., & Chauhan, B. S. (2017). Weed management in
sorghum [Sorghum bicolor (L.) Moench] using crop competition: A
review. Crop Protection,95,74–80. https://doi.org/10.1016/j.cropro.
2016.04.019
Poorter, H., Niklas, K. J., Reich, P. B., Oleksyn, J., Poot, P., & Mommer, L.
(2012). Biomass allocation to leaves, stems and roots: Meta-analyses
of interspecific variation and environmental control. The New Phytolo-
gist,193,30–50. https://doi.org/10.1111/j.1469-8137.2011.03952.x
R Core Team. (2020). R: A language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna, Austria.
https://www.R-project.org/
Raj, S. K., & Syriac, E. K. (2017). Weed management in direct seeded rice:
A review. Agricultural Reviews,38,41–50. https://doi.org/10.18805/
ag.v0iOF.7307
Rao, A. N., Johnson, D. E., Sivaprasad, B., Ladha, J. K., & Mortimer, A. M.
(2007). Weed Management in direct-seeded Rice. Advances in Agron-
omy,93, 153–255. https://doi.org/10.1016/S0065-2113(06)93004-1
Roig-Villanova, I., & Martínez-García, J. F. (2016). Plant responses to vege-
tation proximity: A whole life avoiding shade. Frontiers in Plant Science,
7, 236. https://doi.org/10.3389/fpls.2016.00236
Sakamoto, T., Morinaka, Y., Ohnishi, T., Sunohara, H., Fujioka, S., Ueguchi-
Tanaka, M., Mizutani, M., Sakata, K., Takatsuto, S., Yoshida, S.,
Tanaka, H., Kitano, H., & Matsuoka, M. (2006). Erect leaves caused by
brassinosteroid deficiency increase biomass production and grain yield
in rice. Nature Biotechnology,24, 105–109. https://doi.org/10.1038/
nbt1173
Seavers, G. P., & Wright, K. J. (1999). Crop canopy development and struc-
ture influence weed suppression. Weed Research,39, 319–328.
https://doi.org/10.1046/j.1365-3180.1999.00148.x
Subedi, S. R., Sandhu, N., Singh, V. K., Sinha, P., Kumar, S., Singh, S. P.,
Ghimire, S. K., Pandey, M., Yadaw, R. B., Varshney, R. K., & Kumar, A.
(2019). Genome-wide association study reveals significant genomic
regions for improving yield, adaptability of rice under dry direct seeded
cultivation condition. BMC Genomics,20, 471. https://doi.org/10.
1186/s12864-019-5840-9
Tang, Y., Liu, X., Wang, J., Li, M., Wang, Q., Tian, F., Su, Z., Pan, Y., Liu, D.,
Lipka, A. E., Buckler, E. S., & Zhang, Z. (2016). GAPIT version 2:
An enhanced integrated tool for genomic association and prediction.
Plant Genome,9, plantgenome2015-11. https://doi.org/10.3835/
plantgenome2015.11.0120
Teichmann, T., & Muhr, M. (2015). Shaping plant architecture. Front. Plant
Science,6, 233. https://doi.org/10.3389/fpls.2015.00233
To, H. T. M., Nguyen, H. T., Dang, N. T. M., Nguyen, N. H., Bui, T. X.,
Lavarenne, J., Phung, N. T. P., Gantet, P., Lebrun, M., Bellafiore, S., &
HUBER ET AL.19
Champion, A. (2019). Unraveling the genetic elements involved in
shoot and root growth regulation by Jasmonate in Rice using a
genome-wide association study. Rice,12, 69. https://doi.org/10.1186/
s12284-019-0327-5
Wang, D. R., Agosto-Pérez, F. J., Chebotarov, D., Shi, Y., Marchini, J.,
Fitzgerald, M., McNally, K. L., Alexandrov, N., & McCouch, S. R. (2018).
An imputation platform to enhance integration of rice genetic
resources. Nature Communications,9, 3519. https://doi.org/10.1038/
s41467-018-05538-1
Wang, F., Longkumer, T., Catausan, S. C., Calumpang, C. L. F., Tarun, J. A.,
Cattin-Ortola, J., Ishizaki, T., Pariasca Tanaka, J., Rose, T.,
Wissuwa, M., & Kretzschmar, T. (2018). Genome-wide association and
gene validation studies for early root vigour to improve direct seeding
of rice. Plant, Cell & Environment,41, 2731–2743. https://doi.org/10.
1111/pce.13400
Wang, L., Wang, A., Huang, X., Zhao, Q., Dong, G., Qian, Q., Sang, T., &
Han, B. (2011). Mapping 49 quantitative trait loci at high resolution
through sequencing-based genotyping of rice recombinant inbred
lines. Theoretical and Applied Genetics,122, 327–340. https://doi.org/
10.1007/s00122-010-1449-8
Wang, Q., Tang, J., Han, B., & Huang, X. (2020). Advances in genome-wide
association studies of complex traits in rice. Theoretical and Applied
Genetics,133, 1415–1425. https://doi.org/10.1007/s00122-019-
03473-3
Wang, Q., Tian, F., Pan, Y., Buckler, E.S., & Zhang, Z. (2018). User Manual
for Genomic Association and Prediction Integrated Tool.
Weiner, J., Andersen, S. B., Wille, W. K.-M., Griepentrog, H. W., &
Olsen, J. M. (2010). Evolutionary agroecology: The potential for coop-
erative, high density, weed-suppressing cereals. Evolutionary Applica-
tions,3, 473–479. https://doi.org/10.1111/j.1752-4571.2010.
00144.x
Wing, R. A., Purugganan, M. D., & Zhang, Q. (2018). The rice genome revo-
lution: From an ancient grain to green super rice. Nature Reviews
Genetics,19(8), 505–517. https://doi.org/10.1038/s41576-018-
0024-z
Wolfe, M. S., Baresel, J. P., Desclaux, D., Goldringer, I., Hoad, S.,
Kovacs, G., Löschenberger, F., Miedaner, T., Østergård, H., &
Lammerts Van Bueren, E. T. (2008). Developments in breeding cereals
for organic agriculture. Euphytica,163, 323–346. https://doi.org/10.
1007/s10681-008-9690-9
Worthington, M., & Reberg-Horton, C. (2013). Breeding cereal crops for
enhanced weed suppression: Optimizing allelopathy and competitive
ability. Journal of Chemical Ecology,39, 213–231. https://doi.org/10.
1007/s10886-013-0247-6
Wu, Y., Xi, N., Weiner, J., & Zhang, D. Y. (2021). Differences in weed sup-
pression between two modern and two old wheat cultivars at different
sowing densities. Agronomy,11, 253. https://doi.org/10.3390/
agronomy11020253
Xu, L., Li, X., Wang, X., Xiong, D., & Wang, F. (2019). Comparing the grain
yields of direct-seeded and transplanted rice: A meta-analysis. Agron-
omy,9, 767. https://doi.org/10.3390/agronomy9110767
Yang, W., Guo, Z., Huang, C., Duan, L., Chen, G., Jiang, N., Fang, W.,
Feng, H., Xie, W., Lian,