Genetic diversity of Xanthomonas phaseoli p v. manihotis populations using rep-PCR and VNTR molecular markers

Abstract The objective of this work was to evaluate the genetic diversity of Xanthomonas phaseoli p v. manihotis (Xpm) from eight populations from five cassava producing states in Brazil, through the rep-PCR (BOX-PCR and ERIC-PCR) and variable number of tandem repeat (VNTR) markers. Cassava leaves with symptoms of cassava bacterial blight were collected in eight municipalities, and the Xpm isolates were identified by amplification with primers specific for these isolates. The identity of the Xpm isolates was confirmed with the BOX-PCR, ERIC-PCR, and VNTR markers. The observed selection pressure, together with the mode of reproduction and the mechanisms that increase genetic variability, allows of the pathogen populations to adapt according to microclimate variation, contributing to a differentiated reproductive success. ERIC-PCR and VNTRs are the best markers for evaluating the genetic variability in the eight studied Xpm populations. However, ERIC-PCR is the marker that best separated the groups by population and presented a higher similarity between the isolates of the same population. The study of the genetic diversity of Xpm is key to improve disease monitoring and management strategies in cassava crops.


Introduction
Cassava (Manihot esculenta Crantz) is a staple food that provides a source of income for millions of people in tropical countries (Bayata, 2019;Sonnewald et al., 2020).Brazil is the sixth largest producer in the world, with approximately 18.2 million tons produced in 2021 on 1.2 million hectare, generating R$ 10.9 billion for the cassava agribusiness (FAO, 2023).However, this crop is threatened by several pests and diseases that significantly affect its yield and commercialization, such as cassava bacterial blight (CBB), widely distributed in the different areas where the species is grown (Rache et al., 2019(Rache et al., , 2023;;Zárate-Chaves et al., 2021a).
CBB is caused by Xanthomonas phaseoli pv.manihotis (Xpm), a vascular pathogen that is generally found on the surface of leaves, but may enter the plants through wounds and/or natural openings such as stomata.In the leaves, the main symptoms are darkgreen and angular spots and water-soaked lesions (Veley et al., 2023).Vascular colonization can be associated with gummosis (exudation), vascular necrosis of the stem, and plant wilting and dieback, and one of the main means of pathogen dispersion is the movement of infected but asymptomatic stems (Medina, 2018).In the case of severe infection, premature dehydration and defoliation are common, as well as the death of the non-lignified soft tissue at the top of the growing shoot (Teixeira et al., 2021).
Since, to date, there is no evidence of effective methods to prevent and treat CBB, the use of cultural practices and of resistant varieties is considered the best management strategy against the disease (Bart et al., 2012;Teixeira et al., 2021;Zárate-Chaves et al., 2021b;Ye et al., 2023).Therefore, a good characterization of the Xpm pathogen is important for the development of resistant varieties.
Molecular markers are indispensable tools for studies of plant pathogenic bacteria diversity.The most used are microsatellites or simple sequence repeats (SSRs) and rep-PCR (BOX-PCR and ERIC-PCR) markers (Kapantaidaki et al., 2021;Díaz-Tatis et al., 2022).However, variable number of tandem repeat (VNTR) markers are considered as the most informative for the evaluation of diversity analysis in Xpm populations (Trujillo et al., 2014;Zárate-Chaves et al., 2021a, 2021b).
In this scenario, the lack of information on Xpm population dynamics and the broad range of different edaphoclimatic conditions represented by the different locations where cassava is cultivated still represent a potential risk for crop growers in Brazil.
The objective of this work was to evaluate the genetic diversity of Xanthomonas phaseoli pv.manihotis (Xpm) from eight populations from five cassava producing states in Brazil, through the rep-PCR (BOX-PCR and ERIC-PCR) and VNTR markers.
According to Köppen-Geiger's classification, the climates of the municipalities are: Cfa, humid subtropical, in Londrina and Paranavaí in the state of Paraná; Aw, tropical savanna, in Ocauçu in the state of São Paulo; BWh, hot desert, in Laje and Guanambi in the state of Bahia and Russas in the state of Ceará; and Am, tropical monsoon, in Dourados and Naviraí in the state of Mato Grosso do Sul.
For isolation procedures, sampling was systematically performed using diagonal transects across three to four fields in each location.Specifically, leaves exhibiting distinctive symptoms of CBB or stems with bacterial exudation were carefully selected for bacterial isolation.The quantity of samples collected varied based on the incidence of the disease within each field.Given the extensive survey conducted on a wide range of cassava cultivars and landraces across multiple fields and municipalities, the study did not assign a primary emphasis on evaluating the influence of cassava genotype.
For analyses, the plant samples were taken to the Phytopathology Laboratory at Embrapa Mandioca e Fruticultura.First, the leaf samples were washed in running water, cut into fragments of approximately 1.0 cm 2 , and subjected to a disinfection process in 70% alcohol for 30 s, 50% sodium hypochlorite for 1 min, and, again, in 70% alcohol for 30 s, followed by rinsing three times in sterilized water.
The tissue fragments were macerated using a mortar and pestle in 1.0 mL sterilized distilled water (SDW).A serial dilution (10 -1 , 10 -2 , and 10 -3 ) was performed for each isolate in the suspension obtained in 1.5 µL microtubes containing 450 µL SDW.From this dilution, the tissues were plated in three Petri dishes (one for each dilution), containing the YPG medium with 5.0 g yeast extract, 5.0 g protease (peptone), 5.0 g glucose, 15 g agar, and 1.0 L SDW (Restrepo et al., 2000).
Using a sterilized wooden toothpick, two colonies were scraped from each plate and placed in PCR microtubes containing 100 µL SDW and, then, taken to a thermocycler for 10 min at 95°C; this process was used for cell lysis and subsequent DNA extraction.The samples were kept at 4°C for use in the polymerase chain reactions (PCRs).The molecular characterization of the bacterial isolates was conducted via amplification with specific primers for pathogenic Xpm isolates (Verdier et al., 1998;Melo et al., 2019).
Only the isolates identified as Xpm, based on the presence of a characteristic band for the pathovar obtained through PCR (XV/XK_MOD), were considered.To prepare the inoculum, bacteria were streaked on YPG medium, at 28°C, for 24 hours.A single colony of each Xpm isolate was grown in YPG, at 28°C, with shaking at 230 rpm for 24 hours.Cells were harvested by centrifugation at 3,000 g and resuspended in 10 mmol L -1 MgCl 2 ; the suspension was adjusted to obtain 1×10 6 CFU mL -1 , measured by 0.01 absorbance on a spectrophotometer, as described by Mora et al. (2019).
For inoculation procedures, three wounds were made on the abaxial side of the cassava leaf, and the bacterial suspension was deposited by infiltration using a needleless syringe.During this process, the leaf was pushed against the syringe by the lab technician, who used a finger from their free hand to apply enough pressure for the inoculum to penetrate the intercellular spaces of the leaf blade.Five replicates were carried out per isolate, per cassava cultivar.The cultivars used here were BRS Formosa, BRS Kiriris, and BRS Aramaris, previously described in the literature as moderately susceptible, susceptible, and moderately resistant to CBB, respectively (Teixeira et al., 2021).
After inoculation, the plants were bagged for 24 hours in order to facilitate pathogen penetration and colonization.The experimental design was three randomized complete blocks with 55 plants per block of each variety, totaling 165 plants.At the beginning of the experiment, the plants were 12 weeks old and about 40 cm in height.Assessments were performed every five days.After eight days, symptoms appeared in some plants within the replicates.Subsequently, every three days, new evaluations were conducted from the absence of symptoms until plant death.
For the BOX-PCR reactions, the BOX A 1 R primer was used (5' CTA CGG CAA GGC GAC GCT GAC G 3').The reaction was composed of the following reagents: 10X buffer, 50 mmol L -1 MgCl 2 , 0.2 mmol L -1 dNTPs, 0.2 primer, 1U Taq DNA polymerase, 3.0 µL DNA, and ultrapure water to a final volume of 18 µL.The samples were amplified in a thermocycler, with the following amplification program: 95°C for 2 min, followed by 39 cycles of 94°C for 1 min for denaturation, 51°C for 1 min for annealing, and 72°C for 1 min for extension, ending at a final extension at 72°C for 10 min.PCR products were analyzed on 1% agarose gel in 0.5X TBE buffer by electrophoresis conducted at 70-80V for approximately 4 hours.
For the analysis of the ERIC-PCR primers (ERIC1R and ERIC2F), the same abovementioned composition of BOX-PCR reagents and amplification program, but at an annealing temperature of 48°C, were used.For the PCR reactions of the VNTRs, the final concentrations of the reagents used were: 3.0 µL DNA (5.0ng µL -1 ), 1X Taq buffer, 2.0 mmol L -1 MgCl 2 , 0.2 mmol L -1 dNTPs, and 0.005 mmol L -1 primers for five additional Xpmspecific VNTR loci (XaG1_02, XaG1_29, XaG1_67, XaG1_52, and XaG1_73) (Nakamura et al., 1987).The program used for amplification was 95°C for 4 min, followed by 39 cycles of 94°C for 50 s for denaturation, Δ°C (according to each primer pair, which ranged from 52 to 55°C) for 50 s for annealing, and 72°C for 1 min for extension.Final extension was performed at 72°C for 10 min.PCR products were also analyzed on 1% agarose gel, in 0.5X TBE buffer, by electrophoresis carried out at 70-80V for approximately 4 hours.The primer sequences used in the molecular study of the isolates are presented in Table 1.
The band profiles generated from the BOX-PCR, ERIC-PCR, and VNTR markers were used to construct binary matrices, in which 0 = absence of bands and 1 = presence of bands, and were converted into dissimilarity matrices.Dendrograms for BOX-PCR and ERIC-PCR were constructed based on the Jaccard coefficient, using the unweighted-pair group method with arithmetic mean clustering analysis.
Data from the binary matrix were imported into the POPGENE 1.31 software (Yeh et al., 1999) to perform the combined analysis, where the following diversity parameters were estimated: number of effective alleles, number of observed alleles, Nei's gene diversity (Nei 1973), Shannon diversity index (Lewontin, 1972), number of polymorphic loci, percentage of polymorphic loci (PPL%), allelic diversity, total genetic diversity, genetic diversity of the subpopulation in relation to the total population (G ST ) as in Nei (1973), and gene flow.Genetic similarity and genetic distance between populations were also computed using the model presented in Nei (1978).The analysis of molecular variance (AMOVA) was used to reveal the distribution of genetic diversity within and between populations.In this analysis, the genetic diversity level (F ST ) for the estimate of the genetic structure was calculated, as well as total genetic diversity, which was divided into two distinct hierarchical levels: difference between populations and between individuals within a population.The AMOVA was performed according to Excoffier et al. (2005), using the Arlequin software, version 3.5.2.2.The significance of differentiation was tested with 1,000 permutations, where p denotes the probability of obtaining a random value equal to or greater than the observed value.
The minimum spanning network analysis to represent the genetic distances, as well as the connectivity among haplotypes, was also calculated using the vegan and poppr packages of the R software (R Core Team, 2018).Correlations between the distance matrices generated by markers BOX-PCR + ERIC-PCR, BOX-PCR + VNTR, and ERIC-PCR + VNTR were performed using the Mantel test in the ade4 package, also of the R software.

Results and Discussion
A total of 468 isolates collected in the different producing regions in the Brazilian states were associated with CBB (Figure 1).The expected target fragment of 900 bp was detected in the DNA from 17% (80) of these isolates when subjected to amplification with primers XV/XKmod, confirming the identity of the pathogen (Table 2).The distribution of the presence and absence of symptoms, as well as leaf drop, differed significantly between the varieties by the chi-square test, at 5% probability.After 8 days after inoculation, angular lesions became evident, with necrosis 13 days later, followed by leaf drying and drop.
According to the dendrogram analysis of the Xpm isolates differentiated with the BOX-PCR, ERIC-PCR, and VNTR markers, the isolates clustered by the BOX-PCR and ERIC-PCR markers were divided into seven groups (G1-G7), as shown in Figure 2 A. The isolates from the municipalities of Laje and Ocauçu, in the states of Bahia and São Paulo, respectively, were genetically close according to this marker, which may be due to the shared alleles, indicating a same origin, with migration caused by the exchange of contaminated cuttings by farmers.The data obtained with the ERIC-PCR marker also formed seven groups (G1-G7) (Figure 2 B).However, the isolates from the same population that were separated with BOX-PCR were now grouped in this dendrogram.An intra-group discriminatory capacity was also observed, as the isolates from Guanambi, in the state of Bahia, were separated into two distinct groups despite being in the same node.The VNTR markers also showed a similar pattern to those of the BOX-PCR and ERIC-PCR markers (Figure 2 C), even though they are found in less conserved regions of the genome.In addition, the isolates from Ocauçu and Laje, in the states of São Paulo and Bahia, respectively, analyzed using the VNTR markers showed some type of grouping by the BOX-PCR method.This is an indicative that, despite the geographic separation, the isolates have some genetic similarities that remained unchanged throughout their evolutionary trajectory.Overall, the clusters were similar considering the BOX-PCR, ERIC-PCR, and VNTR molecular techniques (Figure 2).However, the ERIC-PCR primers best separated the groups by population, showing the highest similarity between the isolates from the same populations (Figure 2 B).In general, according to the dendrograms, ERIC-PCR was the marker that best separated the groups by population and presented a higher similarity between the isolates of the same population.
The minimum spanning network (MSN) for the population of the studied isolates according to the BOX-PCR, ERIC-PCR, and VNTR markers and their combination is shown in Figure 3. Regarding the MSN, for the BOX-PCR marker, isolates from Russas, Paranavaí, and Laje in the states of Ceará, Paraná, and Bahia, respectively, share the same genetic profile, with smaller genetic distances (Figure 3 A).Since the isolates were basically distributed according to their own populations, the minimal genetic distance between them was smaller, showing that, for this marker, diversity was lower between and within the populations due to the similar genetic profiles, with less unique haplotypes.
For the ERIC-PCR markers, the MSN revealed that the most frequent haplotypes were from the states of Bahia, São Paulo, and Paraná, forming the largest  3 B).The minimal genetic distance between the isolates was larger between two isolates from Bahia (180 and 184), collected in the same location.
The MSN from the VNTRs had most ramifications (Figure 3 C); however, the distance between the genotypes was low compared with the network generated by the ERIC-PCR primer (Figure 3 B).Although the isolates were different (unique haplotypes), they were closer within each specific population.Moreover, the VNTR markers best discriminated the isolates (Figure 3 C), also showing the variability between these and the ERIC-PCR markers that best distanced the populations (Figure 3 B).
When the combined data of the BOX-PCR, ERIC-PCR, and VNTRs markers was analyzed (Figure 3 D), the populations became very distant and isolates were not shown, since the data used in minimum spanning was selected from the common isolate among the three markers.In this case, the BOX-PCR primer showed the lowest similarity between the groups.
The Mantel test revealed a low positive correlation of 0.19 between the BOX-PCR and VNTR molecular markers.In contrast, there was a positive and relatively high correlation of 0.52 between the ERIC-PCR and VNTR markers and of 0.68 between BOX-PCR and ERIC-PCR.
Considering the analysis of genetic diversity in the populations of Xpm, those with the highest contrast in PPL and N a were observed in the states of Bahia (PPL = 93.33 % and N a = 1.9333) and Ceará (PPL = 62.22% and N a =1.6222) (Table 3).These two populations also presented the highest values of 0.2263 and 0.3686 in Bahia and 0.1826 and 0.2857 in Ceará, respectively, ID#, identification number; +, presence of amplification for the primer/marker; -, absence of amplification for the primer/marker; and NR, not carried out. (2)Municipalities in the Brazilian states of: PR, Paraná; MS, Mato Grosso do Sul; BA, Bahia; SP, São Paulo; and CE, Ceará.similarity was 0.977 between the populations of Ceará and Bahia, and the minimum estimated value was 0.845 between the populations of Mato Grosso do Sul and São Paulo.In contrast, the maximum observed value of genetic distance was 0.069 between the populations of São Paulo and Ceará, whereas the lowest was 0.024 between the populations of Bahia and Ceará.
Genetic variation was estimated among all populations by the G ST parameter, which was 0.1468 for the total population, indicating a moderate genetic differentiation between the subpopulations.The G ST value was used to calculate the number of migrants (N m ), which was equal to 2.9068 for the total population (Table 5).This value could explain the low Pesq.agropec.bras., Brasília, v.58, e03299, 2023 DOI: 10.1590/S1678-3921.pab2023.v58.03299diversity between populations, in alignment with the results obtained with the highest diversity within and not between populations, as also found by Trujillo et al. (2014) among isolates from the Caribbean region of Colombian.This result may be related to the genetic composition in common between the primary inoculum of the sampled areas where the low levels of interpopulation genetic diversity occur, which suggests a continuous gene flow between populations (Goodwin et al., 1993), an important factor in the structure of populations.According to Wright (1949), when N m > 1.0, that is, when one or more individuals migrate per generation, the effects of migration are sufficient to counteract the effects of drift, meaning that the number of migrants per generation prevents divergence among populations.
The AMOVA indicated that 8.83% of total variance is due to differences between populations and 91.17% within populations, showing that the greater genetic differentiation is in the intrapopulation component rather than in the interpopulation component (Table 6).
Therefore, the obtained results show how diversity is distributed in space, how quickly a population of Xpm can evolve, and which factors contribute to change.In the literature, Xpm populations have been characterized using many different DNA markers (Restrepo et al., 2000;Zárate-Chaves et al., 2021b;Díaz-Tatis et al., 2022;Rache et al., 2023).From 1995 to 2000, several prospection studies were conducted resulting in the collection of 906 isolates from Colombia, Venezuela, and Brazil.The characterization of these isolates resulted in the description of 111 haplotypes (Restrepo & Verdier, 1997;Restrepo et al., 2000Restrepo et al., , 2004)).In addition, researches involving molecular characterization have shown that a new population structure can be established in a single harvest cycle (Restrepo et al., 2000(Restrepo et al., , 2004;;Verdier et al., 2004).In the present study, genetic diversity may be associated with rapid changes and the local adaptation of the pathogen since Xpm populations are unstable and can change rapidly in less than a year (Verdier et al., 2004).
High levels of genetic variability in populations of phytopathogens may allow them to adapt to different environments and genotypes of newly introduced resistant hosts (Churchill, 2011).Trujillo et al. (2014), characterizing Xpm populations from the Colombian Caribbean coast using AFLP and VNTRs markers, reported that 80% of genetic variation occurred within populations, which was attributed to the geographic characteristics of the origin of the isolates from each sampled population.Rache et al. (2019), using an improved multiple loci variable number of tandem repeat scheme that targeted 15 VNTR loci (MLVA-15), were able to distinguish 88.9% of haplotypes within the Xpm strain from the Caribbean region of Colombia.Bart et al. (2012) carried out a high-throughput sequencing and identification of effectors to target durable resistance to CBB of temporarily diverse Xam strains from 65 different geographic locations.The authors verified the phylogeny of Brazilian Xam strains, showing a distinct clade that shares a common ancestor with both Colombia and African clades,  Restrepo et al. (2004), when evaluating strains of Xpm from six locations spanning four different edaphoclimatic zones in Colombia from 1995 to 1999, identified 45 different Xam RFLP types or haplotypes.Additionally, the authors concluded that the presence of identical RFLP patterns in several fields of the same or different edaphoclimatic zone indicated pathogen migration.
In the present study, the populations of Xpm from different regions of Brazil resulted in a clonal population structure for each region according to the host and edaphoclimatic conditions regardless of the region.Therefore, it can be assumed that Xpm populations are structurally organized at the regional level of the country, which could be because producers use endemic cassava species in each region, which added to the edaphoclimatic conditions and selection pressure of Xpm populations (Teixeira et al., 2021).Contrastingly, Lelis et al. (2014), evaluating the diversity of 85 Xpm isolates collected in the main producing regions using the AFLP technique, found no correlation between geographic origin and pathogenicity.
Therefore, resistant cassava varieties should be introduced in each region according to the degree of diversity, structure, and aggressiveness of the pathogen.Despite the difference in the number of samples, it is evident that the VNTR markers provide congruent results for populations of Xpm.It is worth noting that this is the first study using VNTR markers in Xmp populations from Brazil.The greater intrapopulation diversity observed indicates the urgent need for adopting resistant varieties and measures to minimize the advancement of the pathogen, as, for example, the use of pathogen-free cuttings.
All this information is of paramount importance for establishing disease management strategies.For this, germplasm with a specific set of Xpm strains that represents the range of genetic diversity within a production area should be evaluated.Furthermore, the variability of Xpm populations and migrations should be considered to improve quarantine measures in order to prevent the importation of infected planting material to areas with a low incidence of the disease.

Conclusions
1. ERIC-PCR and VNTRs are the best markers for assessing genetic variability in eight Xanthomonas phaseoli pv.manihotis (Xpm) populations from five cassava (Manihot esculenta) producing states in Brazil, and ERIC-PCR is the marker that best separated the groups by population and presented a higher similarity between the isolates of the same population.
2. The observed selection pressure, together with the mode of reproduction and mechanisms that increase genetic variability, allows of the pathogen populations to adapt according to the variation of microclimates, contributing to a differentiated reproductive success.

Figure 1 .
Figure 1.Map of Brazil showing the geographical locations (municipalities) where the cassava (Manihot esculenta) leaf samples with cassava bacterial blight symptoms were originally collected and the number of total isolates obtained per location (n).The locations were: the municipality of Russas in the state of Ceará (CE), the municipalities of Laje and Guanambi in the state of Bahia (BA), the municipality of Ocauçu in the state of São Paulo (SP), the municipalities of Naviraí and Dourados in the state of Mato Grosso do Sul (MS), and the municipalities of Paranavaí and Londrina in the state of Paraná (PR).Orange, negative samples for primer XV and Xkmod amplification; and blue, positive samples for XV and Xkmod amplification.

Figure 2 .
Figure 2. Cluster analysis of 50 isolates of Xanthomonas phaseoli pv.manihotis using the BOX-PCR (A), ERIC-PCR (B), and VNTR (C) markers, based on the Jaccard similarity coefficient and dendrogram construction by the unweighted pair group method using arithmetic means.The cut-off threshold is based on Mingoti (2005).G1-G7, seven formed groups.The isolates were obtained from populations from the following Brazilian states: BA, Bahia; SP, São Paulo; PR, Paraná; MS, Mato Grosso do Sul; and CE, Ceará.

Figure 3 .
Figure 3. Minimum spanning network formed by markers BOX-PCR (A), ERIC-PCR (B), VNTRs (C), and their combination (D), showing the relationship between individual genotypes with multiple loci (MLGs) observed in the Xanthomonas phaseoli pv.manihotis populations.Each node represents a different MLG.The size and colors of the nodes correspond to the number of individuals and members of the population, respectively.The thickness and color of the border are proportional to the absolute genetic distance.The lengths of the edges are arbitrary.Brazilian states where the populations were obtained from: BA, Bahia; SP, São Paulo; PR, Paraná; MS, Mato Grosso do Sul; and CE, Ceará.

Table 1 .
Markers used for the molecular analysis.

Table 2 .
Identified isolates associated with cassava bacterial blight, origin of the cassava (Manihot esculenta) variety, and respective collection sites used in the pathogenicity test and in the analysis of genetic diversity(1).
groups, probably ancestral haplotypes (Figure

Table 3 .
(Bart et al., 2012)s in five populations of Xanthomonas phaseoli pv.manihotis(1). the occurrence of a global movement and the need for more globally effective resistance strategies(Bart et al., 2012).

Table 5 .
Estimate of genetic diversity, population differentiation (G ST ), and gene flow estimate for populations of Xanthomonas phaseoli pv.manihotis(1). 1) H T , total diversity; and H S , expected diversity.Number between parentheses represents standard deviation. (

Table 6 .
Analysis of molecular variance in populations (pop.) of Xanthomonas phaseoli pv.manihotis(1). 1) DF, degrees of freedom, with probabilities calculated by 1,023 random permutations; and F ST , level of genetic diversity. (