Evidence of Resistance to QoI Fungicides in Contemporary Populations of Mycosphaerella fijiensis, M. musicola and M. thailandica from Banana Plantations in Southeastern Brazil

Abstract

Yellow and black Sigatoka, caused by Mycosphaerella fijiensis and M. musicola, respectively, are the most important worldwide foliar diseases of bananas. Disease control is heavily dependent on intensive fungicide sprays, which increase selection pressure for fungicide resistance in pathogen populations. The primary objective of this study was to assess the level and spread of resistance to quinone-outside inhibitors (QoI—strobilurin) fungicides in populations of both pathogens sampled from banana fields under different fungicide spray regimes in Southeastern Brazil. Secondly, we aimed to investigate when QoI resistance was confirmed if this was associated with the target-site alteration G143A caused by a mutation in the mitochondrial encoded cytochrome b gene. QoI resistance was detected in fungicide treated banana fields, while no resistance was detected in the organic banana field. A total of 18.5% of the isolates sampled from the pathogens’ populations were resistant to QoI. The newly described M. thailandica was also found. It was the second most abundant Mycosphaerella species associated with Sigatoka-like leaf spot symptoms in the Ribeira Valley and the highest level of QoI resistance was found for this pathogen. The G143A cytochrome b alteration was associated with the resistance to the QoI fungicides azoxystrobin and trifloxystrobin in M. fijiensis, M. musicola and M. thailandica strains. In order to reduce resistance development and maintain the efficacy of QoI fungicides, anti-resistance management strategies based on integrated disease management practices should be implemented to control the Sigatoka disease complex.

1. Introduction

Commercial banana production has been substantially reduced worldwide, for more than two decades, due to the incidence of leaf spots caused by different members of the Sigatoka disease complex. While black Sigatoka is caused by the ascomycetous fungus Mycosphaerella fijiensis M. Morelet (Mf) (syn. Pseudocercospora fijiensis (M. Morelet) Deighton), yellow Sigatoka is associated with M. musicola Leach. (Mm) (syn. Pseudocercospora musae (Zimm.) Deighton) [1,2,3,4,5,6,7]. Mycosphaerella musae (Speg.) Syd. & P. Syd. [1] can also cause leaf spot, as well as the newly described species M. thailandica Crous, Himaman & M. J. Wingf. (Mt) (syn. Parapallidocercospora thailandica (Crous, Himaman & M.J. Wingfield) Videira & Crous) [1,8], which is highly prevalent in banana plantations from Ribeira Valley, São Paulo, Brazil [2]. Mycosphaerella eumusae Crous & X. Mourichon (Me) (syn. Pseudocercospora eumusae Crous & Mour), another member of the Sigatoka disease complex, has a considerably lower incidence in Brazil [6].

Sigatoka diseases are polycyclic, in which high numbers of asexual spores (conidia) are produced and released in multiple cycles, resulting in severe epidemics under favorable (wet) weather conditions. The disease cycle also includes sexual spores (ascospores) that are usually produced once a year through sexual reproduction. Therefore, the pathogens’ populations are expected to have a mixed reproductive system with occurrence of cyclic sexual reproduction and predominance of epidemic clonal dispersal [9,10]. Recent studies from Brazil, Mexico and the Philippines have revealed that Mf and Mm populations have high genotypic diversity from both sexual reproduction and genotype flow from long distance migration input, resulting in high evolutionary potential for both pathogens [4,11,12,13]. Since varietal resistance is practically absent or only partially effective in most widely grown banana cultivars, the intensive and mostly preventive spray application of site-specific systemic fungicides has been the main strategy for the management of the Sigatoka diseases complex, resulting in strong selection pressure for fungicide resistance in the pathogens’ populations [4,14,15].

In the late 1990’s, quinone outside inhibitors (QoI), also known as strobilurins, a class of systemic fungicides with a new single-site mode of action were introduced to the market [16]. Due to superior efficacy, at that time [17], and broad spectrum activity, they were soon also used for managing black and yellow Sigatoka on bananas in Brazil [18,19,20]. The QoIs azoxystrobin, pyraclostrobin, trifloxystrobin and kresoxim-methyl were introduced and formulated as straights or in mixtures with azoles, dehydrogenase inhibitors (SDHIs) and/or multi-site inhibitors (mancozeb and chlorothalonil), totaling more than 20 commercial products [16].

The synthetic molecule design of QoIs was based on natural fungicidal metabolite derivatives of β-methoxy acrylic acid, such as strobilurin A, produced by the basidiomycete wood-rotting fungi Strobilurus tenacellus [21,22]. The binding site of QoIs is the cytochrome b (cyt b), a subunit of the cyt bc₁ (complex located inside the mitochondrial membrane. The mode of action is the inhibition of mitochondrial respiration by specifically binding to the oxidation site of quinol (Qo, or ubiquinol) in cyt b, blocking the transfer of electrons between cyt b and cyt c₁. Therefore, these fungicides prevent the NADH oxidation and the ATP synthesis leading to inhibition of the energy production essential to spore germination and survival [21,23].

Emergence and spread of fungicide resistance rapidly followed the introduction of the QoI fungicides for managing crop diseases worldwide, spanning many distinct pathosystems [24,25]. Because QoI fungicides are classified as a high risk for resistance development [25], their intensive use has resulted in several reports of resistance in many pathosystems, including the Sigatoka complex pathogens [24,26]. For instance, in the early 2000’s, after only three years since the QoI azoxystrobin was labeled in Costa Rica, the intensive spraying of this fungicide on banana fields resulted in the early detection of resistance in Mf populations [27,28]. QoI-resistant populations of Sigatoka pathogens have been reported in Central and South America, Africa and Australia [15,25,26,29,30], while in Brazil there has been only a single report, from a dissertation, of a few QoI resistant strains [2].

In terms of mechanisms of QoI resistance, of the several cyt b mutations that can confer QoI insensitivity, the mutation resulting in the substitution of glycine by alanine at codon 143 (G143A) of cyt b has been associated with poor field performance of QoIs in a range of pathogens [25,31]. This target site alteration was also reported in resistant strains of Mm from Australia [15], Mf from Costa Rica [25,27], and for some Mf strains from Brazil [2].

There is a lack of information about the sensitivity status, and frequency distribution of resistance in Brazilian populations of the Sigatoka diseases pathogens to QoI fungicides. In the face of a high selection pressure due to the historical intensive use of QoI fungicides in local banana plantations we hypothesized that resistance to QoI is pervasive in fields with conventional and/or intensive fungicide spraying systems. This lack of information about the sensitivity status of resistance to QoI fungicides is risky as the continuous spraying of QoIs would only increase the frequency of resistant strains in populations of both Mf and Mm, and their spread across distinct fields via gene flow. As a result, it would eventually contribute to the declining efficacy of the QoI fungicides.

Therefore, the objective of this study was to assess the development and spread of resistance to QoI fungicides in populations of Mf and Mm from bananas in Southeastern Brazil, both phenotypically and genotypically. For hypothesis testing, fields from three contrasting disease management systems were sampled from Vale do Ribeira and Northwestern São Paulo, and from Northern Minas Gerais. If resistance was detected, we also investigated if this was associated with the G143A cyt b target alteration.

2. Materials and Methods

2.1. Yellow and Black Sigatoka Pathogens’ Population Sampling

For obtaining Mm and Mf populations for this study, samples of infected leaves were collected in 2020 from four locations with three contrasting disease management systems: (i) Intensive, in Vale do Ribeira, São Paulo (SPVR-I), from Jacupiranga, Registro and Sete Barras counties (Figure 1 and Figure 2A), and banana varieties Prata (AAB triploid [18], also known as ‘lady finger banana’) and Nanica (AAA, Cavendish subgroup [23]); (ii) Conventional, in Northeastern São Paulo (SPNW-C), from Ilha Solteira county, var. Maçã (AAB [18]), and in Northern Minas Gerais (MGN-C) from Janaúba county, var. Prata and Nanica; (iii) Organic (without fungicide application), in Northeastern São Paulo (SPNW-O), from Ilha Solteira county, with different cultivars and plant ages.

In SPVR-I, the management was based on intensive fungicide sprays, with around 8 to 14 fungicide calendar-based applications per year to control black Sigatoka, mostly because of the extremely predisponent weather conditions for the disease, which includes high humidity (>90%), the presence of leaf wetness and temperatures ranging from 26–28 °C [2,19,32,33,34]. The Vale do Ribeira region represents the highest banana production both in São Paulo State and Brazil, where the susceptible banana varieties Prata and Nanica predominate [7,19,33]. In plantations from these susceptible banana varieties under the predominant highly favorable weather conditions in the Vale do Ribeira, the pathogens’ incubation period could range from as low as 13 to 14 days, whereas under unfavorable weather the incubation period could be extended up to 35 days [7,19,33,34]. Intensive fungicides sprays are usually scheduled to coincide with the continuous emergence of new banana leaves, which are highly susceptible to infection [7]. The SPNW-C and MG-C banana plantations were under reduced fungicide application management, with 4–5 sprays against yellow Sigatoka.

Infected banana leaf fragments with 20 cm² of area presenting typical yellow (Figure 2B) or black Sigatoka (Figure 2C) symptoms were sampled from five plants within a 50 m² radius, from 20 to 40 points randomly distributed in a field. The samples were packaged in paper bags, transported to the lab and kept in a refrigerator until the isolation of the pathogens.

2.2. Mycosphaerella Isolation

Leaf samples were surface cleaned with tap water, cut into 2 cm² pieces, and treated with 75% ethanol and sterile distilled water, for one minute each (Figure 2D). The leaf pieces were then dried in sterile filter paper, transferred to Petri dishes with water–agar (15 g L⁻¹ agar) medium amended with the antibiotics chloramphenicol and streptomycin (50 μgmL⁻¹ each) (Figure 2E), and incubated at 26 °C in the dark until Mf or Mm sporodochia formed and were detected under a stereomicroscope. Using a thin needle, conidia were transferred from the lesions to Petri dishes containing PDA [20.8 g L⁻¹ potato-dextrose (Kasvi, India), 15 g L⁻¹ agar] amended with antibiotics (as above), streak-plated, and kept at 26 °C in the dark until characteristic Mycosphaerella micro-colonies were detected, and typical fungal conidia were observed (Figure 2F–H). Young colonies were transferred to new PDA plates [adapted from 5 and 8]. Long term storage of isolates was performed by initially transferring 0.5 cm² pieces of sterilized filter paper to the top of actively growing 14-day-old Mycosphaerella colonies on PDA. Subsequently, the pieces of filter paper colonized by the fungus mycelium were transferred to cryotubes with silica gel for cryo-preservation at −20 °C.

2.3. Molecular Identification of the Pathogens

To identify the isolates at the species level, a total of 119 fungal isolates (37 isolates from SPNW-O (ISR strains), 34 isolates from SPNW-C (ISC strains), 26 isolates from MGN-C (MG strains), and 22 isolates from SPVR-I (VR strains)), were firstly analyzed if harboring typical morphological characteristics of M. fijiensis or M. musicola [1,6]. Secondly, the species were identified based on PCR assay [27,28]. Mycelial fragments with 2 cm² diameter colonies grown in PDA media for 10 days at 26 °C in the dark were harvested for lyophilization. Genomic DNA was extracted from lyophilized mycelia using the kit Wizard^® Genomic DNA Purification (Promega, Madison, WI), according to the manufacturer’s instructions. DNA was quantified in a Nanodrop^® 2000c spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and diluted to 50 ηg·µL⁻¹ final concentration. The PCR-based species identification was performed using specific primers for Mf (ACTR: 5′-GCAATGATCTTGACCTTCAT-3′; MFactF: 5′-CTCATGAAGATCTTGGCTGAG-3′) and Mm (MMactF2: 5′-ACGGCCAGGTCATCACT-3′; MMactRb: 5′-GCGCATGGAAACATGA-3′), designed from actin gene sequences [2,30,35]. PCR amplifications were conducted in 15 µL volumes containing ultrapure distilled water, 50 ηg of template DNA, 0.3 µM of each primer, 0.2 mM of each dNTP, 2.5 mM MgCl₂, 1.5 µL of 10× PCR reaction buffer and 0.05 U Taq Polymerase (Thermo Fisher Scientific, Waltham, MA, USA). Amplifications were performed in a ProFlex PCR thermal cycler (Applied Biosystems, Waltham, MA, USA), with cycling conditions as follows: initial denaturation at 95 °C for 5 min, followed by 36 cycles of 94 °C for 30 s, 60 °C for 30 s, and at 72 °C for 1 min; with a final extension of 72 °C for 7 min. The amplicons were analyzed by agarose gel electrophoresis, where a 500 bp would indicate Mf, while a 200 bp fragment would indicate Mm identity. Positive Mf and Mm DNA controls were included in the assay, and negative DNA controls from the basidiomycetous fungus Rhizoctonia solani Kühn AG-1 IA, and from the ascomycetous Pyricularia oryzae Cavara lineage Triticum. To confirm the isolates’ identity, we also analyzed the specific amplification of the Mf cyt b and Mm cyt b genes (

Table 1. Specific primers for PCR amplifications, and to evaluate the variation at the cyt b gene in Mycosphaerella fijiensis, M. musicola and M. thailandica.

a Target gene	Primer	Sequence (5′-3′)	Amplicon Size (bp)	Annealing Temperature (°C)
b Mf cyt b	cytb_exon1-1_Mf_F2036	CGTCGCCGTAATGTGGTTC	475	56
	cytb_exon1-1_Mf_R2511	GCCGCAACCTTCTAATATTAG
	cytb_exon2_Mf_F25	CGTGCTTCTGATTCTATTAGGGG	996	56
	cytb_exon2_Mf_R1020	GGCGACTACCAACACAAAT
b Mm cyt b	cytb_Mm_F1251	GTTACCTTTGAAACTTCGGATC	963	58.5
	cytb_Mm_R2213	GACTCAACGTGTTTAGCCC
c Mt cyt b	cytb_Mt_F2	GAAGCATTTAATTCAGTAGAAC	400	58
	cytb_Mt_R2	CAACTATATCTTGTCCTACTC

^a Mf = Mycosphaerella fijiensis; Mm = Mycosphaerella musicola; Mt = M. thailandica. ^b primers designed during this study. ^c primers from Malimpensa [2].

).

Thirteen isolates of Mycosphaerella for which the PCR-based species identification failed were identified based on the sequence analyses of the fungal ITS-rDNA region amplified with ITS1 and ITS4 primers [36] with an annealing temperature of 60 °C, using similar PCR conditions as described above. The PCR products were purified and sequenced by the Center for Biological Resources and Genomic Biology (Crebio, UNESP, Jaboticabal, Brazil), using the ABI 3730xl DNA Analyzer (Applied Biosystems, Waltham, MA, USA). DNA sequences were analyzed and aligned using the Geneious R 9.0.5 (Biomatters, Auckland, New Zealand), and compared with the ITS-rDNA sequences from type species of Mycosphaerella available at the GenBank/NCBI databases. Based on ITS-rDNA sequence similarity, these isolates (Mt ISC57 and Mt ISC86, from SPNW-C; Mt MG60 from MGN-C; Mt FRA1.21b, Mt JA1.34, Mt JA1.4, Mt JA2.13, Mt JA2.15, Mt JA2.16a, Mt JA2.2p, Mt JA2.3a, Mt JA2.6 and Mt T06 from SPVR-I) were identified as M. thailandica.

2.4. QoI Qualitative Fungicide Sensitivity Assays at Discriminatory Dose

One hundred and nineteen Mycosphaerella spp. isolates whose molecular species identification has been confirmed from the SPNW-O (N = 37 Mm strains), SPNW-C (N = 32 Mm, 2 Mt), MGN-C (N = 25 Mm, 1 Mt) and SPVR-I (N = 10 Mf, 2 Mm, 10 Mt) populations were used for fungicide sensitivity experiments. The fungicides Priori (QoI-azoxystrobin at 250 g L⁻¹; Syngenta S.A.) and Flint 500 WG (QoI-trifloxystrobin at 500 g Kg⁻¹; Bayer SA) were diluted in deionized water to produce a 500 µg·mL⁻¹ of stock solutions used in all experiments.

To obtain a suspension of mycelial fragments, a protocol established in this study was used, by which 4 cm² of fungal mycelial disks were taken from fungal colonies grown on PDA at 25 °C for 10 days. The fungal mycelial disks were transferred to microtubes containing 0.5 mL of sterile 2 mm glass beads and 1 mL of sterile distilled water. The tubes were sealed and placed in a Fast Prep FP120 Bio 101 (Thermo Savant, Waltham, MA, USA) for 20 s at speed 4 m·s⁻¹ to disrupt the mycelial mass and to obtain a suspension of mycelial fragments [14].

The qualitative assay to evaluate the QoI fungicide sensitivity by Sigatoka pathogens was conducted using Petri dishes containing PDA with trifloxystrobin (TRI) or azoxystrobin (AZX) at the discriminatory dose of 10 μg·mL⁻¹ [29,30], or no fungicide, all amended with salicylhydroxamic acid (SHAM) at 0.5 mM [25,37] and antibiotics as above. The fungal inoculum was prepared using sterile filter paper discs soaked in mycelial fragments suspension, which were then transferred to the assay media. For each isolate and treatment (no fungicide, TRI, and AZX), four plates containing four inoculum discs were used. These experimental plates were sealed, randomly positioned inside the incubator, and kept for 15 days at 26 °C in the dark for colony growth. The entire assay was repeated once.

The two-proportions z-test based on Pearson’s Chi-square at p ≤ 0.05 was used to compare the equality of two observed proportions of QoI resistant isolates from populations of M. fijiensis, M. musicola and M. thailandica from banana plantations under distinct fungicide management systems [38,39].

Analysis of variance (ANOVA) by the F test and means comparison were performed using the R environment with the statistical libraries agricolae and laercio [39]. For means comparison we applied the Scott–Knott test (at p ≤ 0.05) to contrast between groups of QoI resistant (QoI-R) and QoI sensitive (QoI-S) isolates of M. musicola and M. fijiensis per geographical population of the pathogen.

2.5. QoI Quantitative Fungicide Sensitivity Assays Based on EC₅₀ Values

A qualitative assay to determine the EC₅₀ for TRI and AZX was conducted using six fungicide doses: 0.0, 0.1, 1, 10, 100 and 1000 µg·mL⁻¹. It consisted of a spectrophotometric assay for fungicide sensitivity testing based on the reduction of resazurin to resorufin, a metabolic indicator for fungal respiration activity [40,41,42]. Ten isolates, from the three Mycosphaerella species detected in our survey, were chosen for this assay: three QoI sensitive (Mf CALT1, Mm ISC111 and Mt ISC86) and seven QoI resistant (Mf JA2.24, Mm ISC64, Mm JA1.37, Mm JA1.38c, Mt JA2.13, Mt JA2.2 and Mt JA2.6). A modification of the flat-bottomed 96-well microtiter plate method was adopted to measure the fungal growth in each fungicide dose [37,43], using the mycelial fragments protocol [4,44]. Mycelium fragments were obtained from Mycosphaerella colonies that were grown at 25 °C under 12 h photoperiod for 10 days in PDA supplemented with antibiotics as described above. Fungal mycelial samples from a 10-day old colony from a single agar plate were transferred to 1.5 mL microtubes containing 0.5 mL of 0.1 mm diameter glass beads. A total of 1 mL of distilled water was added to the mixture, and vortexed for two minutes in a Fast-Prep apparatus for 20 s at speed 4 m·s⁻¹. The mycelial fragments suspension recovered from the tubes were mixed and pre-diluted to a final volume of 10 mL, counted under a microscope using a Neubauer chamber and adjusted to 10⁴ fragments mL⁻¹. Each microplate well was filled with 50 μL of inoculum suspension and 100 μL of PD broth (20.7 g·L⁻¹ of potato dextrose (Kasvi, India), prepared with 0.025M phosphate buffer, and final pH adjusted to 5.0) amended with different concentrations of TRI and AZX and in the presence of chloramphenicol and streptomycin (50 mg·L⁻¹ each). Salicylhydroxamic acid (SHAM) at 0.5 mM was also added to the PD broth to suppress alternative oxidase (AOX) activity, following the procedures described by Ma et al. [45] and Vicentini et al. [37]. The microplates with fungal liquid cultures were wrapped in plastic and incubated at 25 °C for 10 days under a 12 h photoperiod, when the fungal growth reached its maximum. After incubation, 50 µL of resazurin at 160 µM was added to each microplate well to obtain a final concentration of 40 µM in a final volume of 200 µL. For spectrophotometric estimates of RZ reduction, initial absorbance readings at 569 ηm (Abs_{569 ηm} at T₀) were taken using a microplate reader (Multiskan™, FC Microplate Photometer, Thermo Scientific^TM, Waltham, MA, USA). Subsequently, the microplates were kept at 25 °C under complete darkness for 24 h, when final absorbance readings were then taken at the same wavelengths (Abs_{569 ηm} at T₂₄). The RZ reduction was estimated as follows:

• RR_{Mycosphaerella} = T₀ − T₂₄, where:
• RR_{Mycosphaerella} = relative reduction of resazurin;
• T = Absorbance reading at 569 ηm;
• T₀ = reading at time zero (immediately after adding resazurin to the fungal 10-day-old liquid culture in buffered PD broth);
• T₂₄ = reading at time 24 (24 h after adding resazurin).

Sensitivity to the two QoI fungicides was determined as 50% effective concentration to inhibit fungal growth (EC₅₀, in µg·mL⁻¹), estimated using the R package ec50estimator coupled with the drc library for analyses of dose response curves [46], from multiple isolates data sets. This package also allows for testing the hypothesis on the most fit model function (fct) for estimating the EC₅₀ by a dose–response curve. Confidence intervals were also estimated.

For EC₅₀ means comparison within each fungicide (TRI or AZX), the experimental design consisted of complete randomized blocks, with four replicates per treatment and experiments in duplicate. Analysis of variance (ANOVA) and the Scott–Knott test (at 5% probability) for means comparison were performed in the R environment using the statistical packages agricolae and ScottKnott [39]).

2.6. Analysis of Allelic Variation in Target Gene cyt b in Mycosphaerella Populations

To examine variation in the target gene cyt b encoding the QoI fungicide binding sites, all 96 Mm from populations SPNW-O (N = 37), SPNW-C (N = 32), MGN-C (N = 25), and SPVR-I (N = 2), ten Mf isolates from population SPVR-I (N = 10) and 13 Mt isolates from populations SPNW- SPNW-C (N = 2), MGN-C (N = 1), and SPVR-I (N = 10) were used for PCR amplifications and further sequencing. PCRs were conducted in 30 µL volume containing ultrapure distilled water, 50 ηg of template DNA, 0.3 µM of each primer, 0.2 mM of each dNTP, 2.5 mM MgCl₂, 3 µL of 10× PCR reaction buffer and 0.05 U Taq Polymerase (Thermo Fisher Scientific, Waltham, MA, USA). The cycling conditions were as follows: initial denaturation at 95 °C for 5 min, followed by 36 cycles of 94 °C for 30 s, 1 min at the specific annealing temperature for each primer pair (Table 1) and at 72 °C for 1 min; with a final extension of 72 °C for 7 min. The PCR products were purified and sequenced by the Crebio (UNESP Jaboticabal Campus, Jaboticabal, SP, Brazil), using the ABI 3730xl DNA Analyzer (Applied Biosystems, Waltham, MA, USA). DNA sequences were analyzed and aligned using the Geneious R 9.0.5 (Biomatters, Auckland, New Zealand) program to identify alleles, haplotypes and distinguish non-synonymous mutations leading to amino-acid changes in inferred protein sequences. For annotation and inference of protein sequences from the cyt b sequences, we included the following reference sequences obtained from the NCBI/GenBank^®: NC037198.1 (Pseudocercospora mori) and LFZO01000638.1 (Mm) for Mm cyt b; NC044132, AF343069 and AF343070 (Mf) for Mf cyt b. There were no sequences from M. thailandica cyt b gene available at GenBank for comparison.

3. Results

Our sampling strategy resulted in 96 isolates of Mm from all four populations sampled, 10 isolates of Mf, only detected in the population SPVR-I, and 13 isolates of Mt obtained from populations SPNW-C, MGN-C and SPVR-I (Figure 1,

Table 2. Distribution of QoI fungicide resistance in populations of Mycosphaerella musicola, M. fijiensis and M. thailandica sampled from four banana plantations with different fungicide management systems ^a.

Species, Type of Management with Fungicides and Population	Average Number of Fungicide Sprays per Crop	Number of Isolates with Growth In Culture Medium Supplemented with 0 or 10 μg·mL−1 of the QoI Fungicides Azoxystrobin or Trifloxystrobin
		μg·mL−1		Resistant Isolates (Proportion)
		0	10
Mycosphaerella musicola
Organic field
SPNW-O	0	37	0	0.0000
Conventional fields
SPNW-C	4	32	6	0.1875
MGN-C	5	25	1	0.0400
Intensive field
SPVR-I	>10	2	2	1.0000
M. fijiensis
Intensive field
SPVR-I	>10	10	1	0.1000
M. thailandica
Conventional fields
SPNW-C	4	2	1	0.5000
MGN-C	5	1	1	1.0000
Intensive field
SPVR-I	>10	10	10	1.0000
Total		119	22	0.1849

^a SPNW-O (North-western São Paulo, organic field), SPNW-C (North-western São Paulo, with fungicide conventional management system), MGN-C (Northern Minas Gerais, with fungicide conventional management system), and SPVR-I (São Paulo, Vale do Ribeira, fungicide intensive management system).

). The discriminatory dose of azoxystrobin (Figure 3B,E) and trifloxystrobin (Figure 3C,F) allowed to identify the resistant strains among Mf, Mm and Mt isolates samples. From the total of 119 isolates screened, 22 (≈18%) were resistant to both QoI fungicides: from population SPNW-C, isolates Mm ISC64 (Figure 3A–C), Mm ISC09, Mm ISC55a, Mm ISC110, Mm ISC116, Mm ISC117, and Mt ISC57; from population MGN-C, isolate Mm MG118, and Mt MG60; and from population SPVR-I, isolates Mm JA1.37, Mm JA1.38c, Mf JA2.24, Mt FRA1.21b, Mt JA1.34, Mt JA1.4, Mt JA2.13, Mt JA2.15, Mt JA2.16a, Mt JA2.2p, Mt JA2.3a, Mt JA2.6, and Mt T06).

The sensitivity assays also revealed cross-resistance to both QoIs fungicides trifloxystrobin and azoxystrobin as all strains showed resistance to both compounds (Table 2). The frequency distribution of resistant and sensitive isolates to QoI fungicides indicated no resistant individuals in the population Mm SPNW-O (organic field) (37 strains tested, whereas 7% (two out of 25) were detected in the Mn population MGN-C (conventional fungicide management), 20% (seven out of 34) in the population SPNW-C (conventional fungicide management), and 100% (two out of two) in the Mn population SPVR-I (intensive fungicide management). For the Mf SPVR-I population, one out of ten strains (10%) was resistant to both QoIs tested. In comparison, 50% QoI resistant individuals (one out of two) were detected in the Mt population MGN-C, 100% (one out of one) in the population SPNW-C, whereas 100% (10 out of 10) were detected in the Mt SPVR-I population.

The two-proportions z-test at p ≤ 0.05 used to compare the equality of two observed overall proportions of QoI resistant isolates of Mf, Mm and Mt (

Table 3. The two-proportions z-test based on Pearson’s Chi-square used to compare the equality of two observed proportions of QoI resistant isolates from populations of Mycosphaerella musicola, M. fijiensis, and M. thailandica from banana plantations under distinct fungicide management ^{a, b}.

	Pearson’s Chi-Square Values below the Diagonal and p Values above the Diagonal
Comparisons	SPNW-O	SPNW-C	MGN-C	SPVR-I
SPNW-O	-	0.012	0.325	0.000
SPNW-C	6.29	-	0.307	0.008
MGN-C	0.97	1.04	-	0.000
SPVR-I	24.71	7.03	12.34	-

^a Test statistic with Yates’ continuity correction. ^b SPNW-O (North-western São Paulo, organic field), SPNW-C (North-western São Paulo, with fungicide conventional management), MGN-C (Northern Minas Gerais, with fungicide conventional management), and SPVR-I (São Paulo, Vale do Ribeira, fungicide intensive management). Gray shading emphasizes the significant pairwise comparisons between populations.

) indicated, in fact, that the population SPNW-O (from an organic banana field), with no resistant isolates, was significantly different from populations SPNW-C and SPVR-I (from conventional or intensive management systems), with proportions varying from 20 to 59% of resistant isolates.

The contrast between groups of QoI resistant (QoI-R) and QoI sensitive (QoI-S) isolates of Mycosphaerella species per geographical population of the pathogen, indicated that only the groups of QoI-R isolates were able to grow at the discriminatory dose of 10 μg·mL⁻¹ of azoxystrobin or trifloxystrobin. In addition, on average, the groups of isolates from the populations SPNW-C and SPVR-I were significantly different from the population MGN-C, showing higher relative growth in QoI-fungicide amended medium (Figure 4).

Based on a QoI quantitative fungicide sensitivity assay, Mycosphaerella isolates were grouped into six different sensitivity phenotype classes for TRI or AZX: sensitive (Mf CALT1 and Mt ISC86, EC₅₀ < 1 μg·mL⁻¹), less sensitive (Mm ISC111, EC₅₀ > 1 and <5 μg·mL⁻¹), moderately resistant (Mm JA1.37, EC₅₀ > 5 and <25 μg·mL⁻¹), resistant (Mm ISC64 and Mm JA1.38c, EC₅₀ > 25 and <75 μg·mL⁻¹), highly resistant (Mf JA2.24 and Mt JA2.13, EC₅₀ > 75 and <150 μg·mL⁻¹), and extremely resistant (Mt JA2.2 and Mt JA2.6A, EC₅₀ >> 150 μg·mL⁻¹) (Figure 5).

Analyses of the cytochrome b (cyt b) partial gene sequences from Mm, Mf and Mt strains showed the known non-synonymous G143A substitution associated with azoxystrobin and trifloxystrobin resistance (QoI-R) in both species (

Table 4. Summary of variation at cytochrome b (cyt b) genes from field isolates of Mycosphaerella musicola strains from Brazil with distinct resistance phenotypes to the QoI fungicides azoxystrobin and trifloxystrobin.

Gene	Reference Sequence (GenBank NCBI)	Length of Amplicon Sequence (bp)	a Synonymous Mutations: Substitutions (s); Deletions (d); Insertions (i)/Position a	Haplo- types detected	Species	Isolates	N	b QoI resistance class	Coding region (bp)	Protein length (aa)	Non-Synonymous Mutations Detected /Position	Non- Synonymous a.a. Change/ Position	a.a Change
Mm cyt b	OP715652, OP715653, and OP715654	963 (from 1251 to 2213)	0	Hm1	Mm	cMm populations	87	S	936	312	0	0
	OP715650, OP715651, and OP715649		0	Hm2	Mm	d ISC9, ISC55a, ISC64, ISC110, ISC116, ISC117, MG118 e JA1.37, e JA1.38c	9	R			f CCT:CGT/1599	f G:A/143	Mm: Grice et al. [15]
	NC037198.1	1173 (from 1172 to 2344)	20 s: A:T/1267; T:A/1312; A:T/1324/T:A/1432; C:G/1450; T:A/1549; A:T/1675;T:A/1816; G:C/1837; A:T/1925; A:T/1957; T:A/1972; C:G/1975; G:C/2018; C:G/2092; T;A/2107; T:A/2156; A:T/2167; A:T/2173; C:G/2194	Hm3	P. mori	-	1	-	1170	390	0	0
	LFZO01000638.1		0	Hm1	Mm	CBS116634	1	-	1173	391	0	0

^a The haplotype could still carry other mutations along the non-coding region of the sequence. ^b Fungicide sensitivity phenotype classes: S = sensitive, R = resistant for both azoxystrobin and trifloxystrobin. ^c Mm populations from North-western São Paulo (SPNW) fungicide conventional management system (SPNW-C): ISC isolates; organic plantation (SPNW-O): ISR isolates; and Northern Minas Gerais conventional management system (MGN-C): MG isolates, Mm SPVR-I population from a fungicide conventional intensive management system in Ribeira Valley, São Paulo. ^d From SPNW-C and MGN-C populations. ^e From SPVR-I population. ^f Non-synonymous mutations and respective nucleotides substitutions are indicated in red.

and

Table 5. Summary of variation at cytochrome b (cyt b) genes from field isolates of Mycosphaerella fijiensis strains from Brazil with distinct resistance phenotypes to the QoI fungicides azoxystrobin and trifloxystrobin.

Gene	Reference Sequence (GenBank NCBI)	Length of Amplicon Sequence (bp)	a Synonymous Mutations: Substitutions (s); Deletions (d); Insertions (i)/Position a	Haplo- Types Detected	Species	Isolates	N	b QoI Resistance Class	Coding Region (bp)	Protein Length (aa)	Non-Synonymous Mutations Detected/Position	Non- Synonymous a.a. Change/ Position	Previous Report for a.a. Change
Mf cyt b exon-1	OP734340 toOP734348	475 (2511-2036)	9 i: GA(TA)3/2058	Hf1-1	Mf	c SPVR population	9	S	475	151	0	0	-
	OP734339		0	Hf1-2	Mf	c JA2.24	1	R			CCT:CGT/ 1599	G:A/143	Mf: Sierotzki et al. [25]
	NC044132	507 (from 2056 to 2562)	2 d: 2525 to 2521. 4 s: A:T/2521; C:G/2515; A:T/2512; C:G/2509. 8 i: GA(TA)3/2058; 2 i: (TA)/2521,2522	Hf1-3	Mf	-	1	-	507	169	0	0	-
	AF343069	459 (2525-2565	0	Hf1-2	Mf	184.97.1	1	R	459 (−2 gaps)	153	d CCT:CGT/ 1599	d G:A/143	Mf: Sierotzki et al. [25]
	AF343070		0	Hf1-4	Mf	185.97.3	1	S	459 (−2 gaps)	153	0	0	-
Mf cyt b exon-2	OP734349 toOP734358	996 (from 25 to 1020)	0	Hf2-1	Mf	c SPVR population	10	S	657	219	0	0	-
	…
	NC044132		0	Hf2-2	Mf	-	1	-			0	0	-
	AF343069		3 s: A:T/431; G:N/481; A:G/808	Hf2-3	Mf	184.97.1	1	R			0	0	-
	AF343070		3 s: A:T/431; G:N/481; A:G/808	Hf2-4	Mf	185.97.3	1	S			AAA:CAA/788	L:F/68	-

^a Synonymous mutations in the coding region of the cyt b gene. The haplotype could still carry other mutations along the non-coding region of the sequence. ^b Fungicide sensitivity phenotype classes: S = sensitive, R = resistant for both azoxystrobin and trifloxystrobin. ^c Mf SPVR-I population from a fungicide conventional intensive management system in Ribeira Valley, São Paulo. ^d Non-synonymous mutations and respective nucleotides substitutions are indicated in red.

, Figure 6). Considering M. musicola in particular (Table 4), 87 Mm QoI-sensitive strains (QoI-S), which do not have this mutation were grouped in the cyt b haplotype Hm1, and nine Mm strains (ISC64, ISC55a, ISC09, ISC110, ISC116, ISC117, MG118, JA1.37 and JA1.38c) carrying the G428C mutation leading to the G143A amino acid substitution belongs to the Hm2. The Mm LFZO01000638.1 cyt b sequence used as reference is also Hm1. The reference sequence NC037198.1 from Pseudocercospora mori presented 20 substitutions, all of them synonymous mutations (Hm3). Both these reference strains have no QoI-sensitivity information available (Table 4).

Looking at M. fijiensis (Table 5, Figure 6), the first cyt b Mf haplotype in exon-1 (Hf1-1) is characterized by nine QoI-S strains, without non-synonymous mutation, but presenting an GA(TA)3 insertion at the 2058-bp position, distinct from the Mf sensitive reference sequence AF343070 (Hf1-4). Reference sequence NC044132 has no phenotypical data, and belongs to Hf1-3, with four non-synonymous substitutions, two deletions, eight insertions GA(TA)3 at the 2058-bp position and two nucleotides TA insertions at 2521 and 2522 positions. The Mf-resistant strain JA2.24 (from SPVR-I) carrying the G143A substitution is similar to AF343069 reference, both Hf1-2. Four haplotypes of the Mf cyt b exon-2 were observed, with none presenting non-synonymous mutations (Hf2-1). The reference sequence NC044132 had no change (Hf2-2), while both AF343069 (QoI-R) and AF343070 (QoI-S) presented three synonymous substitutions (haplotypes Hf2-3 and Hf-2-4, respectively). This last sequence was different due to an A:C substitution at position 788 (Table 5).

Analyses of the cyt b partial gene sequences from QoI resistant Mt isolates (Mt ISC57 from SPNW-C; Mt FRA1.21b, Mt JA1.34, Mt JA1.4, Mt JA2.13, Mt JA2.15, Mt JA2.16a, Mt JA2.2p, Mt JA2.3a, Mt JA2.6 and Mt T06 from SPVR-I) also showed the known non-synonymous G143A substitution associated with QoI resistance (GenBank/NCBI sequences OP796647 to OP796657) (Figure 6). The single QoI sensitive isolate Mt ISC86, from SPNW-C, had no mutation in the cyt b partial gene (OP796645) (Figure 6).

4. Discussion

Resistance to QoI fungicides linked to the G143A substitution and the corresponding mutation in the cyt b gene is a well described mechanism for many plant pathogens, including Mm and Mf [9,14]. Other substitutions such as the G137R in Pyrenophora tritici-repentis, G137S in Venturia effusa or the F129L in pathogens such as Phakopsora pachyrhizi, Pyrenophora teres and Plasmopara viticola also confer QoI resistance [47,48], but so far have not been found in either Mf or Mm [2,25,29,30,49] and are associated with lower levels of resistance. In the present study, a total of 18.5% of all isolates tested (M. fijiensis, M. musicola and M. thailandica) sampled at four different locations in Southern Brazil were QoI-R carrying the G143A substitution in cyt b.

Our primary assay for phenotyping sensitivity to QoI fungicides was based on a qualitative approach using fungal mycelial fragments and measuring inhibition of mycelial growth at the discriminatory dose of 10 μg·mL⁻¹ of azoxystrobin or trifloxystrobin. Similar assays have been used for assessing QoI sensitivity of Mf in Costa Rica, where strains carrying cyt b G143A alleles showed resistance factors (RFs) > 100-fold [25]. We also applied a quantitative approach on a set of ten Mycosphaerella isolates varying in levels of QoI sensitivity, which corroborated the outcome from the qualitative approach (Figure 5). The sensitive isolates showed an EC₅₀ range of >0.12 to 2 μg·mL⁻¹ for azoxystrobin and pyraclostrobin, while resistant isolates showed EC_50s > 20 to 150 μg·mL⁻¹ and extremely resistant isolates an EC_50s ≥ 150 μg·mL⁻¹ (Figure 5).

This qualitative approach for discriminating sensitivity from resistance to QoI is also strongly supported by the studies from Alfaro [30] who detected Mf sensitive isolates, from banana plantations in Costa Rica, with EC₅₀ > 0.001–1 μg·mL⁻¹ for azoxystrobin and pyraclostrobin, while resistant isolates had EC₅₀ >10–100 μg·mL⁻¹ and very resistant isolates ≥ 100 μg·mL⁻¹.

Although the screening for sensitivity to QoIs with the discriminatory dose of 10 μg·mL⁻¹ was useful in detecting Mf, Mm and Mt isolates with the G143A mutation in the cyt b gene in our study, this approach is not suitable if the goal is to determine moderate sensitivity classes, which may fall within EC₅₀ > 1–10 μg·mL⁻¹, and specially for isolates that did not carry the G143A cyt b target site mutation [30]. For such isolates, it is plausible that other target site mutations (e.g., F129L and G137R/S) and/or overexpression of transporters may also be associated with the reduced sensitivity to QoIs, as already reported for other plant pathogens such as Pyricularia oryzae Triticum lineage [50] and Zymoseptoria tritici [51].

Our study was conducted with current populations of both yellow and black Sigatoka pathogens from banana plantations in Southeastern Brazil, with contrasting fungicide management to control these foliar diseases. Pathogen populations from conventional or intensive use of fungicides (Mm/Mt SPNW-C and Mf/Mm/Mt SPVR-I) had a higher frequency of resistant isolates than the population with no fungicide input (Mm SPNW-O) (Table 3, Table 4 and Table 5). The Mm and Mf resistant isolates from our study were able to grow at 10 μg·mL⁻¹ of azoxystrobin or trifloxystrobin. This observation probably reflects the effects of high selection pressure for QoI resistance in the pathogen populations over the years, since no evidence of resistance was formerly detected in populations of the pathogens from Vale do Ribeira sampled as early as 2008 or as recent as 2018 [2,49].

For instance, between 2008 and 2009, the G143A substitution in the cyt b gene was not found in a collection of 36 Mm isolates sampled from banana plantations in Northern and Central-Western Brazil and from ten Mf isolates from Vale do Ribeira, in São Paulo [49]. In 2018, despite the small sample size, Malimpensa [2] reported no QoI resistance nor cyt b substitution in Mf from Vale do Ribeira, and one Mm isolate with the G143A substitution, but with only a reduced sensitivity to strobilurin (at 0.1 μg·mL⁻¹ of the fungicide). However, M. thailandica, which was the most abundant species in the sampling from Vale do Ribeira, 58.3% of the isolates (7 out of 12 total) carried the G143A substitution and were highly resistant to QoI (at doses higher than 10 μg·mL⁻¹) [2]. Similarly, in our survey, M. thailandica was highly prevalent in the population SPVR-I from the same region, representing 45.5% of the isolates obtained (10 out of 22 total), and all these isolates were resistant to QoI fungicides and carried the G143A substitution.

An investigation about the dynamics of QoI resistance in Mf populations sampled in Costa Rica from 2000 to 2003, by quantitative PCR detection of the mutation associated with the G143A substitution, indicated frequencies varying from 9% to 78% [27]. The authors concluded that the populations of the pathogen were not structured at scales < 10 m, and wind-born ascospore dispersal have homogenized Mf populations distributed on scales from 1 to 10 km. Reduction in QoI fungicide applications on individual farms did not cause a rapid decrease in G143A frequency regardless of their own spray practice, since surrounding farms at regional scale keep the QoI selection pressure favoring the resistant Mf populations with G143A mutation. Therefore, evolution of resistance to strobilurins in Mf populations was considered to occur regionally, with single individual farms converging on a regional average frequency of G143A mutation, regardless of their own spray system.

Because our pathogen populations were sampled at regional scales varying from a minimum of >10 km (Ilha Solteira, Northwestern São Paulo, Brazil) to a maximum of 750 (Vale do Ribeira, SP)—1200 km (Janaúba, Northern MG, Brazil), the geographical distances might have led pathogen populations to differentiate considering the effects of spray practices on the selection for fungicide resistance.

As a concluding remark, considering that the overall frequency of QoI resistance in three regional populations of the two major Sigatoka pathogens (Mf and Mm) in Southern Brazil, besides the prevalent Mt in Vale do Ribeira, are already at 18.5% on average, anti-resistance strategies should be adopted to avoid their general spread. An important anti-resistance strategy would be to regionally enforce a limit on QoI fungicide sprays to a minimum, and only in mixtures with protectant, multisite, low-risk fungicides to discourage individual decisions to use only systemic, single site, high risk fungicides, given the evidence that the resistant strains of the pathogen could be dispersed long distance from a source of resistant inoculum.

However, from previous studies there was no evidence for rapid selection against the resistant form of Mf after removing QoI fungicides from the spray program [27]. No information is available on the fitness cost associated with the G143A substitution in both Mm and Mf in competition experiments [52]. No such information exists for Mt either. In the absence of a fitness penalty, resistance to QoI would persist in the agroecosystem.

Future studies should focus on the development of real time disease surveillance and monitoring of fungicide resistance using real-time PCR or loop-mediated isothermal amplification (LAMP) techniques which can be used for on-the-spot detection to help growers to take smart decisions on spraying fungicides in a timely manner rather than based on calendar systems [53,54,55,56,57]. Ideally, this novel system could contribute to the rational and sustainable management of the Sigatoka diseases complex, by limiting the emergence and spread of fungicide resistance.

5. Conclusions

Overall, a total of 18.5% of isolates of M. fijiensis, M. musicola and M. thailandica sampled from Southern Brazil were resistant to QoI fungicides and carried the G143A substitution in cyt b.

Pathogens’ populations from banana plantations with conventional or intensive use of fungicides to control Sigatoka diseases (Mm/Mt SPNW-C and Mf/Mm/Mt SPVR-I) had higher frequency of resistant isolates than the population from fields with no fungicide input (Mm SPNW-O).