Introduction

Cercospora sojina Hara is the causal agent of frogeye leaf spot (FLS) on soybean (Glycine max) (Athow and Probst 1952). The disease was first reported in Japan in 1915, in the United States in 1924, in Brazil in 1971, and in Argentina in 1983 (Melchers 1925; Lehman 1928; Veiga and Kimati 1974; Giorda and Justh 1983). In the United States, the disease historically has been most common in the southern soybean production region, and recently has become more common in the midwestern and northern soybean production regions of the country, including Iowa, Nebraska, North Dakota, and Wisconsin (Yang et al. 2001; Mengistu et al. 2002; Neves et al. 2020, 2022). More common observances in northern states may be explained by the combination of warm temperatures during the winter and the capability of the pathogen to survive for up to 24 months in plant residue remaining on the soil surface by the increasing use of conservation tillage practices (Mian et al. 2008; Cruz and Dorrance 2009; Zhang and Bradley 2014).

Epidemics of FLS have increased in frequency and severity worldwide, and thus have become a very important yield-reducing disease across the major soybean-producing countries. Soybean yield losses caused by FLS epidemics can range from 31% to up to 60% due to reduced photosynthetic leaf area, premature defoliation and reduced seed weight (Dashiell and Akem 1994; Mian et al. 1998). In the United States and Ontario, Canada, the estimated average annual soybean yield losses, caused by FLS, from 2010 to 2019, ranged from 101,467 to 1,453,225 metric tons (Allen et al. 2017; Bradley et al. 2021). Additionally, losses due to FLS during the 2009/10 crop season were estimated at about $2 billion USD in Argentina (Sepulcri et al. 2015). Average yields of non-protected plants against FLS were reduced by 37% in Zambia during the 1997/98 crop season (Mwase and Kapooria 2000). In Brazil, the occurrence of FLS is part of a complex of late-season diseases caused by Cercospora kikuchii, Septoria glycines, and Colletotrichum truncatum, and up to 30% yield losses have been reported (Balardin 2002).

Taxonomy

Domain Eukarya, kingdom Fungi, subkingdom Dikarya, phylum Ascomycota, subphylum Pezizomycotina, class Dothideomycetes, order Mycosphaerellales, family Mycosphaerellaceae, genus Cercospora, species Cercospora sojina (NCBI).

Identification

Morphological characterization

Although Cercospora sojina is recognized as the causal agent of FLS, early literature reported Cercospora daizu as the causal agent of this disease (Athow 1987). Conidia are hyaline, elongate to fusiform and measure 6–8 × 40–60 μm (Wise and Newman 2015). Additionally, conidia can be produced on infected parts of the plant (leaf, stem or seeds) and from infested residue on the soil surface (Cruz and Dorrance 2009). As conidiophores continue to grow, conidia are formed on the tips and are pushed aside (Groenewald et al. 2013; Wise and Newman 2015). In a single lesion, 2 to 25 conidiophores can be produced, and each conidiophore can produce 1 to 11 conidia (Lehman 1928). Conidia can germinate on a leaf surface within an hour of deposition in the presence of water at 25 to 30 °C (Phillips 1999).

Molecular characterization

Conventional polymerase chain reaction (PCR) assays have been successfully developed and used to identify and detect several important plant pathogens. C. sojina can be identified by amplifying a fragment of actin, calmodulin, histone, translation elongation factor, as well as internal transcribed spacer regions and the 5.8S rRNA gene (Groenewald et al. 2013; Neves et al. 2022). However, translation elongation factor and calmodulin genes can be intermixed with other Cercospora species (Groenewald et al. 2013). These genes can be amplified using primers (Table 1), and nucleotide sequences can be compared using the BLAST search on NCBI Genbank.

Table 1 Primers for molecular identification of Cercospora sojina
Full size table

Disease symptoms

Although disease symptoms most commonly appear during reproductive growth stages, FLS lesions can affect leaves, pods and stems at any stage of development (Wise and Newman 2015). Symptoms include small, dark lesions that evolve from tan to brown spots surrounded by a narrow, purple-brown margin (Wise and Newman 2015) (Fig. 1A). The lesion diameters range from 1 to 5 mm (Grau et al. 2004). On the abaxial surface, the formation of clusters of conidia can be observed in the center of mature lesions (Wise and Newman 2015) (Fig. 1B). Stem lesions, which are two to four-times longer and wider than leaf lesions, are less common, but they can appear later in the season (Bisht and Sinclair 1985). Additionally, the fungus can penetrate through the pod walls and infect the seeds (Phillips 1999). Symptoms on seeds include light to dark gray or brown areas that can range from specks to large blotches covering the entire seed coat (Bisht and Sinclair 1985).

Fig. 1
figure 1

Frogeye leaf spot symptoms include round to angular lesions with a dark-brown margin and a tan to light brown center (A); and C. sojina fuzzy gray sporulation in the center of the lesion underside the leaf (B)

Full size image

Disease cycle and epidemiology

Initial inoculum can be produced on infected plant residue where the pathogen can overwinter for up to 24 months, or it can survive on infected seeds (Singh and Sinclair 1985; Cruz and Dorrance 2009; Zhang and Bradley 2014). Conidia are then dispersed throughout the crop canopy from the infested residue by wind or splashing rain. C. sojina infects the plants withbranched hyphae through open stomata. The lesions are not visible for nearly 14 days after infection. Fully expanded leaves are more resistant to infection than young expanding leaves which are highly susceptible. For plants grown in warm (25–30 °C) and humid conditions (> 90% relative humidity), sporulation can occur within 48 h of the appearance of visible symptoms (Wise and Newman 2015). Under favorable conditions, secondary infection of leaves, stems, and pods continue throughout the soybean growing season, characterizing the disease as polycyclic (Fig. 2) (Wise and Newman 2015). In seeds, the fungus can penetrate both indirectly through pores and cracks in the seed coat or directly through hilar tracheids and may grow into seedling tissues during germination and emergence (Singh and Sinclair 1985). Seed transmission can play an important role in disease spread as the disease has been found in fields never planted to soybean or under soybean rotation in Argentina, indicating that the pathogen was likely introduced via infected seeds (Sautua et al. 2018).

Fig. 2
figure 2

Disease cycle of Cercospora sojina in soybeans (Bradley et al. 2016)

Full size image

Pathogenicity

Luo et al. (2018) assembled the genome of C. sojina race 1 and obtained a total assembly size around 40.84 Mb. Additionally, the genome of C. sojina contained 11,655 protein-coding genes, of which a total of 233 proteins were predicted as the putative small (400 amino acids) cysteine-rich proteins (Luo et al. 2018). The authors found 141 putative effectors and more than one third of them were upregulated during starvation suggesting that C. sojina can probably deploy effectors to promote infection (Luo et al. 2018). Despite the fact that most of the species across the Cercospora genus can produce a toxin called cercosporin, it has been disputed if C. sojina produces it (Goodwin et al. 2001). Luo et al. (2018) identified a gene cluster with eight cercosporin biosynthesis genes in the C. sojina genome and observed the increased transcription of the eight genes during infection. These results imply that C. sojina may produce cercosporin during infection. However, authors were unable to detect cercosporin in either cultured mycelium or infected plant tissue (Luo et al. 2018). Finally, in the C. sojina genome, there were around 23.5% potential secreted proteins that were predicted as putative carbohydrate-active enzymes (CAZymes), demonstrating that C. sojina may employ a large group of CAZymes to digest host cell walls during invasion (Luo et al. 2018). Another study sequenced Race 15 of C. sojina and analyzed the comparative genome with respect to Race 1 (Gu et al. 2020). The authors found that the pathogenic reaction patterns of Race 1 and Race 15 were similar.

Genetic diversity

Bradley et al. (2012) used amplified fragment length polymorphism (AFLP) markers to better understand the genetic diversity of a historical collection of 62 C. sojina isolates from Brazil, China, Nigeria, and the United States. The authors found a high degree of genetic diversity with no clear separation of isolates based on their origin. Only two isolates collected from Georgia and two isolates from China were clustered together among the two major clusters and seven sub-clusters obtained. Another study investigated the genetic diversity of a subset of 186 isolates of C. sojina, including historical isolates, which were genotyped for 49 single nucleotide polymorphism (SNP) markers, revealing 35 unique genotypes (Shrestha et al. 2017).

Sexual recombination

Sexual reproduction is a key mechanism through which genetic diversity is produced in many plant-pathogenic fungi (Glass and Kuldau 1992). Although for most Cercospora spp., including C. sojina, a sexual stage has not been observed in either field or laboratory conditions, molecular analyses have shown that Cercospora spp. form a monophyletic group within the teleomorphic genus Mycosphaerella (Goodwin et al. 2001; Crous and Braun 2003; Crous et al. 2004a). In fact, comparative genome analysis of C. sojina with plant pathogen members of the genus Mycosphaerella (M. pini, M. Populorum, Z. tritici [M. graminicola] and M. fijiensis) on different plant hosts (pine, poplar and banana, respectively) found considerable conserved synteny, higher average exon numbers per gene and gene density between C. sojina and Z. tritici compared to the genomes of the other three fungal species in the genus Mycosphaerella (Zeng et al. 2017). These genome features can be explained by the fact that the hosts of C. sojina and Z. tritici, soybean and wheat, have similar characteristics of growing conditions and pathogen resistance, compared with perennial tree species pine, poplar, and banana as hosts of M. pini, M. populorum and M. fijiensis, respectively (Zeng et al. 2017).

When the sexual stage is not known, which is the case of C. sojina, several approaches have been used to provide evidence of cryptic sexual reproduction, including quantification of genetic diversity, population differentiation, and mating-type frequencies (Kim et al. 2013). Typically, populations undergoing sexual reproduction exhibit high genetic diversity and equal mating-type frequencies compared with populations solely or predominantly reproducing asexually (Milgroom 1996). Kim et al. (2013) developed a multiplex PCR assay with specific primers for C. sojina aiming to determine mating types for a collection of 132 C. sojina isolates collected from six fields in Arkansas. Of the 132 C. sojina isolates, 68 isolates had the MAT1-1–1 idiomorph, and 64 isolates had the MAT1-2 idiomorph. No isolates possessed both idiomorphs. An equal proportion of mating-type loci in all populations analyzed and high genotypic diversity (26 to 79%) suggested that populations of C. sojina in Arkansas are most likely undergoing cryptic sexual reproduction (Kim et al. 2013). Another study investigated the genetic diversity of a subset of 186 isolates of C. sojina, including historical isolates, which were genotyped for 49 single nucleotide polymorphism (SNP) markers, revealing 35 unique genotypes (Shrestha et al. 2017). Both mating type alleles (MAT1-1–1 and MAT1-2) were found in individual lesions suggesting opportunity for sexual recombination (Shrestha et al. 2017).

Management

Host genetic resistance and races

In the United States, a total of 12 races of C. sojina were reported from various states (Grau et al. 2004) (Table 2). In Brazil, 25 races have been reported (Yorinori and Klingelfuss 1999), and in Argentina, races 11 and 12 were identified during the 2008/09 and 2009/10 growing seasons (Scandiani et al. 2012) (Table 2). In China, 11 races of C. sojina were identified and, among them, races 1, 7, and 10 were considered the major ones (Huo et al. 1988). However, the total number of races in China increased to 14 and, more recently, a race 15 was reported to be the dominant, occurring 36% more frequently than the previously dominant race 1 (Gu et al. 2020) (Table 2). This has led to a loss of host resistance in many cultivars in China (Gu et al. 2020).

Table 2 Race identification for Cercospora sojina with respective resistant genes (Rcs)
Full size table

Grau et al. (2004) stated that different sets of soybean differential cultivars were used to identify the C. sojina races in the United States, Brazil, and China. Additionally, Mian et al. (2008) pointed out the lack of a universally accepted set of soybean differential cultivars for the classification of C. sojina isolates into races as well as to identify, designate and compare races of this pathogen. Hence, the later authors created a new set of soybean differential cultivars and revised the C. sojina race designations to advance the characterization of C. sojina races and to identify additional FLS resistance genes in soybean (Mian et al. 2008). A total of 93 C. sojina isolates were analyzed for their reaction on 38 putative soybean differential cultivars resulting in 3,534 isolate–differential combinations (Mian et al. 2008). The authors initiated the new race structure with race 5, since there are no known existing cultures of races 1 to 4, and identified 11 unique isolates, designated races as 5 to 15 (Mian et al. 2008).

The approach used by Mian et al. (2008) does not take into account the range of disease severity reaction in each of those differentials. Therefore, Mengistu et al. (2020) proposed a new approach, known as Pathogenicity Group, to address and simplify the current system of C. sojina race designations. The authors evaluated the diversity of 83 C. sojina isolates collected from 2006 to 2009 by using pathogenicity groups among 12 soybean differentials (Davis, Peking, Kent, CNS, Palmetto, Tracy, Lincoln, S-100, Richland, Blackhawk, Hood and Lee). The set of 83 isolates grouped into five pathogenicity groups (PG1 through PG5) representing the virulence diversity present in those isolates collected from various geographical regions (Mengistu et al. 2020). PG1 did not infect eight of the differentials except Blackhawk, Lincoln, S-100, and Lee; PG2 showed low virulence on all differentials except on Davis (hypersensitive reaction); PG3 produced hypersensitive reaction on Davis but with less than moderate reaction to the rest of the differentials; PG4 caused no infection on Davis but moderate infection on Peking; and, PG5 was the most virulent pathotype that infected all genotypes except Davis (Mengistu et al. 2020). Therefore, even the most virulent pathogenicity group could not overcome the resistant Rcs3 gene in Davis and, until now, there are no Rcs3-virulent races reported in the literature. Similarly, a previous study screened 40 isolates of C. sojina collected in 2018 and 2019 across six counties in Georgia, and found no isolates pathogenic on Davis, suggesting that the Rcs3 gene is still an effective source of resistance in Georgia (Harrelson et al. 2021).

The Rcs3 gene is one of the three single dominant genes conditioning resistance to C. sojina recognized by the Soybean Genetics Committee (Mian et al. 2009). The first gene found was Rcs1 in Lincoln, which conferred resistance to race 1 of C. sojina (Athow and Probst 1952). Rcs2 was identified in Kent for resistance to race 2 (Athow et al. 1962). Finally, Rcs3 from Davis was found to condition resistance to race 5 and to all other known races of C. sojina in the United States (Phillips and Boerma 1982; Boerma and Phillips 1983) as well as to all known isolates of C. sojina in Brazil (Yorinori and Klingelfuss 1999) (Table 2). Although other dominant genes for resistance to race 5 were found in the cultivars Ransom, Stonewall and Lee in 1993 (Pace et al. 1993), they were not considered to be important sources of resistance, because, currently, race 5 is not seen as an economic threat to soybean in the United States (Baker et al. 1999). Additionally, another single dominant gene nonallelic to Rcs3 was found from the cultivar Peking and provided resistance against many C. sojina isolates (Baker et al. 1999).

In China, the gene Rcsc7 was assigned to a dominant gene for conditioning resistance to Chinese race 7 (Table 2), but it has not been officially approved by the Soybean Genetics Committee as the allelism between Rcsc7 and other resistance genes is not known (Zou et al. 1999). In Brazil, F1 plants were obtained from the diallel mating of seven soybean cultivars (Bossier, Cristalina, Davis, Kent, Lincoln, Paraná, and Uberaba), and their reactions were evaluated against C. sojina race 4 using a multivariate variable developed from soybean reactions to infection degree, mean lesion diameter, percent of lesioned leaf area, lesions per square centimeter, and disease index (Gravina et al. 2004). The authors reported that Davis, Cristalina, and Uberaba were free of FLS symptoms (Gravina et al. 2004). Mengistu et al. (2011) assessed resistance to C. sojina race 11 by field screening maturity groups I to VI across two locations (Missouri and Illinois). A total of 260 accessions including 12 differentials resulted in 20 remaining resistant accessions that might contain novel loci for FLS resistance as the presence of Rcs3 allele was not found using molecular markers (Mengistu et al. 2011).

Quantitative resistance to race 2 of C. sojina was identified in the greenhouse using recombinant inbred lines derived from the cross of the cultivars Essex and Forrest (Sharma and Lightfoot 2014). Essex is known to be partially resistant while Forrest is partially susceptible to mixed races of C. sojina. The authors inferred that quantitative resistance to C. sojina race 2 involved two major quantitative trait loci (QTL). The two loci were effective at different stages of seedling development, suggesting they were conditional QTL, and, according to the location of the QTL, the loci were not allelic to Rcs3 (Sharma and Lightfoot 2014). Recently, McAllister et al. (2021) also screened 91 recombinant inbred lines from the crossing between Essex and Forrest under greenhouse conditions for FLS resistance to C. sojina race 15 and used single nucleotide polymorphism (SNP) markers to identify associated QTL. Two QTL were mapped, being one QTL reported on chromosome 13 coinciding with the QTL previously reported (Pham et al. 2015), and the QTL on chromosome 19 was novel (McAllister et al. 2021).

Biocontrol

The use of beneficial microorganisms to control plant diseases is an alternative or a supplemental way of reducing the use of chemicals. In the U.S., Lysobacter enzymogenes strain C3 (LeC3) was tested against C. sojina, which effectively inhibited its vegetative mycelial growth and conidial germination on plates (Nian et al. 2021). Moreover, a previous study reported that the application of Trichoderma virens conidial suspensions as a foliar treatment significantly reduced frogeye leaf spot severity of soybean compared to a nontreated control (Lacey 2018). A previous study in Argentina reported reduced mycelial growth of C. sojina in vitro by testing a cell suspension of three indigenous bacterial strains, including BNM297 (Pseudomonas fluorescens), BNM340 and BNM122 (Bacillus amyloliquefaciens) (Simonetti et al. 2012). The authors found that Bacillus BNM122 and BNM340 inhibited the fungus to a similar degree (52–53%). Additionally, a significant inhibition of conidial germination was observed after 24 and 72 h of co-cultivation with cell suspension from BNM297, BNM340 or BNM122, (~ 79%, 79% and 89%, respectively). Biocontrol tests in vivo were conducted and both spray-applied bacteria, BNM340 and BNM122, significantly reduced the disease severity to a similar degree with respect to positive control plants, showing no significant differences between them, while P. fluorescens BNM297 did not affect FLS severity on soybean plants (Simonetti et al. 2012).

Induced systemic resistance (ISR) consists of the activation of a plant defense upon pathogen attack by triggering a cascade of reactions that spread from the site of induction to distant parts of the plant (Kloepper et al. 1992). Previous studies showed that soybean plants inoculated with Bacillus sp. CHEP5 had less FLS severity, with healthier and greener leaves compared to non-inoculated plants (Tonelli and Fabra 2014). Additionally, as Bacillus sp. CHEP5 was applied onto the roots and the response was detected in the shoot system, the bacterial induction of resistance in the plant was considered to be systemic, hence, attributed to ISR (Tonelli and Fabra 2014). The authors also investigated if the mechanism to induce systemic resistance of Bacillus sp. CHEP5 involved the priming of the jasmonic acid dependent pathway. The increased expression of the defense related gene GmAOS in Bacillus sp. CHEP5 plus pathogen challenged plants strongly suggest that the enhanced soybean resistance to C. sojina attack induced by this bacterium occurs in a jasmonic acid dependent pathway (Tonelli and Fabra 2014). Additionally, a following study showed a mutualistic behavior between Bacillus sp. CHEP5 with the nitrogen fixing strain Bradyrhizobium japonicum E109 being more effective in reducing frogeye leaf spot severity than the inoculation of Bacillus sp. CHEP5 alone (Tonelli et al. 2017).

Cultural practices

Cultural practices such as crop rotation and tillage can help to reduce FLS incidence (Grau et al. 2004; Wise and Newman 2015). A previous study recommended that crop rotation with a nonhost of a minimum of two years would be more effective to reduce the level of viable C. sojina inoculum, regardless of the depth of the crop residue in the soil (Zhang and Bradley 2014). Tillage can reduce the inoculum by burying infested plant residues (Mengistu et al. 2014). However, recent studies in Tennessee have not found significant differences in FLS severity, in the absence of fungicide application, between tilled and no-till plots across field trials conducted from 2007 to 2010 (Mengistu et al. 2014), and from 2014 to 2016 (Mengistu et al. 2018). Although tillage alone did not significantly affect disease, fungicide efficacy was greater in tilled compared to no-till plots (Mengistu et al. 2014). Moreover, early planting seems to be favorable to avoid higher FLS pressure, as a previous study reported higher yield reduction due to FLS when planting was delayed two weeks after the optimum planting date (Akem and Dashiell 1994).

Chemical control

The regular use of fungicides in the United States started in 2005 driven by an increase in soybean prices and the potential threat of Asian soybean rust (Phakopsora pachyrhizi) (Phillips et al. 2021). Fungicide applications aiming to control FLS are recommended during reproductive growth stages (Akem 1995). Active ingredients from different fungicide classes available for managing FLS include demethylation inhibitors (DMI), quinone outside inhibitors (QoI), methyl benzimidazole carbamates (MBC), succinate dehydrogenase inhibitors (SDHI) and chloronitriles (Crop Protection Network 2022). MBCs act in the cytoskeleton by inhibiting the formation of the β tubulin assembly during mitosis (Olaya and Geddens 2019). QoIs and SDHIs are fungicides that inhibit respiration (Sierotzki and Scalliet 2013; Fungicide Resistance Action Committee 2020). The QoIs act in the complex III in the mitochondria, binding the activity of the quinol oxidation (Qo) site of the Cytochrome b, which avoid electron transfer between Cytochrome b and Cytochrome c, interrupting ATP synthesis (Bartlett et al. 2002; Sierotzki and Stammler 2019). On the other hand, the SDHIs act in the complex II of the electron transport chain in the mitochondria, also inhibiting the production of ATP (Sierotzki and Scalliet 2013; Klappach and Stammler 2019). The DMIs are compounds that inhibit the sterol biosynthesis in membranes, which can cause cell rupture and electrolyte leakage (Mehl et al. 2019; Kumar et al. 2021). Finally, chlorothalonil is a multi-site fungicide that belongs to the chloronitriles group and is used as a protectant fungicide (Miles et al. 2007; Battaglin et al. 2011).

QoI fungicides, mainly azoxystrobin, pyraclostrobin and trifloxystrobin, have been commercially available and largely used on soybean in the United States, including for FLS management (Sauter et al. 1999; Dorrance et al. 2010; Nelson et al. 2010; Mengistu et al. 2018). However, after the emergence of QoI-resistance, studies have shown that DMIs, MBCs, SDHIs and premixes can be effective for managing FLS (Backman et al. 1979; Dorrance et al. 2010; Butler et al. 2018; Mengistu et al. 2018; Phillips et al. 2021; Viggers et al. 2022). For instance, benomyl (MBC) was very effective in reducing FLS severity among susceptible cultivars in Alabama, United States (Backman et al. 1979) and Zambia (Mwase and Kapooria 2000). However, another study conducted in Zimbabwe in 1996 and 1997 reported that the DMI flusilazole and the mixture of flusilazole + carbendazim were more effective against FLS than benomyl applied alone or as a premix with mancozeb (Galloway 2008). Recently, Mengistu et al. (2018) showed significantly higher efficacies (> 70%) for flutriafol (DMI), thiophanate-methyl (MBC) and the premix azoxystrobin + difenoconazole (QoI + DMI), compared to the single application of pyraclostrobin (27%) and chlorothalonil (30%). Additionally, a previous study summarized data from 66 uniform fungicide trials conducted from 2012 to 2021 across the major soybean-producing states in the U. S. using a meta-analytic approach (Barro et al. 2023). On average, the authors found the most effective fungicides to be the premixes difenoconazole + pydiflumetofen, thiophanate-methyl + tebuconazole, azoxystrobin + difenoconazole and trifloxystrobin + prothioconazole, all with percent control greater than 50%. The poorly performing fungicide was pyraclostrobin (11%). A statistically significant decline in performance over the years was detected for two dual premixes (azoxystrobin + difenoconazole and thiophanate-methyl + tebuconazole) and two single active ingredients (pyraclostrobin and tetraconazole), which can be linked to fungicide resistance issues (Barro et al. Unpublished).

Fungicide application timing and coverage are critical for optimal disease control. Akem (1995) evaluated applications of the fungicide benomyl at six different growth stages, starting from V3 (fully developed leaves, beginning with trifoliate nodes) to R5 (beginning seed), to determine the effect of the fungicide timing on frogeye leaf spot severity and found that applications at R1 (beginning bloom) and R3 (beginning pod) significantly reduced disease severity. Regarding coverage, Butler et al. (2018) conducted field experiments in 2014 and 2015 in Tennessee to evaluate the influence of droplet size on foliar fungicide efficacy. The authors found no significant differences among the industry recommended standard flat fan XR11002VS (XR) nozzle and the drift-reduction nozzle type TTI11002-VP (TTI) but found significant disease reduction after application of azoxystrobin + difenoconazole compared to the nontreated control (Butler et al. 2018). Additionally, results from ten field trials conducted from 2017 to 2020 in Iowa by applying fluxapyroxad + pyraclostrobin using a traditional ground sprayer with an overhead spray boom and a ground sprayer with 360 undercover sprayers showed no statistical difference between fungicide application methods on FLS severity (Viggers et al. 2022).

As mentioned previously, the primary inoculum sources of the disease are infected seeds and plant residue. Therefore, the use of pathogen-free or fungicide-treated seeds is crucial to prevent the introduction and further spread of the disease (Sautua et al. 2018). A previous study in Argentina evaluated the effect of fungicide seed treatments in reducing FLS incidence and found that premixes including benzimidazole fungicides, such as pyraclostrobin + thiophanate-methyl and carbendazim + thiram, were more effective to eradicate the pathogen in seeds (Sautua et al. 2018).

Fungicide resistance

C. sojina isolates with reduced sensitivity to quinone outside inhibitor (QoI) fungicides were first reported from Tennessee in 2010 (Zhang et al. 2012a). The resistance mechanism involved is an amino acid substitution (glycine is replaced with alanine at the codon 143) caused by the G143A mutation in the Cytochrome b gene (Zeng et al. 2015). However, other mutations associated with resistance to QoI fungicides such as the F129L (change from phenylalanine to leucine at codon 129) and G137R (change from glycine to arginine at codon 137), have not been reported in C. sojina (Zeng et al. 2015). Since the first confirmation in 2010, QoI-resistant isolates have become widespread across more than 20 soybean-producing states in the U.S. (Zhang et al. 2012a, b, 2018; Standish et al. 2015; Zeng et al. 2015; Mathew et al. 2019; Zhou and Mehl 2020; Neves et al. 2020, 2021, 2022; Harrelson et al. 2021) (Fig. 3).

Fig. 3
figure 3

States where QoI-resistant isolates of C. sojina were detected due to the amino acid substitution (glycine is replaced with alanine at the codon 143) caused by the G143A mutation in the Cytochrome b gene. Adapted from Harrelson et al. (2021), Mathew et al. (2019), Neves et al. (2020, 2021, 2022), Standish et al. (2015), Zeng et al. (2015), Zhang et al. (2012a, 2012b, 2018) and Zhou and Mehl (2020)

Full size image

Several methods have been used to identify whether C. sojina isolates are sensitive or resistant to QoI fungicides. First, the effect of fungicide in vitro is a standard bioassay to evaluate the influence of chemistries and determine the effective concentration that reduces fungal growth or conidia germination by 50% relative to the non-amended control (EC50) (Fungicide Resistance Action Committee 2020). Based on EC50, the discriminatory dose assay determined for C. sojina was 1 µg/ml of azoxystrobin, 0.1 µg/ml for pyraclostrobin and 1 µg/ml for trifloxystrobin (Zhang et al. 2018). Conidia that germinated on the discriminatory dose will be considered resistant to QoI fungicides. Second, molecular methods also can be used to identify QoI-sensitive and –resistant isolates of C. sojina. Zeng et al. (2015) developed specific primers for PCR assay to recognize a mutation point that confers resistance to QoI fungicides. The primers used to identify QoI-sensitive isolates (Cs-2F/Cs-5R-2) produce a 359 bp fragment whereas the primers used to identify QoI-resistant isolates (Cs-1F/Cs-1R-2) produce a 207 bp fragment (Table 3). Additionally, Mut4-F/Mut4-R primers can amplify a fragment of the Cytochrome b gene that spans the area of F129L, G137R and C143A mutations (Zeng et al. 2015) (Table 3). Standish et al. (2015) developed a polymerase chain reaction-restriction fragment length polymorphism (PCR–RFLP) to identify the G143A mutation in C. sojina using the restriction enzyme Alul. With that, PCR products from QoI-resistant isolates will produce two fragments, while QoI-sensitive isolates will remain intact upon digestion with restriction enzymes. Zhou and Mehl (2020) designed PCR (FLS-F2/FLS-R2) and pyrosequencing (FLS-S2) primers that target the presence of the G143A mutation in the Cytochrome b gene of C. sojina (Table 3).

Table 3 Primers for molecular characterization of resistant isolates of Cercospora sojina to Quinone Outside Inhibittors (QoIs)
Full size table

Since QoI resistant populations of C. sojina in the United States have become widespread, growers have relied more on demethylation inhibitor (DMI) and methyl benzimidazole carbamate (MBC) fungicides, applied alone or as premixes (Zhang et al. 2021). Although DMI and MBC fungicides are classified as medium and high risk, respectively, for fungicide resistance development (Fungicide Resistance Action Committee 2020), C. sojina isolates resistant to these fungicide classes have not yet been reported in the United States. Zhang et al. (2021) investigated the sensitivity to the DMI fungicides, flutriafol and tetraconazole, and the MBC fungicide, thiophanate-methyl, for 145 C. sojina isolates collected prior to 2001 (baseline isolates), and from 2007 to 2012, representing 12 states (AL, GA, IA, IL, LA, MS, SC, WI, KY, TN, AR and NY). No shift towards reduced sensitivity to the DMI and MBC fungicides was found between baseline isolates versus isolates collected from 2007 to 2012 (Zhang et al. 2021).

Therefore, rotating modes of action that have no resistance detected yet is very important as a study reported that the application of a fungicide mix containing a QoI (azoxystrobin) and a DMI (difenoconazole) resulted in a higher proportion of resistant isolates compared to the non-treated plots, suggesting fungicide mixes can still exert selection pressure for QoI resistance (Shrestha et al. 2017).

Conclusions and future directions

In the United States, FLS is an important soybean disease that has been widespread from southern to northern states of the country, causing yield losses as well as raising fungicide resistance issues. However, knowledge of the impact of seed-borne infection on FLS development in the United States and the effectiveness of fungicide seed treatments in seed health are limited. Additionally, many questions relating to population biology, variability and pathotype structure remain unanswered, which is very important for breeding programs to successfully develop resistant cultivars. Studies explaining the mechanism of variation in this species would be extremely important in providing a deep understanding of the role of sexual recombination in creating diversity and will significantly advance our knowledge in areas relating to its pathogenicity, fungicide resistance and management.

The need to identify additional sources of resistance to C. sojina is critical as new variants have shown virulence to most soybean differentials under greenhouse inoculations (Mengistu et al. 2020). Screening in the field and/or greenhouse for evaluating accessions for reaction to C. sojina present many disadvantages such as the time required and the difficulties of inoculating with one/multiple races or relying on natural inoculum (Mengistu et al. 2011). Development of new technologies, such as molecular markers, that can assist the selection for resistance would be useful in developing FLS-resistant soybean cultivars. Moreover, detection of foliar diseases and their severity usually relies on visual assessments, which is subjective and can be influenced by the inherent knowledge and training of raters (Liu et al. 2021). The development of disease monitoring systems automatized in digital platforms can help to reduce or minimize visual bias and optimize disease assessment in the field, mainly for disease phenotyping, which can be used for genetic mapping to identify QTLs for crop genetic resistance and in breeding efforts for resistance to FLS. Liu et al. (2021) analyzed leaf images and hyperspectral reflectance data of healthy and FLS diseased soybean leaves. The models developed by the authors of the study achieved overall accuracies ranging from 91 to 97% and serve as a theoretical reference for improving disease monitoring systems (Liu et al. 2021). Additionally, McDonald et al. (2022) developed an image processing algorithm to evaluate the percent of FLS diseased leaf area and the number of lesions on a leaf. The authors reported that the automated measurement of the percent of diseased leaf area deviated from the manually measured value by less than 0.05% on average, while the automatic lesion counting deviated by an average of 1.6 lesions from the manually counted value (McDonald et al. 2022).

Finally, continued monitoring of fungicide efficacy across field trials as well as C. sojina population sensitivity to QoI, DMI, MBC and SDHI fungicides is critical to support decision making in selecting fungicides for maximizing FLS management. Grower's decisions must take into account not only technical information such as fungicide efficacy and yield return, but also profitability and strategies to mitigate fungicide resistance issues, seeking to preserve the lifetime of site-specific fungicides. Weather-based prediction models can also help build a sustainable fungicide program by applying fungicides only when needed throughout the season. A previous study in Argentina evaluated the FLS epidemic progress in six sites of the Pampas region during the 2009/2010 soybean season and calculated meteorological variables during the nine days before each field observation of disease occurrence for each site, using weather station and satellite data (Sepulcri et al. 2015). Logistic models were used to estimate probabilities of having severe or moderate to null disease. Estimations obtained by the developed model agreed with the observed epidemiological curve for one of the sites during the 2010/2011 season and coincided with the low disease presence recorded during the 2011/2012 season (Sepulcri et al. 2015). Therefore, integrating such weather-based models within a decision support tool determining fungicide spray application can be a sound basis to protect soybean plants against C. sojina.