Streptomycin is used in plant protection against various PPB since the 1950s. The most common target of streptomycin is Erwinia amylovora, the causal agent of fire blight, which infects apple and pear trees, targeting leaves, flowers and shoots. Streptomycin is usually sprayed during bloom (McManus et al., 2002; Sundin and Wang, 2018).

Several mechanisms can be responsible for streptomycin resistance in PPB: (i) a point mutation of the rpsL gene, or (ii) the acquisition of strA-strB genes or aadA genes (Table 2). Other mechanisms have been suggested but they have not yet been confirmed and characterized. In E. amylovora, 95% of a total of 107 strains isolated in Mexico showed a mutation at codon 43 in rpsL, but the other 5% did not show rpsL mutation and did not carry strA-strB or aadA genes. The unknown resistance mechanism was not further investigated (de León Door et al., 2013). In Clavibacter michiganensis, another potential mechanism for streptomycin resistance was also suggested but needs further investigation (Lyu et al., 2019).

The rpsL gene encodes the ribosomal protein S12. A point mutation at codon 43 (mainly changing lysine (Lys) to arginine (Arg), rarely Lys to threonine or Lys to asparagine) prevents the binding of streptomycin to the ribosome. This mutation might be the most prevalent one since it retains a high environmental fitness even in the absence of streptomycin. Another less common mutation has been described at codon 88 (changing Lys to Arg). Both point mutations have been observed in various PPB (e.g., E. amylovora, C. michiganensis subsp. michiganensis, or in vitro in Xanthomonas oryzae pv. oryzicola or Erwinia carotovora, now named Pectobaterium carotovorum; Chiou and Jones, 1995b; Barnard et al., 2010; Zhang et al., 2011; Valenzuela et al., 2019; Escursell et al., 2021). In one streptomycin resistant strain of C. michiganensis, a point mutation at the 128th nucleotide of rpsL (Lys to Arg) was responsible for the resistance (Lyu et al., 2019). The rpsL chromosomal mutations might be the less worrying resistance mechanism because it is not prone to HGT.

Contrary to rpsL mutations, the acquisition of aadA or strA-strB genes requires more attention because of their presence on MGEs. Two alleles of aadA have been described, aadA1, the most prevalent allele, and aadA2. The aadA1 gene was found in isolates of X. oryzae pv. oryzae, whereas aadA2 conferred streptomycin resistance in a Pseudomonas strain (Schnabel and Jones, 1999; Xu et al., 2013). In X. oryzae pv. oryzae, three integrons carrying aadA1 have been described, which was the first report of resistance integrons in PPB (see sections 3.5 and 3.6; Xu et al., 2013).

The key actors in streptomycin resistance in PPB are the strA-strB genes, mainly found associated with Tn5393. This 6.7-kb transposon belongs to the Tn3 family. In its most simple forms, it encodes a putative transposase (TnpA) and a resolvase (TnpR), followed by a putative recombination site (res), an insertion sequence (IS) element (IS1133 in E. amylovora, IS6100 in Xanthomonas campestris), and by the strA-strB genes (Figure 2). These two genes are commonly encountered in human pathogens and are responsible for many infections associated with streptomycin resistant bacteria (Chiou and Jones, 1993; Sundin and Bender, 1995; Förster et al., 2015). They are widely disseminated among Gram-negative bacterial pathogens and commensals from humans, probably because of streptomycin use in clinical contexts. The strA-strB genes are usually found on small plasmids, such as RSF1010 in human pathogens, while they are most commonly encoded on large conjugative plasmids in plant pathogens (Förster et al., 2015).

The strA-strB genes were identified in E. amylovora on plasmid pEA8.7, a plasmid very similar to the broad-host-range plasmid RSF1010, a 8.7-kb IncQ plasmid involved in streptomycin and sulfonamide resistances in bacterial human infections. It is non-conjugative but can be transferred horizontally through mobilization. Another RSF1010-like plasmid has also been observed in Erwinia herbicola in New Zealand, suggesting that IncQ plasmids of the RSF1010 family could be more widespread than previously thought (Palmer et al., 1997). It is hypothesized that the strA-strB sequence on RSF1010 originated from the insertion of Tn5393 into the ancestral plasmid from which RSF1010 evolved. Indeed, the right inverted repeat of Tn5393 (located downstream of strB) is conserved downstream of strB in RSF1010. This IR sequence would be the only vestige of Tn5393 on RSF1010. The strA-strB genes carried by RSF1010 diverged over time from the intact strA-strB genes encoded on Tn5393 from plant pathogens (Sundin, 2000).

Tn5393 is found on a variety of plasmids (Table 3). It was initially discovered on the large conjugative plasmid pEa34 (34 kb) in E. amylovora (Chiou, 1991; Chiou and Jones, 1993). The spread of streptomycin resistance genes among E. amylovora or other bacteria (pathogenic or not) could thus be facilitated by this conjugative plasmid via HGT. Tn5393 was also found on plasmid pEa29 from E. amylovora, where it can insert at several locations. This 29-kb plasmid is, unlike pEa34, unable to self-transfer to other strains, and is not related to pEa34. pEa29 is highly stable in E. amylovora and thus the integration of Tn5393 suggests the stable establishment of the resistance in E. amylovora populations.

Variants of Tn5393 have been described (Figure 2; Table 3). Tn5393a was identified in several strains of Pseudomonas syringae and Pseudomonas marginalis, and lacks the IS1133 of Tn5393 (Sundin and Bender, 1993, 1995; Han et al., 2004). It was also detected in isolates of E. amylovora, located on the conjugative plasmid pEU30 (30 kb; Foster et al., 2004; Förster et al., 2015). Interestingly, the conjugative machinery used by pEU30 most resembles the VirB system from pPSR1 of P. syringae (Foster et al., 2004). It could be hypothesized that the Tn5393a element detected in E. amylovora originated from P. syringae. Tn5393b was identified in X. campestris and contains IS6100 instead of IS1133. The presence of an insertion element increases the level of resistance in comparison to Tn5393a, which contains none (Sundin and Bender, 1995). IS6100 in Tn5393 from X. campestris is 100% identical to an element found in Mycobacterium fortuitum, Pseudomonas aeruginosa and Flavobacterium sp., suggesting its presence outside of PPB (Sundin and Bender, 1995). Also, the tnpR and res of Tn5393, Tn5393a and Tn5393b from E. amylovora, P. syringae and X. campestris, respectively, are identical (Sundin and Bender, 1995).

Other Tn5393 variants have been found in bacteria not pathogenic to plants (Figure 2; Table 3). Tn5393c, reported in the fish pathogen Aeromonas salmonicida (L’Abée-Lund and Sørum, 2000), does not carry any insertion sequence, like Tn5393a, and sequence comparison revealed that Tn5393c and Tn5393a are essentially the same. The presence of Tn5393c in A. salmonicida is surprising because streptomycin is not used to control A. salmonicida in Norway, where the transposon was first described in this bacterium. However, huge amounts of other antibiotics (including oxytetracycline and sulfonamides) have been used and could have selected for the transposon located on pRAS2 which also carries oxytetracycline and sulfonamide resistance determinants (L’Abée-Lund and Sørum, 2000).

Tn5393d was identified in a clinical strain of Alcaligenes faecalis, which can be responsible for human infections related to contamination of the medical equipment (Mantengoli and Rossolini, 2005). Tn5393d carries additional antibiotic resistance genes and is found on pFL424, a 44 -kb non-conjugative plasmid. This transposon seems to have originated from the consecutive insertion of two composite transposons containing aphA6b (conferring resistance to kanamycin, streptomycin and amikacin) and blaPER-1 (PER-1 extended spectrum beta-lactamase) genes (Mantengoli and Rossolini, 2005).

Tn5393e is a variant containing another transposon, Tn6023, and was found on the IncHI2 plasmid pSRC125. It was recovered from a multi-resistant Salmonella enterica serovar Typhimurium isolate of bovine origin. Tn6023 itself contains the aphA1b gene, conferring resistance to various aminoglycosides. Surprisingly, strains containing this transposon did not show resistance to streptomycin, even though the strA-strB genes present on the element did not carry any inactivating mutations, suggesting that these genes were not expressed (Cain and Hall, 2011). Tn5393f and Tn5393g were discovered in unidentified soil bacteria where manure had been applied. Both transposons were found on low GC conjugative plasmids carrying other ARGs such as sul2 or aac3, and Acinetobacter sp. was identified as the putative host (Heuer et al., 2009; Cain and Hall, 2011).

Tn5393h is derived from Tn5393a and includes Tn10, a tetracycline resistance transposon, inserted in the tnpA gene, which likely annihilates its mobility. Tn5393i contains IS1133, but this insertion element is not at the same location as in Tn5393a (namely in tnpR) (Cain and Hall, 2011). A variety of Tn5393 variants sequences are also available in GenBank (Cain and Hall, 2011). Tn5393j from Klebsiella pneumoniae carries another aminoglycoside resistance gene, aphA6, but its tnpR was not functional and its tnpA gene is missing (Espedido et al., 2013). Two variants were named Tn5393k but, in fact, they are not identical; for the sake of clarity, they are referred to as Tn5393k and Tn5393k’ (Figure 2; Table 3). Tn5393k is similar to Tn5393c and Tn5393f, and carries both IS26 and IS1. It is found in S. enterica sv. Typhimurium. The terminal inverted repeats (IRs) of Tn5393k exhibit 100% sequence identity with those in Tn5393, Tn5393c and Tn5393f (Yao et al., 2021). Tn5393k’ on the other hand is extremely similar to Tn5393d; the only difference is that it is lacking the ISPpu17 element (Adamczuk and Dziewit, 2017). Tn5393l carries the tetracycline resistance genes located in Tn10 and is also found in S. enterica sv. Typhimurium (Zhang et al., 2019). Finally, Tn5393n was found in Aeromonas caviae (Luo et al., 2022).

Tn5393 was also found in Corynebacterium striatum, Campylobacter jejuni, P. aeruginosa and Snodgrassella alvi (a honeybee gut symbiont; Sundin et al., 1995; Sundin, 2000; Ludvigsen et al., 2018). Similarly, Tn5393a was recently reported on a IncI1 type conjugative plasmid from S. enterica (Harmer, 2021) while Tn5393b was also found on the R64 plasmid of S. enterica sv. Typhimurium (Cain and Hall, 2011). Our recent BLAST run carried out on the Tn5393 sequence revealed its presence in many other bacteria, human pathogens or not. Among others, Tn5393 was found in Citrobacter koseri, Corynebacterium crudilactis, Enterobacter cancerogenus, Enterobacter cloacae, Enterobacter hormaechei, Escherichia coli, K. pneumoniae, Klebsiella quasipneumoniae, Klebsiella variicola and S. enterica, as well as in the plant root-associated bacterium Pseudomonas putida. However, this does not demonstrate that direct transfer of Tn5393 occurred between PPB and human pathogens but it certainly indicates that Tn5393 is much more widespread than previously thought. Similarly, the fact that this transposon was detected in diverse bacteria (Gram-negative as well as Gram-positive) in distinct geographical locations shows that it might be accessible to a large range of organisms.

The sequence of Tn5393a also retrieved numerous matches in the BLAST analysis, mostly in antibiotic resistant human pathogens, and in two PPB, P. syringae pv. actinidiae (plasmid pMG2_SR198) and Agrobacterium tumefaciens (plasmids pTi and pAt). Finally, Tn5393d displays strong similarities with plasmid p17-84_OXA of Acinetobacter baumannii, a critical multi-resistant pathogen in human health for which new antibiotics are needed according to the WHO (World Health Organization, 2017), and with pOXA58_100004 from Acinetobacter pitii. Tn5393e was found in other human bacteria, among which some are pathogenic.

In relation with human health implications, several important findings are worth mentioning. First, a tigecycline resistance mechanism called “resistance-nodulation-division” (RND) family efflux pump (tmexCD1-toprJ1 pump) was identified in K. pneumoniae isolates from humans and chickens. The corresponding gene tmexCD1-toprJ1 appeared to be found in Tn5393 and the authors suggest that it originated from the chromosome of Aeromonas spp. through Tn5393-mediated translocation (Sun et al., 2020). Second, a tigecycline resistance gene, tet(Y), was identified on a plasmid of a multi-resistant clinical isolate of A. baumannii. This plasmid carried numerous ARGs, including strA, strB, aac(6′)-Ib3, msr(E), mph(E), floR, ARR-3, sul1, dfrA19, tet(39) and tet(Y). Interestingly, the tet(Y) gene was located inside the sequence of Tn5393 and the authors also concluded that Tn5393 played a role in its transmission from Aeromonas spp. (Wang Z. et al., 2021). Lastly, in Australia, a multi-resistant A. baumannii strain that contains a fragment of Tn5393, was isolated, showing again its potential of ARGs spreading (Post and Hall, 2009). It is interesting to note that some species of Aeromonas or Acinetobacter can also be found in the environment and genetic transfers in these genera should be carefully considered.

Recently, the genomic evolution of Tn5393 and kin was examined and different steps were proposed on how Tn5393 could have acquired various MGEs and became more complex.¹ It is also suggested that other variants of Tn5393 could exist (Lima-Mendez et al., 2020; Ross et al., 2021). Additionally, when comparing the different TnpA sequences from Tn5393 and its variants, it is rather complicated to establish their exact filiation and to determine how exchanges between bacterial species have shaped their structures. Still, the complexity of Tn5393 is apparently greater among human or animal pathogens than in PPB (Figure 2; Table 3).

Oxytetracycline belongs to the tetracycline class of antibiotics and is the second most used antibiotic in plant agriculture, mainly against fire blight (E. amylovora) in the case of streptomycin resistance reports. However, it is only used as a second line of defense because it is considered as less effective than streptomycin (McManus et al., 2002; Sundin and Wang, 2018). Various genes can be involved in tetracycline resistance: (i) expression of efflux pumps, (ii) enzymatic inactivation of the antibiotic or (iii) ribosomal protein protecting from the action of tetracyclines (Chopra and Roberts, 2001).

Concerning PPB, cases of oxytetracycline resistance are rarely reported. To the best of our knowledge, oxytetracycline resistance has not been observed yet in field isolates of E. amylovora, although, resistant strains can be obtained in vitro. In fact, the RP1 plasmid carrying an oxytetracycline resistance gene can be transferred from both E. herbicola and P. syringae pv. syringae to E. amylovora. Nevertheless, after pathogenesis, the plasmid phenotype could only be recovered in 0.03% of the isolates, suggesting that it may not be stable in that species (Lacy et al., 1984).

Oxytetracycline resistance cases have been reported in P. syringae (da Silva and Lopes, 1995; Spotts and Cervantes, 1995; Hwang et al., 2005), Xanthomonas arboricola pv. pruni (Herbert et al., 2022) and A. tumefaciens (Luo and Farrand, 1999). The resistance mechanisms were not described, except in X. arboricola pv. pruni, where it was shown that the tetracycline resistance gene tetC was carried by a plasmid, coding for an efflux pump (Herbert et al., 2022). Given the plethora of known tetracycline resistance genes (36 efflux pump genes, 13 ribosomal protection protein genes, 13 enzyme genes, 1 other gene conferring an unknown mechanism of resistance and 11 mosaic genes reported so far; Chopra and Roberts, 2001; Van Hoek et al., 2011; Roberts, 2021), others could be present in PPB but they have not been described yet. When searching for identity while running BLAST on PPB genomes with the sequence of tetC mentioned above, several strains present sequences with high identity percentage, such as X. arboricola pv. pruni, Burkholderia cepacia, Agrobacterium fabrum and several Streptomyces, which could be an indication that tetracycline resistance genes are present in other PPB.

Kasugamycin is an aminoglycoside antibiotic used only in plants, not in human nor veterinary medicine. In E. coli, point mutations in various genes (ksgA, ksgB, ksgC and ksgD) were shown to confer resistance (Sparling, 1970; Sparling et al., 1973; Yoshikawa et al., 1975; Fouts and Barbour, 1981; Yoshii et al., 2012). In PPB, different mechanisms have been identified (Table 2). The presence of aac(2′)-IIa gene in Burkholderia glumae and Acidovorax avenae ssp. avenae was shown to cause the resistance to kasugamycin (Yoshii et al., 2012). The Aac(2′)-IIa acyltransferase inactivates kasugamycin via acetylation of the 2′-amino residue of the antibiotic. In the chromosome of B. glumae, this acetyltransferase is encoded on the IncP genomic island, suggesting that this kasugamycin resistance gene might have been acquired by HGT and that the gene could hypothetically spread among PPB (Yoshii et al., 2012). The aac(2′)-IIa gene was later found on a conjugative IncP-1β plasmid, pAAA83, in an A. avenae ssp. avenae strain, raising even more concern about the potential transmission of this resistance gene (Yoshii et al., 2015). In E. amylovora, resistance to kasugamycin and blasticidin S is induced by the in vitro deletion of both permease genes opp and dpp, suggesting that Opp and Dpp act synergistically to transport these antibiotics. This deletion has not been observed in field isolates, so the relevance of such mechanism in the environment remains to be assessed (Ge et al., 2018). Similarly, in vitro mutation of the ksgA gene in E. amylovora confers kasugamycin resistance but it requires a two-step mutational process and these mutants displayed a considerably reduced fitness (McGhee and Sundin, 2011; Ge et al., 2018).

In PPB, the exact mechanisms involved in OA resistance have not been described in great details (Table 2). In vitro mutants of B. glumae resistant to the OA quinolone were obtained from strains isolated in rice fields in Japan, where the antibiotic is used as PPP. GyrA83 mutation (substitution of serine by arginine or isoleucine at position 83) seemed to be involved in the resistance mechanism (Maeda et al., 2007a,b). Others have also demonstrated that OA resistant B. glumae could be obtained in vitro, however such mutants were unable to survive in paddy fields (Hikichi et al., 1998, 2001).

In Israel, OA was introduced in 1997 when the use of streptomycin was abandoned. About 2 years later, OA resistant E. amylovora and B. glumae were reported. Yet, the exact mechanisms leading to OA resistance in these strains remained unknown, although it was suggested that it might consist into chromosomal mutations rather than gene acquisition through HGT (Manulis et al., 2003; Kleitman et al., 2005).

Resistance to the gentamicin aminoglycoside was only reported in X. oryzae pv. oryzae, where the acquisition of the integron-borne aacA3 gene was incriminated. As indicated in Table 2, aacA3 encodes an aminoglycoside-3’-N-acetyltransferase enzyme that deactivates gentamicin (Xu et al., 2013).

The aadA1 gene was found in X. oryzae pv. oryzae on three integrons and associated with other antibiotic resistance gene cassettes such as aacA3 (conferring resistance to tobramycin, kanamycin, gentamicin and netilmicin) or arr3 (conferring resistance to rifampicin; Xu et al., 2013). Both aadA1 and aadA2 also confer spectinomycin resistance. The co-occurrence of streptomycin and gentamicin resistance is therefore possible and it is then reasonable to speculate that these gene cassettes could be co-transferred under the selective pressure of streptomycin or gentamicin.

In a recent study conducted in South Carolina (United States), strains of X. arboricola pv. pruni were isolated and characterized for their resistance to both streptomycin and oxytetracycline. They carried a plasmid encoding tetC, tetR and strA-strB, with a region similar to Tn5393 (Herbert et al., 2022). It was previously described that, when isolating epiphytic bacteria (not PPB) from Michigan apple orchards, almost all tetracycline resistant strains were also streptomycin resistant and that tetracycline resistance genes were found on plasmids that also carried Tn5393 (Schnabel and Jones, 1999). Similarly, several plasmid-borne tetracycline resistance elements have been found in various epiphytic bacteria and usually the tetracycline resistance was associated with transposons, mainly Tn5393 (carried by an uncharacterized plasmid; Schnabel and Jones, 1999). These data indicate that PPB could become resistant to different antibiotics through the acquisition of resistance genes from other plant-associated bacteria if the genes are present on MGEs.

In 2005, several P. syringae were isolated and assessed for their antibiotic resistance to six antibiotics. One strain of P. syringae pv. syringae was resistant to kanamycin and tetracycline. Eight isolates were resistant to streptomycin, 16 isolates were resistant to rifampicin and 36 to chloramphenicol. Fifty-five strains in total were resistant to ampicillin. Some of these strains were resistant to several antibiotics, and although the resistance mechanisms were not characterized, this study further highlights the possibility of ARGs co-occurrence (Hwang et al., 2005).

Along the same lines, an OA resistant B. glumae was reported to display cross-resistance with other quinolones such as ciprofloxacin (Hikichi et al., 2001), while another isolated B. glumae was resistant to polymyxin B (Paz-Carrasco et al., 2018).

Finally, it is worth mentioning three studies that have addressed resistance to antibiotics mainly used in China. First, it was possible to obtain X. oryzae pv. oryzae strains resistant to phenazine-1-carboxylic acid (PCA), also called shenqinmycin in China, after in vitro exposition to increasing concentrations of the antibiotic (Pan et al., 2018). Second, zhongshengmycin resistant strains of X. oryzae were obtained in vitro and the resistance was related to increasing fatty acid biosynthesis. The exact underlying mechanisms are not known yet (Wang Q. et al., 2021). Third, bismerthiazol resistant mutants of X. oryzae pv. oryzae could also be induced, both in vivo and in vitro (Zhu et al., 2013). However, these mutants were all obtained in vitro or induced, so the natural occurrence of such resistance remains to be demonstrated. Nevertheless, this indicates that X. oryzae pv. oryzae is highly adaptable and may easily develop resistances.

Several studies reported that the application of streptomycin on orchards did not, or not adversely, affect the bacterial population of the soil (Walsh et al., 2013; Shade et al., 2013a) or described minimal alterations with a slight decrease in phylogenetic diversity of bacterial communities of apple tree flowers (Shade et al., 2013b), on the short-term. Another study indicated a decrease of diversity in streptomycin-treated soil (Tolba et al., 2002). When looking at the bacterial communities of apple leaves, a higher frequency of streptomycin resistant bacteria was observed in samples that were not treated with streptomycin (Yashiro and McManus, 2012). However, they only focused on apple leaves and not on the other bacterial communities (soil or roots) that could be affected differently. This work also dealt with the overall bacterial community, and not with specific species that could become resistant without significantly increasing the overall abundance of resistance. Moreover, as indicated before, streptomycin application could select for other ARGs.

The effects of streptomycin and kasugamycin application on bacteria in the apple phyllosphere have been investigated. While the use of streptomycin did not result in any increase of streptomycin resistant E. amylovora, it was correlated with an increase in resistance in other epiphytic bacteria, such as Pantoea agglomerans or Pseudomonas spp. No kasugamycin resistant bacteria could be isolated, however the antibiotic application changed the microbial spectrum in the orchard (Tancos and Cox, 2017). Along the same lines, there was no influence on the abundance of culturable oxytetracycline and gentamicin resistant bacteria when these antibiotics where applied, as compared to untreated soils (Rodríguez-Sánchez et al., 2008). On the opposite, kasugamycin resistant Gram-negative bacteria (mainly residents of the plant phyllosphere or colonists of apple, such as P. agglomerans, Pseudomonas graminis, P. syringae and Stenotrophomonas spp.) could be isolated in another study from orchard soil, apple flowers and leaves treated with kasugamycin (McGhee and Sundin, 2011).

Sundin et al. (1995) observed the distribution of streptomycin resistance transposon Tn5393 in the apple phylloplane and soil of ornamental pear (that had received previous applications of streptomycin) and tomato (no prior exposure to streptomycin; Sundin et al., 1995). The recovery of streptomycin resistant bacteria was generally higher in soil samples, but the highest occurrence was from phylloplane samples from an orchard where streptomycin had been applied. They showed that the use of streptomycin for plant protection can also select for streptomycin resistance in non-target commensal bacteria that inhabit either plant surfaces or the surrounding soil (Chiou and Jones, 1993; Sundin et al., 1995). They investigated the risk of co-selection of other antibiotic resistance genes (tetracycline), but found no evidence that the use of streptomycin increased tetracycline resistance (Sundin et al., 1995). One should however note that streptomycin and tetracycline resistance genes were recently found on the same plasmid (Herbert et al., 2022).

In a bioinformatic analysis, where a total of 127 genomes of E. amylovora from different geographic regions were examined, the greatest number of streptomycin resistant isolates was observed in Western North America (in particular in British Columbia, Canada), where streptomycin was used, at least during the survey (1993–1998; Sholberg et al., 2001; Parcey et al., 2020). Several additional studies showed that the orchards where streptomycin use is the greatest were usually those where the highest numbers of streptomycin resistant bacteria were detected (Chiou, 1991; Norelli et al., 1991; Burr et al., 1993).

Regarding oxytetracycline resistance, the finding of a tetracycline resistant isolate of A. tumefaciens could not be linked to oxytetracycline application (Luo and Farrand, 1999). On the opposite, antibiotic sprays of streptomycin and oxytetracycline were positively correlated with resistance in P. syringae pv. syringae (Spotts and Cervantes, 1995). In epiphytic bacteria, the use of oxytetracycline generally resulted in less abundant bacterial populations than when streptomycin was used, but streptomycin resistant isolates were more common than the tetracycline resistant ones. Moreover, tetracycline resistant strains could be observed in orchards where tetracycline had not been applied, although higher numbers of tetracycline resistant bacteria could be detected in another orchard where oxytetracycline was applied (Schnabel and Jones, 1999).

In Israel, the introduction of OA to replace streptomycin led to the isolation of OA resistant E. amylovora and B. glumae. However, the incidence of OA resistance in E. amylovora was sporadic and irrespective of the number of sprays applied and the severity of the disease (Kleitman et al., 2005). In another study also conducted in Israel after introduction of OA, several OA resistant strains were isolated from different orchards but none was resistant to both streptomycin and OA (Manulis et al., 2003). Concerning gentamicin resistance, the only report available in X. oryzae pv. oryzae in China has not been linked to the use of gentamicin (Xu et al., 2013).

The abundance of streptomycin and tetracycline resistance genes in flowers, leaves and soil samples from orchards treated with streptomycin in Switzerland was analyzed using multiplex qPCR (Duffy et al., 2014). Three orchards that had not been treated with streptomycin prior to the assays were used, and samples were collected over a three-year period. By using qPCR to assess the abundance of ARGs, this study overcame the intrinsic problem linked to culture-dependent assays where unculturable bacteria are overlooked. This method allows for a qualitative assessment of ARGs in the complete apple tree ecosystem but does not identify the species that contain the resistance genes. The results indicated no consistent increase in streptomycin resistance genes in streptomycin-treated samples. Tetracycline resistance genes were also quantified in this study, and no increase was reported with streptomycin use. It is however important to note that due to the nature of the technique used, streptomycin resistance genes, notably strA-strB and aadA, had to be selected for the screening. Other potential resistance mechanisms might have been ignored.

Similarly, the abundance of various tetracycline resistance genetic determinants and several gentamicin resistance genes was investigated under field conditions with five applications of oxytetracycline and gentamicin during 16 months. This study indicated that the occurrence of these genes was not related to the application of antibiotics (Rodríguez-Sánchez et al., 2008).

Little is known about the resilience of an ecosystem to antibiotic resistances, years after the discontinuation of the antibiotic use.

In Israel, streptomycin was removed from the list of approved antibiotics for plant agriculture in 1997 and has not been used since. A decline in the incidence of streptomycin resistance in E. amylovora was observed in the subsequent years (from 57% in 1998 to 15% in 2001). Surprisingly, within 4 years, the resistance had almost disappeared (Manulis et al., 2003).

Other studies have been conducted in a few American states. In California, streptomycin resistance among bacteria from orchards declined from 1973 to 1977. However, streptomycin resistant E. amylovora strains could still be isolated, even though streptomycin had not been applied in these orchards since 1971 (Schroth et al., 1979). In Washington state, streptomycin resistant E. amylovora strains could still be isolated 5 years after termination of streptomycin use (Loper et al., 1991). Schnabel and Jones (1999) showed that streptomycin resistance did not diminish at their study site over the course of 2 years in the absence of the selection pressure (Schnabel and Jones, 1999). In Michigan, a decline in the incidence of streptomycin resistance was observed from 1991 to 1992 when oxytetracycline was used rather than streptomycin. However, streptomycin resistance in E. amylovora quickly reoccurred in 1993 when the use of streptomycin was resumed (McManus and Jones, 1994). This could suggest that streptomycin resistant strains have lower fitness than streptomycin sensitive strains, although Loper et al. (1991) found streptomycin resistant strains where this antibiotic had never been applied, suggesting similar fitness capabilities under no selective pressure (Loper et al., 1991).

It was shown that there is a high presence of streptomycin resistance in agricultural sites without streptomycin treatment history, and that HGT of strA and strB occurred regardless of soil treatment with antibiotics (Tolba et al., 2002). It has also been shown that strA gene is more prevalent in soils from compost, forest or agriculture samples than from vegetable gardens, apple orchards or mixed fruit orchards (Popowska et al., 2012). However, in the same study, strB was less prevalent in compost, forest or agricultural soils than in apple orchards or mixed fruit orchards. The aadA gene followed the same pattern as strB but was only detected in agricultural soil at very low rates, while absent in compost or forest soils. It was rather prevalent in vegetable gardens, apple orchards or mixed fruit orchards.

It is also worth mentioning that the streptomycin resistant transposon Tn5393 was detected in bacteria with no prior exposure to streptomycin. It seems that this transposable element is indigenous to both phylloplane and soil microbial communities (Sundin et al., 1995). In fact, strA-strB and Tn5393 are both present in non-target bacteria (Chiou and Jones, 1993; Sundin et al., 1995). Gentamicin resistance genes are also commonly found in environments associated with sewage effluent or farm animals (Heuer et al., 2002).