The Orchidaceae family is one of the most diverse plant groups. Some of its most outstanding uses are ornamental, aromatic, medicinal, for craftwork and edible. In the in vitro planting process, there are endophytic microorganisms, which contaminate the culture medium. They are commonly found naturally inside the plants, inducing diverse effects, such as the stimulation of seed germination and/or in plant growth (^{Tsavkelova, 2011}; ^{Wilkinson et al., 1989}), yet they can be devastating when the plant tissues are planted in vitro. The search for new alternatives to prevent and/or eradicate the contaminants is an important task (^{Cruz et al., 2006}).

The plant tissue explants may carry contaminants in their surface and/or insides. Those which are carried on the surface can be eliminated by disinfestation, but those found inside the plant are difficult to eradicate, which makes it difficult to meet the basic requirements for the success of micropropagation of any plant species, which is to keep crops free of contaminant microorganisms (^{Ramírez-Villalobos et al., 2000}). However, the microorganisms may play an important role as antagonists of pathogenic agents in plants (^{Ocegeda-Reyes et al., 2020}), although the results of in vitro trials may not necessarily reflect what takes place in vivo (^{Whitaker and Bakker, 2019}).

The presence of microorganisms during the in vitro planting of plant tissues takes place, mostly, when the donating plant grows directly on the field and is exposed to pests, diseases, dust and other agents (^{Ramírez-Villalobos and Salazar 1997}). This phenomenon is also caused by the anatomical characteristics of each species, such as the presence of trichomes on leaves and stems, which stop disinfectants from penetrating. The elimination of these microorganisms may be due to the use of inadequate aseptic techniques in the laboratory (Alvarado, 1998).

Endophytic bacteria are usually difficult to detect, since they remain inside the tissue of the host, and because they are not phytopathogenic, they may go unnoticed, with low rates of multiplication and rooting of plants, and may even lead to their deaths. The presence of microorganisms in the culture medium is also due to the inefficient disinfestation or sterilization, which can be improved with the adequate handling of equipment (^{Red and Tanprasert, 1995}). Based on the above, this investigation had the aims of identifying endophytic bacteria present in Prosthechea citrina, as well as to evaluate antibiotics for their elimination.

Young Prosthechea citrina plants were gathered before flowering in two locations of a pine-oak forest, with the coordinates 17° 20’ 012” N, 96° 30’ 654” W and 17° 19’ 890” N, 96° 30’ 774” W, for sites 1 and 2, respectively, at 2104 masl, in Ixtlán de Juárez, in the Norther Sierra of Oaxaca. The disinfestation began by rinsing with distilled water. The membranous bracts were removed from the bulbs, which were then rinsed, under aseptic conditions, with 70% alcohol for one minute, followed by three rinses with sterile distilled water. They were finally submerged in a 1.5% sodium hypochlorite solution for 30 min and rinsed again, three times, with sterile distilled water.

After the disinfestation, each plant was dissected into pseudobulb, base, midsection and leaf apices, in order to identify possible specific plant organs in which microorganisms could be located. To carry out the isolation, an inoculation loop was rubbed on the internal plat tissue, and planted, using cross-streaking, on plates with an LB medium (Luria Bertani: casein peptone 10 g L^-1, yeast extract 5 g L^-1, NaCl 10 g L^-1 and bacteriological agar 15 g L^-1) (^{Bertani, 1951}), and incubated at 37 °C for 48 h. After this period, the remaining bacterial cultures were purified.

For the non-molecular identification, biochemical tests were carried out following the guidelines by ^{Schaad et al. (2001}). In order to separate the Gram positive and negative bacteria, we performed the Ryu test (Schaad et al., 2001). A drop of 3% potassium hydroxide was placed on a microscope slide and mixed with a bacterial culture. If the mixture formed a thread, there were Gram-negative bacteria, and if the mixture formed no threads, they were Gram-positive. Regarding cell morphology, a drop of sterile distilled water was placed on a microscope slide and mixed with small amounts of bacteria, the bacteria were fixed by burning the slide, and after the water evaporated, a drop of methyl violet at 0.1 N was added and observed under an optical microscope at different magnifications. To determine mobility, after mixing a bacteria sample in a drop of water, a slide was carefully placed on top and observed under an optical microscope. To determine the production of endospores, a bacterial culture was suspended in a test tube with sterile distilled water and placed in a water bath for 30 min. Next an aliquot of the suspension was taken and planted in a Petri dish with LB medium, in order to detect if the bacteria formed endospores. To determine if the bacteria had a strict or facultative aerobic metabolism, the ^{Hugh and Leifson medium (1953}) was used. The bacterium was inoculated by puncturing and incubated under conditions of aerobiosis (without sterile mineral oil) and anaerobiosis (with sterile mineral oil). For the oxidase test, the bacteria were exposed to the reagent N,n,n,n-tetramethyl-p-phenylendiamine (or TMFD) or N,N-Dimethyl-p-phenylendiamine (or DMFD), which change color indicating that the bacteria is aerobic, or transparent if it is anaerobic. A catalase test was also performed, mixing a bacterial suspension with hydrogen peroxide to observe whether the catalase would break it. If so, oxygen bubbles would form, indicating that the bacterium is aerobic.

Molecular identification began with the sequencing of 16 S ribosomal DNA, which was obtained following ^{Chye et al. (2013}). The extraction of bacterial DNA was carried out by following the instructions by the manufacturer of the DNA easy Plant Mini Kit, Qiagen^®, Hilden, Germany. The DNA quantification was carried out assuming that one absorbance unit at 260 nm is equal to 50 ng µL^-1 of double-chain DNA (Rickwood and Hames, 1990). The quality of the DNA was inferred by calculating the spectrophotometric ratio A260/A280 (Genesys 10 UV Scanning, Thermo scientific, Waltham, Massachusetts). The PCR program consisted of predenaturalization at 94 °C for 1 min, denaturalization at 94 °C for 30 s, aligning of forty 10 sec cycles at 95 °C, 30 s at 55 °C, and 30 s at 72 °C and the termination period at 72 °C for 10 min (Matson et al., 2015). The PCR products were sequenced in the MacroGen^® laboratory (10F, 254 Beotkkot-ro, Geumcheon-gu, Seoul 08511, South Korea). The sequences obtained from the region of gene 16S of the ribosomal RNA in this study were compared with sequences of reference organisms, using the BLAST database of the National Center for Biotechnology Information (NCBI, U.S.A.). The complementary of each sequence was obtained with the program CHROMAS. Later, using the program DNASTAR, the consensus sequence was obtained, and the sequences were finally run in the algorithm BLASTN of the NCBI, to determine the degree of homology they held in comparison with the sequences deposited in the database published in the GenBank.

In order to identify what can control the isolated bacteria strains, four antibiotics were evaluated using the disk-plate method: rifampicin (20 µg mL^-1), oxytetracycline (10 µg mL^-1), ampicillin (100 µg mL^-1) and tetracyclin (10 µg mL^-1). The diameters of the bacterial inhibition halos were measured after a growth period of 24 and 48 h. These data were treated using an analysis of variance, in a totally random design, of the four treatments with equal number of repetitions, considering one disk-plate as an experimental unit (ANAVA, SAS). The controls, without antibiotics, were not considered for the statistical analysis, since they served only as a reference of the growths, since they presented no inhibition halos in any of the incubation periods.

Isolations were obtained from different parts of the plant in order to identify possible tissue colonization specificities. Seven types of colonial bacteria were found, which were preliminarily identified using letters (Table 1) and underwent biochemical tests and sequencing for gene ARNr 16S (Table 2). The biochemical tests showed no differences between the isolations included in the tables; they were all Gram negative, bacilli with movement, with no production of endospores and positive for fermentation, oxidation and catalase tests (data not included), therefore they all seem to belong to one same species. However, the colonies of isolations A, B, C, D and E presented a yellowish-brown color, they were convex, bright and with a smooth surface and a foul smell, unlike the other two isolations (F and G), which helped identify the first as Aeromonas hydrophila. The colonies of isolations F and G displayed a beige color, they were convex, opaque in color and with an odorless, wavy edge, which was identified as belonging to the genus Enterobacter (^{Hugh and Leifson, 1953}; ^{Ramos et al., 2007}; ^{Schaad et al., 2001}). This is backed with the comparison of the homology of sequences of rRNA 16S with the GenBank database, since it indicated that isolations A, B, C, D and E corresponded to the species of Aeromonas hydrophila and colonies F and G, to Enterobacter sp (Table 2). After molecular identification was confirmed, the colonization tissue specificity was observed, since both genera and most of the isolations were obtained from the bulb and the base of the leaf, whereas Enterobacter was not isolated from the middle and apical parts of the plants (Table 1). The identification was adjusted with the shapes of the colonies. That is to say that the growths of cultures A, B, C, D and E are similar to each other, belonging to A. hydrophyla, yet different to F and G, which are also similar to each other and belong to Enterobacter.

Regarding the action of the antibiotics, oxytetracycline induced the highest inhibition mean in A. hydrophila, statistically different to the other treatments, followed by tetracycline, at both 24 and 48 h of exposure. Rifampin and ampicillin exerted a limited inhibition on the growth of A. hydrophila. In general, it is worth highlighting that a reduction was observed in the halos for all four antibiotics between 24 and 48 h. That is, the cultures continued growing despite the presence of the antibiotics. For Enterobacter sp., the highest means for inhibition were obtained with ampicillin, significantly different to the others, followed by tetracycline, oxytetracycline, and finally, rifampin. As with A. hydrophila, the inhibition halos decreased in diameter between 24 and 48 h (Table 3).

Reports on Aeromonas hydrophila as endophytic plant bacteria (^{Aytac and Gorris, 1994}; ^{Chye et al., 2013}; ^{Ginestrea et al., 2005}; ^{Pérez-Cordero et al., 2014}) are limited. In this investigation A. hydrophila is reported for the first time as associated to wild orchid Prosthechea citrina, without being pathogenic, since it was isolated from firm tissue, without lesions or decomposition. This association has been explained as a relationship in which A. hydrophila is able to solubilize phosphates (^{Muleta et al., 2013}) and/or exert biological control against phytopathogens (^{Gohel et al., 2006}). The genus Enterobacter has already been reported as associated to orchids and, as in this study, with tissue specificity (^{Fernándes et al., 2011}; ^{Ramos et al., 2007}). We conclude that two species of endophitic bacteria were found, with tissue specificity and which can be controlled with the use of antibiotics.