Association mapping identifies loci for canopy coverage in diverse soybean genotypes

Association mapping identifies loci for canopy coverage
in diverse soybean genotypes
Avjinder S. Kaler & Jeffery D. Ray &
William T. Schapaugh & Marilynn K. Davies &
C. Andy King & Larry C. Purcell
Received: 3 August 2017 /Accepted: 21 March 2018
# Springer Science+Business Media B.V., part of Springer Nature 2018
Abstract Rapid establishment of canopy coverage de-
creases soil evaporation relative to transpiration, improves
water use efficiency and light interception, and increases
soybean competitiveness against weeds. The objective of
this study was to identify genomic loci associated with
canopy coverage (CC). Canopy coverage was evaluated
using a panel of 373 MG IV soybean genotypes that was
grown in five environments. Digital image analysis was
used to determine canopy coverage two times (CC1 and
CC2) during vegetative development approximately 8 to
16 days apart for each environment. After filtration for
quality control, 31,260 SNPs with a minor allele frequency
(MAF) ≥ 5% were used for association mapping with the
FarmCPU model. Analysis identified significant SNP-
canopy coverage associations including 36 for CC1 and
56 for CC2. Five SNPs for CC1 and 11 SNPs for CC2
were present in at least two environments. The significant
SNP associations likely tagged 33 (CC1) and 50 (CC2)
different quantitative trait loci (QTLs). Eleven putative loci
were identified in which chromosomal regions associated
were coincident for CC1 and CC2. Candidate genes iden-
tified using these significant SNPs included those with
reported functions associated with growth, developmental,
and light responses. Favorable alleles from significant
SNPs may be an important resource for pyramiding genes
to improve canopy coverage and for identifying parental
genotypes for use in breeding programs.
Keywords Soybean . Association mapping . Light
interception
Mol Breeding (2018) 38:50
https://doi.org/10.1007/s11032-018-0810-5
Electronic supplementary material The online version of this
article (https://doi.org/10.1007/s11032-018-0810-5) contains
supplementary material, which is available to authorized
users.
A. S. Kaler :M. K. Davies :C. A. King :L. C. Purcell (*)
Department of Crop, Soil, and Environmental Sciences,
University of Arkansas, Fayetteville, AR 72704, USA
e-mail: lpurcell@uark.edu
A. S. Kaler
e-mail: askaelr@uark.edu
M. K. Davies
e-mail: mkdavies@uark.edu
C. A. King
e-mail: chking@uark.edu
J. D. Ray
Crop Genetics Research Unit, USDA-ARS, 141 Experimental
Station Road, Stoneville, MS 38776, USA
e-mail: jeff.ray@ars.usda.gov
W. T. Schapaugh
Department of Agronomy, Kansas State University, Manhattan,
KS 66506, USA
e-mail: wts@ksu.edu

Introduction
Genome-wide association analysis is an alternative ap-
proach to traditional quantitative trait loci (QTL) map-
ping of bi-parental populations and is widely used in
plant and human genetics (Nordborg and Tavaré 2002;
Risch and Merikangas 1996). Genotyping diverse lines
provides thousands of single-nucleotide polymorphisms
(SNPs) across the genome, which allows fine mapping
of complex trait variation down to the nucleotide level
by exploiting historical recombination events (Zhu et al.
2008). The main advantages of genome-wide associa-
tion studies (GWAS) over traditional linkage mapping
include increased mapping resolution, reduced research
time, and greater allele number (Yu et al. 2006).
In soybean, GWAS have identified chromosomal
regions associated with seed protein and oil concentra-
tions (Hwang et al. 2014), carotenoids (Dhanapal et al.
2015a), carbon 13 to 12 ratio (Dhanapal et al. 2015b;
Kaler et al. 2017a), oxygen 18 to 16 ratio (Kaler et al.
2017a), canopy wilting (Kaler et al. 2017b), agronomic
traits (Wen et al. 2014), and ureide concentration (Ray
et al. 2015) as well as other studies. GWAS in soybean
are likely to increase due to recent genotyping of more
than 19,000 accessions of the USDA-ARS Soybean
Germplasm collection that provided over 50,000 SNP
markers. The complete SNP marker dataset is available
at SoyBase (www.soybase.org).
Early establishment of a closed canopy is beneficial
for several reasons. A faster closed canopy increases
water use efficiency by enhancing transpiration relative
to soil evaporation (Purcell and Specht 2004; Rebetzke
et al. 2007; Richards et al. 2007; Slafer et al. 2005). By
reducing the water lost through soil evaporation, quick
canopy establishment ensures that more soil moisture is
available for photosynthesis and biomass production
that otherwise would be lost to evaporation.
A second advantage offered by rapid establish-
ment of canopy coverage is improved canopy solar
radiation interception, which is an important factor
determining crop growth and yield (Edwards and
Purcell 2005; Edwards et al. 2005; Liebisch et al.
2015). The interception of radiation by the canopy
provides the energy required for physiological pro-
cesses including photosynthesis and transpiration
(Liebisch et al. 2015). Capacity of the crop canopy
to intercept solar radiation determines yield, which
depends on the available leaf area, structure, and its
efficiency (Gifford et al. 1984).
Rapid establishment of canopy closure also increases
soybean competiveness, especially for weeds (Bussan
et al. 1997). Herbicide-resistant weeds are becoming a
large problem in crop production (Green and Owen
2011), and this problem is mainly due to the high
selection pressures imposed by widespread use of
mono-herbicide culture (Shaner 1995). As the number
of herbicide-resistant weeds increase, there is a need for
alternative and sustainable approaches to weed manage-
ment. Faster canopy development can suppress early-
season weeds (Fickett et al. 2013; Jannink et al. 2000,
2001), and rapid canopy development can, therefore,
serve as a cultural control method to suppress weed
growth by increasing soybean competiveness.
There are differences in competitiveness among soy-
bean genotypes due to how quickly the canopy is
established (Place et al. 2011a, b) that appear to be
associated with individual seed weight. Place et al.
(2011b) divided seed lots of three cultivars into five
seed weight classes that ranged from 50 to 326 mg
seed−1
. Canopy coverage at 3 weeks after emergence
for all genotypes increased with increasing individual
seed weight. Larger seed classes improved competitive-
ness by increasing plant height and petiole length.
Small-seeded, natto soybean types were generally
slower to establish a canopy compared to large-seeded,
tofu type soybean (Place et al. 2011a).
Purcell (2000) described a method of analyzing dig-
ital images that offers a simple and effective way to
determine canopy coverage. Canopy coverage was mea-
sured as a fraction of green pixels relative to the total
number of pixels in an image, and canopy coverage was
approximately equivalent to the fraction of radiation
intercepted. Digital-image analysis provides an inexpen-
sive and rapid way of measuring canopy coverage over
other methods of estimating radiation interception
(Campillo et al. 2008; Fiorani et al. 2012). The analysis
of digital images has been widely accepted as a high-
throughput method for determining canopy coverage or
the fraction of solar radiation intercepted (De Bruin and
Pedersen 2008; Edward et al. 2005; Gaspar and Conley
2015; Xavier et al. 2017).
Canopy coverage is a quantitative trait that is influ-
enced by genotype, environment, and their interaction
(Xavier et al. 2017). The complexity of this trait arises
from the segregation of alleles at many chromosomal
regions, each with small effects on the phenotype, and
interacting with other alleles and with the environment
(Tuberosa et al. 2007). Therefore, investigation of
50 Page 2 of 12 Mol Breeding (2018) 38:50

genetic control of canopy coverage may be used to
improve crop performance by selecting and pyramiding
favorable loci associated with faster establishment of the
canopy into elite cultivars (Xavier et al. 2017).
To date, Xavier et al. (2017) have conducted the only
genetic evaluation of canopy coverage in soybean. They
phenotyped the soybean nested association mapping
(SoyNAM) population for canopy coverage with a rel-
atively small set of markers (4077 SNPs) and found six
genomic regions that were associated with canopy cov-
erage. The SoyNAM panel was developed from cross-
ing 40 genotypes to one hub parent, and from each
cross, 140 RILs were developed. Although the
SoyNAM panel was selected for diversity, it is likely
that these 41 genotypes may be somewhat limited in
capturing the wide diversity of phenotypes that would
impact canopy coverage. In the present research, 31,260
polymorphic SNPs were utilized for GWAS of canopy
coverage, and canopy coverage was assessed on a panel
of 373 diverse maturity group (MG) IVaccessions. The
objective of this study was to use GWAS to explore the
genetic variation of canopy coverage present within this
panel by identifying significant loci associated with this
trait.
Materials and methods
Germplasm collection and field experiments
A panel of 373 MG IV soybean accessions was selected
from the USDA-ARS Soybean Germplasm Collection
based on genetic diversity and agronomic characteristics
(www.ars.grin.gov). Our selection criteria were based
on relatively high yield, relatively good lodging and
shattering scores, and diverse geographic distribution.
A thorough description of genotype selection criteria
was provided by Dhanapal et al. (2015b).
Field experiments were conducted in five environments
including the Main Arkansas Agricultural Research Cen-
ter, Fayetteville, AR (FY; 36° 9′N, 94° 17′W), in 2016 on a
Captina silt loam; Pine Tree Research Station, Colt, AR
(PT; 35° 7′N, 90° 55′W), in 2016 on a Calloway silt loam;
Rohwer Research Station, Rohwer, AR (RH; 33° 48′N,
91° 17′W), in 2016 on a Sharkey silty clay; Salina, KS
(SA; 38° 70′N, 97° 60′W), in 2016 on a Hord silt loam;
and at the Rice Research and Extension Center, Stuttgart,
AR (ST; 34° 47′N, 91° 51′W), in 2015 on a Crowley silt
loam. Planting dates were 8 June 2015 for ST, 23
May 2016 for RH, 1 June 2016 for FY, 2 June 2016 for
PT, and 15 June 2016 for SA. Seeds were planted at a
density of 37 m−2
at a 2.5-cm depth. At ST, plots were
4.57 m long and two rows wide with 0.76-m row spacing.
At FY, single row plots were 5.48 m in length with a 0.76-
m row spacing. At SA, there were two-row plots that were
3.65 m in length with a 0.76-m row spacing. At PT and
RH, seeds were sown with a drill (19-cm row spacing) and
plots were 1.52 m wide and 4.57 m long. At each envi-
ronment, the experiment was conducted as a randomized
complete block design with two replications.
For each environment, soil water deficit was estimat-
ed for each day beginning at planting as described by
Purcell et al. (2007). Potential evapotranspiration (Eto)
for a given day was determined using a modified
Penman-Monteith approach (Allen et al. 1998) and
multiplied by the fraction of radiation intercepted by
the crop, which served as a crop coefficient (equivalent
to canopy coverage). Estimated soil-water deficits were
cumulated and adjusted with rainfall additions as
needed.
Canopy coverage determination
Canopy coverage was determined by analyzing digital
images that were taken of the canopy with a camera
mounted on a monopod (Purcell 2000). Digital images
were taken at 160 cm above the plots at a resolution of
1280 × 960 pixels. For phenotypic variation of canopy
coverage at ST, FY, and SA, images consisted of a single
row, and at PT and RH, images were taken above the
center of the plots and were composed of seven rows.
For digital images of single rows, care was taken to
ensure that the image was centered directly over the
row and that the image frame extended from the center
line between the two adjacent rows. At the times these
measurements were made, leaves from adjacent rows
did not cross the center line between plots. The first set
of pictures (CC1) was taken 23 to 28 days after emer-
gence when plants were between V2 and V3. A second
set of pictures (CC2) was taken 8 to 16 days later
between V4 and V7. Plant densities were measured in
a 1-m section of row, and results indicated no significant
effect of genotype on the plant density. Weeds poten-
tially interfere with canopy cover estimates, and plots
were weed free before taking pictures. Digital images
were analyzed using SigmaScan Pro (v.5.0, SPSS, Inc.,
Chicago, IL) with a macro that utilizes batch analysis
(Karcher and Richardson 2005). Software measured the
Mol Breeding (2018) 38:50 Page 3 of 12 50

number of green pixels of each image as a fraction of the
total pixel count in the frame. SigmaScan Pro has se-
lectable ranges of hue and saturation values for identi-
fying green pixels in images. For each environment, the
settings for the hue and saturation were determined for
several representative images prior to batch analysis.
Typical ranges that selectively identified green leaves
without soil background were from 50 to 100 for hue
and 0 to 100 for saturation.
Place et al. (2011b) reported a strong association of
early canopy coverage and individual seed weight.
Hence, average phenotypic data of individual seed
weight (i.e., the average single seed weight based upon
100 seeds) from 3 years (2014, 2015, and 2016) were
included in this study to assess a possible association of
canopy coverage with individual seed weight at the
genomic level.
Statistical analysis
Data were tested for heterogeneity of variance by
checking the residual plots, which did not show any
heterogeneity of variance. Genotype and environment
were treated as fixed effects and replication within en-
vironment was considered a random effect for analysis
of variance (ANOVA). The PROC GLIMMIX proce-
dure (α = 0.05) of SAS 9.4 (SAS Institute Inc., Cary,
NC, USA 2013) was used for ANOVA with a model
suggested by Bondari (2003):
yijk ¼ μ þ Gi þ Ej þ GEð Þij þ Bk ijð Þ þ εijk
where μ is the total mean, Gi is the genotypic effect of
the ith
genotype, Ej is the effect of the jth
environment,
(GE)ij is the interaction effect between the ith
genotype
and the jth
environment, Bk(ij) is the effect of replications
within the jth
environment, and εijk is a random error
following N 0; σ2
e
À Á
. Analysis of variance was indepen-
dently applied to CC1 and CC2. Descriptive statistics
and Pearson correlation analysis for canopy coverage
were performed using the PROC UNIVARIATE and
PROC CORR procedures (α = 0.05) of SAS version
9.4, respectively. Broad sense heritability on an entry-
mean basis was calculated as
H2
¼ σ2
G= σ2
G þ
σ2
GE
k

þ
σ2
ε
rk

where σ2
G is the genotypic variance, σ2
GE is the genotype
by environment variance, σ2
ε is the residual variance, k is
the number of environments, and r is the number of
replications. The PROC VARCOMP of SAS 9.4 with
the restricted maximum likelihood estimation (REML)
method was used to estimate the above variance
components.
For association analysis, the best linear unbiased
prediction (BLUP) values were estimated using the
PROC GLIMMIX procedure by environment and
across all environments (AAE). The BLUP values were
used because they reduce the effects of replication (with-
in environment) and the effects of environment (when
conducted across multiple environments).
Genotyping and association analysis
Data for a total of 42,509 SNP markers of all 373
genotypes were obtained from the Illumina Infinium
SoySNP50K iSelect SNP Beadchip (Song et al. 2013),
which are available at SoyBase (www.soybase.org).
Genotype quality controls were applied by eliminating
monomorphic markers, markers with minor allele
frequency (MAF) 5%, and markers with missing rate
higher than 10%. The remaining missing markers in a
set of 31,260 SNPs were imputed using a LD-kNNi
method, which is based on a k-nearest-neighbor geno-
type (Money et al. 2015) and then used in association
analysis. Population structure is a confounding factor in
GWAS that induces false associations. Commonly, the
mixed linear model (MLM) is used to reduce these false
associations. However, these adjustments also compro-
mise true positive associations. As described by Kaler
et al. (2017a, b), the fixed and random model Circulat-
ing Probability Unification (FarmCPU) model, devel-
oped by Liu et al. (2016), effectively controlled false
positives and false negatives using this same genotype
panel. Different models (e.g., compressed mixed linear
model from GAPIT, GLM and MLM models from
TASSEL) were compared based on their Q-Q plots.
The FarmCPU model was superior in controlling the
false discovery rate compared to other models evaluat-
ed. The Q-Q plot of the FarmCPU model resulted in a
sharp deviation from the expected P value distribution
in the tail area, indicating that false positives were
adequately controlled (data not shown) in agreement
with previous research (Kaler et al. 2017b). Therefore,
FarmCPU was used for association analysis in the pres-
ent research. Default parameters were used in FarmCPU
model as recommended by the developer of FarmCPU
(Liu et al. 2016) except in the change of threshold output

from 0.01 to 1, which allowed all markers to be included
in the output Excel file.
Although FarmCPU controlled the false positives
and false negatives, we still used a stringent threshold
value (− log10(P) ≥ 3.5) to declare a significant associ-
ation between SNPs and canopy coverage. This level of
stringency was used to limit the number of associations
to those QTLs likely contributing biologically signifi-
cant effects. This threshold value is more stringent than
that reported in other soybean association mapping
studies of soybean (Dhanapal et al. 2015a, b; Hao
et al. 2012; Hwang et al. 2014; Zhang et al. 2015). We
also eliminated any SNP from consideration if the allelic
effect was less than the standard error of CC1 or CC2
phenotypic measurements. A threshold value of P ≤
0.05 was used to identify significant SNPs present in
more than one environment but only if representative
SNPs met a lower association of P ≤ 0.0003 in at least
one other environment.
Candidate gene identification
Candidate genes were considered when they were with-
in ± 10 kb of a SNP with a significant association. This
distance was chosen because it approximates the aver-
age distance between SNPs (18 kb). Candidate genes
and their associated functional annotation and biological
function were determined using Glyma1.1, Glyma1.0,
and NCBI RefSeq gene models in SoyBase (www.
soybase.org) with consideration for those that may
have an association with growth, developmental, and
light responses.
Results
Phenotype statistics
There was a broad range of CC1 and CC2 values ob-
served within each environment. Overall all 373 acces-
sions, the range (i.e., maximum value–minimum value)
of CC1 values was 0.27 (FY), 0.17 (PT), 0.45 (RH),
0.30 (SA), and 0.23 (ST) (Table 1). For CC2, the range
was 0.40 (FY), 0.28 (PT), 0.62 (RH), 0.29 (SA), and
0.50 (ST). Analysis of variance indicated that genotype,
environment, and their interaction had significant effects
(P ≤ 0.05) on CC1 and CC2. Correlations of canopy
coverage between environments for CC1 or CC2 were
significant and positive (0.10 ≤ r ≤ 0.39) except for the
correlation between RH and ST for CC1 (r = − 0.01)
(Table 2). Correlations between CC1 and CC2 within
environments averaged 0.59 and ranged from 0.76 (FY)
Table 1 Descriptive statistics of canopy coverage (fractional)
over 373 MG IV plant introductions measured at two time points
during vegetative development, CC1 and CC2, for experiments
conducted at Fayetteville, AR (FY); Pine Tree, AR (PT); Rohwer,
AR (RH); Salina, KS (SA); and Stuttgart, AR (ST)
Trait ENV Minimum Maximum Range Mean Heritability (H2
)
CC1
FY 0.06 0.33 0.27 0.18 0.38
PT 0.04 0.21 0.17 0.11 0.52
RH 0.08 0.53 0.45 0.34 0.54
SA 0.43 0.73 0.30 0.58 0.51
ST 0.12 0.35 0.23 0.21 0.52
Overalla
0.58
CC2
FY 0.41 0.81 0.40 0.62 0.46
PT 0.24 0.52 0.28 0.38 0.21
RH 0.35 0.97 0.62 0.80 0.46
SA 0.69 0.98 0.29 0.90 0.33
ST 0.27 0.77 0.50 0.50 0.49
Overall 0.51
a
Broad sense heritability across all environments

to 0.30 (PT) (Table 2). Broad sense heritability by
environment on an entry-mean basis indicated low-to-
moderate heritability for CC1 and CC2 (0.21 ≤ H2
≤
0.54) (Table 1). When considered over all environments,
H2
was 0.58 (CC1) and 0.51 (CC2).
For silt-loam soils in our study (FY, ST, PT, and
SA), irrigation is typically recommended when soil-
moisture deficits exceed 35 mm, and for the clay
soil at RH, irrigation is recommended when soil
moisture deficits exceed 50 mm (Purcell et al.
2007). Using these irrigation thresholds as indicators
of stress, the week prior to CC1, soil moisture was
adequate for ST, RH, and SA but was limiting for
FY (3 out of 7 days) and PT (7 out of 7 days).
Between CC1 and CC2, soil moisture was adequate
at ST every day but limiting at FY (3 out of 14 days),
PT (5 out of 15 days), SA (9 out of 11 days), and
RH (5 out of 9 days). The differences in soil-
moisture availability among environments and be-
tween CC1 and CC2 likely contributed to the G × E
effect.
Genome-wide association analysis
Genome-wide association analysis identified 36
SNPs significantly associated with CC1 at level of
− log10 (P) ≥ 3.5; P ≤ 0.0003 (Table S1 and Fig. S1)
in at least one of the five environments. Of these 36
SNPs, six were significant in at least two environ-
ments. Significant SNPs that were present in a link-
age disequilibrium block on the same chromosome
were considered as one locus. Thus, the 36
significant SNPs comprised 33 putative QTLs
(Table S1). There were 56 significant SNPs associ-
ated with CC2 at a level of − log10(P) ≥ 3.5; P ≤
0.0003 (Table S2 and Fig. S2) in at least one envi-
ronment. Eleven SNPs of these 56 were significant
in at least two environments. These 56 significant
SNPs comprised 47 putative QTLs (Table S2). The
allelic effect (fractional change in canopy coverage
for the major compared to the minor allele) for these
significant SNPs for CC1 ranged from − 0.050 to
0.068 (Table S1) and for CC2 ranged from − 0.048
to 0.086 (Table S2). The positive sign indicates that
the minor allele was associated with increased can-
opy coverage. These allelic effects are in compari-
son to the standard errors of measurement for CC1
and CC2 of 0.004 and 0.001, respectively. The list
of all 33 significant loci for CC1 and 47 significant
loci for CC2, their corresponding MAF, major or
minor allele, allelic effect, R2
value, and common
environments where SNPs were significant are listed
in Tables S1 and S2, respectively. Table 3 shows the
list of significant SNPs associated with CC1 and
CC2 that were present in more than one environ-
ment, common to both CC1 and CC2, and coinci-
dent with previously reported QTLs for canopy cov-
erage (Xavier et al. 2017). Association analysis of
canopy coverage averaged across all environments
identified 11 SNPs and 12 SNPs significantly asso-
ciated with CC1 and CC2, respectively, at level of −
log10 (P) ≥ 3.5; P ≤ 0.0003 (Table S5).
Individual seed weight averaged over 3 years for
these 373 genotypes was significantly (P ≤ 0.05)
Table 2 Correlation of canopy coverage between environments for first (CC1) and second (CC2) measurement dates, between CC1 and
CC2 within each environment, and between canopy coverage and average seed weight (AVG_SW)
FY_CC1 PT_CC1 RH_CC1 SA_CC1 ST_CC1 FY_CC2 PT_CC2 RH_CC2 SA_CC2 ST_CC2
PT_CC1 0.36*
RH_CC1 0.25* 0.39*
SA_CC1 0.26* 0.33* 0.30*
ST_CC1 0.22* 0.17* − 0.01 0.18*
FY_CC2 0.76* 0.33* 0.26* 0.28* 0.21*
PT_CC2 0.20* 0.30* 0.15* 0.26* 0.18* 0.24*
RH_CC2 0.26* 0.27* 0.54* 0.31* 0.10* 0.37* 0.39*
SA_CC2 0.12* 0.24* 0.21* 0.64* 0.19* 0.13* 0.12* 0.22*
ST_CC2 0.23* 0.09 0.03 0.14* 0.73* 0.26* 0.10* 0.18* 0.13*
AVG_SW 0.35* 0.28* − 0.03 0.14* 0.41* 0.28* 0.23* − 0.01 0.05 0.34*
*Significant correlation at P 0.05

Table 3 Significant SNPs associated with canopy coverage at the first (CC1) and second (CC2) measurement dates, at both measurement
dates (CC1 and CC2), and loci that were coincident with previously reported genomic regions
SNP_ID Allelic
effectc
R2
ENVd
Gene namee
Functional annotation (biological function)
CC1
BARC_1.01_Gm_01_51957108_T_Ga
− 0.022 0.025 Glyma01g39090 Serine/threonine kinase activity (meristem
growth)
BARC_1.01_Gm_01_54917573_A_C 0.020 0.069 ST/FA Glyma01g42890 JUMONJI domain-containing protein (meri-
stem growth)
BARC_1.01_Gm_02_5326823_A_Gb
− 0.015 0.014 Glyma02g06610 Protein of unknown function
BARC_1.01_Gm_02_43094876_T_Cb
− 0.033 0.030 Glyma02g40960 Early growth response protein
BARC_1.01_Gm_05_3268626_T_C 0.014 0.056 PT/SA Glyma05g02130 Zinc finger (response to high light intensity)
BARC_1.01_Gm_05_37611048_C_Ta
0.020 0.021 Glyma05g32380 Phosphoenolpyruvate dikinase protein
(meristem growth)
BARC_1.01_Gm_06_7988088_G_Tb
− 0.012 0.006 Glyma06g10540 Glycosidases (plant-type cell wall
organization)
BARC_1.01_Gm_07_18047081_A_Gb
0.048 0.064 Glyma07g18210 Isoamyl acetate-hydrolyzing esterase (lipid
metabolic process)
BARC_1.01_Gm_08_9597333_T_C 0.068 0.057 RH/SA Glyma08g13160 Chaperone binding protein (photosynthesis)
BARC_1.01_Gm_09_3855506_T_Gab
− 0.022 0.028 Glyma09g05020 Peripheral-type benzodiazepine receptor
(abscisic acid stimulus)
BARC_1.01_Gm_11_8840866_G_Ab
0.035 0.049 Glyma11g12341 Plant protein of unknown function (DUF825)
BARC_1.01_Gm_15_1626629_C_T 0.017 0.027 FA/ST Glyma15g02420 Actin-binding protein family
BARC_1.01_Gm_16_5005273_G_Ab
0.026 0.065 Glyma16g05640 Glycerophosphoryl diester phosphodiesterase
(metabolic process)
BARC_1.01_Gm_16_7364708_A_Gb
0.046 0.078 Glyma16g07960 Myb-like DNA-binding domain (gibberellic
acid signaling)
BARC_1.01_Gm_16_30401273_C_Tb
0.049 0.086 Glyma16g25880 Root phototropism protein (response to light
stimulus)
BARC_1.01_Gm_17_8482479_G_A − 0.024 0.013 FA/PT Glyma17g11670 Glycosyl hydrolase family 79 (plant-type cell
wall growth)
BARC_1.01_Gm_20_45740785_C_Tb
− 0.010 0.004 Glyma20g36530 Phosphatase 2a regulatory subunit-related
(meristem growth)
CC2
BARC_1.01_Gm_01_4267470_A_G − 0.037 0.049 FA/PT/SA Glyma01g04616 AUX/IAA Protein (auxin stimulus)
BARC_1.01_Gm_02_4479807_T_Cb
0.014 0.004 Glyma02g05530 Auxin Responsive Protein (auxin stimulus)
BARC_1.01_Gm_02_44256235_A_Gb
− 0.002 0.005 Glyma02g42290 Amino Acid Transporters (multidimensional
cell growth)
BARC_1.01_Gm_02_44522295_G_A − 0.003 0.001 Glyma02g42560 Vesicle Coat Protein Clathrin
(vesicle-mediated transport)
BARC_1.01_Gm_04_3250504_T_C 0.063 0.064 RH/FA/PT Glyma04g04300 Poly-Adenylate Binding Protein (response to
cadmium ion)
BARC_1.01_Gm_05_33832783_T_G 0.063 0.053 RH/PT Glyma05g27670 Myb-like DNA-binding domain
BARC_1.01_Gm_06_6880019_A_Gb
0.019 0.003 Glyma06g09340 Serine/threonine protein kinase (histone
phosphorylation)
BARC_1.01_Gm_06_12426395_T_G 0.009 0.005 SA/ST Glyma06g15755 AAA-type ATPase family protein (chloroplast
organization)
BARC_1.01_Gm_06_14105376_A_Ga
− 0.042 0.109 Glyma06g17710 Fist C domain
BARC_1.01_Gm_07_18047081_A_Gb
0.021 0.051 Glyma07g18210 Isoamyl acetate-hydrolyzing esterase (meta-
bolic process)

correlated with CC1 and CC2 (0.14 ≤ r ≤ 0.41) at 7 of
the 10 possible combinations of CC1/CC2 and environ-
ment (Table 2). Analysis identified 52 SNPs significant-
ly associated with average individual seed weight at a
significance level of − log10 (P) ≥ 3.5; P ≤ 0.0003 (data
not shown). Nine chromosomal regions were identified,
where QTLs associated with canopy coverage and av-
erage seed weight were coincident (six with CC1
(Table S1) and three with CC2 (Table S2)). These asso-
ciations were important because previous studies
showed a positive relationship between individual seed
weight and canopy coverage (Place et al. 2011a, b).
Identifying the common regions would be useful for
further research.
Candidate gene identification
Based on the location of SNPs with significant as-
sociations, nearby genes (within ± 10 kb of the SNP)
were identified that may have association with
growth, developmental, and light response. There
were 36 genes near SNPs tagging CCI and 56 genes
near CC2. Of these genes, 19 (CC1) and 21 (CC2)
had SNPs with significant associations within the
gene. The remaining genes were present within ±
10 kb of the respective SNPs. Based on their bio-
logical function of growth, developmental, and light
response that were reported in SoyBase (www.
soybase.org), 13 genes for CC1 and 19 genes for
Table 3 (continued)
SNP_ID Allelic
effectc
R2
ENVd
Gene namee
Functional annotation (biological function)
BARC_1.01_Gm_07_38128536_G_A 0.034 0.071 SA/PT Glyma07g33260 Ca2+
/calmodulin-dependent protein kinase
(meristem growth)
BARC_1.01_Gm_08_46871422_G_A 0.062 0.088 ST/SA Glyma08g47090 Galactose oxidase/Kelch repeat superfamily
protein
BARC_1.01_Gm_09_773488_T_Cb
0.003 0.026 Glyma09g01250 Plastocyanin-like domain (root hair
elongation)
BARC_1.01_Gm_09_3023789_T_Cb
− 0.001 0.010 Glyma09g04060 Betaine aldehyde dehydrogenase (metabolic
process)
BARC_1.01_Gm_10_38900522_T_C 0.053 0.064 RH/FA Glyma10g29490 Lipoxygenase (growth)
BARC_1.01_Gm_11_8557505_T_C 0.035 0.070 Glyma11g11990 Mate efflux family protein (transmembrane
transport)
BARC_1.01_Gm_13_36385708_G_A − 0.003 0.015 SA/RH Glyma13g33290 Gibberellin 2-beta-dioxygenase (gibberellin
catabolic process)
BARC_1.01_Gm_15_50563545_T_C 0.020 0.060 PT/SA Glyma15g42330 Hexosyltransferases (meristem growth)
BARC_1.01_Gm_16_4707461_C_Tb
− 0.005 0.059 Glyma16g05380 Aspartate kinase (metabolic process)
BARC_1.01_Gm_16_6702694_C_Tb
− 0.001 0.001 Glyma16g07300 Thioredoxin-related protein
BARC_1.01_Gm_16_30654649_C_Tb
0.034 0.065 Glyma16g26100 Mlo family protein (leaf senescence)
BARC_1.01_Gm_18_194608_C_A 0.086 0.098 FA/RH/SA Glyma18g00530 DNA repair protein Rad50 (meristem
structural organization)
BARC_1.01_Gm_20_45740785_C_Tb
− 0.023 0.006 Glyma20g36530 Protein phosphatase 2 regulatory subunit
(meristem growth)
Experiments were conducted at Stuttgart in 2015 (ST), Fayetteville in 2016 (FY), Pine Tree in 2016 (PT), Rohwer in 2016 (RH), and Salina
in 2016 (SA). Significant SNPs were identified using the FarmCPU model with threshold P value (− log10 (P) ≥ 3.5; P ≤ 0.0003)
a
Genomic regions where identified SNPs were coincident with QTLs identified by Xavier et al. (2017)
b
Genomic regions where both CC1 and CC2 were coincident
c
Allelic effect indicates the difference in mean canopy coverage between genotypes with major allele and minor allele. Positive sign
indicates that major allele is associated with increased canopy coverage. Negative sign indicates that minor allele is associated with increased
canopy coverage
d
Common ENV indicates that SNP is present in more than one environment
e
All genes are from the Glyma1.1 assembly (www.soybase.org)

CC2 are potential candidate genes for establishing
faster canopy coverage.
Discussion
There was wide phenotypic variation of canopy cover-
age for both CC1 and CC2 within each environment,
which is important for dissecting complex traits through
association mapping (McCarthy et al. 2008). That there
were significant positive correlations between environ-
ments for CC1 and CC2, and moderate heritability
indicates that canopy coverage can be improved, which
has implications for increasing transpiration relative to
soil evaporation, light interception, and competitiveness
for weeds.
Some of the variability between CC1 and CC2 within
environments and among canopy coverage measure-
ments across environments may be related to soil-
moisture availability prior to measurements. Drought
decreases leaf area development and leaf expansion rate
(Clauw et al. 2015; Manandhar et al. 2017; Tardieu and
Tuberosa 2010). The week before CC1 measurements
there were no soil-moisture limitations at ST, RH, or
SA, but soil-moisture deficits exceeded the threshold for
irrigation for 3 or 7 days before CC1 at FY and PT,
respectively. Between CC1 and CC2 measurements, ST
was the only environment in which daily soil-moisture
deficits were above the irrigation threshold.
Of the nine genomic regions common to both aver-
age seed weight and canopy coverage, eight had the
same directional allelic effect (Tables S1 and S2). For
example, a genomic region on Gm01, where both can-
opy coverage and average seed weight were coincident,
there was a positive allelic effect for canopy coverage of
0.02 (Table S1) and this region also had a positive allelic
effect for average seed weight of 0.05 g seed−1
. Both
these regions were located within the coding region of a
gene, Glyma01g42890, which codes a JUMONJI
domain-containing protein with a biological function
associated with meristem growth (Table S3). Interest-
ingly, the SNP associated with canopy coverage in this
region was present in more than one environment
(Table S1). Similarly, two coincident genomic regions
on Gm05 and Gm16 also had positive allelic effects for
both canopy coverage and average seed weight. These
regions were located close to genes, Glyma05g02130
and Glyma16g25880, which have biological functions
involved with response to light stimulus (Table S3).
For CC1, 19 major alleles out of 41 were linked with
an increase in canopy coverage (positive value of allelic
effect indicates that the major allele was associated with
an increase in canopy coverage) (Table S1). One SNP
on Gm08 that had the largest positive allelic effect
(0.07) was present within the coding region of a gene,
Glyma08g13160, which codes a chaperone binding pro-
tein that has a biological function associated with pho-
tosynthesis (Tables S1 and S3). A SNP on Gm16 that
had the second largest positive allelic effect (0.05) was
also present within the coding region of a gene,
Glyma16g25880, which codes a root phototropism pro-
tein that has a biological function involved with re-
sponse to light stimulus (Tables S1 and S3). Out of 41
SNPs, minor alleles of 22 were associated with an
increase in canopy coverage (negative value of allelic
effect indicates that minor allele was associated with an
increase in canopy coverage) (Table S1). One SNP on
Gm02 that had a large negative allelic effect (− 0.03)
was present within the coding region of a gene,
Glyma02g40960, which codes an early growth response
protein (Tables S1 and S3).
For CC2, there were 38 out of 56 SNPs for which the
major allele associated with an increase in canopy cov-
erage (Table S2). A SNP on Gm18 that had the largest
positive allelic effect (0.09) was present within ± 5 kb
range of a gene, Glyma18g00530, which codes a DNA
repair protein (RAD50) that has a biological function
involved with meristem structural organization
(Tables S2 and S4). Out of 56 SNPs, the minor alleles
of 18 were associated with an increase in canopy cov-
erage (Table S2). One SNP on Gm09 with the largest
negative allelic effect (− 0.05) was present within ±
10 kb range of a gene, Glyma09g30370, which codes
a protein functioning as a glutamine synthetase clone R1
that has a biological function involved with leaf senes-
cence (Tables S2 and S4).
Xavier et al. (2017) identified seven SNPs associated
with canopy coverage, but two of the SNPs on Gm10
were close to one another, which they considered as one
QTL. They, therefore, reported six QTLs for canopy
coverage using the SoyNAM population. Location of
the CC1 and CC2 chromosomal regions identified in
this study were compared with QTLs reported by Xavier
et al. (2017) for canopy coverage. Likewise, we com-
pared genomic regions of CC1 and CC2 to see if they
were coincident (Table 3 and Fig. 1). Four out of six
QTLs reported by Xavier et al. (2017) were located
close to genomic regions that were associated with

CC1 and CC2 in this study (Table 3 and Fig. 1). Twelve
putative loci on Gm02 (2), Gm06, Gm07, Gm09 (3),
Gm11, Gm16 (3), and Gm20 were identified, where
chromosomal regions were associated with both CC1
and CC2 (Table 3 and Fig. 1). These chromosomal
regions have candidate genes with a direct function
associated with response to auxin, response to
gibberellic acid, meristem growth, light-regulated pro-
tein, early growth response protein, and response to light
intensity (Tables S3 and S4). These putative loci may
indicate the stability and importance for improving
faster canopy coverage and may highlight the important
regions of the genome for further investigations.
Conclusions
A high marker density of 31,260 SNPs with a MAF ≥
5% was used in this study for association mapping of
canopy coverage at two dates (CC1 and CC2) in five
environments. There were 36 significant SNPs associ-
ated with CC1 and 56 significant SNPs associated with
CC2 at a significance level of − log10 (P) ≥ 3.5. Five
significant SNPs for CC1 and 11 SNPs for CC2 were
present in at least two environments. The 36 SNPs for
CC1 and 56 SNPs for CC2 likely tagged 33 and 50
QTLs, respectively. Four different putative loci were
located within four genomic regions that were previous-
ly reported (Xavier et al. 2017) as QTLs for canopy
coverage. Twelve putative loci were identified, where
chromosomal regions associated with CC1 and CC2
were coincident. Several of these loci were close to or
within genes related to growth and development. We
identified significant SNPs that were present in more
than one environment, and where chromosomal regions
associated with both CC1 and CC2 were found within
the same genomic location. This information may be
useful in pyramiding alleles for faster canopy coverage
Fig. 1 Location of putative loci
significantly associated with
canopy coverage for both
measurement dates, CC1 and
CC2, and previously reported six
QTLs for canopy coverage

and for improving radiation interception, WUE, crop
growth, and soybean competitiveness.
Acknowledgements The authors gratefully acknowledge partial
funding of this research from the United Soybean Board. Mention
of trade names or commercial products in this publication is solely
for the purpose of providing specific information and does not
imply recommendation or endorsement by the US Department of
Agriculture. The USDA is an equal opportunity provider and
employer.
References
Allen RG, Pereira LS, Raes D, Smith M (1998) Crop evapotrans-
piration: guidelines for computing crop water requirements.
Irrigation and Drainage Paper No. 56, Food and Agriculture
Organization of the United Nations, Rome, Italy
Bondari K (2003) Statistical analysis of genotype × environment
interaction in agricultural research. In: Paper SD15, SESUG.
The Proceedings of the SouthEast SAS Users Group, St Pete
Beach
Bussan AJ, Burnside OC, Orf JH, Ristau EA, Puettmann KJ
(1997) Field evaluation of soybean (Glycine max) genotype
for weed competitiveness. Weed Sci 45:31–37
Campillo C, Prieto MH, Daza C, Moñino MJ, García MI (2008)
Using digital images to characterize canopy coverage and
light interception in a processing tomato crop. Hort Science
43:1780–1786.35
Clauw P, Coppens F, De Beuf K, Dhondt S, Van Daele T, Maleux
K, Storme V, Clement L, Gonzalez N, Inzé D (2015) Leaf
responses to mild drought stress in natural variants of
Arabidopsis. Plant Physiol 167:800–816
De Bruin JL, Pedersen P (2008) Soybean seed yield response to
planting date and seeding rate in the upper Midwest. Agron J
100:696–703
Dhanapal AP, Ray JD, Singh SK, Hoyos-Villegas V, Smith JR,
Purcell LC, King CA, Fritsch FB (2015a) Association map-
ping of total carotenoids in diverse soybean genotypes based
on leaf extracts and high-throughput canopy spectral reflec-
tance measurements. PLoS One 10(9):e0137213
Dhanapal AP, Ray JD, Singh SK, Hoyos-Villegas V, Smith JR,
Purcell LC, King CA, Cregan PB, Song Q, Fritsch FB
(2015b) Genome-wide association study (GWAS) of carbon
isotope ratio (δ13
C) in diverse soybean [Glycine max (L.)
Merr.] genotypes. Theor Appl Genet 128:73–91
Edward JT, Purcell LC, Karcher DE (2005) Soybean yield and
biomass responses to increasing plant population among
diverse maturity groups. II Light interception and utilization
Crop Sci 45:1778–1785
Edwards JT, Purcell LC (2005) Soybean yield and biomass re-
sponses to increasing plant population among diverse matu-
rity groups: I Agronomic characteristics. Crop Sci 45:1770–
1777
Fickett ND, Boerboom CM, Stoltenberg DE (2013) Soybean yield
loss potential associated with EarlySeason weed competition
across 64 site-years. Weed Sci 61:500–507
Fiorani F, Rascher U, Jahnke S, Schurr U (2012) Imaging plants
dynamics in heterogenic environments. Curr Opin Biotech
23:227–235
Gaspar AP, Conley SP (2015) Responses of canopy reflectance,
light interception, and soybean seed yield to replanting sub-
optimal stands. Crop Sci 55(1):377–385
Gifford RM, Thorne JH, Hitz WD, Giaquinta RT (1984) Crop
productivity and photoassimilate partitioning. Science 225:
801–808
Green JM, Owen MDK (2011) Herbicide-resistant crops: utilities
and limitations for herbicide resistant weed management. J
Agric Food Chem 59:5819–5829
Hao D, Cheng H, Yin Z, Cui S, Zhang D, Wang H, Yu D (2012)
Identification of single nucleotide polymorphisms and hap-
lotypes associated with yield and yield components in soy-
bean (Glycine max) landraces across multiple environments.
Theor Appl Genet 124:447–458
Hwang E, Song Q, Jia G, Specht JE, Hyten DL, Costa J, Cregan
PB (2014) A genome-wide association study of seed protein
and oil content in soybean. PLoS Genet 15:1
Jannink JL, Orf JH, Jordan NR, Shaw RG (2000) Index selection
for weed suppressive ability in soybean. Crop Sci 40:1087–
1094
Jannink JL, Jordan NR, Orf JH (2001) Feasibility of selection for
high weed suppressive ability in soybean: absence of
tradeoffs between rapid initial growth and sustained later
growth. Euphytica 120:291–300
Kaler AS, Dhanapal AP, Ray JD, King CA, Fritsch FB, Purcell LC
(2017a) Genome-wide association mapping of carbon iso-
tope and oxygen isotope ratios in diverse soybean genotypes.
Crop Sci 57:3085–3100
Kaler AS, Ray JD, King CA, Schapaugh WT, Purcell LC (2017b)
Genome-wide association mapping of canopy wilting in
diverse soybean genotypes. Theor Appl Genet 130:1–15
Karcher DE, Richardson MD (2005) Batch analysis of digital
images to evaluate turfgrass characteristics. Crop Sci 45:
1536–1539
Liebisch F, Kirchgessner N, Schneider D, Walter A, Hund A
(2015) Remote, aerial phenotyping of maize traits with a
mobile multi-sensor approach. Plant Methods 11:9
Liu X, Huang M, Fan B, Buckler ES, Zhang Z (2016) Iterative
usage of fixed and random effect models for powerful and
efficient genome-wide association studies. PLoS Genet
12(2):e1005767
Manandhar A, Sinclair TR, Rufty TW, Ghanem ME (2017) Leaf
emergence (phyllochron index) and leaf expansion response
to soil drying in cowpea genotypes. Physiol Plantarum 160:
201–208
McCarthy MI, Abecasis GR, Cardon LR, Goldstein DB, Little J,
Ioannidis JPA, Hirschhorn JN (2008) Genome-wide associa-
tion studies for complex traits: consensus, uncertainty and
challenges. Nature Rev Genet 9(5):356–369
Money D, Gardner K, Migicovsky Z, Schwaninger H, Zhong GY,
Myles S (2015) LinkImpute: fast and accurate genotype
imputation for non-model organisms. G3 5(11):23383–
22390
Nordborg M, Tavaré S (2002) Linkage disequilibrium: what his-
tory has to tell us. Trends Genet 18(2):83–90
Place GT, Reberg-Horton SC, Dickey DA, Carter TE (2011a)
Identifying soybean traits of interest for weed competition.
Crop Sci 51:2642–2654

Place GT, Reberg-Horton SC, Carter TE, Smith AN (2011b)
Effects of soybean seed size on weed competition. Agron J
103:175–181
Purcell LC (2000) Soybean canopy coverage and light interception
measurements using digital imagery. Crop Sci 40:834–837
Purcell LC, Specht JE (2004) Physiological traits for ameliorating
drought stress. In: Boema HR, Specht JE (eds) Soybeans:
improvement, production, and uses. 3rd ed. Madison, WI:
American Society of Agronomy. Pp, pp 569–620
Purcell LC, Edwards JT, Brye KR (2007) Soybean yield and
biomass responses to cumulative transpiration: questioning
widely held beliefs. Field Crop Res 101:10–18
Ray JD, Dhanapal AP, Singh SK, Hoyos-Villegas V, Smith JR,
Purcell LC, King CA, Boykin D, Cregan PB, Song Q,
Fritschi FB (2015) Genome-wide association study of ureide
concentration in diverse maturity group IV soybean [Glycine
max (L.) Merr.] accessions. G3 5(11):2391–2403
Rebetzke GJ, Ellis MH, Bonnett DG, Richards RA (2007)
Molecular mapping of genes for coleoptile growth in bread
wheat (Triticum aestivum L.) Theor Appl Genet 114:1173–
1183
Richards RA, Watt M, Rebetzke GJ (2007) Physiological traits
and cereal germplasm for sustainable agricultural systems.
Euphytica 154:409–425
Risch N, Merikangas K (1996) The future of genetic studies of
complex human diseases. Science 273(5281):1516–1517
SAS Institute (2013) The SAS system for Windows. Version 9.3.
SAS Inst. Inc., Cary, NC
Shaner DL (1995) Herbicide resistance: where are we? How did
we get here? Where are we going? Weed Technol 9:850–856
Slafer GA, Araus JL, Royo C, Garcia Del Moral LF (2005)
Promising eco-physiological traits for genetic improvement
of cereal yields in Mediterranean environments. Ann Appl
Biol 146:61–70
Song Q, Hyten DL, Jia G, Quigley CV, Fickus EW, Nelson RL,
Cregan PB (2013) Development and evaluation of
SoySNP50K, a high-density genotyping array for soybean.
PLoS One 8(1):e54985
Tardieu F, Tuberosa R (2010) Dissection and modelling of abiotic
stress tolerance in plants. Curr Opin Plant Biol 13:206–212
Tuberosa R, Salvi S, Giuliani S, Sanguineti MC, Bellotti M, Conti
S, Landi P (2007) Genome-wide approaches to investigate
and improve maize response to drought. Crop Sci 47:120–
141
Wen Z, Tan R, Yuan J, Bales C, Du W (2014) Genome-wide
association mapping of quantitative resistance to sudden
death syndrome in soybean. BMC Genomics 15:809
Xavier A, Hall B, Hearst AA, Cherkauer KA, Rainey KM (2017)
Genetic architecture of phenomic-enabled canopy coverage
in Glycine max. Genetics 206:1081–1089. https://doi.
org/10.1534/116.198713
Yu J, Pressoir G, Briggs WH, Vroh BI, Yamasaki M, Doebley JF,
McMullen MD, Gaut BS, Nielsen HJB, Kresovich S,
Buckler ES (2006) A unified mixed-model method for asso-
ciation mapping that accounts for multiple levels of related-
ness. Nat Genet 38:203–208
Zhang J, Song Q, Cregan PB, Nelson RL, Wang X, Wu J, Jiang
GL (2015) Genome-wide association study for flowering
time, maturity dates and plant height in early maturing soy-
bean (Glycine max) germplasm. BMC Genomics 16:217.
https://doi.org/10.1186/s12864-015-1441-4
Zhu C, Gore MA, Buckler ES, Yu J (2008) Status and prospects of
association mapping in plants. Plant Genome 1:5–20

Association mapping identifies loci for canopy coverage in diverse soybean genotypes

More Related Content

What's hot (20)

Similar to Association mapping identifies loci for canopy coverage in diverse soybean genotypes (20)

More from Avjinder (Avi) Kaler (20)

Recently uploaded (20)

Association mapping identifies loci for canopy coverage in diverse soybean genotypes