Alternative rooting induction of semi-hardwood olive cuttings by several auxin-producing bacteria for organic agriculture systems

Southern Spain is the largest olive oil producer region in the world. In recent years organic agriculture systems have grown exponentially so that new alternative systems to produce organic olive cuttings are needed. Several bacterial isolates, namely Pantoea sp. AG9, Chryseobacterium sp . AG13, Chryseobacterium sp . CT348, Pseudomonas sp . CT364 and Azospirillum brasilense Cd (ATCC 29729), have been used to induce rooting in olive semi-hardwood cuttings of Arbequina, Hojiblanca and Picual cultivars of olive ( Olea europea L). The first four strains were previously selected as auxin-producing bacteria and by their ability to promote rooting in model plants. They have been classified on the basis of their 16S rDNA gene sequence. The known auxin producer A. brasilense Cd strain has been used as a reference. The inoculation of olive cuttings was performed in two different ways: (i) by dipping cuttings in a liquid bacterial culture or (ii) by immersing them in a paste made of solid bacterial inoculant and sterile water. Under nursery conditions all of the tested bacterial strains were able to induce the rooting of olive cuttings to a similar or greater extent than the control cuttings treated with indole-3-butyric acid (IBA). The olive cultivars responded differently depending on the bacterial strain and the inoculation method. The strain that consistently gave the best results was Pantoea sp. AG9, the only one of the tested bacterial strains to express the enzyme 1-aminocyclopropane-1-carboxylate (ACC) deaminase. The results are also discussed in terms of potential commercial interest and nursery feasibility performance of these strains. CFU (colony-forming unit); DAI (days after inoculation); DF (Dworkin and Foster); IAA (indole-3-acetic acid); IBA (indole-3-butyric acid); NB (nutrient broth); PGPR (plant growth promoting rhizobacteria).

Azospirillum brasilense has been widely used as a PGPR with different crops owing, not only to its ability to fix nitrogen, but also to its capacity to produce and release a broad spectrum of plant growth regulators (Holguin & Glick, 2003). For example A. brasilense strain Cd is a very efficient IAA producer (Li et al., 2005).
Olive is the most important crop in the Andalusian region of southern Spain with about 1.5 million ha devoted to its cultivation, making this region the main olive oil producer in the world. 'Arbequina', 'Hojiblanca' and 'Picual' are the three main olive cultivars used in this area (Barranco et al., 2005). In recent years, organic agriculture systems in Spain, especially in Andalusia, have grown exponentially. Conventional plantations of olive trees are easily converted to a system of organic agriculture, so that, the area of organic olive groves in this region includes nearly 50,000 ha. Nowadays, the olive plant propagation system adopted by nursery growers mainly uses semiwoody olive cuttings. Although this system depends on several factors (the olive cultivar, the physical properties and temperature of the rooting media, the type of cutting material, the season, etc.), in general it provides a high level of efficiency. The induction of rooting with indole-3-butyric acid (IBA) is not authorized in organic agriculture. So that, it is necessary to find an alternative system to improve the percentage of natural rooting of the different cultivars. In this sense, some attempts have been made by using different organic products. Thus, Suarez et al. (2002) did not obtain higher rooting percentages with any of the products tested (commercial products included Auxym oligo, Roots and Myco+AA) compared to when no treatment was applied with 'Picual', 'Manzanilla' and 'Picudo'. Centeno & Gomez del Campo (2008) reported the use of some natural auxin sources such as germinating seeds, fungi and alga extract with 'Cornicabra'. However, they did not obtain higher rooting percentages than the IBA control with any of the products used with the exception of 'Terrabal Organico TM ' which is obtained from a soluble fraction of an extract of macerated cereal seeds. Nevertheless this compound could produce a toxic effect on cuttings when treatment duration was increased and the subsequent influence on leaf growth has been not yet established.
Here we report a study of the rooting efficiency in olive cuttings of 'Arbequina', 'Hojiblanca' and 'Picual' by inoculation with different bacteria selected on the basis of their physiological traits as PGPR.

Bacterial strains used and culture conditions
Strains AG9 and AG13 were isolated from plant nursery well water (Montero-Calasanz et al., 2006). Strains CT348 and CT364 were isolated from rhizosphere soil samples collected in an organic olive grove located at Mairena del Aljarafe (Sevilla). A. brasilense Cd (ATCC29729) was obtained from the American Type Culture Collection.
Bacterial isolates were grown in nutrient broth (NB) medium and A. brasilense Cd in modified NFb medium (Döbereiner, 1980). When needed, culture media were supplemented with 100 mg L -1 Ltryptophan as an IAA biosynthesis precursor. All of the bacterial strains were incubated at 28°C on a shaker at 180 rpm for 48 h until they reached about 10 9 colony-forming unit (CFU) mL -1 . IAA production was assessed in the bacterial cultures before they were used in the assays. Bacterial solid peat inoculants were prepared as previously described (Daza et al., 2000). Viable PGPR cells from inoculants bags were estimated by plating 10-fold serial dilutions on NB agar and ranged from 0.5 to 1 × 10 9 CFU g -1 .
In vitro analysis of putative PGPR characteristics IAA production. IAA quantif ication was determined by the modified method described by Patten & Glick (2002). Briefly: 150 mL of H 2 SO 4 were mixed with 250 mL of distilled water and then 7.5 mL of cold 0.5 M FeCl 3 were added. This reagent was added to a culture supernatant (4:1 v/v reagent vs supernatant), mixed gently and, after 20 min in the dark, the absorbance was measured at 535 nm. The concentration of each sample was calculated from a standart plot ranging from 0.01 to 0.4 µg mL -1 pure IAA (Sigma).
Inorganic phosphorus solubilisation assay. The ability of the bacterial strains to solubilise tricalcium orthophosphate was tested on PVK medium (Pikovskaya, 1948). For each isolate 10 µL of a bacterial suspension (about 10 9 CFU mL -1 ) were dropped onto a Petri dish containing solid PVK medium. After incubation at 28°C for 3 days the clearing of the initially turbid medium indicated the presence of a phosphate solubilizing bacterial isolate.
Siderophore production assay. Semi-quantitative analysis of siderophore production by the bacterial strains was performed following the chrome azurol S (CAS) method of Alexander & Zuberer (1991). Ten microlitres of bacterial inoculum was spotted onto Petri plates and incubated at 28°C for three days; the discoloration of the medium (blue to yellow-orange) indicated siderophore-producing bacterial strains.
Determination of ACC-deaminase activity. ACCdeaminase activity was determined according to Penrose & Glick (2003). Bacterial cultures grown overnight (28°C, 180 rpm) in tryptic soybean broth (TSB) were centrifuged and pellets were washed twice with Dworkin and Foster (DF) salts minimal medium without nitrogen. Then, the cells were suspended in DF salts minimal medium with 3.0 mM ACC, which induces ACC deaminase activity, and incubated overnight under the same conditions. Cultures were then washed twice in 0.1 M Tris-HCl, pH 7.6. and finally suspended in 0.1 M Tris-HCl, pH 8.5. Then, 30 mL of toluene was added, to labilize the cell membranes, and the cell suspension was vortexed.
Using two tubes for each bacterial strain, 20 mL of 0.5 M ACC was added in one of the tubes. Both tubes were incubated at 30°C for 15 min. Following the addition of 1 mL of 0.56 M HCl, the mixture was vortexed and centrifuged. One mL of supernatant was mixed with 800 mL of 0.56 mM HCl, and 300 mL of 0.2% 2.4-dinitrophenyl-hydrazine in 2 M HCl was added. The mixture was vortexed and incubated at 30°C for 30 min. Two milliliters of 2 M NaOH were added for colour development which was measured at 540 nm. The absorbance of the tube without ACC was subtracted from the absorbance of the tube plus ACC and then compared to an a-ketobutyrate standard curve to calculate the amount of a-ketobutyrate generated.

Gnobiotic assays
Rooting assay. The effects of bacterial strains on the development of adventitious roots on mung bean cuttings (Vigna radiata) were assessed as described by Mayak et al. (1999), with minor modifications. Mung bean seeds (Wilczek) provided by Penelas Legumes, S.A. (León, Spain) were disinfected by 30 s treatment with 96% ethanol, 6 min with 6% sodium hypochlorite and six washes with sterile distilled water. Then seeds were placed on water agar plates (1% agar) at 28°C for 24-48 h in darkness to stimulate the seed germination.
Seedlings were sowed individually in glass tubes (3 × 25 cm) with cotton plugs containing quartz sand and nutrient solution (Rigaud & Puppo, 1975) under sterile conditions. Tubes were maintained in a growth chamber (Sanyo Electric Co., Japan) at 20-25°C, 80% RH and photoperiod of 12 h (4 LS). After 1 week, stems were cut, above the neck. Each cutting (8-10 cm in length) was placed in a glass tube (2.5 × 15 cm) with cotton plug containing 6 mL sterile distilled water (negative control) or 6 mL bacterial suspensions (bacterial cultures centrifuged, washed twice with sterile distilled water and suspended in the same volume of water). Ten cuttings per treatment were incubated during 10 days in a growth chamber under the environmental conditions described previously. After this period, the number and length of adventitious roots were measured and analysed.
Canola root elongation assay. The effect of bacterial strains on root elongation from canola (Brassica napus) seeds was assessed by the method outlined by Patten & Glick (2002). Canola seeds (Thunder) provided by Department of Biology of the University of Waterloo (Ontario, Canada) were disinfected by 1 min treatment with 70% ethanol, 10 min with 1% sodium hypochlorite and six washes with sterile distilled water. Seeds were incubated with 4 mL of a suspension of each of the bacterial cells (final absorbance at 600 nm equals 0.15) in sterile 0.03 M MgSO 4 or in just sterile 0.03 M MgSO 4 (control) for one hour. Then, seeds were placed into sterile growth pouches (Mega International, Minneapolis, MN, USA) filled with 12 mL of distilled water and placed in a growth chamber under the same environmental conditions previously described. Each bacterial assay (nine growth pouches with eight treated seeds each) with its correspondient control, was performed independently. Two empty growth pouches (water only) were placed at the ends of each rack so that plants at the end of a rack were not subjected to extremes of light or air circulation. Primary root lengths were measured and analysed after 7 days. Seeds that had failed to germinate 3 days after they were sown were marked, and roots that subsequently developed from these seeds were not included in the final root length calculation.
Olive nursery assays. Three nursery assays were carried out in fall of 2006 (Assay 1), winter of 2006 (Assay 2) and summer of 2007 (Assay 3), using 'Arbequina', 'Hojiblanca' and 'Picual' olive semihardwood cuttings (12-15 cm long) with two pairs of leaves at their tops. These were collected from vigorous one-year-old shoots of pathogen-free, 10-15 year old trees in Burguillos (Seville, Spain). The assays were performed in a commercial nursery taking advantage of its propagation squeme. Cuttings were inoculated in two different ways: in treatment 1, the cutting basal ends (1 cm) were submerged for 24 h in a bacterial culture whereas in Treatment 2, they were basally immersed for 4-5 s in a mixture of peat-based solid inoculant and sterile water just to form a film on the cutting basal end. In each assay, negative control by basal immersion in nursery water for 24 h and positive control by basal application of IBA solution (3000 mg L -1 ) for 5 s were used. Cuttings were planted at 2 cm depth in alveolar trays of 130 alveolus of 2 cm long × 2 cm wide × 4 cm depth containing mixed damp coconut fibre and perlite (3:1) (Assays 1 and 3). In Assay 2 only damp coconut fibre was used in order to simplify the transplant from alveolus to pots according to a nurseryman's request.
After planting, cuttings were placed in a propagation tunnel, built inside a climatised greenhouse, provided of electronic controll of air (20°C) and bed (22°C) temperatures. The cuttings were intermittently moistured by spraying water from 0.3 mm nozzles controlled by an electronic sensor. Two leaf treatments of aqueous solution of 0.3% Brotomax ® (8% N, 1.8% Cu, 0.75% Mn, 0.5% Zn and natural plant extract, from Campbells Fertilisers, Australia) were applied on weeks 4 th and 8 th of the mist period.
All bacterial inoculants, plus the negative and positive controls had two replicates of 130 cuttings each, and were randomly distributed in the propagation tunnel. After 60 days ('Arbequina' and 'Picual') or 74 days ('Hojiblanca'), the porcentage of rooted cuttings per treatment was calculated.

Strain identification
Specif ic primers for 16S rDNA (forward primer b341: 5'-CCTACGGGAGGCAGCAG-3' and reverse primer b758: 5'-CTACCAGGGTATCTAATCC-3') (Juck et al., 2000) were used for the identification of each bacterium. Isolation of genomic DNA was done using the 'Wizard Genomic DNA Purification Kit' (Promega, USA) in accordance with the manufacturer's specifications. PCR amplification was performed by using 'Go Taq Master Mix' from Promega and the PCR conditions were: one cycle at 94°C for 1 min, and 29 cycles at 92°C for 40 s, 55°C for 40 s, 75°C for 1.5 min, followed by a final 5 min at 75°C. DNA sequencing was performed by the DNA sequencing facility at the Department of Biology of the University of Waterloo (Ontario, Canada).
The 16S rDNA gene sequence was edited and assembled by BioEdit v.7.0.9.0. software (Hall, 1999), carrying out an alignment with those of phylogenetically closely related organisms (as determined by EzTaxon server) (Chun et al., 2007). The 16S rRNA gene sequences were deposited at DDBJ/EMBL/ GenBank database.

Statistical analyses
Analysis of variance (ANOVA) was performed for the olive nursery assays and for the gnobiotic mung bean assay data, using the software program Statistix (version 8.0, NH Analytical Software, USA) for Windows. When the analysis of variance showed signif icant treatment effect, the least signif icant difference test (LSD, p < 0.05) was applied to make comparisons between the means. A Student's t-test was used to assess differences in root length of canola seeds between inoculated and non-inoculated growth pouches.

Strain characterization and identification
Bacterial isolates were characterised for a range of plant growth promoting traits including auxin and siderophore production, phosphorus solubilization, and ACC deaminase activity ( Table 1). All of the strains tested were auxin producers, with the maximum amount of IAA being accumulated at the stationary growth phase. Strain Pseudomonas sp.CT364 was the only phosphorus solubiliser and the best siderophore producer. Strains Pantoea sp. AG9 and Chryseobacterium sp. CT348 produced siderophores but to a lesser extent. The capacity to produce ACC deaminase was only exhibited by strain AG9.
Comparing the 16S rDNA gene sequences of the isolates with the Eztaxon (http://www.Eztaxon.org) database, the highest similarities were obtained with members of well known plant associated bacterial families as Pseudomonadaceae and Enterobacteriaceae (Table 1).

Gnobiotic assays
Rooting assay. All of the bacterial treatments increased both the number of adventitious roots and the total root length from three to f ive fold when compared with the water control (Table 2). It is noteworthy that roots developed by cuttings placed in a bacterial suspension of strain Pantoea sp. AG9 were much longer (on average, 8.9 mm) than cuttings placed in any of the other bacterial strain suspensions (on average, from 4.2 to 6.2 mm) or in water (on average, 4.9 mm).
Root elongation assay. All of the tested strains were able to signif icantly stimulate the growth of canola seedling primary roots after seven days treatment (Table 3), with plants treated with strain Pantoea sp. AG9 having roots that attained the longest length.   used to inoculate 'Arbequina', 'Hojiblanca' and 'Picual'. The data showed that, after 60 days, some of the bacterial strains produced a significantly higher rooting percentage than observed with IBA treatment for the three cultivars (Table 4).

Olive nursery assays
In 'Arbequina' the best results were obtained when strains Cd and AG9 were applied by dipping the cuttings in the bacterial culture (Method 1). Strain AG9 provided similar results to the IBA control in Method 2. Strain Cd was a poor inducer of rooting in cv. Picual, while strains AG9 (both methods) and AG13 (Method 2) showed a high root induction efficiency. In 'Hojiblanca', strains AG9, AG13 and Cd produced similar or even higher rooting percentages than the IBA control. However, at 60 days after inoculation (dai) bacterial treated cuttings of this last cultivar displayed big callus and scarce emerging roots, requiring two additional weeks in the mist tunnel. This phenomenon was observed in all the performed assays.
Assay 2. All three strains used in Assay 1 and two new isolates obtained from an organic olive grove were included (Table 5). In 'Arbequina' all of the strains yielded good rooting in Method 1; however, only Pantoea sp. AG9 and Chryseobacterium sp. CT348 were as high as the IBA control in Method 2. In 'Picual', the strains AG9 and CT348 were similar to the IBA control in both methods while the other strains were less effective.
In 'Hojiblanca', all the five strains used induced rooting as the IBA control in both treatments.
Assay 3. In this assay only 'Arbequina' was used. In Method 1 strain Pseudomonas sp. CT364 was significantly more efficient than the IBA control while the other isolates did not differ from this positive control Rooting induction of olive cuttings by auxin-producing bacteria 151   (Table 6).

Discussion
The use of model plants provides methodological tools for comprehending the PGPR action mechanisms. Besides, they utilize a simple and idealised system where it is possible to check the biotechnological potential of the studied bacterium. The use of Vigna radiata and Brassica napus have been reported widely by others (Mayak et al., 1999;Penrose & Glick, 2003;Shaharoona et al., 2006;Rashid et al., 2012). They have been mostly used in studies on the involvement of phytohormones, more specif ically auxin and ethylene, in the initiation and development of roots (Mayak et al., 1999). In this work the employ of these model plants has facilitated the pre-screening of a large number of PGPR and the subsequent selection of the more promising strains to be used for rooting in olive cuttings.
The effect of the inoculation method was studied by applying the inoculum in two different ways (Methods 1 and 2). The number of adhered bacteria was similar in both of them, and ranged between 10 5 and 10 7 cells/cutting (data not shown). Although both methods require similar and nursery-manageable amounts of microorganisms, probably Method 2, where cuttings are inmersed in a peat-based inoculant, is easier to commercialise and apply under nursery conditions owing to its similarity to the way in which peat based inoculants have been used since the end of the 19th century (Voelcker, 1896). Moreover, it has been described that some liquid inoculants (Method 1) quickly lose viability (Date, 2000). İsfendiyaroglu et al. (2009) found differences in 'Ayvalik' rooting when they tested twenty-five different types of rooting media. The substrates used in this study were a mixture of coconut fibre and perlite (Assay 1 and 3) or coconut fibre alone (Assay 2), substrates frequently used in olive nurseries. In assay 2, where perlite was removed from the rooting media, the high moisture level observed negatively affected the rooting of the olive cuttings, according to İsfendiyaroglu et al. (2009), being cv. Picual the most affected.
Inoculation with Azospirillum induces the proliferation of lateral roots and root hairs (Glick, 1995;El-Khawas & Adachi, 1999). A. brasilense can also, effectively bind to a variety of surfaces including plant stems, as well as carnation cuttings (Li et al., 2005). In these previous studies, the inoculum was usually applied as a liquid culture. We have obtained good results with strain Cd using this method of inoculation (Method 1) but when it was applied as a solid inoculant (Method 2) rooting was signif icantly reduced in cultivars Arbequina and Picual. Other bacterial strains (AG13, CT348 and CT364) show a similar behaviour with cv. Arbequina. However, all the strains were as effective as the IBA control under both inoculation methods with cultivar Hojiblanca. This phenomenon could be explained in terms of different cultivar/strain interactions. In this sense, we also have observed a strong interaction between the effectiveness of the bacterial strain and the olive cultivar. Some strains that produced a high rooting percentage with one cultivar were nearly ineffective with other cultivars. Similar types of interactions have also been reported with other soil microorganisms (i.e. mycorrhizal fungi) and olive cultivars (Citernersi et al., 1998;Porras-Piedra et al., 2005;Soriano et al., 2006). Moreover, slight differences have been observed in the cultivar Hojiblanca with all of the bacterial strains, including strain Cd. Probably the bacteria-cultivar interaction mechanisms in this cultivar are different than in the others. In fact, our results showed that this cultivar needs more time in the mist tunnel before  suitable rooting is obtained, independently of the season. The strain AG9 is the only strain that always produced high rooting percentages independent of the mode of inoculation and in all of the tested cultivars. As it is not the best auxin producer, its effectiveness is likely explained by other mechanisms. In fact, it is the only strain tested that can synthesize the enzyme ACC deaminase. Bacteria possessing this enzyme have been described as highly effective PGPR owing to the fact that they can reduce the amount of ethylene produced by the plant and thereby promote root elongation (Penrose et al., 2001;Glick et al., 2007). In addition, our elongation results in mung bean cuttings agree with those obtained by Mayak et al., (1999) who observed that bacterial strains that either overproduced IAA or did not produce ACC deaminase yielded a large number of very small adventitious roots. On the other hand, also in agreement with Mayak et al. (1999), the use of bacteria with normal levels of IAA and ACC deaminase yielded plants with a moderate number of much longer roots. Strain AG9 was most similar to Pantoea sp (99%) or Erwinia sp (> 97%). In fact, some species of these genera have been renamed (e.g., Erwinia herbicola has been changed to Pantoea agglomerans) (Loiret et al., 2004) and its actual taxonomic position is not well established.
Our findings support the possible use of alternative methods for olive mist propagation under a context of 'organic agriculture system'. Although additional studies are needed, the results presented here strongly suggest that it is possible to obtain effective rooting induction of olive tree cuttings by the treatment of those cuttings with some plant growth-promoting bacteria that minimally synthesize and secrete the phytohormone IAA and actively produce the ethyleneregulating enzyme ACC deaminase. The method described here is a good starting point for a simple and straightforward approach that could easily be optimized by the olive nursery industry.