Professional Documents
Culture Documents
Genome-Wide Association Analyses of Quantitative Disease Resistance in Diverse Sets of Soybean (Glycine Max (L.) Merr.) Plant Introductions
Genome-Wide Association Analyses of Quantitative Disease Resistance in Diverse Sets of Soybean (Glycine Max (L.) Merr.) Plant Introductions
1 Center for Applied Plant Science and Center for Soybean Research, The Ohio State University, Columbus,
Ohio, United States of America, 2 Department of Horticulture and Crop Science, The Ohio State University,
a1111111111 Columbus, Ohio, United States of America, 3 Department of Plant Pathology, The Ohio State University,
a1111111111 Wooster, Ohio, United States of America
a1111111111
a1111111111 ¤ Current address: Corteva Agriscience, Napoleon, Ohio, United States of America
a1111111111 * [email protected]
Abstract
OPEN ACCESS Phytophthora sojae is one of the costliest soybean pathogens in the US. Quantitative dis-
Citation: Rolling W, Lake R, Dorrance AE, McHale ease resistance (QDR) is a vital part of Phytophthora disease management. In this study,
LK (2020) Genome-wide association analyses of QDR was measured in 478 and 495 plant introductions (PIs) towards P. sojae isolates
quantitative disease resistance in diverse sets of
soybean [Glycine max (L.) Merr.] plant
OH.121 and C2.S1, respectively, in genome-wide association (GWA) analyses to identify
introductions. PLoS ONE 15(3): e0227710. https:// genetic markers linked to QDR loci (QDRL). Populations were generated by sampling PIs
doi.org/10.1371/journal.pone.0227710 from the US, the Republic of Korea, and the full collection of PIs maintained by the USDA.
Editor: Istvan Rajcan, University of Guelph, Additionally, a meta-analysis of QDRL reported from bi-parental studies was done to com-
CANADA pare past and present findings. Twenty-four significant marker-trait associations were identi-
Received: August 29, 2019 fied from the 478 PIs phenotyped with OH.121, and an additional 24 marker-trait
Accepted: December 25, 2019
associations were identified from the 495 PIs phenotyped with C2.S1. In total, 48 significant
markers were distributed across 16 chromosomes and based on linkage analysis, represent
Published: March 20, 2020
a total of 44 QDRL. The majority of QDRL were identified with only one of the two isolates,
Copyright: © 2020 Rolling et al. This is an open and only a region on chromosome 13 was consistently identified. Regions on chromosomes
access article distributed under the terms of the
Creative Commons Attribution License, which
3, 13, and 17 were identified in previous GWA-analyses and were re-identified in this study.
permits unrestricted use, distribution, and Five QDRL co-localized with P. sojae meta-QDRL identified from QDRL reported in previous
reproduction in any medium, provided the original biparental mapping studies. The remaining regions represent novel QDRL, in the soybean-
author and source are credited.
P. sojae pathosystem and were primarily identified in germplasm from the Republic of
Data Availability Statement: All relevant data are Korea. Overall, the number of loci identified in this study highlights the complexity of QDR to
within the paper and its Supporting Information
P. sojae.
files.
Department of Agriculture Hatch projects for development, including saturated soils and temperatures around 25 ˚C [2–4]. Under these
Development of Disease Management Strategies conditions, zoospores of P. sojae chemotactically swim towards soybean roots [5,6]. Successful
for Soybean Pathogens in Ohio OHO01303 to AD,
infection of susceptible plants results in seed rot and damping off at early growth stages, and
and Genetic Analysis of Soybean Added-Value
Traits and Soybean Variety Development for Ohio wilting, stem lesions, and plant death at later growth stages. Genetic resistance is considered
OHO01279 to LM. the most effective strategy for preventing or reducing the impact of Phytophthora diseases
[4,7].
Competing interests: The authors have declared
that no competing interests exist. Soybean-breeding programs have relied on resistance conferred by single, dominantly
inherited Resistance to Phytophthora sojae (Rps)-genes. More than 30 Rps-genes/alleles have
been mapped (S1 Table) [8–25]. The effectiveness of Rps-mediated resistance is limited as
these genes only confer immunity towards specific races of P. sojae. Over time this results in
selection within pathogen populations, leading to adaptation [3,4] and limited Rps-gene life-
spans of 8–20 years [26]. A recent survey completed in the North Central region of the US
demonstrated that none of the deployed Rps-genes confer resistance to all P. sojae isolates,
with field populations having complex virulence patterns [27].
Quantitative disease resistance (QDR) is an additional form of genetic resistance, function-
ing through the action of multiple genes each conferring a portion of the overall phenotype
[28–30]. Though QDR does not confer complete immunity, the lack of race-specificity is pre-
dicted to place less selection on pathogen populations; thus, QDR is expected to be more dura-
ble than race-specific resistance [31–33]. For example, wheat growers have used QDR for
stripe and leaf rust resistance since the early 20th century, with no observed increase in patho-
gen virulence [34]. Similarly, QDR for the rice pathogen Magnaporthe oryzae contributed by
Pi21 has conferred effective resistance for over a century [35]. Used in combination, QDR can
also increase the durability of Resistance-gene (R-gene) mediated resistance as highlighted by
examples of pepper-potato virus Y [36] and root-knot nematode pathosystems of potato
[3,37].
In the soybean-P. sojae pathosystem, QDR has also played an important role in disease
resistance breeding [3]. QDR protects soybean yields when Rps-gene(s) are ineffective and the
environment is conducive to disease [7,38]. Breeding efforts to introduce QDR (and other
traits) for cultivar development involve large populations. Field testing can be misleading as
the pathotypes of P. sojae populations within fields are highly variable, the presence of an Rps-
gene could mask the effects of QDR [39], and QDR is not expressed within the first five days
following germination [40]. Marker-assisted selection (MAS) can provide an alternative
method for selecting cultivars with higher levels of QDR. Thus, studies to map the genetic posi-
tions of QDR can be applied to increase the efficiency of selection for these traits via MAS [41].
Numerous biparental mapping studies identified QDR loci (QDRL) towards P. sojae in US
cultivars Conrad, Sloan, and V71-370 [42–48]. Recently, the soybean nested associated map-
ping populations were used to study P. sojae resistance, identifying four QDRL [49]. In addi-
tion to US cultivars, QDRL have been mapped in landraces originating from China and the
Republic of Korea (South Korea, SK) using biparental mapping populations derived from
crosses with eight accessions [50–54]. The majority of these mapping studies identify several
QDRL within each population; most were small-effect loci explaining less than 10% of the vari-
ation (S2 Table).
Mapping studies have also been completed using genome-wide association (GWA) analy-
ses. This includes a study of 797 plant introductions (PIs) from SK [55] which had previously
been shown to be a source of Rps-genes and high levels of QDR against P. sojae [56], 175 Chi-
nese soybean breeding lines [57], and 279 breeding lines from the Yangtze-Huai Chinese
breeding program [58]. Recently, a collection of 169 Brazilian cultivars was used to identify
QDRL in regional germplasm [59]. Similar to loci identified by biparental mapping, these
GWA studies identified small-effect QDRL.
Although soybean growers deploy QDR to P. sojae in the US [3], US germplasm has never
been evaluated in a GWA analysis for QDR to P. sojae. A population derived from US germ-
plasm could be useful in identifying alleles in germplasm most adapted for the region. Though
alleles from adapted germplasm may be easier to introgress or maintain in elite cultivars, iden-
tifying novel QDRL from diverse germplasm remains a crucial aspect of resistance breeding.
Thus, the objective of the current research was to complete genome-wide association (GWA)
analyses by sampling US, SK, and the full diversity of the USDA soybean PI collection. Two
isolates were used to assay QDR because P. sojae is diverse [27] and previous studies have
mapped different QDRL depending on the isolate [47,48].
We phenotyped 547 (495 used in GWA analysis) PIs for QDR using isolate C2.S1 of P. sojae
and also measured QDR in a different set of 549 PIs (478 used in GWA analysis) with the sec-
ond isolate OH.12108.6.3 (OH.121) of P. sojae. In total, 48 SNPs associated with QDR to P.
sojae were identified and consolidated into 44 QDRL, with 21 QDRL identified with OH.121
and 23 QDRL identified with C2.S1. In addition to GWA analyses, we completed a meta-
QDRL analysis to consolidate previous results and identify any colocalization between the loci
identified in this study by GWA analysis and those QDR previously identified from biparental
mapping studies. The 44 QDRL reported here represent a mixture of both novel QDRL and
those that confirm results of previous mapping studies of this complex pathosystem.
The OH set included 478 PIs (S4 Table) in GWA analysis composed of 170 US PIs, 158 SK PIs,
and 150 PIs sampled from the GRIN population.
In addition to the PIs, cultivars with characterized QDR against P. sojae (‘L83-570—Rps3a’,
‘Williams 79—Rps1c’, ‘Conrad—rps’, ‘Sloan—rps’, ‘Resnik—Rps1k’,‘OX20-8—Rps1a’), were
included as checks. The P. sojae isolates were expected to be virulent on these lines despite the
presence of Rps-genes, with the exception of OH.121 on ‘Williams 79—Rps1c’. Before disease
assays, all seeds were surface sterilized using chlorine gas [64].
a Rps-gene response and removed from GWA analyses. PIs that developed a lesion 70–90% of
the time were further scrutinized by checking the scores in the layer test to ensure a lack of an
Rps-gene response, and PIs with IRRS scores less than 1.5 were removed from the GWA
analyses.
GWA analyses
The C2 set and the OH set were each analyzed separately; four GWA analyses were completed
for C2 and OH set, for a total of 8. An analysis was completed using all PIs for a set as well as
the three populations (C2-US, C2-SK, C2-GRIN, OH-US, OH-SK, and OH-GRIN) that com-
prise each set. GWA analyses were implemented using the R [66] package GAPIT [68] using a
multiple locus mixed model (MLMM). General linear model, mixed linear model, compressed
mixed linear model, and FarmCPU were also tested, but did not yield appropriate Q-Q plots
or were not as able to resolve significant associations (S1 Fig). When implemented in GAPIT,
MLMM does not readily provide an estimate of percent variation explained. Instead, PVE was
estimated using CMLM. The Q-matrix resulting from fastStructure analyses was used as a
covariate. The significance threshold was determined using a modified Bonferroni adjustment
by calculating the Meff using SimpleM [69]. Genome-wide significance thresholds were deter-
mined by α/Meff where α = 0.05.
Haploview [70] was used for haplotype block assembly using the four-gamete rule [71]. A
haplotype block possessing a marker(s) with a significant marker-trait association was consid-
ered a single QDRL. Haplotype blocks for markers that were unlinked to other markers were
delimited according to the positions of the flanking markers.
Meta-QDRL analysis
We collected pertinent information for reported QDRL towards P. sojae identified prior to Jan
1, 2019. Collected data included (1) linkage maps, (2) R2 of genetic effect, (3) significance
(LOD), (4) linkage group, (5) confidence interval (CI), (6) predicted position of QDRL, (7)
generation and (8) size of populations from 16 biparental mapping studies (S2 Table). Due to
inconsistent reporting of these values, estimates were made for LOD values and confidence
intervals (CIs) when not provided. LOD values were calculated using the formula LOD = R2/
1.5. If the CI was not provided it was calculated using the equation CI = 163/(N x R2) where N
is the population size [72].
Meta-analysis was completed using the software BioMercator V3.1 [73,74]. The initial step
was to create a project-specific consensus map from genetic maps found in the biparental stud-
ies; this was aided by the inclusion of the consensus map [75] downloaded from Soybase [61].
QDRL were projected onto the project-specific consensus map. Studies prior to 2010 were
removed and only one mapping study per generation of each biparental cross was included.
The projected QDRL were used to identify position and CI based on Akaike-type criteria val-
ues. The physical positions of meta-QDRL were determined by selecting the markers flanking
the meta-QDRL and locating the positions of these markers in the Wms.82 genome (Wms.82.
a2.v1) [76].
Results
Population structure analysis and linkage disequilibrium
The population structure of the USDA germplasm collection was analyzed in order to generate
population-groups from which our GWA populations would be sampled. Population structure
analyses of the 20,087 PIs with SoySNP50K genotypic data resulted in marginal likelihood
Fig 1. Estimated population structure of soybean plant introductions (PIs) genotyped with the SoySNP50K iSelect BeadChip. (A) Population
structure analysis of the 20,087 genotyped PIs maintained by the USDA Soybean Germplasm Center (Urbana, IL) was carried out at k = 12. (B) The C2
set consists of 495 PIs phenotyped with P. sojae isolate C2.S1. (C) The OH set consists of 478 PIs phenotyped with P. sojae isolate OH.121. The 12
population-groups identified by fastStructure are indicated by lower case letters. The United States (US) and South Korean (SK) population-groups of
interest are indicated with labels “b” and “j”, respectively. To create C2-US, C2-SK, OH-US, and OH-SK, the US and SK populations-groups (b, j) were
sampled.
https://doi.org/10.1371/journal.pone.0227710.g001
PIs in the germplasm collection were represented in both the C2 set and OH set (Fig 1B and
1C). Additional population structure analyses were completed to identify subpopulations in
the PIs comprising C2-US and OH-US as well as the C2-SK and OH-SK populations. These
analyses identified six and nine subpopulations within the US and SK populations, respectively
(S2 and S3 Figs). The majority of the subpopulations are present in the respective C2-US and
OH-US populations and C2-SK and OH-SK populations, with only slightly different propor-
tions of each subpopulation represented (S2 and S3 Figs).
A total of 33,234 (C2 set) and 34,248 (OH set) SNP-markers were used in GWA analyses.
There was an average density of one marker every 29 kb for both the C2 and OH sets. Due to
linkage disequilibrium (LD), SNP-markers could be consolidated to an effective marker num-
ber [69] of approximately 11,000 in the OH set and C2 set, as well as the OH-GRIN and
C2-GRIN populations. The effective marker numbers were nearly 5,000 for the C2-SK and
OH-SK populations, 4,051 in C2-US population, and 3,291 in the OH-US population (S4 Fig).
In the combined 974 individuals from the C2 and OH sets, where LD is expected to be lower
than the less diverse populations (i.e., C2-US, C2-SK, OH-US, and OH-SK) [77], LD decayed
to an r2 of 0.2, a common threshold of significant LD [78], at an approximate distance of
350kb in euchromatic regions and greater than 500kb in heterochromatic regions (S4 Fig).
Given that each population has an estimated marker density < 300 kb/marker, we anticipate
that the majority of the soybean genome was captured in each GWA analysis (S4 Fig).
Table 1. Heritability and genetic variance for layer test traits with Phytophthora sojae isolates C.2.S.1 and OH.121
for the C2 and OH sets of plant introductions, respectively.
C2 set OH set
a 2b 2c 2
Trait H σg H σg2
��� d
IRRS 0.58 0.94 0.65 1.98�
���
IRW 0.78 0.19 0.45 0.10�
���
ISW 0.65 0.23 0.68 0.27���
���
IPH 0.58 7.70 0.77 14.00���
ΔRW 0.52 0.11 ���
0.38 0.09���
ΔSW 0.56 0.12��� 0.43 0.16���
NRW 0.75 0.23��� 0.39 0.14�
���
NSW 0.66 0.25 0.65 0.30���
���
NPH 0.55 6.57 0.63 8.12���
a
, IRRS, inoculated root rot score; IRW, inoculated root weight; ISW, inoculated shoot weight; IPH, inoculated plant
height; ΔRW, change in root weight; ΔSW, change in shoot weight; NRW, non-inoculated root weight; NSW, non-
inoculated shoot weight; NPH, non-inoculated plant height
b
, H2, broad-sense heritability
c
, σg2, genetic variance;
d �
, , significant at P value < 0.05;
���
, significant at P value < 0.001.
https://doi.org/10.1371/journal.pone.0227710.t001
NRW. For the most part, correlations between different types of traits were low to moderate,
ranging from -0.21 for ΔRW and NPH to 0.56 for ISW and NSW. However, IRRS was not sig-
nificantly correlated to the non-inoculated traits, indicating that IRRS is independent of non-
inoculated plant size and weights.
Table 2. Significant correlations between layer test traits assessed within the C2 set (unshaded) and OH sets (shaded) of plant introductions.
Inoculated Combined Non-inoculated
Traita IRRS IRW ISW IPH ΔRW ΔSW NRW NSW NPH
b
IRRS 0.63 -0.54 -0.53 -0.57 -0.59 -- -- --
IRW -0.71 0.87 0.73 0.38 0.37 0.48 0.51 0.38
ISW -0.65 0.83 0.75 0.30 0.43 0.45 0.56 0.36
IPH -0.77 0.76 0.72 0.23 0.31 0.37 0.41 0.43
ΔRW -0.38 0.38 0.26 0.25 0.79 -0.43 -0.31 -0.21
ΔSW -0.51 0.44 0.48 0.39 0.58 -0.27 -0.35 -0.24
NRW -- 0.41 0.36 0.36 -0.45 -0.43 0.83 0.59
NSW -- 0.40 0.51 0.34 -0.30 -0.49 0.78 0.69
NPH -- 0.33 0.3 0.43 -0.19 -0.35 0.63 0.65
https://doi.org/10.1371/journal.pone.0227710.t002
478 PIs. The BLUP values for NRW and NSW had normal distributions (S7 Fig). The other
traits had nearly normal distributions with a slight positive skew to IRRS, IRW, and ISW; a
slight negative skew for IPH and NPH; and heavy tailing in the ΔRW and ΔSW traits (S7 Fig).
All nine traits had significant genetic variance (Table 1) with moderate to high heritability,
ranging from 0.38 for ΔRW to 0.77 for IPH. Both the heritability and genetic variance were
comparable between isolates, with OH.121 traits on average having slightly higher genetic vari-
ance, higher residual variance, and marginally lower heritability (Table 1).
The average BLUP values of the three populations comprising the OH set (OH-GRIN,
OH-SK, OH-US) were very similar for each trait, with a greater standard deviation of BLUP
values in the OH-GRIN and OH-SK populations relative to the OH-US population (S7 Fig).
Genetic variance was lowest in the US population, and heritability was highest in the SK popu-
lation. The US and SK populations had similar BLUP values for IRRS, IRW, ISW, and IPH
and were significantly more resistant than the GRIN population. The US population had sig-
nificantly higher BLUP values for the non-inoculated traits (sans NSW compared to SK) indi-
cating, once again, that the US population averaged larger plants than the SK and GRIN
populations.
There were significant correlations among nearly all traits in the OH set (Table 2). Correla-
tions were strongest among traits of the same type (inoculated, non-inoculated, or combined).
Correlations among inoculated traits were moderate to high, ranging from -0.65 for IRRS and
ISW to 0.83 for ISW and IRW. The correlation between the combined traits ΔSW and ΔRW
was moderate at 0.58, and correlations among non-inoculated traits were moderate to high,
ranging from 0.63 for NPH and NRW to 0.78 for NSW and NRW. For the most part, correla-
tions between different types of traits were low to moderate, ranging from -0.19 for ΔRW and
NPH to -0.51 for ISW and NPH. However, similar to data from the C2 set, IRRS was not sig-
nificantly correlated to the non-inoculated traits, indicating that IRRS is independent of non-
inoculated plant size and weights.
https://doi.org/10.1371/journal.pone.0227710.t003
https://doi.org/10.1371/journal.pone.0227710.t004
set. Little overlap was observed between the results of the OH set and C2 set. Only QDRL C13-
5, C13-6, C13-7, O13-4, and O13-5 (Tables 3 and 4; Figs 2 and 3) were identified in similar
regions (<500 kb apart) between OH and C2 sets.
Within a given set (C2 or OH), different QDRL were detected depending on the popula-
tion. For example, the C2-SK population identified different QDRL than the C2-US and
C2-GRIN. Similar differences were also observed among the QDRL identified from the popu-
lations within the OH set. In this case, the populations within a given set are structured and
allele frequencies are expected to contribute to the differences in the QDRL that were detected.
Indeed, there were significant differences in allele frequencies between populations (P
value < 0.01) at all identified QDRL (Fig 4).
Fig 2. Manhattan plot of genome wide association of layer test traits from the C2 set. Marker associations for the full C2 set and the C2-GRIN,
C2-SK, and C2-US populations for each trait are overlaid. For significant associations, markers for each population and trait are differentiated by the
shape and color of the marker, respectively. Significance thresholds were calculated for each population using SimpleM, only the highest (C2 set) and
lowest (C2-US) thresholds are displayed. Red shading of non-significant markers indicates the genetic regions associated with known Rps-genes.
Genomic positions previously identified significantly associated with quantitative resistance to P. sojae in previous GWA analyses are indicated by the
black vertical bars. Grey highlighted regions represent meta-QDRL, with the darker grey highlight indicating QDRL identified in exotic germplasm,
and the lighter grey representing QDRL identified from US cultivars.
https://doi.org/10.1371/journal.pone.0227710.g002
There were also some differences in allele frequencies from the corresponding populations
within the C2 and OH sets (i.e., C2-US vs. OH-US, C2-SK vs. OH-SK and C2-GRIN vs.
OH-GRIN), despite sampling the same population-groups from the population structure anal-
ysis (Fig 1). While the majority of allele frequencies were very similar and not significantly dif-
ferent, eight alleles were significantly different between populations (Fig 5) with five
significant differences in allele frequencies between the US populations and three significant
differences between the C2 set and OH set. The ability to detect QDRL is affected by allele fre-
quency [79] and could contribute to the differences in identified QDRL.
Fig 3. Manhattan plot of genome wide association of layer test traits from the OH set. Marker associations for the full OH set and the OH-GRIN,
OH-SK, and OH-US populations for each trait are overlaid. For significant associations, markers for each population and trait are differentiated by the
shape and color of the marker, respectively. Significance thresholds were calculated for each population using SimpleM, only the highest (OH set) and
lowest (OH-US) thresholds are displayed. Genomic locations of Rps-genes, regions identified as significantly associated with quantitative resistance to
P. sojae in previous GWA analyses, and meta-QDRL are indicated as described for Fig 3.
https://doi.org/10.1371/journal.pone.0227710.g003
The QDRL markers were often significantly associated with QDR towards either OH.121
or C2.S1 in their respective populations. Using single marker analyses, most markers (37 of
44) had a significant marker effect for the isolate, population, and trait in which the association
was originally identified by GWA analysis. More surprising is that a large number (26 of 44) of
the single marker tests also had significant effects for QDR to the alternate isolate in the corre-
sponding population. Thus, in addition to the direction of allelic effects being the same
between the two isolates and their corresponding populations, these allelic effects were often
significant based on single marker analyses.
For most of the corresponding populations, the significant markers had a similar allele fre-
quency in both populations (e.g. C2-GRIN and OH-GRIN) (Figs 5 and 6). However, for nine
QDRL, the allele frequency was significantly different between populations or monomorphic
in one population. Of the 18 QDRL that were not confirmed in the relevant population for
resistance to the alternate isolate, one was monomorphic and three had significantly different
allele frequencies in the relevant population corresponding to the alternative isolate. In these
instances, differences in the marker effect may be due to population sampling rather than iso-
late or environment.
In both the C2 and OH sets, there was a significant correlation (P value <0.001) between
the number of resistant alleles and an overall resistance rank (Fig 7a and 7c). In the C2 set, the
most resistant quintile averaged 25.2 resistance alleles (of 44), significantly higher than the
other quantiles (Fig 7b). The OH set displayed a similar trend with the most resistant quintile
Fig 4. Allele frequencies of SNPs in the combined US, SK, and GRIN populations in the C2 and OH sets of PIs.
The frequency of SNP alleles was calculated by adding the total number of individuals with the first allele (first in
alphabetical order of SNP) divided by the total number of individuals in the population. The US population is
represented by grey dots, the SK population by black dots and the GRIN population with red dots. A χ2 test was
completed to test for significant differences in allele frequencies. All allele frequencies were significantly different (P
value < 0.01).
https://doi.org/10.1371/journal.pone.0227710.g004
averaging 25.8 resistance alleles, significantly more resistance alleles than the other quintiles
(Fig 7d).
Meta-QDRL analysis
Comparison to previously identified QDRL would provide further evidence to validate our
findings as well as distinguish the novel QDRL identified in this study. While over a hundred
QDRL for resistance to P. sojae have been previously identified from bi-parental mapping
studies, many were identified in only a single study. Thus, in order to focus on the robust
QDRL, we completed a meta-analysis of QDRL.
In addition to four GWA studies (S8 Table), 16 biparental mapping studies have also been
completed (S2 Table), identifying 143 QDRL positions (77 positions on Soybase) [61]. The
studies were filtered such that only a single mapping study from the same biparental cross and
generation is represented and studies prior to 2010 were removed. The remaining 11 studies
describe 114 QDRL and were included in this meta-analysis. The meta-analysis of the QDRL
Fig 5. Allele frequencies and average BLUP values of PIs with resistant and susceptible alleles at QDRL identified
in GWA analyses in the US, GRIN, and set populations. The frequency of the resistant allele is indicated in the
“Freq” column. Significantly different allele frequencies (χ2 test, α = 0.05) and instances where one population is
monomorphic were indicated with grey shading. For resistant (red) and susceptible (black) alleles of each QDRL,
means and standard errors were calculated from the population in which the significant association was identified and
are indicated with points and bars, respectively. P values from t-test are presented in the “Pheno” column with P
value < 0.05 indicated with � , < 0.01 indicated with �� and < 0.001 indicated with ��� . In order to consistently have
the resistance indicated as a higher BLUP value, BLUP values for inoculated root rot score (IRRS) were multiplied by
“-1”. Eight QDRL had a greater distribution of BLUP values and were displayed on a separate chart to aid readability.
(A) US population allele frequency and average BLUP values. (B) GRIN population allele frequency and average BLUP
values. (C) Set population allele frequency and BLUP values.
https://doi.org/10.1371/journal.pone.0227710.g005
Fig 6. Allele frequencies and average BLUP values of PIs with resistant and susceptible alleles at the QDRL identified in GWA analyses in the
C2-SK and OH-SK populations. The frequency of the resistant allele is indicated in the “Freq” column, none were significantly different (χ2 test, α =
0.05). For resistant (red) and susceptible (black) alleles of each QDRL, means and standard errors were calculated from the population in which the
significant association was identified and is indicated with points and bars, respectively. P values from t-test are presented in the “Pheno” column with
P value < 0.05 indicated with � , < 0.01 indicated with �� and < 0.001 indicated with ��� . In order to consistently have the resistance indicated as a
higher BLUP value, BLUP values for inoculated root rot score (IRRS) were multiplied by “-1”. Eight QDRL had a greater distribution of BLUP values
and were displayed on a separate chart to aid readability. Significantly different allele frequencies were identified with a χ2 test and are indicated with
grey shading.
https://doi.org/10.1371/journal.pone.0227710.g006
Fig 7. Relationship between PI QDR level and the number of resistance alleles at the identified QDRL. The total
number of resistant alleles from all QDRL was calculated for all PI in the C2 set (a,b) and OH set (c,d). An overall
resistance score was calculated by ranking each PI on the basis of a summed rank of the PI for each inoculated trait.
The linear relationship between rank and resistant allele number was calculated in the C2 set (a) and OH set (c). The
average rank and resistant allele number was calculated in quintiles in the C2 set (b) and OH set (d). Different letters
indicate significant differences in rank between quintiles as determined by Fisher’s protected least significant
difference (P value < 0.05).
https://doi.org/10.1371/journal.pone.0227710.g007
consolidated these results into 22 robust meta-QDRL distributed across ten chromosomes
(Table 5). Based on the reference genome (Wms.82.a2), the meta-QDRL are associated with an
average physical region spanning 5547kb (Table 5).
Table 5. Phytophthora sojae QDRL identified in meta-analyses of previously completed biparental mapping studies.
Namea Flanking Markersb Flanking Positionsc
m01-1 BARC-035199-07136—BARC_2.0_Gm01_3237203 1031156–3237203
m01-2 BARC_2.0_Gm01_3342559—Sat_414 3342559–52336556
m02-1 BARC-013499-00502—BARC-063263-18286 2411666–4581169
m02-2 BARC-047945-10443—BARC-043983-08572 15054401–15694465
m04-1 BARC-030751-06938—BARC-019015-03051 2062813–5241170
m06-1 BARC-05997-16280—ss1235978364 5449370–7564115
m06-2 BARC-053603-11920—Sat316 47944572–48016485
m06-3 BARC-063259-18282—ss1235977679 11363772–11755372
m08-1 BARC-032503-08989—BARC-057257-14650 7780999–9484485
m08-2 BARC-017983-02492—BARC-040893-07862 10473841–11762085
m08-3 BARC-059367-15768—BARC-050015-09290 40735744–42086643
m09-1 BARC-064511-18708—Satt417 11755372–18102265
m13-1 Sat_297—Sat_298 19951364–26976064
m13-2 BARC-018521-02928—Satt335 31794058–31831950
m13-3 AW756935—BARC-016585-02149 41606169–44344337
m15-1 Satt411—BARC-008231-00112 2517404–3964389
m15-2 BARC_2.0_Gm15_6841404—BARC_2.0_Gm15_9983279 6841404–9983279
m16-1 BARC_2.0_Gm16_298020—BARC_2.0_Gm16_759723 298020–759723
m16-2 BARC_2.0_Gm16_27864182—BARC_2.0_Gm16_58887013 58887013–27864182
m18-1 BARC-020839-03962—ss1235985359 981868–2396395
m18-2 Sat_141—BARC-005806-00273 2418920–3276029
m18-3 BARC_2.0_Gm18_5151897—Satt325 5151897–8587948
a
, The name of the meta-QTL, with the m prefix indicating that it is a meta-QDRL, followed by “-##”in which the two digit number represents the chromosome to
which the QDRL is located, and “-#” in which the number represents the order the QDRL are found on the chromosome.
b
, Left and right flanking markers determined by meta-QDRL analyses
c
, basepair position of flanking markers QDRL
https://doi.org/10.1371/journal.pone.0227710.t005
Some QDRL identified in this study were novel in their association towards P. sojae disease
resistance, but were in regions previously associated with resistance to other soybean patho-
gens and pests. QDRL OH-03-3 is in a region associated with resistance to whiteflies and soy-
bean cyst nematode (SCN) [80]. QDRL OH-05-1 is in a region associated with resistance to
the toxins produced by Fusarium virguliforme [81,82]. QDRL OH-06-1 is near the position of
a QDRL associated with resistance to sudden death syndrome (SDS) [83–85]. QDRL C2-08-3
was within regions associated with Sclerotinia and SDS resistance [86,87]. QDRL C2-11-1 is
near QDRL for SCN [88]. QDRL OH-20-1 is in a region associated with QDR to SDS [82].
OH-20-2 in a region associated with QDR to SCN [80] (Table 6). Finally, the remaining 14
QDRL are in regions not previously associated with resistance to fungal, viral, or nematode
pathogens based on information available in Soybase [61] (March 1st, 2019).
Discussion
A large and diverse population of soybean accessions led to the
identification of 44 QDRL
In this study, six traits for QDR were measured for PIs sampled from the USDA collection,
representing germplasm from the US, the Republic of Korea, as well as PIs that were sampled
without geographic consideration. PIs were phenotyped following inoculation with either P.
Table 6. Summary of colocalizations of identified QDRL with previously identified QDRL, Rps-genes, or other pathogens.
QDRL Other QDRL Rps-genes Other Pathogensa Citation(s)
b
C2-02-1 Satt634 RpsZS18 [57]/S1 Table
C2-03-1 Rps1 S1 Table
C2-03-2 Rps1 S1 Table
C2-03-3 ss715585728 [55]
OH-03-3 SCN [80]
C2-04-1 14–7c S2 Table
OH-05-1 SDS [81,82]
OH-06-1 SDS [83–85]
C2-08-2 m08-1 Table 5
C2-08-3 Sclerotinia & SDS [86,87]
C2-11-1 SCN [82]
c
C2-13-2 1–1, 2–1, 4–1, 3–1, 9–2 S2 Table
C2-13-3 1–1, 2–1, 4–1, 3–1, 9–2c S2 Table
C2-13-4 m13-1 Table 5
C2-13-5 Rps3 S1 Table
C2-13-6 Rps3 S1 Table
C2-13-7 ss715615031b Rps3 [55]
OH-13-1 Rps3 S1 Table
OH-13-2 Rps3 S1 Table
OH-13-3 Rps3 S1 Table
OH-13-4 Rps3 S1 Table
OH-13-5 Rps3 S1 Table
C2-15-1 m15-1 Table 5
C2-16-1 m16-2 Table 5
C2-17-1 Rps10 S1 Table
C2-17-2 Satt301b Rps10 [57]/S1 Table
OH-18-1 m18-1 Table 5
OH-18-2 Rps6 S1 Table
OH-20-1 SDS [82]
OH-20-2 SCN [80]
a
, identified resistance loci to other pathogens when no P. sojae resistance colocalized with identified QDRL. Colocalization with other pathosystems was not searched if
colocalization was identified with other P. sojae resistance.
b
, Significant marker referenced in GWA analysis.
c
, Name of QDRL as stored in Soybase.
https://doi.org/10.1371/journal.pone.0227710.t006
sojae isolate C2.S1 or OH.121, forming the two sets of 495 PIs in the C2 set and 478 PIs in the
OH set, respectively. Identification and comparison of QDRL for multiple isolates of P. sojae
can be informative for making breeding decisions, as it is genetically complex [89], likely con-
tributing to not only diverse virulence patterns but also other aspects of pathogenesis. Though
it would have been interesting to phenotype all PIs with both isolates, a limited number of
seeds were available from the USDA collection.
GWA analyses were completed separately depending on the aforementioned experimental
design, and though the traits were significantly correlated, unique QDRL were identified for
each measurement. Though correlated, it remains pertinent to evaluate multiple traits, as each
QDRL could be functioning though different molecular mechanism or different tissue and
may only be detected by measuring a specific trait [90]. Overall, 44 QDRL were identified, and
though GWA analyses are expected to identify large-effect and common alleles [91], here we
identified 44 small-effect QDRL explaining less than 14% of the variation, consistent with
results from previous mapping studies (S2 Table).
Superficially, the GWA analyses here appear to have identified many more QDRL than pre-
vious studies. For example, GWA analyses for QDR towards P. sojae have been completed in
more specific samples of germplasm and individually have identified a maximum of 16 mark-
ers associated with resistance [55]. Biparental mapping studies have also identified similar
numbers of QDRL [42–48]. For example, a cross of Conrad x Hefeng 25 identified eight
QDRL [44]; the cross of Conrad x Sloan identified 10 QDRL [48]; and crosses of PI 3998841 or
PI 407861A with OX20-8, identified 3 and 7 QDRL, respectively [51–53]. There were a total of
16 genetic regions associated with QDR to P. sojae identified using six nested populations in a
joint linkage mapping study with an average of 1.5 QDRL identified in individual populations
and 6 additional QDRL detected with joint linkage analyses [53].
However, both the diversity of PIs and the completion of analyses in multiple populations
likely contributed to the large number QDRL identified. Though the total number of QDRL
exceeded previous GWA results, analyses within individual population are comparable to pre-
vious research. For example, 11 QDRL were identified in the C2-SK population, and seven
QDRL in the OH-SK population. These results are similar to the seven QDRL identified in the
study completed in SK germplasm by Schneider et al. (2016) [55]. GWA analyses in Chinese
breeding lines have resulted in identifying three or one QDRLs [57,58], likewise the C2-US
and OH-US populations, consisting primarily of US cultivars, resulted in the identification of
3 QDRL.
The high level of soybean diversity in the full C2 set or OH set, may allow for the detec-
tion of novel QDRL, but it also increases the population structure, and, therefore, the risk of
false positives [92]. However, efforts to reduce false positives in GWA analyses were applied
by using models that incorporated population structure and kinship among PIs [93]. The
false-positive rate was also increased because multiple GWA analyses were completed for
each population within each set (C2 and OH) of PIs. The methodology in this study was
similar to that of Bandillo et al. (2015) [94], where a broad collection of PIs was initially
sampled for GWA analysis, followed by analyses of four subsets from the initial collection to
identify QTL for seed protein and oil content. Though testing subsets of PIs in addition to
the full set does increase the false positive rate, we applied appropriate and stringent signifi-
cance thresholds specific to each population based on the respective effective marker num-
bers [71].
conditions [47], in this study we consolidated these QDRL into 22 robust meta-QDRL distrib-
uted across 10 chromosomes. In total, twelve QDRL, or 27%, colocalized with previously iden-
tified loci from GWA analyses, non-meta-QDRL for P. sojae resistance or meta-QDRL. This is
similar to previous GWA studies where approximately 33% of QDRL co-localized with results
from other studies [55,57,59]. Five of the 22 meta-QDRL colocalized with QDRL we identified
in this study, strongly supporting the involvement of these regions in QDR to P. sojae. Though
only 5 QDRL colocalized with meta-QDRL, this analysis greatly facilitated the consolidation of
the growing number of identified QDRL for resistance towards P. sojae.
Rps-gene regions often overlap with QDRL regions [53,55]. In this study, 14 QDRL over-
lapped with or were within 500kb of previously reported Rps-gene regions. Though pheno-
typically distinct, the colocalization of Rps-genes is conspicuous, coaxing the research
question: Are canonical R-genes functioning in a QDR? Indeed, researchers have previously
proposed that QDR may be an R-gene-mediated response that does not confer complete
resistance, resulting in race-specific QDRL [95]. While studies have previously identified
QDRL that have been significant for one isolate and not other isolates [48], at this time, no
significant race or isolate × QDRL interaction has been reported, which would be the hall-
mark of a bona fide race-specific QDRL. Few studies on QDR to P. sojae have specifically
addressed race-specificity of QDRL, and those that have tested for race-specific QDRL did
not identify any significant isolate × QDRL interaction [96]. In this study, isolate, popula-
tion, as well as environment, were all confounded. Thus, we were unable to test the race or
isolate-specificity of the QDRL.
There was little overlap in QDRL identified between the C2 set and OH set
The C2 set and OH set were both comprised of US, SK and GRIN populations sampled from
the USDA germplasm collection. Though the germplasm was similar, the two sets varied for
isolate and the greenhouse environment, and while similar numbers of QDRL were identified
in both the C2 set and OH set, there was little overlap between the QDRL identified with each
isolate, among populations, or for each trait. With the exception of a region on Chr13, the
QDRL identified in either the C2 set or OH set were unique to the set.
As in other host-pathosystems, QDR is often a complex trait [28]. This type of genetic com-
plexity can be observed in the foliar disease of maize, Southern Corn Leaf Blight, where GWA
analysis for resistance in the maize NAM population resulted in the identification of 32 QDRL
[97]. Similarly, over 100 QDRL have been identified for the well-studied Fusarium head blight
resistance in wheat [98]. Based on current and previous mapping studies for resistance to P.
sojae, we can estimate that greater than 60 loci are contributing to QDR. The majority of iden-
tified QDRL toward P. sojae in soybean explain less than 15% of the variation in the population
in which they were mapped. The small individual effect of each allele contributes to the com-
plexity of the resistance phenotype itself, as well as to the genetic analysis of the trait. In fact,
while many QDRL have been reported as isolate-specific or specific to an individual resistance
trait or assay [47,48,53,55], it is unclear whether these specific QDRL are due to the marginal
significance of small effect loci, or the functional complexity of the QDR. Interestingly, though
QDRL were only identified with one isolate of P. sojae, there were significant differences in the
response trait between the resistant and susceptible alleles both within the populations in
which the QDRL were identified as well as within the corresponding population assayed with
the alternate isolate. This provides independent corroboration for these QDRL and indicates
that many, if not all, of the 44 QDRL are valid.
Among the populations that made up each isolate set, we observed unique QDRL, with no
overlap in the genetic positions of QDRL identified among populations within each isolate set.
As expected, we also observed unique alleles and genetic structure for each population within
an isolate set. Populations were sampled from specific groups, which were determined by a
population structure analysis, and possessed differences in the number of segregating markers
as well as significantly different allele frequencies at QDRL. Thus, the lack of overlap between
QDRL identified in each population is likely a result of some combination of differences in
resistance allele polymorphism, allele frequency, and/or LD.
Supporting information
S1 Table. Positions of Rps-genes/alleles in the soybean genome.
(XLSX)
S2 Table. Published quantitative disease resistance towards Phytophthora sojae mapping
studies.
(XLSX)
S3 Table. Overall ranking of resistance for plant introductions in the C2 set. PIs were
ranked from most resistance to least resistant for the four inoculated traits, ΔCRW and ΔCSW,
ranks were summed and re-ranked.
(XLSX)
S4 Table. Overall ranking of resistance for plant introductions in the OH set. PIs were
ranked from most resistance to least resistant for the four inoculated traits, ΔCRW and ΔCSW,
ranks were summed and re-ranked.
(XLSX)
S5 Table. FastStructure population number selection.
(XLSX)
S6 Table. Genetic variance and heritability in each population.
(XLSX)
S7 Table. Significant marker-trait associations identified for non-inoculated traits.
(XLSX)
S8 Table. Results of previously completed GWA analyses for P. sojae quantitative disease
resistance.
(XLSX)
S1 Fig. Comparison of QQ-plots of different genome-wide association analysis models
implemented in GAPIT. Models represented (1) GLM; general linear model, (2) MLM;
mixed linear model, (3) CMLM; compressed mixed linear model, (4) MLMM; multiple linear
mixed model.
(TIFF)
S2 Fig. Population clustering using fastStructure indicated nine subpopulations within the
SK populations. (A) Nine subpopulations identified in the C2-SK PIs. (B) Nine subpopula-
tions identified in the OH-SK PIs (C) Model complexities from one to ten were tested identify-
ing a plateauing marginal likelihood value at k = 9.
(TIFF)
S3 Fig. Population clustering using fastStructure indicated six subpopulations within the
US populations. (A) Six subpopulations identified in the C2-US population. (B) Six subpopu-
lations identified in the OH-US population. (C) Model complexities from one to eight were
tested identifying a plateauing marginal likelihood at k = 6.
(TIFF)
S4 Fig. Linkage disequilibrium decay in heterochromatic and euchromatic regions and sig-
nificance thresholds calculated using Meff. Linkage between markers (r2) as a function of
physical distance (base pair) calculated in the 974 plant introductions used in GWA analyses.
Significance thresholds include: a, effective marker number; b, calculated significance threshold
[-log10(0.05/effective marker number]; c, estimated effective marker density in markers per
kilobases.
(TIFF)
S5 Fig. Histogram distribution and qq-normality plots of the 495 PIs in C2 set used for
GWA mapping. IRRS, inoculated root rot score; IRW, inoculated root weight; ISW, inocu-
lated shoot weight; IPH, inoculated plant height; ΔRW, change in root weight; ΔSW, change in
shoot weight; NRW, non-inoculated root weight; NSW, non-inoculated shoot weight; NPH,
non-inoculated plant height.
(TIFF)
S6 Fig. Comparison of BLUP values between the US, SK, and GRIN populations in the C2
set. The GRIN population (red) the SK population (white) boxplots, and the US population
(grey) populations are represented for all nine traits: IRRS, inoculated root rot score; IRW,
inoculated root weight; ISW, inoculated shoot weight; IPH, inoculated plant height; ΔRW,
change in root weight; ΔSW, change in shoot weight; NRW, non-inoculated root weight; NSW,
non-inoculated shoot weight; NPH, non-inoculated plant height. Significantly differences in
population determined using Fisher’s protected LSDs (P value <0.05) are indicated with letters.
(TIFF)
S7 Fig. Histogram distribution and qq-normality plots of the 478 PIs in OH set used for
GWA mapping. IRRS, inoculated root rot score; IRW, inoculated root weight; ISW, inocu-
lated shoot weight; IPH, inoculated plant height; ΔRW, change in root weight; ΔSW, change in
shoot weight; NRW, non-inoculated root weight; NSW, non-inoculated shoot weight; NPH,
non-inoculated plant height.
(TIFF)
S8 Fig. Comparison of BLUP values between the US, SK, and GRIN populations in the OH
Set. The GRIN population (red) the SK population (white) boxplots, and the US population
(grey) populations are represented for all nine traits: IRRS, inoculated root rot score; IRW,
inoculated root weight; ISW, inoculated shoot weight; IPH, inoculated plant height; ΔRW,
change in root weight; ΔSW, change in shoot weight; NRW, non-inoculated root weight;
NSW, non-inoculated shoot weight; NPH, non-inoculated plant height. Significantly differ-
ences in population were tested using Fisher’s protected LSDs (P value <0.05) are indicated
with letters.
(TIFF)
S9 Fig. Manhattan plot results of non-inoculated traits in the C2 set. Population the
marker-trait association was identified in via the shape of the significant marker. The specific
trait is identified by the color of the marker. Significance thresholds calculated using SimpleM
are displayed for the C2 set, and the C2-US Population.
(TIFF)
S10 Fig. Manhattan plot results of non-inoculated traits in the OH set. The figure shows
which population the marker-trait association was identified in via the shape of the significant
marker. The specific trait is identified by the color of the marker. Significance thresholds calcu-
lated using SimpleM are displayed for the OH set, and the OH-US Population.
(TIFF)
Acknowledgments
The authors thank Dr. Colin Davis, Dr. Stephanie Karhoff, Dr. Kyujung Van, Aliya Ansari,
Lily Thompson, Emma Kline, Brandon Bowers, and Hunter Jarosz for technical assistance in
greenhouse assays. The authors also thank the Ornamental Plant Germplasm Center for their
generous provision of greenhouse space.
Author Contributions
Conceptualization: William Rolling, Leah K. McHale.
Investigation: William Rolling, Rhiannon Lake.
Writing – original draft: William Rolling, Anne E. Dorrance, Leah K. McHale.
Writing – review & editing: William Rolling, Rhiannon Lake, Anne E. Dorrance, Leah K.
McHale.
References
1. Kaufmann MJ, Gerdemann J. Root and stem rot of soybean caused by Phytophthora sojae n. sp. Phy-
topathology. 1958; 48: 201–208.
2. Dorrance AE, Lipps P, Mills D. Phytophthora damping off and root rot of soybean. Columbus: Ohio
State University Extension Fact Sheet. 2000. www.soybeanresearchinfo.com/pdf_docs/PRR_Ohio_
09.pdf
3. Schmitthenner A. Problems and progress in control of Phytophthora root rot of soybean. Plant Dis.
1985; 69: 362–368.
4. Dorrance AE. Management of Phytophthora sojae of soybean: a review and future perspectives. Can
J Plant Pathol. 2018; 40:210–219.
5. Morris PF, Ward E. Chemoattraction of zoospores of the soybean pathogen, Phytophthora sojae, by
isoflavones. Physiol Mol Plant Pathol. 1992; 40: 17–22.
6. Morris PF, Bone E, Tyler BM. Chemotropic and contact responses of phytophthora sojae hyphae to
soybean isoflavonoids and artificial substrates. Plant Physiol. 1998; 117: 1171–1178. https://doi.org/
10.1104/pp.117.4.1171 PMID: 9701573
7. Dorrance AE, Robertson A, Cianzo S, Giesler L, Grau C, Draper M, et al. Integrated management
strategies for Phytophthora sojae combining host resistance and seed treatments. Plant Dis. 2009;
93: 875–882. https://doi.org/10.1094/PDIS-93-9-0875 PMID: 30754536
8. Zhang J, Changjian X, Xiaoming W, Duan C, Sun S, Wu X, et al. Genetic characterization and fine
mapping of the novel Phytophthora resistance gene in a Chinese soybean cultivar. Theor Appl Genet.
2013; 126: 1555–1561. https://doi.org/10.1007/s00122-013-2073-1 PMID: 23467992
9. Cheng Y, Ma Q, Ren H, Xia Q, Song E, Tan Z, et al. Fine mapping of a Phytophthora-resistance gene
RpsWY in soybean (Glycine max L.) by high-throughput genome-wide sequencing. Theor Appl Genet.
2017; 130: 1041–1051 https://doi.org/10.1007/s00122-017-2869-5 PMID: 28246754
10. Lohnes D, Schmitthenner A. Position of the Phytophthora resistance gene Rps7 on the soybean
molecular map. Crop Sci. 1997; 37: 555–556.
11. Demirbas A, Rector B, Lohnes D, Fioritto R, Graef G, Cregan P, et al. Simple sequence repeat mark-
ers linked to the soybean genes for Phytophthora resistance. Crop Sci. 2001; 41: 1220–1227.
12. Weng C, Yu K, Anderson TR, Poysa V. Mapping genes conferring resistance to Phytophthora root rot
of soybean, Rps1a and Rps7. J Hered. 2001; 92: 442–446. https://doi.org/10.1093/jhered/92.5.442
PMID: 11773256
13. Sun S, Wu XL, Zhao JM, Wang YC, Tang QH, Yu DY, et al. Characterization and mapping of
RpsYu25, a novel resistance gene to Phytophthora sojae. Plant Breed. 2011; 130:139–143.
14. Gordon SG, St Martin SK, Dorrance AE. 8 Maps to a resistance gene rich region on Soybean Molecu-
lar Linkage Group F. Crop Sci. 2006; 46: 168–173.
15. Fan AY, Wang XM, Fang XP, Wu XF, Zhu ZD. Molecular identification of Phytophthora resistance
gene in soybean cultivar Yudou 25. Acta Agron Sinica. 2009; 35: 1844–1850.
16. Yu A, Xu P, Wang J, Zhang S, Wu J, Li W, et al. Genetic analysis and SSR mapping of gene resistance
to Phytophthora sojae race 1 in soybean cv Suinong 10. Chinese Journal of Oil Crop Sciences. 2010;
32: 462–466.
17. Wu X, Zhang B, Shi S, Zhao J, Feng Y, Na G, et al. Identification, genetic analysis and mapping of
resistance to Phytophthora sojae of Pm28 in soybean. Agricultural Sciences in China. 2011; 10:
1506–1511.
18. Lin F, Zhao M, Ping J, Johnson A, Zhang B, Abney TS, et al. Molecular mapping of two genes confer-
ring resistance to Phytophthora sojae in a soybean landrace PI 567139B. Theor Appl Genet. 2013;
126: 2177–2185. https://doi.org/10.1007/s00122-013-2127-4 PMID: 23689748
19. Sun J, Li L, Zhao J, Huang J, Yan Q, Xing H, et al. Genetic analysis and fine mapping of RpsJS, a
novel resistance gene to Phytophthora sojae in soybean [Glycine max (L.) Merr.]. Theor Appl Genet.
2014; 127: 913–919. https://doi.org/10.1007/s00122-014-2266-2 PMID: 24419901
20. Zhang J, Xia C, Duan C, Sun S, Wang X, Wu X, et al. Identification and candidate gene analysis of a
novel Phytophthora resistance gene Rps10 in a Chinese soybean cultivar. PloS one. 2013; 8: e69799.
https://doi.org/10.1371/journal.pone.0069799 PMID: 23936102
21. Ping J, Fitzgerald JC, Zhang C, Lin F, Bai Y, Wang D, et al. Identification and molecular mapping of
Rps11, a novel gene conferring resistance to Phytophthora sojae in soybean. Theor Appl Genet.
2016; 129: 445–451. https://doi.org/10.1007/s00122-015-2638-2 PMID: 26660465
22. Niu J, Guo N, Sun J, Li L, Cao Y, Li S, et al. Fine mapping of a resistance gene RpsHN that controls
Phytophthora sojae using recombinant inbred lines and secondary populations. Front Plant Sci. 2017;
8: 538. https://doi.org/10.3389/fpls.2017.00538 PMID: 28443124
23. Sahoo DK, Abeysekara NS, Cianzio SR, Robertson AE, Bhattacharyya MK. A novel Phytophthora
sojae resistance Rps12 gene mapped to a genomic Region that contains several Rps Genes. PloS
one. 2017; 12: e0169950. https://doi.org/10.1371/journal.pone.0169950 PMID: 28081566
24. Zhong C, Sun S, Yao L, Ding J, Duan C, Zhu Z. Fine mapping and identification of a novel Phy-
tophthora Root Rot resistance locus RpsZS18 on chromosome 2 in Soybean. Front Plant Sci. 2018; 9:
44. https://doi.org/10.3389/fpls.2018.00044 PMID: 29441079
25. Zhong C, Li Y, Sun S, Duan C, Zhu Z. Genetic mapping and molecular characterization of a broad-
spectrum Phytophthora sojae resistance gene in Chinese soybean. Int J Mol Sci. 2019; 20: 1809.
26. Grau, CR, Dorrance AE, Bond J, Russin JS. Fungal Diseases. Pages 679–763 In: editors. Soybeans:
Improvement, Production and Uses. 3rd ed. Agronomy Monograph no 16. Boerma HR, Specht JE eds
27. Dorrance A, Kurle J, Robertson A, Bradley C, Giesler L, Wise K, et al. Pathotype diversity of Phy-
tophthora sojae in eleven states in the United States. Plant Dis. 2016; 100: 1429–1437. https://doi.org/
10.1094/PDIS-08-15-0879-RE PMID: 30686193
28. Kou Y, Wang S. Broad-spectrum and durability: understanding of quantitative disease resistance.
Curr Opin Plant Biol. 2010; 13: 181–185. https://doi.org/10.1016/j.pbi.2009.12.010 PMID: 20097118
29. Poland JA, Balint-Kurti PJ, Wisser RJ, Pratt RC, Nelson RJ. Shades of gray: the world of quantitative
disease resistance. Trends Plant Sci. 2009; 14: 21–29. https://doi.org/10.1016/j.tplants.2008.10.006
PMID: 19062327
30. Nelson R, Wiesner-Hanks T, Wisser R, Balint-Kurti P. Navigating complexity to breed disease-resis-
tant crops. Nature Rev Genet. 2018; 19: 21. https://doi.org/10.1038/nrg.2017.82 PMID: 29109524
31. McDonald BA, Linde C. Pathogen population genetics, evolutionary potential, and durable resistance.
Annu Rev Phytopathol. 2002; 40: 349–379. https://doi.org/10.1146/annurev.phyto.40.120501.101443
PMID: 12147764
32. St. Clair DA. Quantitative disease resistance and quantitative resistance loci in breeding. Annu Rev
Phytopathol. 2010; 48: 247–268. https://doi.org/10.1146/annurev-phyto-080508-081904 PMID:
19400646
33. Mundt CC. Durable resistance: a key to sustainable management of pathogens and pests. Infect
Genet Evol. 2014; 27: 446–455. https://doi.org/10.1016/j.meegid.2014.01.011 PMID: 24486735
34. Krattinger SG, Lagudah ES, Spielmeyer W, Singh RP, Huerta-Espino J, McFadden H, et al. A putative
ABC transporter confers durable resistance to multiple fungal pathogens in wheat. Science. 2009;
323: 1360–1363. https://doi.org/10.1126/science.1166453 PMID: 19229000
35. Fukuoka S, Saka N, Koga H, Ono K, Shimizu T, Ebana K, et al. Loss of function of a proline-containing
protein confers durable disease resistance in rice. Science. 2009; 325: 998–1001. https://doi.org/10.
1126/science.1175550 PMID: 19696351
36. Palloix A, Ayme V, Moury B. Durability of plant major resistance genes to pathogens depends on the
genetic background, experimental evidence and consequences for breeding strategies. New Phytol.
2009; 183: 190–199. https://doi.org/10.1111/j.1469-8137.2009.02827.x PMID: 19344475
37. Fournet S, Kerlan M, Renault L, Dantec J, Rouaux C, Montarry J. Selection of nematodes by resistant
plants has implications for local adaptation and cross-virulence. Plant Pathol. 2013; 62: 184–193.
38. Dorrance A, McClure S, St. Martin S. Effect of partial resistance on Phytophthora stem rot incidence
and yield of soybean in Ohio. Plant Dis. 2003; 87: 308–312. https://doi.org/10.1094/PDIS.2003.87.3.
308 PMID: 30812766
39. Dorrance A, McClure S, DeSilva A. Pathogenic diversity of Phytophthora sojae in Ohio soybean fields.
Plant Dis. 2003; 87: 139–146. https://doi.org/10.1094/PDIS.2003.87.2.139 PMID: 30812918
40. Dorrance A, McClure S. Beneficial effects of fungicide seed treatments for soybean cultivars with par-
tial resistance to Phytophthora sojae. Plant Dis. 2001; 85: 1063–1068. https://doi.org/10.1094/PDIS.
2001.85.10.1063 PMID: 30823277
41. Xu Y, Crouch JH. Marker-assisted selection in plant breeding: from publications to practice. Crop Sci.
2008; 48: 391–407.
42. Burnham K, Dorrance A, VanToai T, St Martin S. Quantitative trait loci for partial resistance to Phy-
tophthora sojae in soybean. Crop Sci. 2003; 43: 1610–1617.
43. Han Y, Teng W, Yu K, Poysa V, Anderson T, Qiu L, et al. Mapping QTL tolerance to Phytophthora root
rot in soybean using microsatellite and RAPD/SCAR derived markers. Euphytica. 2008; 162: 231–
239.
44. Li X, Han Y, Teng W, Zhang S, Yu K, Poysa V, et al. Pyramided QTL underlying tolerance to Phy-
tophthora root rot in mega-environments from soybean cultivars ‘Conrad’and ‘Hefeng 25’. Theor Appl
Genet. 2010; 121: 651–658. https://doi.org/10.1007/s00122-010-1337-2 PMID: 20390244
45. Tucker D, Maroof S, Mideros S, Skoneczka J, Nabati D, Buss G, et al. Mapping quantitative trait loci
for partial resistance to Phytophthora sojae in a soybean interspecific cross. Crop Sci. 2010; 50: 628–
635.
46. Wang H, Waller L, Tripathy S, St Martin SK, Zhou L, Krampis K, et al. Analysis of genes underlying
soybean quantitative trait loci conferring partial resistance to Phytophthora sojae. Plant Gen. 2010; 3:
23–40.
47. Wang H, St Martin SK, Dorrance AE. Comparison of phenotypic methods and yield contributions of
quantitative trait loci for partial resistance to in soybean. Crop Sci. 2012; 52: 609–622.
48. Stasko AK, Wickramasinghe D, Nauth BJ, Acharya B, Ellis ML, Taylor CG, et al. High-density mapping
of resistance QTL toward Phytophthora sojae, Pythium irregulare, and Fusarium graminearum in the
same soybean population. Crop Sci. 2016; 56: 2476–2492.
49. Scott K, Balk C, Veney D, McHale LK, Dorrance AE. Quantitative disease resistance loci towards Phy-
tophthora sojae and three species of Pythium in six soybean nested association mapping populations.
Crop Sci. 2019; 59: 605–623.
50. Nguyen V, Vuong T, VanToai T, Lee J, Wu X, Mian M, et al. Mapping of quantitative trait loci associ-
ated with resistance to and flooding tolerance in soybean. Crop Sci. 2012; 52: 2481–2493.
51. Lee S, Mian MR, McHale LK, Wang H, Wijeratne AJ, Sneller CH, et al. Novel quantitative trait loci for
partial resistance to Phytophthora sojae in soybean PI 398841. Theor Appl Genet. 2013; 126: 1121–
1132. https://doi.org/10.1007/s00122-013-2040-x PMID: 23354974
52. Lee S, Mian R, McHale LK, Sneller CH, Dorrance AE. Identification of Quantitative Trait Loci Condi-
tioning Partial Resistance to in Soybean PI 407861A. Crop Sci. 2013; 53: 1022–1031.
53. Lee S, Mian MR, Sneller CH, Wang H, Dorrance AE, McHale LK. Joint linkage QTL analyses for partial
resistance to Phytophthora sojae in soybean using six nested inbred populations with heterogeneous
conditions. Theor Appl Genet. 2014; 127: 429–444. https://doi.org/10.1007/s00122-013-2229-z
PMID: 24247235
54. Abeysekara NS, Matthiesen RL, Cianzio SR, Bhattacharyya MK, Robertson AE. Novel sources of par-
tial resistance against Phytophthora sojae in soybean PI 399036. Crop Sci. 2016; 56: 2322–2335.
55. Schneider R, Rolling W, Song Q, Cregan P, Dorrance AE, McHale LK. Genome-wide association
mapping of partial resistance to Phytophthora sojae in soybean plant introductions from the Republic
of Korea. BMC Genomics. 2016; 17: 607. https://doi.org/10.1186/s12864-016-2918-5 PMID:
27515508
56. Dorrance A, Schmitthenner A. New sources of resistance to Phytophthora sojae in the soybean plant
introductions. Plant Dis. 2000; 84: 1303–1308. https://doi.org/10.1094/PDIS.2000.84.12.1303 PMID:
30831872
57. Sun J, Guo N, Lei J, Li L, Hu G, Xing H. Association mapping for partial resistance to Phytophthora
sojae in soybean (Glycine max (L.) Merr.). J Genet. 2014; 93: 355–363. https://doi.org/10.1007/
s12041-014-0383-y PMID: 25189230
58. Li L, Guo N, Niu J, Wang Z, Cui X, Sun J, et al. Loci and candidate gene identification for resistance to
Phytophthora sojae via association analysis in soybean [Glycine max (L.) Merr.]. Mol Gen Genomics.
2016: 291: 1095–1103
59. Ludke WH, Schuster I, Silva FLd, Montecelli TDN, Soares BdA, Oliveira ABd, et al. SNP markers
associated with soybean partial resistance to Phytophthora sojae. Crop Breed Appli Biotechnol. 2019;
19: 31–39.
60. Song Q, Hyten DL, Jia G, Quigley CV, Fickus EW, Nelson RL, et al. Development and evaluation of
SoySNP50K, a high-density genotyping array for soybean. PloS one. 2013; 8: e54985. https://doi.org/
10.1371/journal.pone.0054985 PMID: 23372807
61. Grant D, Nelson RT, Cannon SB, Shoemaker RC. SoyBase, the USDA-ARS soybean genetics and
genomics database. Nucleic Acids Res. 2010; 38: D843–6. https://doi.org/10.1093/nar/gkp798 PMID:
20008513
62. Scheet P, Stephens M. A fast and flexible statistical model for large-scale population genotype data:
applications to inferring missing genotypes and haplotypic phase. Am J Hum Genet. 2006; 78: 629–
644. https://doi.org/10.1086/502802 PMID: 16532393
63. Raj A, Stephens M, Pritchard JK. fastSTRUCTURE: variational inference of population structure in
large SNP data sets. Genetics. 2014; 197: 573–589. https://doi.org/10.1534/genetics.114.164350
PMID: 24700103
64. Olhoft PM, Donovan CM, Somers DA. Soybean (Glycine max) transformation using mature cotyledon-
ary node explants. Agrobacter Protoc. 2006; 343 395–396
65. Dorrance AE, Berry SA, Anderson TR, Meharg C. Isolation, storage, pathotype characterization, and
evaluation of resistance for Phytophthora sojae in soybean. Plant Health Progress. 2008; 10: 1094.
66. R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical
Computing. 2012.
67. Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. J Stat Softw.
2015; 67: https://doi.org/10.18637/jss.v067.i01
68. Zhang Z, Ersoz E, Lai C, Todhunter RJ, Tiwari HK, Gore MA, et al. Mixed linear model approach
adapted for genome-wide association studies. Nat Genet. 2010; 42: 355–360. https://doi.org/10.1038/
ng.546 PMID: 20208535
69. Gao X, Starmer J, Martin ER. A multiple testing correction method for genetic association studies
using correlated single nucleotide polymorphisms. Genet Epidemiol. 2008; 32: 361. https://doi.org/10.
1002/gepi.20310 PMID: 18271029
70. Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps.
Bioinformatics. 2005; 21: 263–265. https://doi.org/10.1093/bioinformatics/bth457 PMID: 15297300
71. Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B, et al. The structure of haplo-
type blocks in the human genome. Science. 2002; 296: 2225–2229. https://doi.org/10.1126/science.
1069424 PMID: 12029063
72. Darvasi A, Soller M. A simple method to calculate resolving power and confidence interval of QTL map
location. Behav Genet. 1997; 27: 125–132. https://doi.org/10.1023/a:1025685324830 PMID: 9145551
73. Arcade A, Labourdette A, Falque M, Mangin B, Chardon F, Charcosset A, et al. BioMercator: integrat-
ing genetic maps and QTL towards discovery of candidate genes. Bioinformatics. 2004; 20: 2324–
2326. https://doi.org/10.1093/bioinformatics/bth230 PMID: 15059820
74. Sosnowski O, Charcosset A, Joets J. BioMercator V3: an upgrade of genetic map compilation and
quantitative trait loci meta-analysis algorithms. Bioinformatics. 2012; 28: 2082–2083. https://doi.org/
10.1093/bioinformatics/bts313 PMID: 22661647
75. Song Q, Marek L, Shoemaker R, Lark K, Concibido V, Delannay X, et al. A new integrated genetic link-
age map of the soybean. Theor Appl Genet. 2004; 109: 122–128. https://doi.org/10.1007/s00122-004-
1602-3 PMID: 14991109
76. Schmutz J, Cannon SB, Schlueter J, Ma J, Mitros T, Nelson W, et al. Genome sequence of the palaeo-
polyploid soybean. Nature. 2010; 463: 178–183. https://doi.org/10.1038/nature08670 PMID: 20075913
77. Slatkin M. Linkage disequilibrium–understanding the evolutionary past and mapping the medical
future. Nat Rev Genet. 2008; 9: 477–485. https://doi.org/10.1038/nrg2361 PMID: 18427557
78. Breseghello F, Sorrells ME. Genetics. Association mapping of kernel size and milling quality in wheat
(Triticum aestivum L.) cultivars. 2006; 172:1165–1177
79. Rosenberg NA, Huang L, Jewett EM, Szpiech ZA, Jankovic I, Boehnke M. Genome-wide association
studies in diverse populations. Nat Rev Genet. 2010; 11: 353–366.
80. Jiao Y, Vuong TD, Liu Y, Meinhardt C, Liu Y, Joshi T, et al. Identification and evaluation of quantitative
trait loci underlying resistance to multiple HG types of soybean cyst nematode in soybean PI 437655.
Theor Appl Genet. 2015; 128: 15–23. https://doi.org/10.1007/s00122-014-2409-5 PMID: 25316311
81. Arahana VS, Graef GL, Specht JE, Steadman JR, Eskridge KM. Identification of QTLs for resistance
to Sclerotinia sclerotiorum in soybean. Crop Sci. 2001; 41: 180–188.
82. Swaminathan S, Abeysekara NS, Liu M, Cianzio SR, Bhattacharyya MK. Quantitative trait loci under-
lying host responses of soybean to Fusarium virguliforme toxins that cause foliar sudden death syn-
drome. Theor Appl Genet. 2016; 129: 495–506. https://doi.org/10.1007/s00122-015-2643-5 PMID:
26678962
83. Moellers TC, Singh A, Zhang J, Brungardt J, Kabbage M, Mueller DS, et al. Main and epistatic loci
studies in soybean for Sclerotinia sclerotiorum resistance reveal multiple modes of resistance in multi-