Kaler et al 2018 euphytica

Association mapping identifies loci for canopy temperature
under drought in diverse soybean genotypes
Avjinder S. Kaler . Jeffery D. Ray . William T. Schapaugh . Antonio R. Asebedo .
C. Andy King . E. E. Gbur . Larry C. Purcell
Received: 19 December 2017 / Accepted: 12 July 2018
Ó Springer Nature B.V. 2018
Abstract Drought stress is a global constraint for
crop production, and improving crop tolerance to
drought is of critical importance. Because transpira-
tion cools a crop canopy, a cool canopy under drought
indicates a genotype still has access to soil moisture.
Because measurements of canopy temperature may be
increased in scale in field environments, it is partic-
ularly attractive for large-scale, phenotypic evalua-
tions. Our objectives were to identify genomic regions
associated with canopy temperature (CT) and to
identify extreme genotypes for CT. A diverse panel
consisting of 345 maturity group IV soybean acces-
sions was evaluated in three environments for CT.
Within each environment CT was normalized (nCT)
on a scale from 0 to 1. A set of 31,260 polymorphic
single nucleotide polymorphisms (SNPs) with a minor
allele frequency C 5% was used for association
mapping of nCT. Association mapping identified 52
SNPs significantly associated with nCT, and these
SNPs likely tagged 34 different genomic regions.
Averaged across all environments, eight genomic
regions showed significant associations with nCT.
Several genes in the identified genomic regions had
reported functions related to transpiration or water
acquisition including root development, response to
abscisic acid, water deprivation, stomatal complex
morphogenesis, and signal transduction. Fifteen of the
SNPs associated with nCT were coincident with SNPs
for canopy wilting. Favorable alleles from significant
SNPs may be an important resource for pyramiding
genes, and several genotypes were identified as
sources of drought-tolerant alleles that could be used
in breeding programs for improving drought tolerance.
Keywords Drought tolerance Á High throughput
phenotyping Á Infrared canopy temperature
Abbreviations
AAE Across all environments
BLUP Best unbiased linear prediction
CT Canopy temperature
GEBV Genomic estimated breeding value
GWAM Genome-wide association mapping
LD Linkage disequilibrium
MAF Minor allele frequency
nCT Normalized canopy temperature
A. S. Kaler Á C. A. King Á L. C. Purcell (&)
Department of Crop, Soil, and Environmental Sciences,
University of Arkansas, Fayetteville, AR 72704, USA
e-mail: lpurcell@uark.edu
J. D. Ray
Crop Genetics Research Unit, USDA-ARS, 141
Experimental Station Road, Stoneville, MS 38776, USA
W. T. Schapaugh Á A. R. Asebedo
Department of Agronomy, Kansas State University,
Manhattan, KS 66506, USA
E. E. Gbur
Agriculture Statistics Lab, University of Arkansas,
Fayetteville, AR 72701, USA
123
Euphytica (2018)214:135
https://doi.org/10.1007/s10681-018-2215-2

QTLs Quantitative trait loci
SNPs Single nucleotide polymorphisms
TAGV True additive genetic value
Introduction
Drought is a major global constraint for crop produc-
tivity in rain-fed areas, which will make it difficult to
meet predicted food demand for a world population
that is expected to double by 2050 (Foley et al. 2011).
Currently, the average rate of increased cereal pro-
duction yield per year (1.3%) is lower than that
required (2.4%) to meet the future food demand (Ray
et al. 2013). Climate change not only affects temper-
ature, but it also affects the magnitude and distribution
of rainfall, which can result in a decrease in water
availability for critical times of the crop cycle (Feng
et al. 2013). Climate change also decreases the
predictability of rainfall and leads to increased
frequency of drought and flooding conditions (Dou-
glas et al. 2008). Developing drought-tolerant culti-
vars is a high priority for improving crop performance
in water-scarce environments. Soybean [Glycine max
(L.) Merr.] is among the most widely grown crops in
the world and is valuable because of its high oil and
protein concentration. Drought adversely affects soy-
bean yield to some degree at most developmental
stages, particularly during reproductive development
(Oya et al. 2004).
Direct selection of genotypes for grain yield under
water-limited environments has limited utility because
of low heritability, polygenic control, epistatic effects,
and genotype by environment interactions (Piepho
2000). Stomatal conductance regulates transpiration to
maintain the plant water balance (Gollan et al. 1986).
An early response of plants to drought stress is
stomatal closure, which serves to reduce water loss
through transpiration (Cornic and Massacci 1996).
Porometry is a method to determine stomatal response,
however, this approach is slow and laborious for a
large number of genotypes in a breeding program
(Jones 1979; Leport et al. 1999). Evaporative cooling
through transpiration is related to stomatal conduc-
tance, and variation in canopy temperature (CT) can
be used as an indicator for transpiration and stomatal-
conductance differences among genotypes (Jackson
et al. 1981; Jones et al. 2009). Genotypes that have a
faster-growing and deeper root system may extract
water from deeper in the soil profile than genotypes
with more shallow roots, resulting in greater soil
moisture availability, resulting in a cooler canopy
during drought. Access to soil moisture deep in the
profile may thereby stabilize yield in water-scarce
environments (Blackman and Davis 1985). Genotypes
may also conserve soil moisture when it is abundant in
the soil by limiting transpiration and then drawing
upon that conserved moisture during a drought when
other genotypes have exhausted their soil moisture
(King et al. 2009; Ries et al. 2012). Relative to
genotypes that exhaust their soil water supply quickly,
genotypes that conserve soil moisture when it is
abundant would expectantly have an increased CT
when soil moisture was abundant but a cooler canopy
during the onset of drought due to the conserved soil
water.
Field measurement of CT of a large number of
genotypes is difficult because many environmental
factors such as air temperature, humidity, wind speed,
solar radiation, as well as stomatal aperture affect leaf
temperature. Infrared thermography has been effective
in evaluating drought in different crops including
soybean and cotton (Gossypium hirsutum L.)
(O’Shaughnessy et al. 2011), and maize (Zea mays
L.) (Zia et al. 2011). Aerial infrared image analysis has
an advantage over the use of conventional IR
thermometers for screening of CT because a large
number of genotypes can be captured in a single image
(Merlot et al. 2002). Aerial thermal images provide
more rapid and accurate measurements of CT than
ground-based images, and this method does not
interfere with stomatal responses (Guilioni et al.
2008; Jones et al. 2009). Therefore, CT can be used
as a selection criterion to screen genotypic variation in
stomatal conductance in a breeding program under
drought stress conditions.
Traits related to drought tolerance are complex and
multi-genic and interact with the environment (Carter
et al. 1999). Complexity is mainly due to segregation
of alleles at many chromosomal regions, each with
small additive effects, and their interaction with other
alleles and with the environment (Tuberosa et al.
2007). Dissection of genetic control of CT can identify
the loci controlling CT variation and can be used to
improve crop productivity by selecting and pyramid-
ing those favorable loci into elite cultivars (Blum
123
135 Page 2 of 18 Euphytica (2018)214:135

2005). The identification of quantitative trait loci
(QTLs) associated with CT is one way to dissect the
genetic control (Dixit et al. 2014).
Advancement in high-throughput genotyping and
sequencing technologies have provided fast and low-
cost molecular markers, particularly single nucleotide
polymorphisms (SNPs) (Syvanen 2005). Genome-
wide association mapping (GWAM) is an alternative
approach to linkage mapping of bi-parental popula-
tions and can provide high mapping resolution for
complex trait variation (Nordborg and Tavare´ 2002;
Risch and Merikangas 1996). GWAM is based on
linkage disequilibrium (LD), due to non-random
association of alleles between genetic loci across the
genome (Zhu et al. 2008). The detection of QTLs
through GWAM depends on the level of LD between
functional loci and markers. In soybean, several
GWAM studies have been reported that identified
chromosomal regions associated with seed protein and
oil concentrations (Hwang et al. 2014), carotenoids
(Dhanapal et al. 2015a), carbon isotope ratio (d13
C)
(Dhanapal et al. 2015b) and oxygen isotope ratio
(d18
O) (Kaler et al. 2017a), canopy wilting (Kaler et al.
2017b), canopy closure or the fraction of light
intercepted (Kaler et al., 2018), agronomic traits
(Wen et al. 2014), ureide concentrations (Ray et al.
2015), and the fraction of N derived from the
atmosphere (Dhanapal et al. 2015c). The number of
GWAM studies in soybean is likely to increase due to
recent genotyping of more than 19,000 accessions of
the USDA-ARS Soybean Germplasm collection that
provided approximating 50,000 SNPs (Song et al.
2013) that are available at SoyBase (www.soybase.
org).
To date, there has been no report of mapping CT
either in bi-parental populations or GWAM panels in
soybean. However, there are mapping studies of CT in
other crop species including wheat (Triticum aestivum
L.; Rebetzke et al. 2013), rice (Oryza sativa L.; Liu
et al. 2005), and maize (Zea mays L.; Liu et al. 2011).
In this research, a set of 31,260 polymorphic SNPs was
used for GWAM. Our objectives of this research were
to use association mapping to explore the genotypic
variation of nCT in a panel of 345 diverse maturity
group IV accessions, to identify the significant SNPs
associated with nCT, and identify extreme genotypes
for nCT.
Materials and methods
Field experiments
A panel of 345 maturity group IV soybean accessions
was evaluated in three environments including the
Pine Tree Research Station, AR (35°70
N, 90°550
W) in
2016 (PT16), Rohwer Research Station, AR (33°480
N,
91°170
W) in 2016 (RH16), and Salina, KS (38°700
N,
97°600
W) in 2016 (SA16). These accessions were
selected from the USDA-ARS Soybean Germplasm
Collection based on GRIN (Germplasm Resources
Information Network, www.ars-grin.gov) data. We
elected to evaluate maturity group IV accessions
because they are widely grown in the midsouthern
U.S. (Salmeron et al. 2016), and they represent a
bridge between northern and southern germplasm
pools in the U.S. Genotypes were selected for geo-
graphic diversity and for having acceptable agronomic
traits for yield, lodging, and shattering as discussed by
Dhanapal et al. (2015b). These diverse accessions
originated from 10 different nations including South
Korea, China, Japan, North Korea, Georgia, Russia,
Taiwan, India, Mexico, and Romania.
The 345 accessions were grown in a randomized
complete block design with two replications at each
environment. These accessions were sown on 23 May
2016 at RH16 on a Sharkey silty clay, 2 June 2016 at
PT16 on a Calloway silt loam, and 15 June 2016 at
SA16 on a Hord silt loam. Seeds were planted at a
density of 37 m-2
at a depth of 2.5 cm. At SA16, there
were two-row plots that were 3.65 m in length with
0.76 m row spacing. At PT16 and RH16, plots
consisted of seven rows, 19-cm apart and 4.57 m in
length. Herbicides and insecticides were applied as
recommended to control weeds and insects.
Soil water deficit was estimated for each environ-
ment from the day of planting as described by Purcell
et al. (2007). The Penman–Monteith approach was
used to determine potential evapotranspiration (Eto)
for a given day (Allen et al. 1998), and Eto was
multiplied by the estimated fraction of radiation
intercepted by the crop for that day, which served as
a crop coefficient (equivalent to canopy coverage).
Estimated soil–water deficits were cumulated and
adjusted with rainfall additions as needed.
123
Euphytica (2018)214:135 Page 3 of 18 135

Canopy temperature evaluation
Aerial thermal infrared image analysis was imple-
mented to evaluate nCT. At PT16 and RH16, a
tethered balloon, which was approximately 2 m in
diameter with a lifting capacity of 1.5 kg when filled
with helium (www.giant-inflatables.com), was used as
an aerial platform to take infrared images from a
height of approximately 75 m when wind speed
was B 2 m s-1
. A thermal infrared camera, FLIR Tau
2 640 (FLIR, Goleta CA) with 640 9 512 resolution
and with a 13 mm lens, collected data for wavelengths
from 7.5 to 13.5 lm. This camera is small and light
weight (110 g) with a Noise Equivalent Differential
Temperature (NEdT) less than 50 mK at f/1.0 with
FLIR proprietary noise reduction. The video was
recorded using a digital video recorder (www.
foxtechfpv.com, model DV02) that was mounted on
a picavet (http://www.armadale.org.uk/kitebasic.
htm), which reduced the motion of the camera when
suspended from the balloon. Images were evaluated
based on the 256 different shades of gray values that
differed by approximately 0.05 °C (i.e., 50 mK), with
a range of approximate 12.8 °C (256 * 0.05 °C) at a
specific focal plane temperature. Due to high sensi-
tivity, the FLIR Tau 2 640 can detect small differences
in temperature but does not provide absolute temper-
ature values. Therefore, differences in the shades of
gray among plots from the infrared images represent
relative temperature differences.
Aerial canopy temperature at Salina (SA16) was
measured on August 16th at 3:00 p.m. using a DJI
S1000 octocopter outfitted with a FLIR VUE Pro R
(FLIR, Goleta CA) with a 13 mm lens and 640 9 512
resolution that recorded wavelengths from 7.5 to
13.5 lm. The FLIR VUE Pro R recorded 14-bit TIFF
images in one second intervals and tagged images with
GPS location using an onboard GPS. Flight was
conducted using autonomous flight mode with altitude
set to 120 meters above ground level, 85% side
overlap for flight lines, and a forward flight speed of
4 m s-1
. Ground control points were established at the
field corners and throughout the study area using
calibration tiles approximately 1 m2
to ensure accu-
rate extraction of plot-level canopy temperature
measured in brightness values. Plot thermal brightness
values were extracted from each plot using ArcMAP
10.5 (ESRI 2017. ArcGIS Desktop: Release 10.5
Redlands, CA: Environmental Systems Research
Institute).
Because canopy temperature was measured differ-
ently at SA16, data were normalized for all environ-
ments on a scale from 0 to 1. Normalized CT (nCT)
was calculated as:
nCT ¼
xi À min xð Þ
max xð Þ À min xð Þ
where xi represents the ith CT measurement in
environment X and min(x) and max(x) represent the
minimum and maximum CT values for environment
X, respectively.
Phenotype statistics
Descriptive statistics and Pearson correlation analysis
for nCT were performed using the PROC UNIVARI-
ATE and PROC CORR procedures (a = 0.05) of SAS
version 9.4 (SAS Institute 2013), respectively. Anal-
ysis of variance (ANOVA) was conducted using the
PROC MIXED procedure (a = 0.05) of SAS 9.4,
based on a model suggested by Bondari (2003), yijk ¼
l þ Gi þ Ej þ GEð ÞijþBk ijð Þ þ eijk; where l is the total
mean, Gi is the genotypic effect of the ith genotype, Ej
is the effect of the jth environment, GEð Þij is the
interaction effect between the ith genotype and the jth
environment, Bk ijð Þ is the effect of replication within
the jth environment, and eijk is a random error
following N 0; r2
e
À Á
. Genotype was treated as a fixed
effect and replication within the environment was
considered as a random effect.
On an entry-mean basis, broad sense heritability
was calculated as H2
¼ r2
G= r2
G þ
r2
GE
k

þ
r2
e
rk

;
where r2
G is the genotypic variance, r2
GE is the
genotype by environment variance, r2
e is the residual
variance, k is the number of environments, and r is the
number of replications. These variance components
were estimated using the PROC VARCOMP proce-
dure of SAS 9.4 with the REML (Restricted Maximum
Likelihood Estimation) method. The Best Linear
Unbiased Prediction (BLUP) values for each inde-
pendent environment and across all environments
were estimated by using the R package ‘‘lme4’’, and
BLUP values were then used in GWAM analysis.
123

Genotyping
The Illumina Infinium SoySNP50 K iSelect SNP
Beadchip provided 42,509 SNPs for the 345 genotypes
used in this experiment (www.soybase.org). Markers
that were monophorpic, had a minor allele frequency
(MAF) 5%, or had a missing rate larger than 10%
were excluded, leaving 31,260 SNPs for further
analysis. Imputation of remaining missing SNPs of the
31,260 SNPs used in the analysis was applied using a
LD-kNNi method, which is based on a k-nearest-
neighbor-genotype (Money et al. 2015). These 31,260
polymorphic SNPs were then used for association
testing to identify SNPs significantly associated with
nCT.
Genome-wide association analysis
Association analysis using a diverse population can
induce false positive associations due to population
stratification. A mixed linear model (MLM) is most
commonly used to reduce false positives by incorpo-
rating the family relatedness and population structure
in the model (Yu et al. 2006; Zhang et al. 2010).
However, these adjustments can also compromise true
positive associations (Liu et al. 2016). Previously, we
reported that the Fixed and random model Circulating
Probability Unification (FarmCPU), developed by Liu
et al. (2016), effectively controlled both false positives
and false negatives (Kaler et al. 2017b), and this model
was used in the present research. A threshold value
(- Log10(P) C 3.5), which is equivalent to a P
value B 0.0003, was used to declare a significant
association of SNPs with nCT. This threshold level is
more stringent than that reported in other soybean
GWAM studies (Dhanapal et al. 2015a, b; Hao et al.
2012; Hwang et al. 2014; Zhang et al. 2015).
Significant SNPs present in more than one environ-
ment were identified using a threshold value of
P B 0.05 but only when the SNP had an association
of at least P B 0.0003 in a second environment.
Extreme genotypes identification
Extreme genotypes for nCT were selected based on
genomic estimated breeding values (GEBVs), which
utilize a genomic-relationship matrix and phenotypic
data (Clark and van der Werf 2013; Zhang et al. 2007)
along with true additive genetic values (TAGVs) for
genotypes. The GEBVs were estimated using Efficient
Mixed Model Association (EMMA) algorithms in
‘‘sommer’’ R package (Covarrubias-Pazaran 2016).
Allelic effects of all significant SNPs (P B 0.0003)
were used to calculate the TAGV of each accession
(Falconer and Mackay 1996; Kaler et al. 2017b).
Allelic effects were calculated by taking a difference
in mean nCT between genotypes with major allele and
minor allele. Major alleles were considered as favor-
able if they were associated with a reduction in the
nCT. To estimate the TAGV for each accession, the
absolute value of the allelic effect of each significant
SNP was considered as a negative value if an
accession had a favorable allele of a significant SNP
at that location (that is, if the allelic effect decreased
nCT). Otherwise, if the allelic effect was unfavorable
(i.e., increased nCT), the allelic effect for a SNP was
considered as a positive value. All positive and
negative allelic values were summed to estimate the
TAGV of each accession.
Candidate gene identification
All significant SNPs at level of - Log10(P) C 3.5
were used to identify candidate genes for each
environment and across all environments. Candidate
genes, their associated functional annotation, and
biological function were identified within ± 10 kb
using Glyma2.0 and NCBI RefSeq gene models in
Soybase (www.soybase.org) with consideration for
those that may have direct association with CT, tran-
spiration, and water transport. The ± 10 kb distance
was chosen because it approximates the average dis-
tance between SNPs (18 kb).
Results
Phenotype descriptions
Measurements of nCT were made on 345 maturity
group IV soybean accessions for three environments
(SA16, RH16, and PT16) between full bloom and
early podfill when the canopy was closed. Environ-
mental conditions including solar radiation, maximum
and minimum temperature, and daily rainfall were
collected at each environment. On the day of mea-
surement, maximum and minimum temperature was
34 and 23 °C at PT16, 35 and 24 °C at RH16, and 34
123

and 26 °C at SA16, respectively. On the measurement
date, photosynthetically active radiation was
26.2 MJ m-2
at PT16, 27.8 MJ m-2
at RH16, and
29.5 MJ m-2
at SA16. Prior to nCT measurement,
there had been no rainfall for 13 days at PT16, 19 days
at RH16, and 9 days at SA16. This resulted in an
estimated soil moisture deficit exceeding 52 mm at
RH16, 60 mm at PT16, and 50 mm at SA16. Irrigation
is recommended at 35 mm for silt loam soils (PT16,
SA16) and 50 mm for clay soils (RH16) (Purcell et al.
2007); hence, there was considerable drought at all
locations on the measurement days.
Canopy temperature was normalized in the range
[0, 1] so that data are on the same scale for each
environment. There was a broad range of nCT within
each environment, indicating wide phenotypic varia-
tion. Figure 1a shows that nCT data were normally
distributed for each environment and when averaged
across environments. BLUP values of nCT for each
environment and averaged across all environments
were calculated to reduce the effect of extreme values.
These BLUPs were also normally distributed but had
less variation than phenotypic values (Fig. 1b). Anal-
ysis of variance indicated that genotype, environment,
and their interaction had significant effects (P B 0.05)
on nCT (data not shown). There was a weak significant
(P B 0.05) positive correlation for nCT between
SA16 and PT16 (r = 0.13); however, there was no
significant correlation for nCT between SA16 and
RH16 or between PT16 and RH16. Broad sense
heritability of nCT was 45% for PT16, 55% for RH16,
26% for SA16, and 19% across all environments.
The GEBVs were determined using a genomic-
relationship matrix and phenotypic data of 345
accessions. The 345 accessions were ranked from
lowest to highest based on the average GEBVs of nCT
across all environments (Table 1). Based on the
average GEBV ranking, the 15 accessions with lowest
GEBV for nCT and 15 accessions with highest GEBV
for nCT were selected. Accessions with low GEBVs
(cool nCT) also had large negative TAGVs (- 2.33 to
- 0.08), and in contrast, accessions with high GEBVs
(warm nCT) had large positive TAGVs (1.65–3.02)
(Table 1). One extreme accession, PI 592940, had a
large negative GEBV (- 0.095), a relatively cool nCT
(0.35) and a large negative TAGV (- 1.27) (Fig. 2).
In contrast, PI 398640 had a large positive GEBV
(0.068), a relatively warm nCT (0.64), and a large
positive TAGV (2.94) (Fig. 2). The 15 accessions with
the lowest GEBVs and cooler nCT averaged across all
environments were from China (9 accessions), South
Korea (2 accessions), and one each from Mexico,
North Korea, Japan, and Georgia (Table 1). The 15
accessions with the highest GEBVs and warmer nCT
averaged across all environments were from South
Korea (11 accessions), Japan (3 accessions), and
Georgia (1 accession) (Table 1).
Genome-wide association analysis
Association mapping of BLUP nCT identified 52
significant SNPs in three environments associated
Fig. 1 Distribution of normalized canopy temperature for each
of the three environments (Pine Tree 2016 (PT16), Rohwer 2016
(RH16), and Salina 2016(SA16) and average across all
environments (AAE). The normalized means (a), Best linear
unbiased predictions (BLUPs) (b)
123

with BLUP values of nCT at a significance level of
- Log10(P) C 3.5; P B 0.0003 (Table 2; Fig. 3). Of
the 52 SNPs, four were significant in more than one
environment. Significant SNPs that were closely
spaced and present within the same LD block were
considered as one locus. These 52 significant SNP
associations likely identified 34 putative loci
(Table 2). Two putative loci on Gm03 and Gm04
were each identified by four closely spaced SNPs;
three putative loci on Gm14, Gm15, and Gm18 were
each identified by three closely spaced SNPs; six
putative loci on Gm02, Gm03, Gm04, Gm07, Gm08,
and Gm14 were each identified by two closely-spaced
SNPs. The remaining loci were identified by one SNP
(Table 2). The allelic effect (difference in mean nCT
between genotypes with major allele and minor allele)
for these significant SNPs ranged from - 0.070 to
0.161 (Table 2). A positive sign of an allelic effect
indicates that the minor allele was favorable and
associated with reduced nCT, and a negative sign
indicates that the major allele was favorable and
associated with reduced nCT.
Association analysis of BLUP nCT averaged across
all environments (AAE) identified eight significant
Table 1 The 15 accessions
with the lowest and highest
ranking for canopy
temperature (CT) based on
genomic estimated breeding
values (GEBVs) of
averaged normalized
canopy temperature (nCT)
across all environments
(AAE)
a
TAGV true additive
breeding values
b
Rank ranking of accessions
based on the genomic
estimated breeding values
of averaged normalized
canopy temperature
c
CW canopy wilting
presented as genomic
estimated breeding values
Accession Province Country AAE TAGVa
GEBVs Rankb
CWc
Cooler canopy temperature
PI592940 Sichuan China 0.35 - 1.27 - 0.095 1 - 5.53
PI567620B Henan China 0.42 - 1.94 - 0.093 2 - 4.89
PI592937 Sichuan China 0.33 - 2.27 - 0.092 3 - 8.94
PI602501 Jiangsu China 0.36 - 2.33 - 0.088 4 - 6.77
PI432359 Jalisco Mexico 0.35 - 0.66 - 0.085 5 - 3.82
PI603174A unknown North Korea 0.38 - 1.23 - 0.085 6 0.24
PI424405B Cholla Puk South Korea 0.31 - 0.63 - 0.084 7 - 4.93
PI424159B Kyongsang Puk South Korea 0.35 - 1.33 - 0.083 8 - 4.56
PI404167 Unknown China 0.53 - 1.04 - 0.083 9 - 4.07
PI567540A Shandong China 0.32 - 2.29 - 0.081 10 - 5.12
PI567500 Hebei China 0.37 - 1.98 - 0.077 12 - 3.85
PI417278 Unknown Japan 0.51 - 1.98 - 0.075 11 - 3.84
PI603543B Shanxi China 0.42 - 0.20 - 0.075 13 - 8.84
PI407735 Beijing China 0.35 - 1.32 - 0.073 14 - 6.69
PI567201D Unknown Georgia 0.46 - 0.08 - 0.071 15 - 5.31
Warmer canopy temperature
PI398772 Chungchong Nam South Korea 0.68 2.76 0.041 331 - 0.69
PI423890C Akita Japan 0.63 2.72 0.042 332 15.12
PI423888 Akita Japan 0.63 1.65 0.044 333 9.90
PI398298 Kyonggi South Korea 0.55 2.79 0.045 334 - 2.17
PI424381 Chungchong Puk South Korea 0.58 2.88 0.047 335 13.80
PI442012B Kyonggi South Korea 0.63 3.02 0.049 336 12.90
PI404159 Unknown Georgia 0.68 2.72 0.051 337 15.36
PI424549A Kyongsang Puk South Korea 0.71 2.79 0.053 338 - 1.92
PI423796B Kangwon South Korea 0.65 3.02 0.055 339 - 1.67
PI399036 Kyongsang Nam South Korea 0.64 3.02 0.055 340 7.91
PI424435 Cholla Nam South Korea 0.66 2.63 0.056 342 5.17
PI398939 Cholla Puk South Korea 0.60 2.63 0.056 341 5.18
PI274423 Miyagi Japan 0.58 2.63 0.058 343 5.20
PI424263 Kangwon South Korea 0.64 3.02 0.064 344 - 1.97
PI398640 Chungchong Puk South Korea 0.64 2.94 0.068 345 4.72
123

SNP associations at - Log10 (P) C 3.5; P B 0.0003
(Table 2; Fig. 3). Of these eight SNPs, three were
common to the 52 significant SNPs identified in the
three individual environments (Table 2). These SNPs
likely tagged eight different loci (Table 2). The allelic
effect (difference in mean nCT between genotypes
with major allele and minor allele) of nCT for these
SNPs ranged from - 0.014 to 0.095 (Table 2). The list
of 52 significant SNPs in three environments and eight
significant SNPs averaged across environments, their
corresponding MAF, major or minor allele, allelic
effect, and common environments are listed in
Table 2.
Candidate gene identification
A total of 52 SNPs significantly associated with nCT
from three environments and eight SNPs significantly
associated with AAE were used to identify the
potential genes within ± 10 kb of the respective
SNPs in SoyBase (www.soybase.org) using gene
models including Glyma2.0 and NCBI RefSeq. Based
on these significant SNPs, 52 genes for SNPs identi-
fied from three environments and eight genes for SNPs
identified for AAE were identified. A list of closely
located SNP ID, gene symbols, their associated func-
tional annotation, and biological function are reported
in Table 3. Based on this identification, 23 significant
SNPs out of 52 from single environments and three
significant SNPs out of eight for AAE were located
within genes, and the remaining SNPs were present
within ± 10 kb of the genes on genomic regions.
Genes potentially associated with nCT included
annotated biological function for root hair elongation,
root development, response to abscisic acid stimulus
and water deprivation, stomatal complex morpho-
genesis, and signal transduction (Table 3).
Discussion
In this research, a panel of 345 maturity group IV
soybean accessions were evaluated for nCT in three
environments. To our knowledge, this is the first
identification of molecular markers associated with
canopy temperature in soybean and the first report of
the association between canopy temperature and
canopy wilting. Canopy temperature measurements
are a potentially powerful means of phenotyping crops
for drought tolerance because it is amenable to high-
throughput methodologies (e.g., Araus and Cairns
2014; Chapman et al., 2014) and because it is an
indirect measure of stomatal conductance and the
access of plants to available soil moisture (Yousfi
et al., 2016).
Fig. 2 Distribution of average normalized canopy temperature
(nCT) across all environments (a), genomic estimated breeding
values (b), and true additive genetic values of accessions (c).
Both extremes were selected based on the canopy wilting; PI
592940 was slowest wilting genotype and PI 398640 was the
fastest wilting genotype in our GWAM panel
123

Table 2 List of significant SNPs associated with normalized
canopy temperature (nCT) in three environments (ENV)
including Pine Tree in 2016 (PT16), Rohwer in 2016
(RH16), Salina in 2016 (SA16), and averaged across all
environments (AAE) using the fixed and random model
circulating probability unification (FarmCPU) model with
threshold P value of (- Log10(P) C 3.5; P B 0.0003)
Locus CHRa
Location SNP_ID Alleleb
- Log10
(P)
Allele
effectc
ENV Common
ENVd
Single ENV
1 1 32,846,138 BARC_1.01_Gm_01_32846138_A_G A/G 4.30 0.079 RH16
2 1 39,939,520 BARC_1.01_Gm_01_39939520_A_G A/G 10.60 - 0.070 PT16
3 2 9,744,668 BARC_1.01_Gm_02_9744668_T_C T/C 4.30 0.060 RH16
2 9,776,807 BARC_1.01_Gm_02_9776807_G_A G/A 4.80 0.066 RH16
4 3 164,959 BARC_1.01_Gm_03_164959_T_G T/G 5.20 0.074 RH16
3 168,228 BARC_1.01_Gm_03_168228_A_G A/G 4.30 0.063 RH16
5 3 2,456,859 BARC_1.01_Gm_03_2456859_A_G G/A 3.50 - 0.028 SA16
6 3 3,827,087 BARC_1.01_Gm_03_3827087_G_A A/G 7.10 0.102 PT16
7 3 4,957,847 BARC_1.01_Gm_03_4957847_T_G G/T 4.50 - 0.010 SA16
8 3 40,278,033 BARC_1.01_Gm_03_40278033_G_A G/A 4.40 0.081 RH16
3 40,466,433 BARC_1.01_Gm_03_40466433_C_T C/T 4.60 0.082 RH16
3 40,467,180 BARC_1.01_Gm_03_40467180_G_A G/A 4.60 0.082 RH16
3 40,516,071 BARC_1.01_Gm_03_40516071_A_G A/G 4.30 0.080 RH16
9 4 7,957,588 BARC_1.01_Gm_04_7957588_G_T T/G 4.30 0.060 RH16
4 8,017,920 BARC_1.01_Gm_04_8017920_T_C C/T 4.30 0.062 RH16
4 8,019,074 BARC_1.01_Gm_04_8019074_G_A A/G 4.90 0.065 RH16
4 8,023,658 BARC_1.01_Gm_04_8023658_C_T T/C 4.30 0.062 RH16
10 4 43,390,997 BARC_1.01_Gm_04_43390997_A_C C/A 4.00 - 0.005 PT16
11 4 46,083,177 BARC_1.01_Gm_04_46083177_C_T T/C 4.60 0.059 RH16
4 46,086,046 BARC_1.01_Gm_04_46086046_G_A A/G 4.60 0.059 RH16
12 6 12,426,395 BARC_1.01_Gm_06_12426395_T_G G/T 3.50 0.021 SA16
13 7 3,234,327 BARC_1.01_Gm_07_3234327_T_G G/T 4.00 0.113 SA16
7 3,851,184 BARC_1.01_Gm_07_3851184_A_G A/G 6.00 0.120 PT16
14 7 7,536,244 BARC_1.01_Gm_07_7536244_C_A C/A 4.40 0.112 RH16
15 8 45,270,892 BARC_1.01_Gm_08_45270892_A_G G/A 3.80 0.013 PT16
8 45,671,888 BARC_1.01_Gm_08_45671888_A_C C/A 4.00 0.161 SA16
16 9 5,057,308 BARC_1.01_Gm_09_5057308_C_T T/C 4.00 0.019 PT16 PT16/RH16
17 9 40,407,114 BARC_1.01_Gm_09_40407114_C_T T/C 4.40 0.013 SA16
18 9 43,595,722 BARC_1.01_Gm_09_43595722_A_G A/G 4.10 0.038 PT16
19 10 38,249,878 BARC_1.01_Gm_10_38249878_T_G G/T 4.20 0.062 RH16
20 10 41,100,669 BARC_1.01_Gm_10_41100669_A_G G/A 4.40 0.069 RH16
21 11 7,251,966 BARC_1.01_Gm_11_7251966_C_T T/C 6.80 0.097 PT16
22 11 36,244,289 BARC_1.01_Gm_11_36244289_A_G A/G 4.20 0.008 PT16 PT16/RH16
23 14 2,221,273 BARC_1.01_Gm_14_2221273_T_C T/C 4.60 0.059 RH16
14 2,311,158 BARC_1.01_Gm_14_2311158_G_A G/A 5.00 0.067 RH16
24 14 4,064,786 BARC_1.01_Gm_14_4064786_C_T C/T 4.80 0.067 RH16
14 4,430,386 BARC_1.01_Gm_14_4430386_G_T G/T 5.20 0.084 RH16
14 4,853,955 BARC_1.01_Gm_14_4853955_A_G G/A 4.30 - 0.041 PT16
25 14 7,052,209 BARC_1.01_Gm_14_7052209_A_G A/G 5.50 - 0.068 PT16
26 14 47,305,241 BARC_1.01_Gm_14_47305241_T_G T/G 5.70 0.035 SA16
27 15 15,726,428 BARC_1.01_Gm_15_15726428_C_T T/C 5.50 0.090 RH16 RH16/PT16
123

We found a wide phenotypic variation within each
environment for nCT and such variation is important
for dissecting complex traits through association
mapping (McCarthy et al. 2008). The 15 accessions
with the lowest (i.e. coolest) nCT in ranking also had
considerably lower GEBVs (- 0.09) and large nega-
tive TAGVs (- 2.33, Table 1). In contrast, the 15
accessions with the highest (i.e. warmest) nCT in
ranking had considerably higher GEBVs (0.06) and
large TAGVs (3.02, Table 1).
Extreme genotypes were selected using TAGVs
and GEBVs of nCT, and these genotypes were also
extremes for canopy wilting (Kaler et al. 2017b). In
addition to the nCT measurement, we also rated
canopy wilting (CW) at each of these experiments
using the following scale: 0 (no wilting), 20 (slight
wilting with leaf wilting and rolling in the upper
portion of the canopy), 40 (moderate wilting in the
upper portion of the canopy and some wilting
throughout the canopy), 60 (severe wilting throughout
the canopy), 80 (severe wilting with some leaf
dessication), and 100 (severe wilting throughout the
canopy and dying plants) (King et al. 2009). There was
a significant (P B 0.05) positive correlation between
CW and nCT when phenotypic data were averaged
over environments (r = 0.25). The correlation coeffi-
cient between CW and nCT increased when using
Table 2 continued
Locus CHRa
Location SNP_ID Alleleb
- Log10
(P)
Allele
effectc
ENV Common
ENVd
15 15,729,124 BARC_1.01_Gm_15_15729124_T_C C/T 5.50 0.090 RH16 RH16/PT16
15 15,742,691 BARC_1.01_Gm_15_15742691_C_A A/C 4.20 0.081 RH16
28 16 35,807,551 BARC_1.01_Gm_16_35807551_G_A G/A 4.50 - 0.033 SA16
29 17 11,546,048 BARC_1.01_Gm_17_11546048_A_G A/G 5.50 0.019 PT16
30 17 38,712,454 BARC_1.01_Gm_17_38712454_G_A A/G 4.90 0.096 PT16
31 18 11,947,921 BARC_1.01_Gm_18_11947921_C_T C/T 5.20 0.107 RH16
32 18 13,037,246 BARC_1.01_Gm_18_13037246_G_A G/A 4.20 0.088 RH16
18 13,041,332 BARC_1.01_Gm_18_13041332_G_A G/A 4.20 0.088 RH16
18 13,117,752 BARC_1.01_Gm_18_13117752_G_T G/T 4.30 0.088 RH16
33 18 46,218,075 BARC_1.01_Gm_18_46218075_G_A A/G 4.20 0.059 RH16
34 18 60,748,254 BARC_1.01_Gm_18_60748254_C_T C/T 5.30 - 0.053 PT16
AAE
1 1 4,720,160 BARC_1.01_Gm_01_4720160_C_T C/T 7.30 0.074 AAE
2 3 3,497,393 BARC_1.01_Gm_03_3497393_A_C C/A 4.60 - 0.014 AAE
3 4 8,019,074 BARC_1.01_Gm_04_8019074_G_A A/G 10.60 0.052 AAE RH16
4 7 43,182,856 BARC_1.01_Gm_07_43182856_G_A G/A 6.70 0.013 AAE
5 13 24,858,209 BARC_1.01_Gm_13_24858209_A_G A/G 4.30 0.001 AAE
6 13 34,845,629 BARC_1.01_Gm_13_34845629_A_G G/A 3.50 0.095 AAE
7 15 15,729,124 BARC_1.01_Gm_15_15729124_T_C C/T 5.20 0.074 AAE RH16/PT16
8 18 13,037,246 BARC_1.01_Gm_18_13037246_G_A G/A 4.00 0.078 AAE RH16
a
CHR Glycine max chromosome number
b
Allele major/minor alleles of single nucleotide polymorphism
c
Allelic effect difference in mean normalized CT between genotypes with major allele and minor allele. Negative sign indicates that
major allele is associated with reduced CT. Positive sign indicates that minor allele is associated with reduced CT
d
Common ENV Indicates that SNP is present in more than one environment. Bold area represents the common environment
cFig. 3 Manhattan plots of - Log10 (P) versus chromosomal
position of significant SNP associations of normalized canopy
temperature (nCT) for three environments; a Pine Tree 2016,
b Rohwer 2016, c Salina 2016, and d averaged nCT across all
environments (AAE) using the FarmCPU model. Red line
represents the association threshold (- Log10 (P) C 3.5;
P B 0.0003)
123

123

Table 3 List of significant SNPs associated with normalized
canopy temperature (nCT) and potential genes based on 52
SNPs identified from three environments and eight SNPs for
nCT averaged across all environment (AAE). Bold areas
indicates SNPs that were located within genes
Locus SNP_ID Gene symbola
Functional annotations (biological function)
Single environment
1 BARC_1.01_Gm_01_32846138_A_G Glyma.01g102500 NAD(P)-linked oxidoreductase superfamily protein
(response to water deprivation)
2 BARC_1.01_Gm_01_39939520_A_G Glyma.01g120300 Leucine Rich Repeat (stomatal complex morphogenesis)
3 BARC_1.01_Gm_02_9744668_T_C Glyma.02g103600 Ribosomal protein S9 (translation)
BARC_1.01_Gm_02_9776807_G_A Glyma.02g103800 Ubiquitin protein ligase (ubiquitination)
4 BARC_1.01_Gm_03_164959_T_G Glyma.03g001400 RNA-binding protein
BARC_1.01_Gm_03_168228_A_G Glyma.03g001500 Syringolide-induced protein
5 BARC_1.01_Gm_03_2456859_A_G Glyma.03g023400 Uncharacterized protein
6 BARC_1.01_Gm_03_3827087_G_A Glyma.03g032900 Topoisomerase-related protein (embryo development)
7 BARC_1.01_Gm_03_4957847_T_G Glyma.03g039700 Peroxidase (response to oxidative stress)
8 BARC_1.01_Gm_03_40278033_G_A Glyma.03g192000 Uncharacterized protein (cell differentiation)
BARC_1.01_Gm_03_40466433_C_T Glyma.03g193800 Inosine-uridine preferring nucleoside hydrolase (uridine
catabolic process)
BARC_1.01_Gm_03_40467180_G_A Glyma.03g193800 Inosine-uridine preferring nucleoside hydrolase (uridine
catabolic process)
BARC_1.01_Gm_03_40516071_A_G Glyma.03g194500 Ribonuclease (aging)
9 BARC_1.01_Gm_04_7957588_G_T Glyma.04g090400 Aryl-alcohol dehydrogenase (oxidation–reduction
process)
BARC_1.01_Gm_04_8017920_T_C Glyma.04g091200 Rhamnogalacturonate lyase
BARC_1.01_Gm_04_8019074_G_A Glyma.04g091200 Rhamnogalacturonate lyase
BARC_1.01_Gm_04_8023658_C_T Glyma.04g091200 Rhamnogalacturonate lyase
10 BARC_1.01_Gm_04_43390997_A_C Glyma.04g173400 ENOLASE (response to abscisic acid stimulus)
11 BARC_1.01_Gm_04_46083177_C_T Glyma.04g190100 Ribonucleoprotein (leaf morphogenesis)
BARC_1.01_Gm_04_46086046_G_A Glyma.04g190100 Ribonucleoprotein (leaf morphogenesis)
12 BARC_1.01_Gm_06_12426395_T_G Glyma.06g152500 AAA-TYPE ATPase family protein (chloroplast
organization)
13 BARC_1.01_Gm_07_3234327_T_G Glyma.07g039400 GRAS family transcription factor (regulation of
transcription)
BARC_1.01_Gm_07_3851184_A_G Glyma.07g046000 Pectinesterase (cell wall modification)
14 BARC_1.01_Gm_07_7536244_C_A Glyma.07g082000 2-Hydroxyacid Dehydrogenase (oxidation–reduction
process)
BARC_1.01_Gm_08_45671888_A_C Glyma.08g340500 Eukaryotic Translation Initiation Factor 4 Gamma
(response to abscisic acid stimulus)
16 BARC_1.01_Gm_09_5057308_C_T Glyma.09g056300 Plasma membrane H 1 -ATPase (response to water
deprivation)
17 BARC_1.01_Gm_09_40407114_C_T Glyma.09g179400 Harpin-induced protein 1 (Hin1) (root hair elongation)
18 BARC_1.01_Gm_09_43595722_A_G Glyma.09g212000 Domain of unknown function (DUF1995) (photosynthesis)
19 BARC_1.01_Gm_10_38249878_T_G Glyma.10g147200 Serine-type peptidase activity (response to hypoxia)
21 BARC_1.01_Gm_11_7251966_C_T Glyma.11g095600 Hexokinase (root hair cell development)
22 BARC_1.01_Gm_11_36244289_A_G Glyma.11g257000 Transcription factor jumonji (jmjC) domain-containing
protein
23 BARC_1.01_Gm_14_2221273_T_C Glyma.14g030600 Pleckstrin homology (PH) domain-containing protein
(signal transduction)
123

GEBV (r = 0.35) and TAGV (r = 0.47). One geno-
type, PI 592940, had a relatively cool nCT (0.35), the
lowest GEBV (- 0.095), and a large negative TAGV
(- 1.27), (Table 1; Fig. 2) and this genotype also had
a low canopy wilting score and a low GEBV (- 5.53)
for canopy wilting (Kaler et al. 2017b). At the other
extreme, PI 398640 had warm nCT (0.64), the largest
GEBV for nCT (0.068), and a large TAGV (2.94)
(Table 1; Fig. 2); this accession also had a high
canopy wilting score with a high GEBV (4.72). The
Table 3 continued
Locus SNP_ID Gene symbola
Functional annotations (biological function)
BARC_1.01_Gm_14_2311158_G_A Glyma.14g030600 Pleckstrin homology (PH) domain-containing protein
(signal transduction)
24 BARC_1.01_Gm_14_4064786_C_T Glyma.14g031700 WD domain(gravitropism)
BARC_1.01_Gm_14_4430386_G_T Glyma.14g055800 Temperature sensing protein-related (response to heat)
BARC_1.01_Gm_14_4853955_A_G Glyma.14g059900 Cytochrome P450 (root hair elongation)
25 BARC_1.01_Gm_14_7052209_A_G Glyma.14g081500 Ca2 ? -independent phospholipase A2 (salicylic acid
mediated signaling pathway)
26 BARC_1.01_Gm_14_47305241_T_G Glyma.14g207200 ABC transporter (root development)
27 BARC_1.01_Gm_15_15726428_C_T Glyma.15g172200 Lycopene cyclase protein (stomatal complex
morphogenesis)
BARC_1.01_Gm_15_15729124_T_C Glyma.15g172200 Lycopene cyclase protein (stomatal complex
morphogenesis)
BARC_1.01_Gm_15_15742691_C_A Glyma.15g172300 Serine/threonine-protein Kinase 38 (protein
phosphorylation)
28 BARC_1.01_Gm_16_35807551_G_A Glyma.16g196600 Ubiquitin carboxyl-terminal hydrolase family protein (lateral
root morphogenesis)
29 BARC_1.01_Gm_17_11546048_A_G Glyma.17g142000 Asparagine–tRNA ligase (chloroplast stroma
organization)
30 BARC_1.01_Gm_17_38712454_G_A Glyma.17g232200 Glycosyl hydrolase family 10(xylan biosynthetic process)
31 BARC_1.01_Gm_18_11947921_C_T Glyma.18g106600 Transposase-like protein (plasmodesma organization)
32 BARC_1.01_Gm_18_13037246_G_A Glyma.18g110400 Receptor-like protein kinase 1 (response to jasmonic acid
stimulus)
BARC_1.01_Gm_18_13041332_G_A Glyma.18g110400 Receptor-like protein kinase 1 (response to jasmonic acid
stimulus)
BARC_1.01_Gm_18_13117752_G_T Glyma.18g110800 MATE efﬂux family protein (response to jasmonic acid
stimulus)
33 BARC_1.01_Gm_18_46218075_G_A Glyma.18g191500 Histone-lysine N-methyltransferase ATXR2
34 BARC_1.01_Gm_18_60748254_C_T Glyma.18g302300 Microtubule associated protein (MAP65ASE1) (microtubule
cytoskeleton organization)
AAE
1 BARC_1.01_Gm_01_4720160_C_T Glyma.01g043300 WRKY DNA -binding domain (defense response)
2 BARC_1.01_Gm_03_3497393_A_C Glyma.03g031100 Ribosomal protein S2 (photosynthesis)
3 BARC_1.01_Gm_04_8019074_G_A Glyma.04g091200 Rhamnogalacturonate lyase family
4 BARC_1.01_Gm_07_43182856_G_A Glyma.07g255400 Glycogen synthase kinase-3 (signal transduction)
5 BARC_1.01_Gm_13_24858209_A_G Glyma.13g136000 Adenylate kinase (root development)
6 BARC_1.01_Gm_13_34845629_A_G Glyma.13g238200 Translation initiation factor 3 (translational initiation)
7 BARC_1.01_Gm_15_15729124_T_C Glyma.15g172200 Lycopene cyclase protein (stomatal complex
morphogenesis)
8 BARC_1.01_Gm_18_13037246_G_A Glyma.18g110400 Receptor-like protein kinase 1 (response to jasmonic acid
stimulus)
a
All genes are from the Glyma2.0 assembly (www.soybase.org)
123

genotypes with low average nCT represent new
genetic sources for the cool-canopy temperature trait
with potential alternative alleles or different mecha-
nisms to achieve cool canopy temperature. Bai and
Purcell (2018) reported that lines selected from a cross
between Benning and PI 417937 as extremes for
canopy wilting also segregated for CT with slow-
wilting lines generally having cooler CT under
drought than fast-wilting lines.
Of the 52 SNPs significantly associated with nCT,
there were 44 that had minor alleles associated with a
decrease in the nCT (positive sign of allelic effect
indicates that minor allele was associated with a
decrease in the CT) (Table 2). There was a SNP on
Gm08 with a minor allele, that had the largest positive
allelic effect (0.161), and that was present within the
coding region of a gene, Glyma.08g340500. This gene
codes a eukaryotic translation initiation factor (4
GAMMA protein) that has a biological function
associated with the response to abscisic acid (ABA)
stimulus (www.soybase.com, Table 3). Up-regulation
of Glyma.08g340500 when soil-moisture is plentiful
may increase the ABA accumulation in roots, closing
stomata and conserving soil moisture for use later
during drought, resulting in a cooler canopy during
drought. Eight of the 52 SNPs had a major allele that
Fig. 4 Location of SNPs
significantly associated with
normalized canopy
temperature (nCT) in three
environments and across
environments with
identified significant SNPs
for canopy wilting as
described by Kaler et al.
(2017b). Yellow Oval
represents the genomic
regions where canopy
wilting and nCT were
coincident
123

was associated with reduction in nCT (Table 2).
A SNP for which the major allele was associated with
the largest reduction in nCT (- 0.070) was located on
Gm01. This SNP was present within the coding region
of Glyma.01g1203005, which has a leucine rich repeat
protein and has a biological function involved with
stomatal complex morphogenesis (Table 3). Of the
eight SNPs with significant associations with AAE,
seven had the minor allele associated with decreased
nCT. Whereas one SNP had the major allele associ-
ated with decreased in nCT (Table 2). Based on the
reported biological functions from SoyBase, SNPs
from this GWAM identified genes with functions
including root hair elongation, root development
response to abscisic acid stimulus and water depriva-
tion, stomatal complex morphogenesis, and signal
transduction (Table 3). Examining variation within
these genes as well as pyramiding favorable alleles of
these genes could be promising for future research and
in stress-tolerance breeding programs.
There have been no previous QTL or association
mapping studies in soybean for nCT, although there
have been several reports of mapping QTLs for
delayed wilting in soybean (Abdel-Haleem et al.
2012; Charlson et al. 2009; Du et al. 2009; Hwang
et al. 2015, 2016). Previously, Kaler et al. (2017b)
described the genomic regions that were associated
with canopy wilting variation in the same GWAM
panel reported in the present research. The genomic
regions associated with canopy wilting were compared
with SNPs associated with nCT to determine if they
are located in the same chromosomal regions (Fig. 4).
In this study, there were 15 chromosomal regions on
Gm01, Gm02, Gm04 (2), Gm07 (2) Gm08, Gm09 (2),
Gm14, Gm16, Gm17, and Gm18 where loci of canopy
wilting and nCT were coincident (Fig. 4). Two
chromosomal regions on Gm02 and Gm17 were also
coincident with meta-QTLs of canopy wilting identi-
fied by Hwang et al. (2016). It is not surprising that
many CW and nCT putative loci were coincident
giving the relationship between transpiration rate,
wilting, and canopy temperature.
These coincident regions contain genes that have
annotated functions associated with crop water bal-
ance including stomatal complex morphogenesis,
ABA stimulus, root hair elongation, and root devel-
opment (Table 3). For example, Glyma.02g103800 on
Gm02 encodes an ubiquitin protein ligase, and this
protein has been reported to regulate drought tolerance
via an ABA signaling pathway in hot pepper (Cap-
sicum annuum L.) (Lim et al. 2017). Timely up-
regulating Glyma.02g103800 may serve to conserve
soil moisture prior to drought that can then be drawn
upon during drought. Another gene, Gly-
ma.09g179400, on Gm09 encodes a Harpin-induced
protein that has a biological function involved with
root hair elongation. Up-regulation of Gly-
ma.09g179400 may increase root hair elongation in
plants, providing access to more soil moisture and
resulting in a cooler canopy. These loci, where
chromosomal regions for nCT and canopy wilting
were coincident, may indicate the stability and
importance of these loci for improving drought
tolerance and highlight regions of the genome for
further investigations.
Conclusions
This research used the high-density marker data of
31,260 SNPs with MAF C 5% to explore nCT vari-
ation in soybean with GWAM. There were 52
significant SNPs associated with nCT variation from
three environments and eight significant SNPs asso-
ciated with nCT averaged across all environments at a
significance level of - Log10(P) C 3.5. These 52
SNP-nCT associations from individual environments
and eight SNP-nCT associations from AAE likely
tagged 34 and eight different loci, respectively. Of the
52 SNPs, four had significant associations in more
than one environment. Based on the TAGVs and
GEBVs of accessions, PI 592940 was ranked very low
for both nCT and canopy wilting compared to other
genotypes. Genomic regions for nCT with regions for
canopy wilting variation were coincident at 15 chro-
mosomal regions. Several genotypes were identified
as potential donors for alleles leading to cooler
canopies and delayed wilting during drought.
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 U.S. Department of Agriculture. The
USDA is an equal opportunity provider and employer.
Appreciation is also extended to Marilynn Davies and Jody
Hedge for excellent technical assistance (Grant No. 1820-172-
0118-A).
123

Compliance with ethical standards
Conflict of interest The authors declare that they have no
conflict of interest.
References
Abdel-Haleem H, Carter TE Jr, Purcell LC, King CA, Ries LL,
Chen PC, Schapaugh W Jr, Sinclair TR, Boerma HR (2012)
Mapping of quantitative trait loci for canopy-wilting trait in
soybean [Glycine max (L) Merr]. Theor Appl Genet
125:837–846
Allen RG, Pereira LS, Raes D, Smith M (1998) Crop evapo-
transpiration: guidelines for computing crop water
requirements. Irrigation and Drainage Paper No 56, Food
and Agriculture Organization of the United Nations, Rome
Araus JL, Cairns JE (2014) Field high-throughtput phenotyping:
the new crop breeding frontier. Trends Plant Sci 19:52–61
Bai H, Purcell LC (2018) Aerial canopy temperature differences
between fast-slow wilting soybean genotypes. J Agron
Crop Sci 204:243–251
Blackman PG, Davis WJ (1985) Root to shoot communication
in maize plants of the effects of soil drying. J Exp Bot
36:39–48
Blum A (2005) Drought resistance, water-use efficiency, and
yield potential-are they compatible, dissonant, or mutually
exclusive? Aust J Agric Res 56:1159–1168
Bondari K (2003) Statistical analysis of genotype x environment
interaction in agricultural research. Paper SD15, SESUG:
The Proceedings of the SouthEast SAS Users Group, St
Pete Beach
Carter TE Jr, De Souza PI, Purcell LC (1999) Recent advances
in breeding for drought and aluminum resistance in soy-
bean. In: Kauffman H (ed) Proceedings of the sixth world
soybean research conference, Chicago. Superior Printing,
Champaign, pp 106–125
Chapman SC, Merz T, Chan A, Jackway P, Hrabar S, Dreccer
MF, Holland E, Zheng B, Ling TJ, Jimenez-Berni J (2014)
Pheno-copter: a low-altitude, autonomous remote-sensing
robotic helicopter for high-throughput field-based pheno-
typing. Agronomy 4:279–301
Charlson DV, Bhatnagar S, King CA, Ray JD, Sneller CH,
Carter TE Jr, Purcell LC (2009) Polygenic inheritance of
canopy wilting in soybean [Glycine max (L) Merr]. Theor
Appl Genet 119:587–594
Clark SA, van der Werf J (2013) Genomic best linear unbiased
prediction (gBLUP) for the estimation of genomic breed-
ing values. Genome-Wide Association Studies and Geno-
mic Prediction. Springer, Berlin, pp 321–330
Cornic G, Massacci A (1996) Leaf photosynthesis under
drought stress. In: Baker NR (ed) Photosynthesis and the
environment. Kluwer Academic Publishers, Dordrecht
Covarrubias-Pazaran G (2016) Genome-assisted prediction of
quantitative traits using the R package sommer. PLoS
ONE 11:e0156744. https://doi.org/10.1371/journal.pone.
0156744
Dhanapal AP, Ray JD, Singh SK, Hoyos-Villegas V, Smith JR,
Purcell LC, King CA, Fritsch FB (2015a) Association
mapping of total carotenoids in diverse soybean genotypes
based on leaf extracts and high-throughput canopy spectral
reflectance measurements. PLoS ONE 10:e0137213.
https://doi.org/10.1371/journal.pone.0137213
Purcell LC, King CA, Cregan PB, Song Q, Fritsch FB
(2015b) Genome-wide association study (GWAM) of
carbon isotope ratio (d13
C) in diverse soybean [Glycine
max (L) Merr] genotypes. Theor Appl Genet 128:73–91
Purcell LC, King CA, Fritsch FB (2015c) Genome-wide
association analysis of diverse soybean genotypes reveals
novel markers for nitrogen traits. Plant Genome-US 8:3.
https://doi.org/10.3835/plantgenome2014.11.0086
Dixit S, Huang BE, Sta Cruz MT, Maturan PT, Ontoy JCE,
Kumar A (2014) QTLs for tolerance of drought and
breeding for tolerance of abiotic and biotic stress: an
integrated approach. PLoS ONE 9:e109574. https://doi.
org/10.1371/journal.pone.0109574
Douglas I, Alam K, Maghenda M, McDonnell Y, McLean L,
Campbell J (2008) Unjust waters: climate change, flooding
and the urban poor in Africa. Environ Urban 20:187–205
Du W, Yu D, Fu S (2009) Detection of quantitative trait loci for
yield and drought tolerance traits in soybean using a
recombinant inbred line population. J Integr Plant Biol
51:868–878
Falconer DS, Mackay TFC (1996) Introduction to quantitative
genetics, 4th edn. Longmans Green, Harlow
Feng X, Porporato A, Rodriguz-Iturbe I (2013) Changes in
rainfall seasonality in the tropics. Nat Clim Change
3:811–815
Foley JA, Ramankutty N, Brauman KA, Cassidy ES, Gerber JS,
Johnston M, Mueller ND, Connell C, Ray DK, West PC,
Balzer C, Bennett EM, Carpenter SR, Hill J, Monfreda C,
Polasky S, Rockstro¨m J, Sheehan J, Siebert S, Tilman D,
Zaks DPM (2011) Solutions for a cultivated planet. Nature
478:337–342. https://doi.org/10.1038/nature10452
Gollan T, Passioura JB, Munns R (1986) Soil water status affects
the stomatal conductance of fully turgid wheat and sun-
flower leaves. Aust J Plant Physiol 13:459–464
Guilioni L, Jones HG, Leinonen I, Lhomme JP (2008) On the
relationships between stomatal resistance and leaf tem-
peratures in thermography. Agric For Meteorol
148:1908–1912
Hao D, Cheng H, Yin Z, Cui S, Zhang D, Wang H, Yu D (2012)
Identification of single nucleotide polymorphisms and
haplotypes associated with yield and yield components in
soybean (Glycine max) landraces across multiple envi-
ronments. 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 pro-
tein and oil content in soybean. PLoS Genet 15:1. https://
doi.org/10.1186/1471-2164-15-1
Hwang S, King CA, Chen P, Ray JD, Cregan PB, Carter TE Jr,
Li Z, Abdel-Haleem H, Matson KW, Schapaugh W Jr,
Purcell LC (2015) Confirmation of delayed canopy wilting
QTLs from multiple soybean mapping populations. Theor
Appl Genet 128:2047–2065
Hwang S, King CA, Chen P, Ray JD, Cregan PB, Carter TE Jr,
Li Z, Abdel-Haleem H, Matson KW, Schapaugh W Jr,
Purcell LC (2016) Meta-analysis to refine map position and
reduce confidence intervals for delayed-canopy-wilting
123

QTLs in soybean. Mol Breed 36:91. https://doi.org/10.
1007/s11032-016-0516-5
Jackson RD, Idso SB, Reginato RJ, Pinter PJ Jr (1981)
Canopy temperature as a crop water stress indicator. Water
Resour Res 17:1133–1138. https://doi.org/10.1029/
WR017i004p01133
Jones HG (1979) Stomatal behaviour and breeding for drought
resistance. In: Mussell H, Staples R (eds) Stress physiology
in crop plants. Wiley, New York, pp 408–428
Jones HG, Serraj R, Loveys BR, Xiong LZ, Wheaton A, Price
AH (2009) Thermal infrared imaging of crop canopies for
the remote diagnosis and quantification of plant responses
to water stress in the field. Funct Plant Biol 36:978–989
Kaler AS, Dhanapal AP, Ray JD, King CA, Fritschi FB, Purcell
LC (2017a) Genome-wide association mapping of carbon
isotope and oxygen isotope ratios in diverse soybean
genotypes. Crop Sci 57:1–16. https://doi.org/10.2135/
cropsci2017030160
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
130:2203–2217
Kaler AS, Ray JD, Schapaugh WT, Davies MK, King CA,
Purcell LC (2018) Association mapping identifies loci for
canopy coverage in diverse soybean genotypes. Mol Breed
38:50. https://doi.org/10.1007/s11032-018-0810-5
King CA, Purcell LC, Brye KR (2009) Differential wilting
among soybean genotypes in response to water deficit.
Crop Sci 49:290–298
Leport L, Turner NC, French RJ, Barr MD, Duda R, Davies SL,
Tennant D, Siddique KHM (1999) Physiological responses
of chickpea genotypes to terminal drought in a Mediter-
ranean-type environment. Eur J Agron 11:279–291
Lim CW, Park C, Kim JH, Joo H, Hong E, Lee SC (2017) Pepper
CaREL1, an ubiquitin E3 ligase, regulates drought toler-
ance via the ABA-signalling pathway. Sci Rep. https://doi.
org/10.1038/s41598-017-00490-4
Liu H, Zou G, Liu G, Hu S, Li M, Yu X, Mei H, Luo L (2005)
Correlation analysis and QTL identification for canopy
temperature, leaf water potential and spikelet fertility in
rice under contrasting moisture regimes. Chin Sci Bull
50:317–326
Liu Y, Subhash C, Yan J, Song C, Zhao J, Li J (2011) Maize leaf
temperature responses to drought: thermal imaging and
quantitative trait loci (QTL) mapping. Environ Exper Bot
71:158–165
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:e1005767. https://doi.org/10.1371/journal.pgen.
1005767
McCarthy MI, Abecasis GR, Cardon LR, Goldstein DB, Little J,
Ioannidis JPA, Hirschhorn JN (2008) Genome-wide asso-
ciation studies for complex traits: consensus, uncertainty
and challenges. Nat Rev Genet 9:356–369. https://doi.org/
10.1038/nrg2344
Merlot S, Mustilli A-C, Genty B, North H, Lefebvre V, Sotta B,
Vavasseur A, Giraudat J (2002) Use of infrared thermal
imaging to isolate Arabidopsis mutants defective in stom-
atal regulation. Plant J 30:601–609
Money D, Gardner K, Migicovsky Z, Schwaninger H, Zhong
GY, Myles S (2015) LinkImpute: fast and accurate geno-
type imputation for non-model organisms. G3-Genes
Genom Genet 5:23383–23390
Nordborg M, Tavare´ S (2002) Linkage disequilibrium: what
history has to tell us. Trends Genet 18:83–90
O’Shaughnessy SA, Evett SR, Colaizzi PD, Howell TA (2011)
Using radiation thermography and thermometry to evalu-
ate crop water stress in soybean and cotton. Agric Water
Manag 98:1523–1535
Oya T, Nepomuceno AL, Numaier N, Farias JRB, Tobita SS, Ito
S (2004) Drought tolerance characteristics of Brazilian
cultivars—evaluation and characterization of drought tol-
erance of various Brazilian soybean cultivars in the field.
Plant Prod Sci 7:129–137
Piepho HP (2000) A mixed-model approach to mapping quan-
titative trait loci in barley on the basis of multiple envi-
ronment data. Genetics 156:2043–2050
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 DK, Mueller ND, West PC, Foley JA (2013) Yield trends
are insufficient to double global crop production by
2050. PLoS ONE 8:e66428. https://doi.org/10.1371/
journalpone0066428
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-Genes Genom
Genet 5:2391–2403. https://doi.org/10.1534/g3.115.
021774
Rebetzke GJ, Condon AG, Rattey AR, Farquhar GD, Richards
RA (2013) Genomic regions for canopy temperature and
their genetic association with stomatal conductance and
grain yield in bread wheat (Triticum aestivum L). Funct
Plant Biol 40:14–26
Ries LL, Purcell LC, Carter TE Jr, Edwards JT, King CA (2012)
Physiological traits contributing to differential canopy
wilting in soybean under drought. Crop Sci 52:272–281
Risch N, Merikangas K (1996) The future of genetic studies of
complex human diseases. Science 273:1516–1517
Salmeroń M, Gbur EE, Bourland FM, Buehring NW, Earnest L,
Fritschi FB, Golden BR, Hathcoat D, Lofton J, Miller TD,
Neely C, Shannon G, Udeigwe TK, Verbree DA, Vories
ED, Wiebold WJ, Purcell LC (2016) Yield response to
planting date among soybean maturity groups for irrigated
production in the US Midsouth. Crop Sci 56:747–759
SAS Institute (2013) The SAS system for windows version 93.
SAS Inst Inc, Cary
Song Q, Hyten DL, Jia G, Quigley CV, Fickus EW, Nelson RL,
Cregan PB (2013) Development and evaluation of
SoySNP50 K, a high-density genotyping array for soy-
bean. PLoS ONE 8:e54985. https://doi.org/10.1371/
journal.pone.0054985
Syvanen AC (2005) Toward genome-wide SNP genotyping. Nat
Genet 37:5–10
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
123

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 Genom 15:809. https://
doi.org/10.1186/1471-2164-15-809
Yousfi S, Kellas N, Saidi L, Benlakehal Z, Chaou L, Siad D,
Herda F, Karrou M, Vergara O, Gracia A, Araus JL, Serret
MD (2016) Comparative performance of remote sensing
methods in assessing wheat performance under Mediter-
ranean conditions. Agr Water Manag 164:137–147
Yu J, Pressoir G, Briggs WH, Vroh BI, Yamasaki M, Doebley
JF, McMullen MD, Gaut BS, Nielsen DM, Holland JB,
Kresovich S, Buckler ES (2006) A unified mixed-model
method for association mapping that accounts for multiple
levels of relatedness. Nat Genet 38:203–208
Zhang Z, Todhunter RJ, Buckler ES, Van Vleck LD (2007)
Technical note: use of marker-based relationships with
multiple-trait derivative-free restricted maximal likeli-
hood. J Anim Sci 85:881–885
Zhang Z, Ersoz E, Lai CQ, Todhunter RJ, Tiwari HK, Gore MA,
Bradbury PJ, Yu J, Arnett DK, Ordovas JM, Buckler ES
(2010) Mixed linear model approach adapted for genome-
wide association studies. Nat Genet 42:355–360
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
soybean (Glycine max) germplasm. BMC Genom 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-US 1:5–20
Zia S, Sophrer K, Du W, Spreer W, Romano G, Xiongkui H,
Mu¨ller J (2011) Monitoring physiological responses to
water stress in two maize varieties by infrared thermogra-
phy. Int J Agric Biol Eng 4:7–15
123

Kaler et al 2018 euphytica

More Related Content

What's hot (20)

Similar to Kaler et al 2018 euphytica (20)

More from Avjinder (Avi) Kaler (20)

Recently uploaded (20)

Kaler et al 2018 euphytica