Carboxylic Acid Amide but Not Quinone Outside Inhibitor Fungicide Resistance Mutations Show Clade-Specific Occurrence in Pseudoperonospora cubensis Causing Downy Mildew in Commercial and Wild Cucurbits

Fungicide resistance has significantly limited control efforts for many oomycete diseases of agricultural importance. Oomycete pathogens such as Phytophthora capsici Leonian and Pseudoperonospora cubensis (Berk. & M. A. Curtis) Rostovzev can develop resistance to fungicides in just a few years, creating the need for researchers, companies, and growers to constantly identify new chemistries for crop protection (Blum et al. 2011; D'Arcangelo et al. 2021; Gisi et al. 2002; Kousik et al. 2017; Parada-Rojas and Quesada-Ocampo 2022). For many fungicide groups, the targets of a particular mode of action are unknown and the only way to identify fungicide resistance phenotypes in pathogen populations is by challenging isolates with the fungicide through field evaluations or laboratory experiments with amended media (Bagi et al. 2014; Bittner et al. 2017; Wang et al. 2013). Nonetheless, in recent years, diagnostic assays to detect fungicide resistance mutations have been developed for some fungal and oomycete pathogens (Salcedo et al. 2021).

One widely studied group of fungicides are the quinone outside inhibitor (QoI), strobilurin or FRAC 11 fungicides. These products bind to the outer quinone-oxidizing pocket of the cytochrome bc1 enzyme complex III, inhibiting mitochondrial respiration by blocking electron transfer which terminates cellular respiration and the production of ATP (Gisi et al. 2002). Gisi et al. (2002) reported a single nucleotide polymorphism (SNP) that causes an amino acid substitution from glycine to alanine at position 143 (G143A) in the cytochrome b (cytb) gene. Within P. cubensis, the G143A substitution was associated with high or complete resistance based on sequencing of the cytb gene and species-specific markers were developed (Gisi and Sierotzki 2008). Similarly, two studies determined a range of substitutions at the 1105 position within the CesA3 gene that resulted in fungicide resistance to carboxylic acid amide (CAA) or FRAC 40 fungicides within Plasmopara viticola and Pseudoperonospora cubensis isolates. CAA fungicides inhibit cellulose biosynthesis in Peronosporales by targeting the cellulose synthase 3 (CesA3) gene (Blum et al. 2012; Sierotzki et al. 2011). Shortly thereafter, Blum et al. (2011) expanded on this research and found that an SNP that caused an amino acid substitution from a conserved glycine to an alanine (A), serine (S), valine (V), or tryptophan (W) at position 1105 (G1105A/S/V/W) or a substitution from a valine to a leucine (L) or methionine (M) at position 1109 (V1109L/M) in the CesA3 gene was most commonly associated with CAA resistance. For P. cubensis isolates in vitro, it was determined that the 1105W and 1105V were the most common mutations to occur, but their frequency and distribution in P. cubensis populations remains unknown (Blum et al. 2011). It has been reported that CAA resistance is based on recessive gene(s) and resistant phenotypes are only expressed in homozygous isolates while heterozygous isolates are sensitive (Thind 2012).

P. cubensis, causal agent of cucurbit downy mildew (CDM), has experienced drastic changes in fungicide sensitivity since its reemergence in 2004 in the United States (Cohen et al. 2015; Holmes et al. 2015; Quesada-Ocampo et al. 2012; Ojiambo et al. 2015). A disease previously controlled in cucumber using host resistance and in other cucurbits via modest fungicide applications, is now considered the most economically devastating disease that infects the family Cucurbitaceae (Cohen et al. 2015; Holmes et al. 2015; Salcedo et al. 2020). Frequent fungicide applications are required to control CDM (D'Arcangelo et al. 2021). Host-adapted clades exist in P. cubensis with clade 1 isolates most frequently infecting summer squash (Cucurbita pepo), butternut squash (Cucurbita moschata), pumpkin (Cucurbita maxima), and watermelon (Citrullus lanatus) and clade 2 isolates most frequently infecting cucumber (Cucumis sativus) and muskmelon or cantaloupe (Cucumis melo) (Rahman et al. 2021; Wallace et al. 2020). Wild cucurbits, such as Cucurbita foetidissima, Momordica balsamina, and M. charantia can also become infected by P. cubensis (Wallace et al. 2014, 2015), but it remains unclear what role wild hosts may be playing in serving as reservoirs for P. cubensis isolates containing fungicide resistance mutations (Wallace and Quesada-Ocampo 2017; Wallace et al. 2020). In recent years, loss of sensitivity in P. cubensis to mefenoxam (FRAC 4), QoIs (FRAC 11), CAAs (FRAC 40), fluopicolide (FRAC 43), and propamocarb (FRAC 28) have been reported in different cucurbit production regions in the United States leaving growers with few options for disease control (Adams and Quesada-Ocampo 2014; Blum et al. 2011; Gisi and Sierotzki 2008; Hausbeck and Linderman 2014; Rahman et al. 2017; Thomas et al. 2018). Additionally, as researchers from various states in the United States conduct annual evaluations of fungicide efficacy in the field, it is not uncommon to observe different fungicide sensitivities by region and/or by cucurbit host (D'Arcangelo et al. 2021; Keinath et al. 2019). Typically, fungicide sensitivity in P. cubensis is evaluated on cucumber in limited states and then recommendations are extrapolated to other crops and regions. However, considering the two host-adapted P. cubensis clades infecting cucurbits (Wallace et al. 2020) and the seasonality described for those clades using spore trapping in North Carolina (NC) (Rahman et al. 2021), it would be beneficial to investigate how clade composition impacts fungicide sensitivity. Thus, using microsatellite markers (Wallace et al. 2020; Wallace and Quesada-Ocampo 2017), molecular diagnostic assays for QoIs (G143A) and CAAs (G1105W/V) resistance mutations (Rahman et al. 2017), and a P. cubensis clade diagnostic assay (Rahman et al. 2021), in this study we aimed to (i) evaluate the occurrence of fungicide resistance mutations to QoI and CAA fungicides in P. cubensis isolates infecting commercial cucurbits in NC; (ii) determine if wild cucurbits can serve as reservoirs for P. cubensis isolates with fungicide resistance mutations; and (iii) establish if the occurrence of QoI and/or CAA fungicide resistance mutations are associated to a particular P. cubensis clade and what impacts this may have on management practices. Our findings indicate that CAA but not QoI fungicide resistance mutations show clade-specific occurrence in P. cubensis, which highlights the importance of understanding the population dynamics of P. cubensis clades by crop and region to design effective disease management strategies.

Materials and Methods

Sample collection, DNA extraction, and microsatellite amplification

Isolates of P. cubensis were collected from infected leaves of commercial and wild cucurbit hosts in field plots from Eastern, Central, and Western NC at two time points (summer and fall) per year in 2013 and 2014 as previously described by Wallace et al. (2020). Field plots with commercial cucurbit hosts contained 10 plants per plot and wild cucurbit plots contained five plants per plot. Commercial cucurbits included Cucumis sativus ‘Straight 8’ (cucumber), Cucumis melo ‘Hale's Best Jumbo’ (cantaloupe), Cucurbita pepo ‘Table Ace’ (acorn squash), Cucurbita maxima ‘Big Max’ (pumpkin), Cucurbita moschata ‘Waltham Butternut’ (butternut squash), and Citrullus lanatus ‘Micky Lee’ (watermelon). The wild cucurbits planted included M. charantia (bitter melon), M. balsamina (balsam apple), Cucurbita foetidissima (buffalo gourd), and Lagenaria siceraria (bottle gourd). Infected leaves from each plot were collected in gallon-sized plastic bags.

In the lab, sporulation of P. cubensis was verified under a dissecting microscope. A single lesion was removed from each leaf sample with a sterile scalpel and was placed into a microcentrifuge tube at −80°C until DNA extraction was completed. In total, 425 to 600 µM acid-washed glass beads (Sigma Life Sciences, St. Louis, MO) and 2.3 mm zirconia/silica beads (BioSpec Products, Inc., Bartlesville, OK) were added to each microcentrifuge tube and tissue was disrupted using an Omni BeadRuptor-24 (Omni International, Inc., Kennesaw, GA). DNA was extracted using a standard phenol-chloroform method (Wallace and Quesada-Ocampo 2017; Wallace et al. 2020), suspended in 1× TE buffer, and quantified using a NanoDrop ND 1000 spectrophotometer and NanoDrop 2.4.7c software (NanoDrop Technologies, Inc., Wilmington, DE). Integrity was confirmed using gel electrophoresis with the presence of a band greater than 12,000 bp.

Eleven microsatellites (Supplementary Table S1) were amplified and subjected to fragment analysis as previously described by Wallace and Quesada-Ocampo (2017) and Wallace et al. (2020). Forward primers were labeled with a partial M13 tail on the 5′ end for downstream fragment analysis (Schuelke 2000). Labeled PCR products were submitted to the NC State Genomic Sciences Laboratory (GSL, Raleigh, NC) for genotyping with a 3730xl DNA Analyzer (Applied Biosystems, Foster City, CA). Raw data were individually analyzed by calling peaks and binning them with the Geneious Microsatellite Plug-in (Kearse et al. 2012). Once genotyped, samples with greater than 20% missing data were removed from analysis, allowing the total data set to have 385 isolates with an average of 4.6% missing data (Wallace et al. 2020). In the current study, 236 of the 385 isolates from Wallace et al. (2020) were used (Supplementary Table S2). All genotyping data used for analyses are available at DOI: 10.6084/m9.figshare.19754344.

Amplification of fungicide resistance mutations

Allele-specific TaqMan qPCR assays were further developed from PCR and digital droplet PCR (ddPCR) assays used in Rahman et al. (2017) based on mutations previously known to be associated with QoI (FRAC 11) and CAA (FRAC 40) resistances in other oomycetes (Blum et al. 2012; FRAC 2020; Ishii et al. 2001). All reactions were completed in a CFX 96 Touch qPCR system equipped with Bio-Rad CFX Manager software (Bio-Rad, Hercules, CA) using hard-shell 96-well plates (Bio-Rad). For the QoI assay, a 25-µl reaction was utilized containing 12.5 μl of PerfeCta qPCR ToughMix (QuantaBio, VWR, Beverly, MA), 0.5 μl of MgCl₂ (25 mM, NewEngland BioLabs, Ipswich, MA), 1.25 μl of the forward and reverse primers (10 μM, kd947, kd948), 0.25 μl of each TaqMan probe (10 μM, LGC Biosearch Technologies, Novato, CA, 10 μM, kd966, kd967), and 1 μl of P. cubensis DNA (5 ng) extracted from leaf tissue and sterile distilled water (Table 1). The probes used included one for detection of the sensitivity genotype (G143) and another for detection of an SNP in amino acid position 143 in the cytochrome b gene which causes a substitution from a glycine to an alanine (G143A) (Supplementary Fig. S1). For the QoI assay, the following protocol was used: an initial denaturation period for 5 min at 95°C followed by cycles of 95°C for 15 s and 65°C for 1 min for 44 cycles. For the CAA assay, a 25-µl reaction was also used including the same reagents except for the primer set of interest (as775, as776) and three TaqMan probes at 10 nM (kd1496, kd1515, and kd1516; Table 1). The probes utilized detected the sensitivity genotype (G1105), an SNP at amino acid position 1105 in the cellulose synthase 3 (CesA3) gene which induces a substitution from a glycine to a tryptophan or valine (G1105W/V) (Supplementary Fig. S2). For the CAA assay, the protocol was as follows: an initial denaturation period for 5 min at 95°C followed by cycles of 95°C for 15 s and 68°C for 1 min for 49 cycles.

TABLE 1. Primers and probes used in this study to determine the distribution of host-adapted Pseudoperonospora cubensis clade and characterize the frequency of single nucleotide polymorphisms associated with the development of fungicide resistance^a

Assay	ID	Description	Sequence (5′ to 3′)	Product size (bp)
Carboxylic acid amide (CAA)	as775	CAA reverse primer	CGCGTAGTATACTTGCCACAGCTG	185
	as776	CAA forward primer	CGTGGAATTACGTGTCAGCAATGTT
	kd1496	CAA G1105V probe	Quasar670-ACCACCAGCAACGACACAAACACGTTT–BHQ-2	N/A
	kd1515	CAA G1105 probe	FAM-ACCACCAGCAACGACCCAAACACGTTT–BHQ-1	N/A
	kd1516	CAA G1105W probe	HEX-ACCACCAGCAACGACCAAAACACGTTT–BHQ-1	N/A
Quinone outside inhibitor (QoI)	kd947	QoI forward primer	TGGTTAATTAGATATACTCATGCG	221
	kd948	QoI reverse primer	GGATTATCTACAGCAAAACCG
	kd966	QoI G143 probe	FAM-CAAATGAGTTTTTGGGGTGCAACTG–BHQ-1	N/A
	kd967	QoI G143A probe	HEX-CAAATGAGTTTTTGGGCTGCAACTG–BHQ-1	N/A
Clade	ar1517	Universal forward primer	AACGGTGATCCATGCAGCTTTA	N/A
	ar1531	MSU-1 or clade 2 reverse primer	CCATCAAGCCAGCAACTTGTT	93
	ar1495	SC1982 or clade 1 reverse primer	CGCAGTGGTTGGGTGTGT	111
	ar1529	MSU-1 or clade 2 probe	FAM-TGGTCGAGC+ATTGACAAGAGCCTATCC-IABkFQ	N/A
	ar1497	SC1982 or clade 1 probe	HEX-CAG+TAGCA+TAACCCAAG+ACTTCGT-IABkFQ	N/A

^a Product sizes are indicated for primers used for amplification. For probes, +A or +T indicate locked nucleic acids in the probes. Primers and probes for CAA and QoI assays were designed herein, while the clade assay was developed by Rahman et al. (2021).

View as image HTML

Clade differentiation

A clade-specific TaqMan qPCR assay, which included locked nucleic acid (LNA) probes, was used for clade differentiation (Rahman et al. 2021). LNA probes utilized were dye-labeled (ar1497, ar1529) and had Iowa BlackÒ (IABkFQ) as a quencher (Table 1). The 20-μl multiplex qPCR reaction included 4 μl of PerfeCta Multiplex qPCR ToughMix (QuantaBio), 1.6 μl of 25 mM MgCl₂, 1.5 μl of the universal forward primer (10 μM, ar1517), 1.0 μl of each reverse primer (10 μM, ar1495 and ar1531), 0.2 μl of each LNA probe (10 μM), and 1 μl of template DNA (5 ng) extracted from leaf tissue (Table 1) and sterile distilled water. A CFX 96 Touch qPCR system (Bio-Rad) with Bio-Rad CFX Manager software was used to run all reactions in hard-shell 96-well plates (Bio-Rad) with the following reaction protocol: an initial denaturation for 5 min at 95°C followed by 38 cycles of 95°C for 15 s and 65°C for 1 min.

Population structure analyses

For confirmation that the reduced data set from Wallace et al. (2020) still had two optimal genetic clusters, the admixture model with correlated allele frequencies was used in the program STRUCTURE 2.3.4 and individuals (N = 236) were assigned to determine genetic clusters (Pritchard et al. 2000). A burn-in of 30,000 and a Monte Carlo Markov Chain of 500,000 was used. Ten runs were performed for each K (1-40) and 20 for each K (1-5). Once raw STRUCTURE data was obtained, STRUCTURE Harvester was used to determine the optimal number of genetic clusters (K).

Estimation of genetic diversity, genetic differentiation, and linkage disequilibrium analyses in relation to fungicide resistance genotype distributions

Following guidelines introduced by Quesada-Ocampo et al. (2012) and Naegele et al. (2016), categories with at least seven individuals were included in analyses to examine genetic diversity and linkage disequilibrium analyses (N = 108) in relation to CAA and QoI fungicide resistance genotypes. For CAA genotypes, groupings with more than seven individuals included clade 1 sensitive (G1105, Clade1-S, N = 45), clade 2 sensitive (G1105, Clade2-S, N = 15), and clade 2 resistant (G1105W, Clade2-Rw, N = 48). Regarding QoI analyses, groups with more than seven individuals included clade 1 sensitive (G143, Clade1-S, N = 13), clade 1 resistant (143A, Clade1-R, N = 32), and clade 2 resistant (Clade2-R, N = 63). Similar to Wallace et al. (2020), locus-based statistics such as Simpson's index, expected heterozygosity, and evenness were generated for the aforementioned groupings through the function “locus_table” using R package Poppr (Kamvar et al. 2014). Additionally, genetic diversity metrics were calculated with the “poppr” function within described clade and fungicide resistance categories.

The index of association (I_A) and standardized index of association (rbarD) were calculated to test for random recombination of P. cubensis isolates within the described groupings above using 999 resamplings from a clone-corrected data set. The index of association is a measure of multilocus linkage disequilibrium by quantifying recombination among microsatellite loci and detecting association between alleles. If isolates are freely recombining, which is consistent with random mating, I_A and rbarD are expected to be zero, while indices are greater than zero if isolates are not freely recombining; therefore, indicating nonrandom mating (Kamvar et al. 2014). A standardized index of association (rbarD) was an additional statistic utilized because it is considered a more robust test that considers the number of loci tested. Statistical significance (P ≤ 0.001) indicated an occurrence of non-random mating with no evidence of recombination, while nonsignificant values indicated random mating and evidence of recombination (Agapow and Burt 2001).

Pairwise comparisons of F_ST values were generated to determine population differentiation across clade and fungicide resistance groupings using GenAlEx (Peakall and Smouse 2006). Using 999 permutations, P values and significance (P ≤ 0.05) were calculated for pairwise F_ST values. Standards of genetic differentiation described by Hartl and Clark (2006) were used as a measure to interpret differentiation as low (<0.10), moderate (0.10 to 0.20), or high (>0.20). Additionally, a discriminant analysis of principle components (DAPC) was completed to examine population differentiation within the F_ST populations described above (Jombart et al. 2010). DAPC is a multivariate statistical approach that partitions variance in the sample into a between- and within-group component to maximize discrimination between predetermined groups. Using 1,000 replicates, the function ‘xvalDapc’ from the R package adegenet was used to select the correct number of principle components (PC) as DAPC is sensitive to the number of PCs used within analysis. The chosen number of PCs was based on the highest average percentage of successful reassignment and lowest root mean squared error (Jombart et al. 2010) per grouping.

Results

P. cubensis populations are structured into two host-adapted clades

To allow for more accurate and robust comparisons between host-adapted clades and fungicide resistance genotypes, 10 isolates showing a mixed clade via qPCR assay were removed from the original data set (N = 236) for a total of 226 P. cubensis isolates collected from wild and commercial hosts belonging to a single clade. For the 226 isolates retained for analyses, Bayesian clustering identified two genetic clusters (K1 and K2) (Supplementary Fig. S3). All but five isolates had greater than a 60% membership to one of the two K clusters and evidence of host-adaptation was observed. K2 isolates were more frequently found in Cucumis sativus, Cucumis melo, and the wild host Lagenaria siceraria in contrast to K1, which included isolates collected from Cucurbita pepo, Cucurbita maxima, Cucurbita moschata, Citrullus lanatus, and the wild hosts M. charantia and M. balsamina.

CAA and QoI fungicide resistance genotypes are more abundant in commercial cucurbits

When examining isolates infecting commercial hosts for the presence of CAA resistance mutations, we found that most genotypes containing 1105W resistance alleles occurred in isolates from cucumber and cantaloupe (Table 2), which mostly belong to clade 2 (Table 3). The 1105W resistant genotype was observed in 78.3% of cucumber isolates (N = 60), 60.0% of cantaloupe isolates (N = 30), and in only 6.1% of pumpkin isolates (Table 2). Within commercial hosts, isolates infecting species in the genus Cucurbita had the greatest occurrence of the 1105V resistance allele and a greater content of G1105 sensitive genotypes. In contrast, isolates infecting Cucumis species exhibited the G1105W resistant genotype and the resistant genotype 1105V was not present in any of the isolates from the Cucumis hosts (Table 2). Watermelon isolates (N = 14) all had the G1105 sensitive genotype, while greater than 40% of acorn squash (N = 28) and pumpkin (N = 33) isolates also had the G1105 sensitive genotype. The heterozygous genotype G1105-1105V was present in 11.8, 25, and 39.4% of isolates of butternut squash, acorn squash, and pumpkin, respectively, and was only observed in 1.7% of cucumber isolates (Table 2). Additionally, isolates from 2013 had the greatest occurrence of G1105 sensitive genotypes, while isolates from 2014 had a greater occurrence of both the G1105-1105V and G1105-1105V-1105W genotypes (Table 4).

TABLE 2. Occurrence of carboxylic acid amide (CAA) and quinone outside inhibitor (QoI) genotypes expressed as percentages within Pseudoperonospora cubensis isolates infecting commercial and wild cucurbits in North Carolina separated by host^a

	Commercial hosts												Wild hosts
	Cucumis sativus		Cucumis melo		Cucurbita pepo		Cucurbita maxima		Cucurbita moschata		Citrullus lanatus		Momordica charantia		Momordica balsamina		Cucurbita foetidissima		Lagenaria siceraria
Genotype	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%
CAA genotype
G1105	3	5.0	6	20.0	12	42.9	14	42.4	13	38.2	14	100.0	2	33.3	3	60.0	2	20.0	0	0.0
1105W	47	78.3	18	60.0	0	0.0	2	6.1	0	0.0	0	0.0	1	16.7	1	20.0	3	30.0	4	66.7
1105V	0	0.0	0	0.0	1	3.6	0	0.0	2	5.9	0	0.0	0	0.0	0	0.0	0	0.0	0	0.0
G1105-1105V	1	1.7	0	0.0	7	25.0	13	39.4	4	11.8	0	0.0	1	16.7	1	20.0	1	10.0	0	0.0
G1105-1105W	1	1.7	1	3.3	0	0.0	3	9.1	6	17.6	0	0.0	1	16.7	0	0.0	1	10.0	0	0.0
1105V-1105W	2	3.3	0	0.0	2	7.1	0	0.0	0	0.0	0	0.0	0	0.0	0	0.0	1	10.0	1	16.7
G1105-1105V-1105W	6	10.0	5	16.7	6	21.4	1	3.0	9	26.5	0	0.0	1	16.7	0	0.0	2	20.0	1	16.7
Total	60		30		28		33		34		14		6		5		10		6
QoI genotype
G143	0	0.0	1	3.3	3	10.7	2	6.3	2	5.9	6	42.9	0	0.0	5	100.0	2	20.0	0	0.0
G143-143A	22	36.7	6	20.0	7	25.0	7	21.9	12	35.3	0	0.0	1	16.7	0	0.0	1	10.0	0	0.0
143A	38	63.3	23	76.7	18	64.3	23	71.9	20	58.8	8	57.1	5	83.3	0	0.0	7	70.0	7	100.0
Total	60		30		28		32		34		14		6		5		10		7

^a CAA genotypes include G1105 (Sensitive, S), 1105W (Resistant, Rw), 1105V (Resistant, Rv), G1105-1105V (S-Rv), G1105-1105W (S-Rw), G1105-1105W (Rv-Rw), and G1105-1105V-1105W (S-Rv-Rw). QoI genotypes include G143 (Sensitive, S), G143-143A (Resistant-S, R-S), and 143A (R). Hosts include Cucumis sativus, Cucumis melo, Cucurbita pepo, Cucurbita maxima, Cucurbita moschata, Citrullus lanatus, M. charantia, M. balsamina, Cucurbita foetidissima, and Lagenaria siceraria.

View as image HTML

TABLE 3. Occurrence of carboxylic acid amide (CAA) and quinone outside inhibitor (QoI) genotypes expressed as percentages within Pseudoperonospora cubensis isolates infecting commercial and wild cucurbits in North Carolina separated by clade^a

	Clade 1		Clade 2
Genotype	No.	%	No.	%
CAA genotype
G1105	54	45.8	15	13.9
1105W	6	5.1	70	64.8
1105V	3	2.5	0	0.0
G1105-1105V	24	20.3	3	2.8
G1105-1105W	9	7.6	5	4.6
1105V-1105W	2	1.7	4	3.7
G1105-1105V-1105W	20	16.9	11	10.2
Total	118		108
QoI genotype
G143	17	14.4	4	3.7
G143-143A	31	26.3	25	23.1
143A	70	59.3	79	73.1
Total	118		108

^a CAA genotypes include G1105 (Sensitive, S), 1105W (Resistant, Rw), 1105V (Resistant, Rv), G1105-1105V (S-Rv), G1105-1105W (S-Rw), G1105-1105W (Rv-Rw), and G1105-1105V-1105W (S-Rv-Rw). QoI genotypes include G143 (Sensitive, S), G143-143A (Resistant-S, R-S), and 143A (R).

View as image HTML

TABLE 4. Occurrence of carboxylic acid amide (CAA) and quinone outside inhibitor (QoI) genotypes expressed as percentages within Pseudoperonospora cubensis isolates infecting commercial and wild cucurbits in North Carolina separated by clade, year, and collection time^a

	Clade 1 (N = 118)				Clade 2 (N = 108)				Clade 1 (N = 118)				Clade 2 (N = 108)
	2013		2014		2013		2014		Summer		Fall		Summer		Fall
Genotype	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%
CAA genotype
G1105	39	61.9	15	27.3	10	20.4	5	8.5	23	52.3	31	41.9	8	13.8	7	14.0
1105W	5	7.9	1	1.8	31	63.3	39	66.1	4	9.1	2	2.7	36	62.1	34	68.0
1105V	2	3.2	1	1.8	0	0	0	0.0	0	0.0	3	4.1	0	0.0	0	0.0
G1105-1105V	7	11.1	17	30.9	0	0	3	5.1	1	2.3	23	31.1	0	0.0	3	6.0
G1105-1105W	5	7.9	4	7.3	4	8.2	1	1.7	7	15.9	2	2.7	3	5.2	2	4.0
1105V-1105W	1	1.6	1	1.8	1	2	3	5.1	0	0.0	2	2.7	2	3.4	2	4.0
G1105-1105V-1105W	4	6.3	16	29.1	3	6.1	8	13.6	9	20.5	11	14.9	9	15.5	2	4.0
Total	63		55		49		59		44		74		58		50
QoI genotype
G143	14	22.2	3	5.5	1	2.0	3	5.1	12	27.3	5	6.8	3	5.2	1	2.0
G143-143A	17	27.0	14	25.5	10	20.4	15	25.4	9	20.5	22	29.7	11	19.0	14	28.0
143A	32	50.8	38	69.1	38	77.6	41	69.5	23	52.3	47	63.5	44	75.9	35	70.0
Total	63		55		49		59		44		74		58		50

^a CAA genotypes include G1105 (Sensitive, S), 1105W (Resistant, Rw), 1105V (Resistant, Rv), G1105-1105V (S-Rv), G1105-1105W (S-Rw), G1105-1105W (Rv-Rw), and G1105-1105V-1105W (S-Rv-Rw). QoI genotypes include G143 (Sensitive, S), G143-143A (Resistant-S, R-S), and 143A (R).

View as image HTML

QoI resistance mutations were much more prevalent across all commercial hosts (Table 2) regardless of clade (Table 3). Isolates with the 143A resistance genotype were observed in greater than 50% of all commercial hosts (Table 2) in both clades (Table 3), with watermelon isolates exhibiting the greatest amount of the G143 sensitive genotype (42.9%). The heterozygous genotype G143-143A was most prevalent in cucumber (36.7%) and butternut squash (35.3%) and was detected in all commercial hosts except watermelon (Table 2).

Wild cucurbits also have P. cubensis isolates with fungicide resistance mutations

Wild hosts of P. cubensis can in fact serve as reservoirs for fungicide resistant isolates. Considering CAA resistance, all wild hosts harbored different combinations of genotypes where at least one resistance allele was present (G1105-1105V, G1105-1105W, 1105V-1105W, and G1105-1105V-1105W) although this may not always result in a resistant phenotype due to CAA resistance inheritance patterns being recessive. The CAA resistant 1105W genotype was most prevalent in Lagenaria siceraria isolates (66.7%) (Table 2), while isolates from M. charantia and M. balsamina had a greater occurrence of the G1105 sensitive genotype (33.3 and 60.0%, respectively). The wild perennial host to NC, Cucurbita foetidissima, harbored at least one isolate with every resistance allele (Table 2). Wild hosts also harbor isolates with QoI resistance genotypes. M. charantia had 83.3% isolates with QoI resistant143A genotypes and Cucurbita foetidissima had 70.0% (Table 2). Comparatively, 100% of L. siceraria isolates had 143A resistant genotypes, while M. balsamina had 100% isolates with the G143 sensitive genotype.

CAA resistance mutations are more prevalent in clade 2 isolates while QoI resistance mutations are present in both clades of P. cubensis

When evaluating genotypes of P. cubensis isolates related to CAA resistance, we found that most isolates with G1105 sensitive genotype were in clade 1 compared with clade 2 (78.3 and 21.7%, respectively) (Table 3). Additionally, the 1105V resistant genotype and heterozygous genotype G1105-1105V were exclusive to clade 1, while the 1105W resistant genotype and heterozygous genotype 1105V-1105W were predominant in clade 2 (92.1 and 66.7%, respectively). Interestingly, the heterozygous genotypes G1105-1105W and triploid genotype G1105-1105V-1105W were mostly in clade 1 isolates when compared with clade 2 (64.3 and 64.5%, respectively) (Table 3). Additionally, some genotypes were more commonly found in a particular clade and year, for example, G1105-1105V-1105W genotypes were more abundant in clade 2 in 2013 but not in 2014, and 1105V-1105W genotypes were more abundant in clade 2 in 2014 but not in 2013 (Table 4). Evaluation of QoI resistance mutations revealed that most isolates with G143 sensitive genotypes were in clade 1 (14.4%) compared with clade 2 (3.7%) (Table 3). Interestingly, the heterozygous genotype G143-143A and 143A resistant genotype were almost equally distributed among both clades. We also observed in 2013 that clade 2 isolates had a greater abundance of the 143A resistant genotype and less of the G143 sensitive genotype; however, in 2014, similar genotype occurrence of G143, G143-143A and 143A were observed for both clade 1 and clade 2 isolates (Table 4).

Clade 1 isolates with CAA and QoI sensitivity genotypes decrease in occurrence from summer to fall

Regarding sampling time, clade 1 isolates with CAA G1105 resistant genotypes decreased from 52.3% in the summer to 41.9% in the fall (Table 4). Meanwhile, the occurrence of the G1105-1105V genotype increased when comparing summer (2.3 and 0%) to fall (31.1 and 6.0%) in clade 1 and clade 2, respectively, although clade 2 occurrence remains minimal. Additionally, the G1105-1105V-1105W genotype was more prevalent in clade 1 at both time points (20.5 and 14.9%), whereas the 1105W sensitive genotype was present in similar distributions in clade 2 at both sampling times (62.1 and 68.0%) (Table 4). Although the 143A and G143-143A genotypes were widespread among both clades, the most isolates with G143 sensitive genotypes were observed in clade 1 in the summer (27.3%), with a decrease in G143 alleles into the fall (6.8% (Table 4). Across sampling time and between years, the majority of isolates with the G143 sensitive genotype were detected in clade 1 in the summer of 2013 (37.9%), but otherwise were not prevalent (Table 4).

CAA and QoI isolates with sensitivity genotypes decrease in clade 1 when moving from west to east NC

When comparing the occurrence of genotypes in clade 1 across west, central, and east NC, sensitivity genotypes (G1105/G143) isolate occurrence decreases from west to east NC regarding CAAs (53.1, 49.1, and 31.0% respectively) as well as QoIs (46.9, 3.5, and 0%, respectively) (Table 5). Interestingly, the occurrence of sensitive genotypes in clade 2 increased from west to east NC (7.4, 9.3, and 23.7%), although the 1105W resistant genotype remained predominant at each location (>60%) (Table 5). The G1105-1105V, G1105-1105W, and G1105-1105V-1105W genotypes are mostly in clade 1 in eastern NC (27.6, 13.8, and 20.7%) while clade 2 has less or no occurrence of these genotypes (Table 5). More G1105 sensitive genotypes were observed in both clades in 2013 in relation to CAAs at all locations compared with 2014. Additionally, the G1105-1105V genotype became more prevalent in clade 1 isolates at all locations in 2014, while the 1105W resistant genotype remained in abundance in clade 2 at all locations and both years (Table 5). QoI-resistant genotypes (143A) were once again widespread at all locations, but sensitivity was greatest in the west in clade 1 and clade 2 (46.9 and 11.1%) and was detected in clade 1 isolates in the east (Table 5). In comparison, clade 2 isolates from central NC were mostly 143A and G143-143A genotypes, with 2.6% of clade 2 isolates in the east being G413 genotypes (Table 5).

TABLE 5. Occurrence of carboxylic acid amide (CAA) and quinone outside inhibitor (QoI) genotypes expressed as percentages within Pseudoperonospora cubensis isolates infecting commercial and wild cucurbits in North Carolina separated by clade and location^a

	Clade 1 (N = 118)						Clade 2 (N = 108)
	West		Central		East		West		Central		East
Genotype	No.	%	No.	%	No.	%	No.	%	No.	%	No.	%
CAA
G1105	17	53.1	28	49.1	9	31.0	2	7.4	4	9.3	9	23.7
1105W	2	6.3	3	5.3	1	3.4	18	66.7	26	60.5	26	68.4
1105V	1	3.1	2	3.5	0	0.0	0	0.0	0	0.0	0	0.0
G1105-1105V	5	15.6	11	19.3	8	27.6	0	0.0	3	7.0	0	0.0
G1105-1105W	1	3.1	4	7.0	4	13.8	3	11.1	1	2.3	1	2.6
1105V-1105W	1	3.1	0	0.0	1	3.4	1	3.7	3	7.0	0	0.0
G1105-1105V-1105W	5	15.6	9	15.8	6	20.7	3	11.1	6	14.0	2	5.3
Total	32		57		29		27		43		38
QoI
G143	15	46.9	2	3.5	0	0	3	11.1	0	0	1	2.6
G143-143A	10	31.3	14	24.6	7	24.1	6	22.2	12	27.9	7	18.4
143A	7	21.9	41	71.9	22	75.9	18	66.7	31	72.1	30	78.9
Total	32		57		29		27		43		38

^a CAA genotypes include G1105 (Sensitive, S), 1105W (Resistant, Rw), 1105V (Resistant, Rv), G1105-1105V (S-Rv), G1105-1105W (S-Rw), G1105-1105W (Rv-Rw), and G1105-1105V-1105W (S-Rv-Rw). QoI genotypes include G143 (Sensitive, S), G143-143A (Resistant-S, R-S), and 143A (R).

View as image HTML

Differences in genetic diversity and linkage disequilibrium estimates were detected in relation to fungicide resistance genotypes

Genetic diversity estimates were calculated based on groupings related to combinations of clade and fungicide resistance genotypes (N = 108, Table 6). We found that clade 2 isolates with CAA resistance genotypes (Clade2-Rw) had the greatest value of isolates, MLGs, Shannon-Wiener (H), Stoddart and Taylor's (G), and Simpson's (Lambda) indices for diversity (Table 6). Interestingly, although Clade2-Rw isolates had the greatest value for MLGs followed by clade 1 sensitive (Clade1-S) isolates, all groups examined had the same value for expected MLGs (eMLGs). Clade1-S isolates had the greatest gene diversity according to Nei's unbiased gene diversity estimate (Hexp), followed by Clade2-Rw, and then Clade2-S isolates. When evaluating the index of association (I_A) and the standardized I_A estimates within CAA groupings, results from both the original and clone-corrected datasets were significant indicating an occurrence of nonrandom mating with no evidence of recombination.

TABLE 6. Genetic diversity and index of association estimates for Pseudoperonospora cubensis isolates from the original data set (N = 226) grouped by clade and carboxylic acid amide (CAA) as well as quinone outside inhibitor (QoI) genotypes with greater than seven isolates per group (N = 108)^a

		All individuals									Clone-corrected
Clade × fungicide	Group	Nb	MLGc	eMLGd	He	Gf	Lambdag	Hexph	IAi	rbarDj	IAi	rbarDj
Clade × CAA	Clade1-S	35	35	15	3.56	35	0.971	0.431	0.6*	0.07*	0.721*	0.085*
	Clade2-S	15	15	15	2.71	15	0.933	0.371	1.344*	0.183*	1.344*	0.183*
	Clade2-Rw	44	44	15	3.78	44	0.977	0.386	0.709*	0.08*	0.798*	0.09*
	Total	108	92	14.952	4.514	90.163	0.989	0.466	1.116	0.129	1.332	0.154
Clade × QoI	Clade1-S	13	11	11	2.31	8.89	0.888	0.376	0.161*	0.023	0.04	0.006
	Clade1-R	32	26	11.96	3.19	22.26	0.955	0.419	0.881*	0.114*	0.832*	0.107*
	Clade2-R	63	58	12.77	4.02	52.92	0.981	0.377	0.904*	0.102*	0.803*	0.091*
	Total	108	92	12.68	4.439	73.823	0.986	0.458	1.332	0.154	1.166	0.134

^a Groups analyzed within CAA include clade 1 sensitive isolates (Clade1_S) as well as clade 2 sensitive and resistant isolates (Clade2_S and Clade2_Rw, respectively). While groups analyzed within QoI include clade 1 sensitive and resistant isolates (Clade1_S and Clade1_R, respectively) as well as clade 2 resistant (Clade2_R).

^b N: number of individuals.

^c MLG: number of multilocus genotypes.

^d eMLG: number of expected MLG at the smallest sample size based on rarefaction.

^e H: Shannon-Wiener index of MLG diversity (Shannon 2001).

^f G: Stoddart and Taylor's index of MLG diversity (Stoddart and Taylor 1988).

^g Lamba: Simpson's index (Simpson 1949), corrected for population size.

^h Hexp: Nei's unbiased gene diversity (Nei 1978).

ⁱ I_A: index of association for all individuals and for clone-corrected data (Brown et al. 1980). An asterisk indicates statistical significance (significant with P ≤ 0.001, P = 0.01) compared with 999 resamplings (Agapow and Burt 2001).

^j rbarD: standardized index of association for all individuals and for clone-corrected data (Brown et al. 1980). An asterisk indicates statistical significance (significant with P ≤ 0.001, P = 0.01) compared with 999 resamplings (Agapow and Burt 2001).

View as image HTML

When evaluating genetic diversity estimates for clade by QoI genotype groups, we found that clade 2 resistant (Clade2-R) isolates had the greatest number of isolates, MLG, and eMLG values as well as the greatest estimates for the Shannon-Wiener (H), Stoddart and Taylor's (G), and Simpson's (Lambda) indices. Clade 1 resistant (Clade1-R) isolates had the second greatest values across all categories and the greatest value for Nei's unbiased gene diversity (Hexp). Values for MLG, eMLG, the Shannon-Wiener (H), Stoddart and Taylor's (G), and Simpson's (Lambda) indices were observed to be the lowest within Clade 1 sensitive (Clade1-S) isolates. Also, Clade1-S isolates random mating and evidence of recombination was observed based on I_A and rbarD values in the original and the clone-corrected dataset (Table 6).

Genetic differences were observed in the distribution of fungicide resistance genotypes

Genetic differentiation among populations was examined using the pre-defined populations by clade and fungicide resistance genotypes mentioned above using DAPC and pairwise comparisons (F_ST). When evaluating pre-defined CAA populations using DAPC, we found that Clade1-S and Clade2-Rw isolates clustered separately, while Clade2-S isolates clustered among Clade1-S and Clade2-Rw isolates and had overlap of isolates in both clusters (Fig. 1). When a pairwise F_ST analysis was performed, these results were corroborated as the greatest genetic differentiation occurred between Clade1-S and Clade2-Rw isolates, while low genetic differentiation was observed between both Clade2-S and Clade2-Rw isolates as well as between Clade2-S and Clade1-S isolates (Fig. 1). Evaluation of predefined QoI populations using DAPC analysis revealed a separation of the Clade2-R cluster from Clade1-R and Clade1-S cluster, although some Clade2-R isolates extended into these other groupings (Fig. 2). Additionally, an overlap in clusters was observed between Clade1-S and Clade1-R isolates (Fig. 2). Results of pairwise F_ST analysis showed medium differentiation between Clade1-R and Clade2-R and between Clade1-S and Clade2-R, while low differentiation was observed between Clade1-S and Clade1-R (Fig. 2).

Download as PowerPoint

Download as PowerPoint

Discussion

When evaluating isolates from commercial cucurbit hosts, most isolates containing the CAA resistant genotype 1105W were present in clade 2 hosts (Cucumis spp.; cucumber and cantaloupe), while clade 1 hosts (Cucurbita/Citrullus spp.; pumpkin, acorn squash, butternut squash, and watermelon) had mostly isolates with sensitivity G1105 genotypes to CAAs, the 1105V genotype, or the heterozygous genotype G1105-1105V. To our knowledge, this is the first report of the distribution and frequency of P. cubensis isolates carrying the 1105W and 1105V alleles related to CAA resistance in field isolates collected from multiple hosts, regions, and across years. The presence of heterozygous genotypes is not unexpected as P. cubensis is a diploid organism and thus has two alleles at every one locus. This enables a wider range of responses to selective pressures such as fungicides and increases the ability of adaptation (Lebeda and Cohen 2011). To characterize phenotypic expression of the observed heterozygous genotypes, bioassays to determine baseline sensitivities would need to be performed (Wang et al. 2009; Wong and Wilcox 2000). CAA resistance is based on a nuclear recessive gene, suggesting resistant phenotypes are only expressed in homozygous recessive isolates while heterozygous isolates would express a sensitive phenotype (Blum et al. 2011; Gisi et al. 2007; Thind 2012).

The nature of triploid genotypes related to CAA remains unknown, but one possible explanation for this phenomenon is heterokaryosis, or the state of having multiple genetically distinct nuclei in a single cell (Fletcher et al. 2019). While heterokaryosis is a trait well-known in some true fungi, this occurrence is less understood in oomycete organisms although genomic signatures of heterokaryosis have been detected in the lettuce downy mildew pathogen Bremia lactucae (Fletcher et al. 2019) and states of transient heterokaryosis have been detected in some Phytophthora spp. (Catal et al. 2010; Long and Keen 1977). Oomycetes such as P. cubensis are thought to be less conducive for the propagation of stable heterokaryons because the production of zoospores breaks up heterokaryons every asexual generation (Long and Keen 1977), but reports of doubly resistant isolates to metalaxyl and hygromycin derived from single nucleate zoospores have been made in Phytophthora infestans (Judelson and Yang 1998) and may give insight into the processes occurring within P. cubensis. Aneuploidy, or the loss or gain of chromosomes in a nucleus to create substantial variation in gene expression and develop novel phenotypic variation in clonal populations could also be considered as it has been described in the oomycete Phytophthora ramorum (Kasuga et al. 2016); however, studies must be conducted in the future to understand these occurrences in P. cubensis specifically. Similarly, the reason for a mixed clade detection in 10 isolates removed from analyses is unknown. Isolates that showed a mixed clade result did not have mixed genotypes for the other markers we evaluated in this study, providing some level of confidence that the mix clade result was not due to a mixed infection in the single lesion used for analyses. It is unclear if the mixed clade result for 10 isolates, which has been seen in other studies (Wallace et al. 2020), is due to a genetic event, such as hybridization, since that is outside of the resolution provided by markers used in this study.

Regarding the occurrence of SNPs related to QoI resistance, greater than 40% of isolates of all commercial hosts, irrespective of clade, were resistant to QoIs (143A). This result was unsurprising as QoI fungicides have been regularly used for control of agricultural plant pathogens since 1996 and resistance in the field has been documented in a variety of different pathosystems, including P. cubensis (Gisi and Sierotzki 2008; Gisi et al. 2002; Ishii et al. 2001). In general, the G143A substitution is often associated with high or complete resistance while the presence of other substitutions in the ctyb gene such as G137R and F129L exhibit moderate resistance (Standish et al. 2019; Vielba-Fernández et al. 2018). Interestingly, we observed the heterozygous genotype, G143-143A, in all commercial hosts except watermelon. Due to the multi-copy, mitochondrial nature of the cytb gene, one possible explanation for the detection of these heterozygous genotypes is heteroplasmy, or mitochondrion carrying QoI-sensitive (G143) and QoI-resistant (143A) alleles simultaneously within an individual cell (Vielba-Fernández et al. 2018). Heteroplasmy has been described in multiple fungi in relation to the G143-143A genotype (Ishii et al. 2001, 2009; Villani and Cox 2014,). In the downy mildews in particular, data is lacking due to the laborious nature of bioassays and difficulties of culturing; however, evidence of heteroplasmy has been observed in the oomycete grape downy mildew pathogen, Plasmopara viticola, using an allele-specific PCR assay and sequencing methods (Fontaine et al. 2019). While it can be assumed that any isolate carrying a 143A allele in our study would be phenotypically resistant based on dominant inheritance patterns of QoIs, further research would need to be conducted to characterize potential levels of resistance as some studies have suggested that only when 143A frequency exceeds 70% is the highly resistant phenotype expressed (Vielba-Fernández et al. 2018).

When evaluating the occurrence of fungicide resistance and sensitivity genotypes related to the location of hosts within the state, we found that clade 1 isolates with CAA and QoI sensitivity genotypes decreased in occurrence when moving from western to eastern NC and from the summer to fall sampling period. Traditional management practices within each host system and local cucurbit production may impact the distribution of fungicide resistance mutations across a region due to stratification of the population based on the host-adapted clade (Ojiambo et al. 2015; Wallace et al. 2020). While clade 1 hosts are much less devastated by CDM each season, clade 2 hosts, such as cucumber, are extensively scouted and managed with fungicides and as a result may experience higher selection pressures resulting in population shifts towards fungicide resistance (D'Arcangelo et al. 2021; Hollomon 2015), which can rapidly disperse long distances aerially due to profuse asexual reproduction and large population sizes (Savory et al. 2011). Clade 2 isolates in this study were mostly composed of resistant genotypes to both CAAs, with the 1105W genotype, and QoIs, with the 143A genotype, across the entire state of NC and throughout sampling periods. Comparatively, clade 1 isolates had an increase in QoI, but not CAA, resistance alleles when moving east, which has more prominent cucumber (clade 2 isolate) production. In NC in particular, it has been well documented within research plots that clade 1 P. cubensis sporangia are more common in the fall, but not as commonly detected in summer, even though reports were made to the CDM ipmPIPE on clade 1 hosts by early July in the state, whereas clade 2 sporangia were detected in both seasons (Rahman et al. 2021; Wallace et al. 2020). While host availability could be one of the major reasons why clade 2 is seen earlier in the growing season in NC, it is unknown if this is the case in other southeastern states. CDM management is dependent on chemical control annually, and when P. cubensis is reported to be present in an adjacent state, regardless of the host it has been reported on, prophylactic sprays begin. As P. cubensis inoculum disperses north up the east coast each season in a polycyclic nature, every fungicide treatment applied in each state influences the genetic diversity, population structure, and occurrence of fungicide resistance alleles as the epidemic progresses (Wyenandt et. al 2017). More specifically, these fungicide exposure events play a crucial role in impacting the amount of selective pressure exerted on pathogen populations over time, ultimately shifting them towards resistance (Hollomon 2015). A better understanding of the state-by-state impacts of differing cucurbit production and management practices in population structure as the epidemic progresses may allow for designing fungicide programs that are crop and clade specific and do not favor establishment of fungicide resistance.

When evaluating wild hosts, we determined that isolates from all wild cucurbit hosts were capable of harboring fungicide resistance mutations related to CAAs and QoIs except for M. balsamina, which showed 100% sensitivity genotypes for QoIs (N = 5). Cucurbita foetidissima, unlike the other wild hosts examined, is found in many regions of the United States as an unmanaged weed and unlike commercial cucurbits, it is a perennial. Interestingly, of the 10 Cucurbita foetidissima isolates examined, six were clade 2 and four were clade 1 with at least one of all combinations of CAA-resistance alleles detected (1105W, G1105-1105V, G1105-1105W, 1105V-1105W, and G1105-1105V-1105W) except 1105V. This is of importance because P. humuli, a very closely related organism to P. cubensis, can infect the perennial, commercial host, Humulus lupulus, allowing it to overwinter systemically (Purayannur et al. 2020, 2021). If Cucurbita foetidissima is surviving year-round, it may pose as an overwintering reservoir for P. cubensis and could be a continuous source of inoculum harboring isolates with fungicide resistance mutations; possibly resulting in earlier downy mildew outbreaks in temperate regions. According to spatiotemporal analysis, there has been about a 5% discrepancy of CDM outbreak patterns when compared with expected wind patterns (Ojiambo et al. 2015); which could in part be due to wild cucurbit reservoirs, pointing to the potential risk of Cucurbita foetidissima. However, outside of greenhouse production (Naegele et al. 2016), there is no evidence that P. cubensis can survive winter above 30° latitude in Florida as the pathogen needs a living host to survive (Cohen et al. 2015; Lebeda and Cohen 2011). Thus, inoculum contribution of wild cucurbits, overwintering potential, and subsequent occurrence of fungicide resistance mutations at the onset of the yearly CDM epidemic in the United States must be explored further. Additionally, if trends observed in NC are confirmed for the rest of the United States cucurbit production, the forecasting system used for CDM monitoring annually (CDM ipmPIPE) could be modified to factor in differences in host preference and fungicide sensitivity between the clades and potential overwintering sources (Ojiambo et al. 2011).

Like Wallace et al. (2020), the greatest genetic diversity estimates based on MLGs were in clade 2 isolates, but interestingly in this study, it was related to genotypically resistant isolates (CAA: Clade2-Rw, and QoI: Clade2-R), but not clade 2 sensitive isolates (CAA/QoI: Clade2-S). Further, when comparing predefined populations related to fungicide resistance, the greatest gene diversity was observed in clade 1 isolates sensitive to CAAs, while evidence of recombination and random mating was observed only in the analysis of clade 1 isolates sensitive to QoIs when evaluating the index of association (I_A). The greatest gene diversity and evidence of recombination in clade 1 isolates reiterates the results found by Wallace et al. (2020) and could be grounded in the idea that a reproductive barrier, such as host preference, might exist between the two clades, and other environmental factors, such as fungicide exposure and sensitivity, are influential to their reproductive process.

Genetic differentiation in the form of two host-adapted clades (Wallace et al. 2020) and changes in fungicide efficacy (D'Arcangelo et al. 2021; Goldenhar and Hausbeck 2019; Keinath et al. 2019) are well documented in P. cubensis. Because clade is a confounding factor in P. cubensis population analyses, we separated and analyzed categories by clade to control for this. We saw instances where clade 1 and clade 2 isolates displayed low differentiation based on fungicide resistance genotype. For example, when assessing differentiation between CAA genotypes, DAPC analysis revealed cluster overlap between Clade2-S isolates with both Clade1-S and Clade2-Rw isolates and F_ST analysis confirmed low differentiation between these pre-defined populations regardless of the high differentiation when based solely on clade. Meanwhile, Clade1-S and Clade2-Rw isolates had no cluster overlap within DAPC analysis and exhibited high differentiation from one another. As each P. cubensis clade can infect multiple cucurbit hosts, host may not be the only factor structuring the population, and therefore, these results could give insight into how fungicide resistance or sensitivity genotypes are playing a role in aspects of reproduction, recombination, and population stratification. In other oomycetes such as Phytophthora spp., oospores are capable of forming when isolates are exposed to hormones or molecules of other mating types or even different species (Shattock et al. 1986; Tomura et al. 2017), and thus, more research must be done to determine if true mating and recombination is truly occurring in P. cubensis, if hybridization events are possible, or if oospores are simply forming as a response to hormone exposure without actual recombination or hybridization (Corredor-Moreno and Saunders 2020; Rouxel et al. 2013). Controlled oospore crosses between clade 1 and clade 2 isolates and self-crosses in combination with genotypic information could determine if progeny are truly recombinant and not simply parental genotypes, but must be explored to better understand how host-adaptation and fungicide sensitivity structure populations.

Cucurbit growers rely on data from annual cucumber fungicide efficacy field trials (clade 2 host) to determine what will be the most effective approach for control of CDM (D'Arcangelo et al. 2021; Holmes et al. 2015). Findings of our study reveal that there are prevalent differences in the occurrences of fungicide sensitivity and resistance genotypes between the host-adapted clades of P. cubensis that must be considered. While this is the first characterization of these genotypes in natural P. cubensis populations to our knowledge, variation in fungicide efficacy between hosts has been observed in the field with some active ingredients, such as propamocarb, resulting in more effective control in winter squash (clade 1 host) when compared with cucumber (clade 2 host) in the same year (Adams et al. 2018a, b). We observed a majority of clade 2 isolates with the resistant 1105W genotype for CAAs, but clade 1 isolates had a greater prevalence of the sensitivity genotype, G1105. Variation in fungicide efficacy has widely been reported in the P. cubensis system throughout the years (D'Arcangelo et al. 2021; Goldenhar and Hausbeck 2019; Keinath et al. 2019). When clade is not taken into account when selecting a management program, products that are no longer effective on one of the two clades may be repeatedly applied or products with more than one active ingredient where only one is effective for a particular clade may in reality be functioning as a single-active ingredient product, which would result in the build-up of fungicide resistant isolates and the loss of fungicide efficacy (Deising et al. 2008; Keinath 2016). By modifying the current system used to make product recommendations to include P. cubensis clade, or host, applications of infective products would be reduced or eliminated; in turn decreasing the development of fungicide resistance in populations over time; something especially of concern as options of effective products for use in cucumber are already reduced compared with clade 1 hosts. It is vital that fungicide efficacy, resistance development, and clade−host relationships be evaluated on a case-by-case basis in different cucurbit growing regions due to documented variability among pathogen populations (Quesada-Ocampo et al. 2012; van den Bosch et al. 2014; Wallace et al. 2020) as it could eventually lead to reliable, but varying repertoires of effective products at the clade-level to use for management.

Acknowledgments

We thank all the members of the Quesada lab for their valuable help and Aidan Shands for his technical assistance in validating the fungicide resistance assays.