A majority of the cucurbit seed associated endophytes expressed functional traits known to enhance plant disease resistance. Approximately 67% (113/169) and 62% (104/169) of the tested endophytic bacteria produced acetoin and/or diacetyl in vitro and displayed ribonuclease (RNase) activity, respectively (Figures 2, 3 and Supplementary Table S1). Cucumber seeds, specifically, were the primary source of the positive candidates accounting for 44% (49/113) of the acetoin/diacetyl producers and 37% (38/104) of strains with RNase activity. The genus Bacillus accounted for 59% (67/113) of acetoin/diacetyl producers and 73% (76/104) of RNase positive isolates. Furthermore, the vast majority of the 83 Bacillus isolates displayed acetoin/diacetyl activity (81%, 67/83) and RNAse activity (92%, 76/83) (Figure 3). In addition, 49% (18/37) and 27% (10/37) of the Paenibacillus isolates demonstrated acetoin/diacetyl production and RNase activity in vitro, respectively. Regarding the identified Enterobacteriaceae family members, 75% (15/20) were acetoin/ diacetyl producers and 30% (6/20) showed RNase activity. All isolated bacterial endophytes of the genera Cronobacter, Pantoea, Microbacterium, and Staphylococcus displayed these traits.

The bacterial endophytes were tested for their ability to suppress growth in vitro of a spectrum of plant pathogens with diverse host ranges including cucurbits. In total, 43% (73/169) of isolates showed anti-pathogenic traits in vitro including against both oomycete pathogens (P. capsici, P. aphanidermatum) and fungal pathogens (R. solani, F. graminearum) of which F. graminearum (control) is not known to cause disease in cucurbits (Figure 2). For each pathogen tested, antagonistic endophytes were identified that originated from diverse, major cucurbits. For example, the potential R. solani antagonists originated from the five cucurbit genera and 15 different varieties. The seeds of diverse varieties of melon (C. melo), in particular variety Cantaloupe delicious (Figure 2 and Supplementary Table S1), hosted the highest number and broadest spectrum (fungal, oomycetes) of in-vitro antagonistic endophytic bacteria. For example, 82% (18/22) of cucurbit family endophytes that displayed anti-F. graminearum activity originated from melon seed; this number was 55% (24/44) for P. capsici and 42% (23/55) for R. solani (Supplementary Table S1). Despite desiccation and long-term storage, dry seeds (variety names without asterisks, Figure 2 and Supplementary Table S1) were the main supplier of bacterial endophytes in this study (113/169) and constituted the majority of the antagonists (77%, 56/73 of antagonists). Fresh seeds were also sources of antagonists, however, (e.g., melon, C. melo; Figure 2 and Supplementary Table S1).

With respect to the tested fungal pathogens, of the 55 strains that antagonized R. solani in vitro, 52 belonged to the genera Bacillus and Paenibacillus though they only constituted 71% (120/169) of the endophyte library (Figure 3 and Supplementary Table S1). Similarly, of the 22 strains that antagonized F. graminearum, all belonged to Bacillus and Paenibacillus. By contrast of the 20 Enterobacteriaceae endophytes tested, none suppressed the fungal pathogens (Figure 3 and Supplementary Table S1).

A broader taxonomic spectrum of endophytes suppressed growth of the tested oomycete pathogens. Nevertheless, of the 44 endophytes that suppressed growth of P. capsici in vitro, 64% (28/44) belonged to Bacillus and Paenibacillus, with the remainder belonging to 5 other genera including Lactococcus (16%, 7/44). There were 19 taxonomically diverse strains that suppressed P. aphanidermatum in vitro, of which Lactococcus endophytes were the most prevalent (41%, 7/19), whereas only 5 Bacillus and Paenibacillus strains (combined) were antagonists. It is noteworthy that all 7 isolates of the genus Lactococcus showed in vitro anti-oomycete activity (Figure 3 and Supplementary Table S1).

The endophytes were also tested for antagonism against Podosphaera fuliginea, the causative agent of cucumber powdery mildew disease. As the pathogen is obligate, the bioassay employed leaf disks (Figures 4A–D). The disease index (DI%) and preventive effect (PE%) were calculated for all tested endophytes (Supplementary Tables S2B,C). In total, 37% (62/169) of the tested endophytic bacteria demonstrated significant suppression of powdery mildew symptoms (p < 0.05, mean DI%) in comparison to disks treated with LB media (negative control) (Figures 4D, 5A and Supplementary Tables S3, S4). Noteworthy, these promising 62 strains originated from the five tested cucurbit genera (Figure 5B and Supplementary Table S2A). Interestingly, most of the remaining endophytes displayed adverse effects on cucumber leaf health in the presence of the pathogen (Figure 4E and Supplementary Tables S2A–C). The beneficial endophytes displayed mean positive PE values of up to 89%. Of these beneficial strains, 66% (41/62) and 79% (49/62) were statistically equal or more effective (One-Way ANOVA Dunnett’s multiple comparisons) than the fungicide, prothioconazole [(positive control, concentration optimized in a pre-experiment (see Materials and Methods; data not shown)] (Figure 4A) and the biofungicide, Bacillus subtilis strain QST 713 (Figure 4B), respectively.

Unlike in vitro antimicrobial activity, fresh cucurbit seeds were the major sources (61%, 38/62) of antagonists that significantly suppressed powdery mildew in planta (Supplementary Tables S2A, S3). Furthermore, cucumber, pumpkin, cantaloupe and watermelon seeds contributed many of the antagonists, accounting for 23% (14/62), 19% (12/62), 18% (11/62) and 16% (10/62) of the significantly (p < 0.05) effective endophytes, respectively. With respect to endophytic taxonomy, the strains that significantly restrained symptoms of the disease belonged to 14 different bacterial genera (Figure 5B and Supplementary Tables S3, S4), of which the most prevalent (in rank order) were Bacillus (37%, 23/62), Paenibacillus (16%, 10/62) and Unclassified Enterobacteriaceae (14.5%, 9/62). However, amongst the most effective antagonists were Pediococcus and Pantoea endophytes (Figure 4C).

To independently validate the leaf disk results and to make the assay more robust, the endophytes that were effective against powdery mildew in comparison to prothioconazole fungicide (41 strains, Supplementary Table S2D) were re-tested using detached whole cucumber leaves infected with the pathogen (Figures 6A–E and Supplementary Figure S1). In total, 66% (27/41) of endophytes were confirmed to inhibit powdery mildew symptoms visually in comparison to leaves treated with LB media (negative control), with PE values ranging from 8–46% (Figure 6E, Supplementary Figures S1A–E, and Supplementary Table S2D). However, only six strains showed significant suppression of powdery mildew disease when compared to the negative control (independent t-test, one-tailed, unequal variance, using visual scores at p < 0.1) (Figure 6E and Supplementary Figure S1). As above, the majority of the remaining endophytes displayed adverse effects on cucumber leaf health in the presence of the pathogen (Supplementary Figure S1E and Supplementary Table S2D). The fresh seeds were the major sources of beneficial microbes, accounting for 89% (24/27) of disease antagonists (Figure 6E). Pumpkin seeds were the main source of promising antagonists accounting for 30% (8/27) of candidates, followed by cucumber seeds 22% (6/27), cantaloupe (19%, 5/27) and watermelon seeds (19%, 5/27). Nevertheless, 50% (3/6) of strains that exhibited significant disease suppression originated from fresh cucumber seeds and belong to the Enterobacteriaceae family (Figure 6E). Bacillus endophytes were the most prevalent antagonists, comprising 26% (7/27) of the positive strains. However, a Pediococcus endophyte from fresh cantaloupe seeds displayed the highest PE% against the disease (46%), equivalent to the positive controls (Supplementary Table S2D).

Bacillus and Paenibacillus are cosmopolitan aerobic bacterial genera in agroecosystems (McSpadden Gardener, 2004). Their ubiquitous existence has been confirmed by culture dependent techniques, while metagenomic analyses have revealed additional species diversity (Borneman et al., 1996; Mahaffee and Kloepper, 1997; Garbeva et al., 2003). Moreover, they are known for specific attributes that favor them as crop inoculants including their ability: (1) to form stress tolerant endospores; (2) to secrete diverse antimicrobial secondary metabolites (peptide antibiotics); (3) to secrete extracellular enzymes and signaling molecules; (4) to scavenge nutrients from the environment; (5) to biosynthesize phytohormones; and (6) to induce host systemic defense. Various beneficial strains of Bacillus and Paenibacillus are commercially available as biopesticides and biofertilizers (McSpadden Gardener and Fravel, 2002; McSpadden Gardener, 2004; Choudhary and Johri, 2009; Böhme et al., 2016).

Despite an earlier focus of biocontrol research on the use of Gram negative bacteria such as Pseudomonas, Agrobacterium and Erwinia, later studies starting in 1999 introduced Bacillus species as alternative biocontrol agents (Emmert and Handelsman, 1999; Nagórska et al., 2007; Rybakova et al., 2016). The above-noted biological traits of Bacilli facilitated their commercial formulation as marketable biopesticides (Ongena and Jacques, 2008; Rybakova et al., 2016). For example, in a recent study (Karimi et al., 2016), 4/8 Bacillus strains (including B. amyloliquefaciens, B. pumilus and B. siamensis) isolated from the sugar beet rhizosphere, and shoots and roots of apple, showed significant suppression of pathogenic R. solani infecting sugar beet under greenhouse conditions.

Interestingly, here, 43% (36/83) of Bacillus strains isolated from cucurbit seeds exhibited antagonism in vitro against diverse fungal and oomycete pathogens including R. solani (isolated from sugar beet), F. graminearum (from maize), P. capsici (from pepper) and P. aphanidermatum (from cucumber). These Bacillus isolates originated from seeds of all tested cucurbit species (Figure 2) and were predicted to be diverse species (Khalaf and Raizada, 2016). In a previous study, various Bacillus strains (B. subtilis, B. amyloliquefaciens, and B. cereus) isolated from cucurbit fruits (but not seeds) (C. melo, C. callosus, Citrullus lanatus and Bryonia cretica) demonstrated in vitro anatagonism against four fungal pathogens of melons including F. oxysporum F. sp. melonis and F. oxysporum, and F. sp. radicis-cucumerinum (Glassner et al., 2015). Furthermore, 71% of tested Bacillus microbes against Fusarium wilt in cucumber were effective (Raza et al., 2017). Fusarium wilt disease is a devastating disease of cucumber, caused by Fusarium oxysporum (Raza et al., 2017). We do not know if the anti-F. graminearum strains identified in this study showed cross-antagonism toward other Fusarium species, but it may explain the reason why cucurbits possess endophytes that antagonize a non-cucurbit pathogen.

With regard to the role of Bacillus in managing cucurbit powdery mildew, in a previous study, four Bacillus strains isolated from the healthy phyllosphere and rhizosphere of cucurbits infected with powdery mildew demonstrated antagonism against the fungus Podosphaera fusca on detached melon leaves and seedlings; the suppressive efficacy was up to 80% (Romero et al., 2004). In another study, Bacillus subtilis strain UMAF6639 demonstrated reduction of powdery mildew disease severity in infected melon leaves (García-Gutiérrez et al., 2012). The antagonism involved three major mechanisms including induction of host systemic resistance (García-Gutiérrez et al., 2012), stimulation of plant signaling pathways mediated by jasmonate and salicylic acid, and finally induction of host defense mediated by a surfactin lipopeptide. Generally, however, powdery mildew antagonism by biocontrol agents has been shown to be caused by microbial production of the antifungal compounds, iturin and fengycin (García-Gutiérrez et al., 2013).

Since the first characterization of the genus Paenibacillus 20 years ago, numerous studies of its role as an endophyte have been reported (Rybakova et al., 2016). For example, P. polymyxa strains identified in our lab from seeds and roots of maize exhibited anti-F. graminearum activity in vitro as well as under greenhouse conditions (Mousa et al., 2015). The displayed antagonism was confirmed by molecular detection of the fusA gene and biochemical detection of produced fusaricidin derivatives. Similarly, 3 out of 4 isolates belonging to the genus Paenibacillus (and one isolate belonging to the genus Pantoea) identified from wheat seeds of a commercial variety exhibited anti-F. graminearum activity in the dual culture assay (Díaz Herrera et al., 2016). Moreover, a P. polymyxa strain isolated from the vinegar industry showed effective suppression of Fusarium wilt in cucumber and promoted other beneficial microbes to successfully compete against the causative agent (F. oxysporum, a soil pathogen) for colonizing the rhizospheric niche of cucumber (Shi et al., 2017). P. polymyxa PB71 identified from seeds of Styrian oil pumpkin demonstrated antagonism against fungal diseases caused by Didymella bryoniae under greenhouse condition as well as suppression of powdery mildew during field trials, and improved pumpkin yield (Fürnkranz et al., 2012). In a recent study (Liu et al., 2017), endophytic bacterial strain RSE1, isolated from rice (Oryza sativa L.) seeds and identified as P. polymyxa, showed in vitro antagonism against the fungal pathogen Ustilaginoidea oryzae, the causative agent of rice false smut. The genetic basis of the biocontrol trait was predicted through computational analysis of the draft genome of the endophytic strain as a glucanase gene. In our study, 17/37 (46%) of Paenibacillus strains isolated from commercial cucurbit seeds exhibited in vitro antagonism when results of all tested crop pathogens were combined (Supplementary Table S1). These candidates included 16 antagonists of R. solani and P. capsici, and 14 strains against F. graminearum in vitro (Figure 3). Intriguingly, the R. solani antagonists were isolated from all tested cucurbit genera and hence may have been inadvertently selected. Almost all of these promising candidates were acetoin and/or diacetyl producers (Supplementary Table S1). Seven strains of Paenibacillus recovered from fresh seeds of pumpkin showed significant reduction of powdery mildew symptoms (p < 0.01–0.0001) in the leaf disk bioassay (Supplementary Tables S3, S4). Our results are congruent with the reported patents of Paenibacillus species as biocontrol agents antagonizing a wide range of plant pathogens comprising pathogenic species of Fusarium, Rhizoctonia, Pythium, Phytophthora, Botrytis, Penicillium, Sclerotinia and Cladosporium (Rybakova et al., 2016).

The genus Pantoea was first reported in the literature as a plant pathogen (Muraschi et al., 1965; Nadarasah and Stavrinides, 2014), but more recently various strains have been shown to exert beneficial biocontrol activity to host plants through various mechanisms including competitive colonization, production of antibiotics and/or induction of host systemic defense. Pantoea species have been shown to target a wide spectrum of plant pathogens including bacteria, fungi, oomycetes and parasitic nematodes via secretion of antimicrobials such as pantocins, herbicolins, microcins, and phenazines (Walterson and Stavrinides, 2015). Commercially formulated Pantoea-based biopesticides are currently available such as BlightBan C9-1 (P. vagans C9-1) and Bloomtime Biological (P. agglomerans E325) which are used for combating fire blight in apples and pears (Johnson and Stockwell, 1998; Pusey, 2002). In a recently published paper, three endophytic bacteria isolated from surface-sterilized rice (Oryza sativa L.) seeds promoted in vitro rice seedling development along with protection against F. oxysporum. Using 16S rRNA sequencing, the strains were identified as Pantoea dispersa, along with Enterobacter asburiae and Pseudomonas putida (Verma et al., 2017). In our study, we isolated two Pantoea strains from fresh cucumber seed mucilage that exhibited in vitro antagonism against the tested oomycetes (Supplementary Table S1) and showed significant suppression of powdery mildew symptoms in the leaf disk bioassay (p < 0.05) and detached whole leaf bioassay (p < 0.1) (Figures 4C, 6C and Supplementary Tables S2, S3). These strains were also producers of acetoin and diacetyl (Supplementary Table S1). Our findings suggest that Pantoea may hold potential as inoculants that can target a new, wider anti-pathogenic spectrum.

Another recent study (Sandhya et al., 2017) showed the potential of seed endophytes as biocontrol agents. In that study, maize seeds and roots were used as sources of bacterial endophytes; 10 out of 39 endophytic bacteria demonstrated diverse in vitro activities comprising growth promotion and antagonism against 6 fungal phytopathogens (Macrophomina phaseolina, Rhizoctonia solani, Fusarium oxysporum, Sclerotium rolfsii, Pythium aphanidermatum and Alternaria sp.) under drought stress. These endophytic bacteria were identified to the genus level as Pseudomonas, Acinetobacter, Enterobacter, and Sinorhizobium.

In the food industry, various species of lactic acid bacteria (LAB) belonging to the genera Enterococcus, Lactobacillus, Lactococcus, Pediococcus and Leuconostoc, are well known safe biopreservatives and biological control agents (Gajbhiye and Kapadnis, 2016; Porto et al., 2017). Their antagonism is mediated by secretion of cyclic dipeptides, diacetyl, organic aids, ethanol, antifungal metabolites (such as phenyllactic acid or fatty acids), bacteriocins and antibiotics (such as reutericyclin) (Akbar et al., 2016; Gajbhiye and Kapadnis, 2016). In our study, the results revealed 10 antifungal LAB strains belonging to the genera Lactococcus, Pediococcus and Leuconostoc. All Lactococcus strains showed in vitro antagonism against oomycetes (Supplementary Table S1) whereas four strains antagonized Podosphaera fuliginea in planta (Supplementary Table S2A). The two identified Pediococcus strains were shown here to exhibit a broad antimicrobial target spectrum including against R. solani, P. capsici and Podosphaera fuliginea, but not F. graminearum. However, in a previous study, a Pediococcus pentosaceus strain isolated from dairy products antagonized F. graminearum growth (Sellamani et al., 2016). Moreover, the broad antifungal spectrum of Pediococcus acidilactici LAB 5 strain was first reported in a study by Mandal et al. (2007) and included antagonism against Alternaria solani, Aspergillus fumigaus, A. parasiticus, Cladosporium herbarum, Colletotrichum acutatum, Curvularia lunata, Fusarium oxysporum, Microsporium sp. Mucor sp. and Penicillium sp. (Mandal et al., 2007). In this study, a single Leuconostoc strain demonstrated antagonism toward R. solani and P. aphanidermatum, though in an earlier study, the intimate association between R. solani and Leuconostoc bacteria in sugar beet roots was interpreted as a synergistic interaction that promoted root rot (Strausbaugh, 2015). Combining these previous reports with our study, it would appear that the probiotic characteristics of LAB are not limited to food preservation, but may extend to counteract plant pathogens during crop cultivation.

As noted earlier, acetoin (3-hydroxy-2-butanone) and its major bioanalogs (2,3-butanediol and diacetyl), are important volatile organic compounds (VOCs) produced by several microbes including bacteria. In earlier research, B. subtilis, B. amyloliquefaciens and P. polymyxa demonstrated growth promotion and induction of systemic resistance in Arabidopsis thaliana by secretion of these VOCs (Ryu et al., 2003, 2004; Lee et al., 2012). In our study, perhaps surprisingly, production of acetoin and diacetyl volatiles correlated to antagonism of fungi and oomycetes despite the lack of a plant host in our in vitro assay. For example, 95% (21/22), 82% (45/55), 77% (34/44) and 58% (11/19), respectively, of F. graminearum, R. solani, P. capsici and P. aphanidermatum in vitro antagonists, were acetoin and diacetyl producers (Supplementary Table S1). Furthermore, 43.5% (27/62) of promising candidates that significantly (p < 0.05) reduced disease severity against powdery mildew in the in planta leaf disk bioassay were VOC producers (Supplementary Table S2). It may be that these endophytes have been selected to have multiple anti-pathogen mechanisms, including both direct antibiosis (evident in our dual-culture assays) and induction of host defense. The majority of the promising VOC-producing antagonists in this study belong to the phylum Firmicutes, in particular the genera Bacillus and Paenibacillus (Figure 3). Similarly, in an earlier study, four different VOC-producing B. amyloliquefaciens subspecies demonstrated in vitro antagonism against several plant pathogenic fungi (Asari et al., 2016). Therefore, our findings point to the promise of using endophytic bacteria displaying VOC activity in biological control applications (Vespermann et al., 2007; Asari et al., 2016), though further confirmation of the impact of the VOCs produced by the cucurbit endophytes will be required. Specifically, VOCs of promising biocontrol candidates could be collected in sealed plates and applied to the plant tissue in the presence of the relevant pathogen.

In our study, of the 104 RNase positive isolates, 91% (20/22) of F. graminearum antagonists displayed RNase activity followed by 73% (40/55) for R. solani, 55% (24/44) for P. capsici, 42% (8/19) of P. aphanidermatum, and 53% (33/62) for powdery mildew (of those that significantly suppressed the disease in the leaf disk bioassay at p < 0.05). Interestingly, 83% (86/104 RNase positive strains) belonged to Bacillus (76/104) and Paenibacillus (10/104). In previous studies, Bacilli have been shown to produce RNAses, for example, Bacillus amyloliquefaciens has been shown to produce barnase, a ribonuclease and antibiotic protein (Khan et al., 2016). Noteworthy, the antifungal activity of some pathogenesis-related (PR) proteins secreted by host plants has been associated with their ribonuclease activity, including the PR-4 and PR-10 families (Filipenko et al., 2013; Miranda et al., 2017; Moosa et al., 2017). The associated mechanism(s) of action of these ribonucleases is not clear but may include direct antagonism (by penetrating fungal cells and degrading native mRNA) or indirect antagonism (by inducing host programmed cell death) (Filipenko et al., 2013; Miranda et al., 2017; Moosa et al., 2017). Further experiments will be required to determine whether the extracellular RNase activity of any of the endophytes plays a role in their phytopathogen suppression, and if so, their mechanism of action.

In our previous study (Khalaf and Raizada, 2016), we reported that extracellular lytic enzyme activities, including cellulase, pectinase and protease function, were displayed by numerous cucurbit seed-associated bacterial endophytes. Our previous observation may be relevant here, as several biocontrol agents have been shown to exert their antagonistic activity through secretion of lytic enzymes that protect the plant either directly or indirectly. The direct mechanism has been shown to involve breaking down of essential complex polymers within the pathogen such as chitin, protein, cellulose and DNA (Pliego et al., 2011). By contrast, the lysis products (e.g., chitin fragments) can be indirectly employed in plant protection by eliciting host defence responses (Heydari and Pessarakli, 2010; Pliego et al., 2011; Duran-Flores and Heil, 2016). Further experiments will be needed to verify this hypothesis.

Surprisingly, we noticed adverse effects from some endophytic bacteria in the cucumber leaf disk bioassay (Supplementary Table S2B), suggesting that they were being recognized as pathogens or added to plant stress, perhaps due to their high titre in combination with plant wounding. Some endophytes have been shown to be helpful to specific hosts but pathogenic to others (Brader et al., 2017). Alternatively, some of the observed variation between the endophytes may have been due to differences in their growth phase. In a study by Romero et al. (2004), bacterial cells in stationary phase showed relatively greater inhibition than log phase cells with respect to growth of the obligate pathogen Podosphaera fuliginea (Romero et al., 2004). With respect to the strain biosafety, promising endophytes must be subject to further study to evaluate their impacts on human health and the surrounding environment. Potential risks could be identified through the rules and regulations of the safety assessment charter laid down by the American Biological Safety Association (ABSA) (Bharti et al., 2017).

Our study represents only the first step toward the development of novel biocontrol inoculants. Apart from powdery mildew, the study was limited to in vitro screening, and likely revealed endophytes that exhibit direct antagonism, along with traits such as VOCs pertinent to the induction of the host defense response. Other implicated biocontrol mechanisms such as pathogen/endophyte-plant interactions and competition with natural microflora will require in planta assays (Eljounaidi et al., 2016). More generally, future biological control applications of promising antagonists will require replicated field and greenhouse trials, which are forthcoming.

The focus of this study was bacterial, not fungal, endophytes as potential biocontrol agents, because vegetable growers spray with fungicides, and hence bacterial biocontrol agents can be integrated with chemical approaches. Furthermore, bacteria are easily handled and rapidly grow to permit inoculant preparation (Whipps, 2001). Although very limited, a small number of successful trials of seed bacterial endophytes have been reported for the management of fungal phytopathogens (Mukhopadhyay et al., 1996; Loper and Henkels, 1999; Cottyn et al., 2001; Ruiza et al., 2011; Gagne-Bourgue et al., 2013; Truyens et al., 2015), suggesting that the microbes identified in this study may have future biocontrol potential. Optimistically with the advent of next generation sequencing technology, omics tools could help in performing studies of comparative genomics between de novo inoculants and their biocontrol cousins of the same genus. These studies may facilitate the selection of promising candidates for further in vivo testing based on their genomes encoding biocontrol genes (Douriet-Gámez et al., 2017).

Cucurbit fruits and their seeds have been widely grown by humans for thousands of years, able to grow in diverse climates where fungi and oomycetes can thrive. As with all plants, cucurbit seeds germinate in the soil. Here we demonstrate that the majority of seed associated endophytes of cultivated cucurbits have the potential to suppress fungal and oomycete pathogens that affect early seed germination as well as a later foliar disease. Furthermore, we demonstrate that the majority of these antagonistic microbes are Bacillus and Paenabacillus strains. Bacilli are widely used as commercial biocontrol agents of vegetable crops, and it is interesting to speculate whether these biocontrol agents are mimicking the ecological niche established by their endophytic cousins. The endophytes characterized in this study may assist with the implementation of alternative pest management strategies to address growing concerns for safe food production globally.

One hundred and sixty nine endophytic bacteria were previously cultured from seeds of diverse cucurbits and classified based on their 16S rRNA sequences (Supplementary Table S1) (Khalaf and Raizada, 2016).

Data of leaf disk bioassay were analyzed using GraphPad Prism 7. One-way analysis of variance (ANOVA) was applied, and mean values of the treatments were compared to controls using Dunnett’s multiple comparison test. Data from the detached whole leaf bioassay were analyzed using t-tests of visual disease scores (one-tailed and unequal variance).