In Mexico, roselle (Hibiscus sabdariffa) is an economically important crop; it is cultivated for domestic consumption and for export. The state of Guerrero is the major roselle producer with 14,294 hectares cultivated, which represent approximately 70% of the national area (^{SIAP, 2019}).

In recent years, the disease known as spotted of roselle, induced by the fungus Corynespora cassiicola, has reached 100% incidence and caused the total loss of the crop (^{Ortega-Acosta et al., 2015}; Ortega-Acosta et al., 2019) with severity values higher than 50% for leaves and calyces; at this level of severity, the commercial product (calyces) is usually discarded thus causing economic losses to the producers (^{Ortega-Acosta et al., 2019}; ^{Ortega-Acosta et al., 2020a}). In addition to roselle, C. cassiicola causes severe diseases in diverse crops such as cotton (Gossypium hirsutum), soybean (Glycine max), rubber tree (Hevea brasiliensis), among others, and, worldwide, it has been isolated from more than 530 plant species (^{Nghia et al., 2008}; ^{Dixon et al., 2009}; ^{Fortunato et al., 2015}).

Currently, conventional and alternative agrochemicals are used for controlling C. cassiicola (^{Vawdrey et al., 2008}; ^{Ortega-Acosta et al., 2019}; ^{Zhu et al., 2020}). However, despite their effectiveness, the frequent use of these substances can cause serious environment and health issues (^{Hahn 2014}; ^{Ghosal and Hati, 2019}; ^{Weber and Hahn, 2019}). For this reason, one of the alternatives for sustainably controlling plant diseases is the biological control with antagonistic microorganisms in order to limit the development of pathogens or reduce the severity of crop damages (^{Blakeman and Fokkema, 1982}; ^{Köhl et al., 2019}).

In previous studies about biological control to reduce disease damages caused by C. cassicola, bacteria such as Pseudomonas flourescens, Bacillus subtilis, B. cereus and B. thuringiensis have been evaluated in cucumber, tomato and rubber trees (^{Romeiro et al., 2010}; ^{Rahman et al., 2010}; ^{Manju et al., 2019}, ^{Köhl et al., 2019}). No studies about the use of biological controls for spotted of roselle caused by the fungus were found. Therefore, the objectives of this study were to determine the ability of bacterial strains to inhibit C. cassiicola growth in vitro and evaluate their control in leaves and calyces under biospace conditions.

For the study, 100 bacterial strains from the biobank of the Molecular Microbiology and Environment Biotechnology Laboratory, Autonomous University of Guerrero, were used; the strains were isolated from diverse environments: food (20), amphibian skin (41), tailings (22), air (7) and rhizosphere (10). For in vitro studies, the C. cassiicola CCHFR strain (GenBank: KM207768), reported by ^{Ortega-Acosta et al. (2015}), was used. All the bacterial strains were used in antagonistic in vitro tests based on dual confrontations to select only those able to inhibit the growth of the CCHFR strain; the bacteria were identified through micro- and macroscopic morphology using the Vitek2^® automated system (^{Vargas et al., 2005}) for their subsequent evaluation in biological control essays of spotted of roselle under biospace conditions.

The essay to determine the inhibition percent of mycelial development in vitro consisted of a mycelial fragment 1 cm² in size obtained from a seven-day culture of C. cassiicola (CCHFR), which was placed in the middle of a Petri dish containing agar PDA culture medium (Potato- Dextrose-Agar Difco^®); then, the bacteria were inoculated using 6 mm sterile filter paper disks impregnated with 20 µL of a bacterial suspension at a density of 1×10⁸ UFC mL^-1, placed at each cardinal point at a distance of 2.5 cm from the fungus (^{Petatán-Sagahón et al., 2011}) and incubated at 25±2 °C for 21 days; disks with sterile distilled water were used as the negative control. The essays in triplicate were applied only to strains with >50% inhibition. Photographs of each essay were taken, and the total area of the mycelial development was obtained using the ImageJ v.1.8.0 software (^{Cuervo-Parral et al., 2011}).

The percent inhibition of the mycelial growth was calculated using the following formula:

Where Dc: Mycelial growth of the control plates, and Dt: Mycelial growth in the Petri dishes treated (^{Durairaj et al., 2018}).

Since in a preliminary essay under biospace conditions, the CCHFR strain had a low level of pathogenicity (data not shown), a sampling in the field was conducted at El Pericón, municipality of Tecoanapa, Guerrero, where roselle leaves with spotted of roselle symptoms were collected in September 2019, following the methodology described by ^{Ortega-Acosta et al. (2015}); a pathogenically active C. cassiicola strain called CCPER4 was isolated and identified.

The vegetal material used consisted of three-month-old roselle plants of a “Creole” variety, which were sown in polyethylene pots (15×25 cm) using a mixture of soil and compost (1:1) as substrate. The plants were fertilized twice using a 45-30-20 of NPK formulation (^{Alejo, 2017}). No chemical products were applied for controlling pests and diseases. To evaluate the biological control of C. cassiicola (CCPER4), only the seven bacteria which inhibited ≥50% mycelial development were used; each strain was considered as a treatment with six replications; each replication consisted of four pots with roselle plants (2 plants/pot) arranged in a completely randomized design, and the experiment unit consisted of two central/replication pots. For inoculation, a bacterial suspension of 1×10⁸ UFC mL^-1 was prepared and sprayed on three-month-old plants using a manual sprayer on the point of dripping. After 24 h, a suspension of 2×10⁵ conidia mL^-1 of C. cassiicola (CCPER4) was prepared and inoculated using a manual sprayer; all the plants were covered with transparent polyethylene bags for 72 h. For the positive control, roselle plants were sprayed only with C. cassiicola (CCPER4) conidia.

To estimate the severity of the disease induced by C. cassiicola (CCPER4) in each treatment, the plants were stratified in three levels (low, intermediate and high) (^{Villanueva-Couoh et al., 2005}), and then four leaves or calyces per stratum were evaluated. To estimate the extent of disease severity, two diagrammatic scales were used for leaves: 0=0, 1= (>0-2 to 4), 2=(>4-7 to 12), 3= (>12-19 to 29), 4= (>29-42 to 57) and 5=(>57-70 to ≤100), and for calyces: 0=0, 1=(>0-3 to 5), 2=(>5-10 to 18), 3=(>18-30 to 46), 4=(>46-63 to 77) and 5=(>77-87 to ≤100) (^{Ortega-Acosta et al., 2016}). The estimation of the severity was evaluated from October to November, for leaves, and from November to December for calyces. Four evaluations were made to leaves and calyces at six-day intervals, according to the phenological stage. The temperature and relative humidity within the biospace were estimated using an Extech^® RHT10 digital hygrothermograph.

The data obtained from the antagonism in vitro tests were subjected to an analysis of variance (ANOVA) and a separation of means (Tukey, p=0.01). Using the estimated data of severity in roselle leaves and calyces, the Area Under the Disease Progress Curve (AUDPC) per treatment was estimated based on the method described by ^{Campbell and Madden (1990}), followed by an analysis of variance and a separation of means (Tukey, p=0.01). Statistical analyses were performed using SAS v 9.4 statistical software.

Of the 100 strains evaluated, seven that inhibited ≥50% of C. cassiicola (CCHFR) were selected. Of these, the strain Serratia marcescens (M13ACD) showed 100% inhibition (Figure 1, Table 1), and the strains of Acinetobacter lwoffii (A5), Klebsiella pneumoniae (M10-1 y M10-10), Sphingomonas paucimobilis (NF21), Serratia liquefaciens (M8ACD) and Acinetobacter sp. (5H2) had statistical differences compared to the control; these strains were selected for control essays under biospace conditions (Table 1).

The CCPER4 strain produced dense, compact and black-greyish mycelium, and straight-to-curve conidia with short light-brown pseudosepta that measured 57.0-308.2 x 10-20 µm (length x width, respectively), which are distinctive characteristics of C. cassiicola (^{Qi et al., 2011}; ^{Ortega-Acosta et al., 2020b}).

The artificial inoculation of C. cassiicola (CCPER4) induced the spotted of roselle disease under biospace conditions. The estimated severity values of AUDPC for spots on roselle leaves and calyces allowed to determine an effect on disease reduction in all the bacterial treatments compared to the control treatment.

However, the effect was differential on leaves depending on the inoculated bacteria (p=0.01), while in calyces, all showed the same effect on the decrease of severity compared to the control treatment (p=0.01). For example, in leaves, the treatments inoculated with the strains M10-1 and M10-10 (K. pneumoniae) had the lowest severity values with 9.5 and 12.6 of AUDPC, while the strains NF21 (S. paucimobilis), A5 (A. lwoffii), 5H2 (Acinetobacter sp.), M13ACD (S. marcescens) and M8ACD (S. liquefaciens) reduced the severity of AUDPC with values of 14.5 to 30.9, and the control treatment had the greatest severity with 42.1 of AUDPC (Table 1).

Figure 1. Antagonistic activity in PDA culture medium after 21 days in a culture medium at 27±2 °C. A= Serratia marcescens strain (M13ACD) in C. cassiicola. B= Control treatment (C. cassiicola- CCHFR). 

On the other hand, in calyces, there was a significant decrease in severity in all the bacterial strains with AUDPC values between 8.6 and 17.1 and were statistically similar but different compared to the control with 42.1 of AUDPC (Table 1). The temperature and relative humidity conditions during the evaluations were 28 °C and 67%, respectively.

Based on the evaluation of the 100 bacterial strains, seven had an effect on inhibiting the development of C. cassiicola (CCHFR) in vitro, and significantly reduced the severity of spotted of roselle under biospace conditions. The results indicate that the studied bacteria have the potential to be evaluated in further studies as biological control agents of spotted of roselle under field conditions and inoculum natural pressure.

Under in vitro conditions, S. liquefaciens (M8ACD) inhibited 68% of C. cassiicola (CCHFR) and 100% of S. marcescens (M13ACD), and was the best under in vitro conditions (Table 1); inhibition can be possibly due to the surfactants (serrawatina type) produced by the strains (^{Shekhar et al., 2015}).

Regarding severity on leaves, S. marcescens (M13ACD) and S. liquefaciens (M8ACD) had 30.7 and 30.9 of ABCPE, respectively. These results suggest that the Serratia spp. strains had a better effect in vitro and that this result is similar to that reported by ^{Sabu et al. (2017}), who stated that the mentioned genus is able to control Pythium myriotylum in vitro, as well as other pathogens in ginger crops (Zingiber officinale). However, the effect exerted under greenhouse conditions was significantly lower compared to the other strains used (^{Mustafa et al., 2019}).

Regarding K. pneumoniae strains (M10-1 and M10-10), these were the species with the greatest ability to reduce severity on leaves (Table 1), but in calyces the effect was similar to that of the other bacterial isolates. In this regard,, ^{Dey et al. (2019}) evaluated the effect of a Klebsiella sp. strain on in vitro and in vivo control of root rot in black bean crops (Vigna mungo), and determined a good control of the disease, and for this reason, the authors recommend its use as a biocontrol agent.

On the other hand, A. lwoffii (A5) inhibited 62.7% of C. cassiicola growth in vitro, which had an effect similar to that exerted by K. pneumoniae strains. However, A. lwoffii reduced the disease severity in leaves with 18.8 of AUDPC, which was statistically similar to the effect exerted by strains of K. pneumoniae (M10-1 and M10-10), S. marcescens (M13ACD) and S. liquefaciens (M8ACD). In calyces, the severity caused by A. lwoffii showed 9.1 of AUDPC, which was similar to the other bacterial treatments (Table 1). In this regard, ^{Trotel-Aziz et al. (2008}) mentioned that this bacterial genus is able to induce resistance in grapevine plants (Vitis vinifera) and that this is a way to prevent Botrytis cinerea colonization and, consequently, reduces damage. Under in vitro conditions, S. paucimobilis (NF21) showed 80.9% inhibition. Regarding severity in leaves and calyces, the ABCPE values were 14.5 and 13.1, respectively. ^{Medina-De la Rosa et al. (2016}) mentioned that this species had an antagonistic effect on Fusarium sp.

The Acinetobacter sp. (5H2) strain inhibited 69.4% of C. cassiicola and the AUDPC value for severity in leaves was 26.9, which was similar to that of S. liquefaciens (M8ACD); the AUDPC value for severity in calyces was 13.6, which was statistically similar to the other bacterial treatments. These results are in agreement with those reported by other authors, who mention that the use of bacteria exerts control on diseases caused by C. cassiicola (^{Romeiro et al. 2010}; ^{Manju et al., 2019}).

No reports of the bacterial species evaluated as biological control of C. cassiicola agents were found, and, therefore, this research shows the first evidences of biological control of C. cassiicola, the causal agent of spotted of roselle in Guerrero, by using and applying bacterial species such as K. pneumoniae, A. lwoffii, S. paucimobilis, S. marcescens and S. liquefaciens, isolated from diverse environments. For this reason, future studies could focus on the evaluation of these bacteria under inoculum natural pressure of the spotted of roselle disease, and also integrate other sustainable control strategies such as the removal of weed reservoirs of C. cassiicola (^{Hernández-Morales et al., 2018}), rotation with chemical alternatives (^{Ortega-Acosta et al., 2019}), among others, in order to decrease the impact of the damage caused by the pathogen in roselle cropping.

For the study, seven bacteria were isolated from air, tailings, and amphibian skin with in vitro antagonistic potential against C. cassiicola, the causal agent of spotted of roselle. The strains of M13ACD (S. marcescens) and NF21 (S. paucimobilis) were the most effective to inhibit the development of C. cassiicola in vitro with 100 and 80.9% inhibition, respectively. Under biospace conditions, M10-1 and M10-10 (K. pneumoniae), were the most effective in reducing the severity of spotted of roselle with AUDPC values of 9.5 and 12.6, compared to the control treatment (45.7 of AUDPC). In calyces, the seven strains selected decreased the levels of disease severity with AUDPC values of 8.6 to 17.1, while the control treatment had the highest levels of severity with an AUDPC value of 42.1.