Seven farms, located near Ugento (LE, Italy), provided access to their olive trees, which age ranged from fifteen to centenary, and were mainly from cv. Ogliarola di Lecce, although afew were also from cv.Leccino (Table 1).

A commercial symbiotic BF consortium (Micosat F ®, CCS-Aosta, Figure 1) was used at a dose of 20 kg ha^-1. The root inoculation was a 15 cm deep scarification in which the microbial fertilizer was distributed. A furrow (Figure 1), which allows the surface roots of the plants to be inoculated, was made close to the trees. ¹Micro Organisms (MOs) from finely ground cultivated Sorghum sudanensis roots, containing spores and hyphae of Funneliformis coronatus GO01 and GU53, F. caledonium GM24, F. intraradices GB67 and GG32, F. mosseae GP11 and GC11, F. viscosum GC41; saprotrophic fungi: Streptomyces spp. ST60, Streptomyces spp. SB14, Streptomyces spp. SA51, Beauveria spp. BB48, Trichoderma viride, Trichoderma harzianum TH01, Trichoderma atroviride TA28, Trichoderma spp.; rhizosphere bacteria: Bacillus subtilis BA41, Pseudomonas fluorescens PN53, Pseudomonas spp. PT65 and Pochonia chlamidosporia, in the relative percentage of 40% crude inoculum and 21.6% bacteria, and saprotrophic fungi.

A four-point survey card was drawn up (Table 2)

The codes were processed on the farms using Friedman’s unpaired tests (StatBox V6.5, Grimmersoft, Paris).

According to the indications of previous works 12,14,15, the in vivo raw leaf pH measurements were carried out using a BORMAC pH-meter "XS pH 70" (www.giorgiobormac.com), range pH 0 ÷ 14, two decimals, supplied with a Hamilton Peek Double-PoreF, / Knick glass-plastic combination electrode, dimensions (LxØ) mm 35 × 6. The measurement was conducted on a packet of leaves pierced by a gimlet and a sensitive electrode was inserted into the thus obtained hole. Other types of electrodes were found to be unsuitable for this type of measurement. In order to understand the direction of the variations in the variables, it seemed appropriate to express the concentration of hydrogen ions [H ⁺], or H = 10 ^ -pH * 10 ^ 6, rather than the potency (₁₀) of ^H+ (pH). In this way, an increase in acidity - linked to the greater vigor of the plant - algebraically corresponds to an increase in [H⁺] and not to a decrease in pH.

The NIR spectrum of the upper leaf blade was detected using a smart-NIR-SCÏO^TM spectrometer (Consumer-Physics, Tel Aviv), which operates in the NIR 740-1070 nm band. The chemometric elaborations were carried out by means of the SCÏO-Lab software, which operates using AKA (As Known As) recognition matrices, built by means of a Random-Forest algorithm, and provides a percentage of recognition of the cells of the matrix (Table 3).

A series of leaf composition models were obtained from NIR-SCÏO^TM foliar spectra, which had previously been obtained in a sorghum experiment 15. The equations were applied to the olive leaf spectra to obtain an estimate of several constituents. The predicted values were analysed by means of a bi-factorial GLM linear model (SAS 9.1).

As described in a recent paper 16, the litter-bag probes of ground hay (Figure 2) were buried close to secondary and superficial roots in a treated area (S- Symbiotic) or in an equivalent untreated area (C- Control). After being buried for about sixty days, the probes were removed from the ground, brushed and dried at a medium temperature (<60 °C). They were then examined by means of NIR-SCÏO^TM. The chemometric elaborations were identical to those shown above for the leaves.

It was possible to have access to an NIR-SCÏO^TM equation, taken from unpublished results of an experimental trial on tomato plants, that provides an indirect estimate of the respiration capacity of the soil (symbol R), measured according to Anderson and Domsch 17. After having added sugar to the soil sample, it was possible to measure the aerobic activity, induced by the microbial consortium, using the infrared meter. The correlation between the estimated and measured data resulted to be sufficiently high (R²_{cross-validated} 0.86) to be considered reliable under comparable conditions.

Table 4 presents the results obtained from the monitoring of 484 plants on the seven farms, from which 379 determinations of the leaf pH, originating from 1516 leaf samples, were made with 3048 NIR-SCÏO^TM spectra, while 380 NIR-scans were obtained from the litter-bags from the soil.

On the basis of the elaboration of the codes carried out on each farm, by means of the unpaired Friedman’s tests reported in Table 4, the degree of disease severity was significantly reduced in two cases: C (-40%) and D (-11%), while it became significantly worse in two others, namely G (+ 8%) and - especially- F (+ 164%), a farm where the great stoniness of the soil made it possible to only inoculate the part of the field that was most affected by the disease. The quantity of basal suckers decreased significantly following the treatment in two cases (C and F), while it increased abnormally (+ 278%) in the G groves. The regrowth of the main branches increased significantly in two cases (D and F) and almost significantly on farm A, while their presence was reduced by 43% on farm C. The regrowth of the secondary branches significantly decreased in two groves (C and G) and on average increased in B.

Overall (Table 5), the S leaves were recognized as S at 66% (P <0.0001), a significantly greater value than the 54% found in the controls, which, however, was still statistically significant as there were numerous samples. However, higher-level fingerprints appeared on individual farms, with an average 74% for SS and 66% for CC.

A greater reflectance was observed for the leaves of the Symbiotic olive trees (Figure 3).

In addition to a physical difference, a chemical variation also appeared in the leaves three months after the treatment. Their chemical composition resulted differentiated for all of the seventeen considered traits (Table 6), all of which showed significant Farm * MO interactions; however, when the symbiotic effect was considered, 7/17 of the variables responded to BF with an increase (water, ADF, non-digestible NDF, ADL, ether extract), while 5/17 responded with a decrease (cellulose, hemicellulose, total digestibility and crude protein). Among all the variables, the crude protein in the leaf emerged because of its high correlation (+0.90) with the variation in the disease severity degree (last column in Table 6), as clearly shown in Figure 4, where a positive relationships linked the two traits, which means that a decrease in the Ln(S/C) of protein favored a reduction in the disease.

The a priori percentage threshold was 50% for CC and for SS. It is shown, in Table 8, that the S symbiotic samples were recognized as S at 70% (P <0.0001), a significantly greater value than the 57% found in the C controls recognized as C, which is almost significant (P = 0.0569).

However, higher-levels appeared on individual farms , with an average fingerprint of 80% for SS and of 73% for CC.

The average spectra of the two BF types are displayed in Figure 5. It appears that a greater reflectance of radiation distinguished the S litter-bags. The average S/C ratio increase is constant at around + 8% (Table 9). On the other hand, the standard deviation of the S spectra is reduced by 16%, compared with the corresponding C, and this is a sign of greater homogeneity. Figure 6 shows the two trends of the upper level and a minor dispersion pertaining the S spectra.

The respiration capacity of the soil, as estimated by means of the NIRS of the litter-bags, increased significantly after the symbiotic treatment at the roots and in the soil on farm E (+ 38%), while it appeared somewhat decreased on G (-17%) and B (- 11%) (Table 7).

A descending parabolic curve (Figure 7) shows that the disease decreased (favorable) when the fingerprint of the S leaves recognized as S was high, and vice versa.

Therefore, the symbiotic treatment increased the homogeneity of the leaf.

The litter-bag fingerprint appeared very different from the foliar NIRS fingerprint. In fact, an ascending parabolic trend is shown in Figure 8: when the disease decreased, the fingerprint of the Control litter-bags was lower. Vice versa, when the fingerprint of the Control litter-bags increased, the incidence of the disease increased. Therefore, the symbiotic treatment increased the heterogeneity of the litter-bags.

All the variables provided in the study are summarized in Table 10. Excluding farm F, the remaining six are represented in a row in the table, in the form of the symbiotic effect d_S/C equivalent to Ln(S/C). The holistic PLS regressive model for the disease severity degree is reported in Table 11, and the scatterplots of the measured vs. predicted values are reported in Figure 9.

The mitigation effect on the disease appeared to be highly determined (R² =0.96), and the model passed the cross-validation process, obtaining an R² =0.87. There were six characteristic factors:

1) the acidity differential, with standardized factor d_H = -0.155, had a negative sign as the factors were opposite: symbiotic BF lowered the pH, raised the H+ and therefore reduced the disease;

2) the fingerprint of the CC litter-bags (L_CC = +0.281) had a positive sign: when the value was reduced, the pathological degree diminished, a sign that the BF had produced some effects;

3) the fingerprint of the SS leaves (F_SS= -0.301) had a high value and a negative sign: when the value was increased, the disease was reduced;

4) the soil respiration had a favorable negative sign (-0.209): when the respiration increased, the incidence of the disease decreased;

5) the water content of the leaves accounted for -0.133 units, which means that a greater quantity of water flowed and remained in the olive leaves during mitigation and recovery;

6) the crude protein accounted for +0.350 units, the highest contribution to the fitting.

The other dependent variables of the visual appraisal were less predictable. The standardized coefficients in Table 11 show a modest adaptation for the regrowth of the main (R²= 0.72) and secondary branches (0.59) and for the quantity of the basal suckers (0.48). However, none of these models passed the cross-classification.

The attempt to obtain a direct calibration of the average NIR spectra of the Control litter-bags from six farms was successful (Figure 10): the partial least squares reached an R² in calibration of 0.945 (SEC 3.8%) and was stable in cross-validation mode (0.716; SECV 9.3%). The correlation (not shown) was non-uniform and showed seven nm ranges with higher resolution of the relationships.

Overall, the soil biota played a key role by provide information and improved insights in this work. There are two environments in nature in which the highest known microbial densities are reached, that is, the human intestine 18and plant roots 11 (1 x 10¹³ and 1011per ml/g, respectively) 19. The inoculated BF appeared to influence in an almost paradoxical manner being in a very minor with respect to the microbial mass of the soil. According Fausto et al. 20, bacterial communities are more abundant than other communities in the soil of olive groves and reach the leaves through the xylem sap. They observed differences between a conventional and a more sustainable management system in the composition of soil bacterial communities. Hartman et al 21 also found that cropping practices can manipulate the abundance patterns of the root and soil microbiome species , and thus pave the way toward smart farming.

The response to symbiotic inoculation was first evaluated indirectly, considering the respiratory capacity of the soil in the control and in the inoculated modes; the results of the pairwise comparisons resulted in a negative sign (-0.209), in agreement with a favorable descent of the disease in the holistic model.

Among the agronomic measures of coexistence with X. fastidiosa, the working of the soilappeared beneficial 22,23. The soil in control plots have not been cut in the present work. Faced with this protocol anomaly, it was observed that, in the presence of disease 10, harrowing the soil of the control plants did not slow down the decline in the disease, and a Dentamet ® treatment was necessary to improve the survival of the trees.

The symbiotic treatment increased the homogeneity of the composition of the S litter-bags, as revealed by the higher fingerprint of the S type which were more prominent than the C types (70% vs. 57%). Therefore, the BF inoculum overall increased the functional homogeneity of the biota in the soil responsible for the litter-bag degradation. However, this soil-mechanism was unrelated to the mitigation of the disease, and this variable was in fact excluded from the holistic model. On the other hand, the fingerprinting of the control litter-bags recognized as the control ones was included in the holistic model with a positive sign (coeff. +0.281) and appeared as unfavourable for a reduction in the pathological intensity. Moreover, the observed parabolic trend means that when the percentage of fingerprinting CC was low, probably because of a numerical reduction and / or non-compliance of the MOs in the soil, the inoculation may have led to more favorable mitigation of the disease results: the first key for a successful inoculation has therefore been identified as a probably existing high degree of variability in the soil, which is favorable for welcoming the BF consortium. On the other hand, where a “strong” autochthonous biota or unfavorable settlement conditions oppose the acceptance of BF, a positive result is less probable and can also generate negative results, as on farm G.

The close relationships among the average NIR spectra of the litter-bags obtained without BF and the future results expected from a BF inoculation represents a possible extension for this quick method for a rapid soil fertility evaluation: a kind of “symbiometer”.

In the holistic model, the fingerprints of the Symbiotic leaves revealed a -0.301 value vs. the disease severity evolution, a sign that the MOs have had a maximum effect on both the leaves with higher fingerprinting, and on the plants, by reducing the symptoms. Instead, where the S leaves were fingerprinted less, the results were non-significant, and on one occasion were negative.

Previous research 10 showed that Dentamet® treated trees always had a mean NDVI index per tree greater than 0.60, with maximum relative peaks in parts of the tree canopies of 0.85, in both cv. Cellina di Nardò and cv.Ogliarola salentina trees; on the other hand, untreated trees, but still with vegetation, showed a mean NDVI index of 0.45. As far as the plant vigor in the present work is concerned, we did not use the visible red range, and the true indexes, such as NDVI and NDRE, are therefore not available for comparisons. However, the greater absorption by the leaves of the Symbiotic olive trees indicates a greater presence of substances capable of interacting with NIR radiation.

Accurate disease detection vs. healthy plant results have recently been obtained, by means of coordinated research on remote and local sensing 24, and these results have been confirmed by means of the quantitative polymerase chain reaction; when high-resolution fluorescence, quantified by three-dimensional simulations and thermal stress indicators, were coupled with photosynthetic traits sensitive to rapid pigment dynamics and degradation, 80% was exceeded. Moreover, it was found that trees that had visually been found asymptomatic, but then remotely labelled as diseased, later developed X. fastidiosa symptoms at almost double the rate of the asymptomatic trees that had been signalled as safe.

In the present work, we showed that several spectral plant-trait alterations in the foliar part of the plant (not the canopy) are linked to - or independent of – the mitigation or recruitment of the symptoms. The key mechanisms shown in the present work were reductions in leaf protein, cellulose and hemicellulose, which were balanced by an increase in ADF and lignin, together with an enhanced water content. Although the latter finding may be accounted for by considering a greater fluidity in the xylem vessels and / or increased osmotic pressure in the phloem, which may be attributable to pathogen inactivation, the other variations remain open questions. In the case of grapevines with Pierce’s diseases, symptomatic leaves developed lower water potentials than healthy plants around noon each day, which was related to a reduction in the osmotic potential of the leaf. The discovery that water stress was associated with X. fastidiosa, even though wilting had not been observed, provided significant clues to help understand the disease 25.

In this regard, a recent work 26has shown that ABA is not the key factor in controlling the stomatal closure of AM-inoculated olive plants under drought conditions. In fact, other AM-related factors are involved in the control of stomata regulation in mycorrhizal olive plants exposed to severe drought. These factors act specifically on the ‘Zarrazi’ drought-resistant cultivar, thereby permitting suitable stomata behaviour.

The reinforcement of cell wall parts during the mitigation processes can help protect the leaf from water exchanges. Moreover, along with the progression of disease severity 27, hydroxytyrosol glucoside disappeared (- 92%), while quinic acid increased more than three-fold, thus suggesting that changes in the levels of the two phenolic compounds could be reputed as markers of the bacterial infection in olive trees.

The protein reduction was in contrast with the expected effects of AM, as observed in a previous study with Sorghum15. Moreover, a test on AM inoculation in cv. Picholine marocaine olive plantlets28 showed that the leaf protein increased by about +21% under normal water conditions and by about +145% under severe water stress; in fact, in that case, the AM had responded to water stress with -24%, while the control plants registered -62%. A negative protein sign is generally indicative of a pathogenic attack: in greenhouse trials, leaf spot injury, caused by inoculation with various pathogens, reduced the crude protein content of infected alfalfa leaves by 22%, compared with a healthy control 29.

According to Lavee and Avidan 30, an alternate bearing of olive trees can modify the protein content in the leaves, in the sense of a lowering in the foliar protein in "off" trees compared to "on" trees. In our experiment, the attenuation of the foliar protein structures would seem to indicate a plant mitigating of the disease to one that starts to have a condition of limited production.

An increased concentration of expansins was observed in the vessels of the diseased plants 31. It is plausible to assume that, as a result of the synthesis of these proteins in the stem, the plant recycles leaf amino acids transferred by means of the phloem pathway.

Contrary to the general expectations about a pH decrease after BF inoculation, and reinforced by our early published result 12on healthy olive plants (-4% in BF inoculated symbiotic plants), the leaf pH here did not decrease, and in fact increased in one case. However, the general sense of the [H⁺] - complement of the pH - was negative (-0.155), thus in agreement in the holistic model with mitigation.

The rapid methods applied in this experiment do not make it possible to define whether the good results are due to an effective mycorrhization of the roots or to some positive microbial interaction in the rhizosphere with endophytic MOs, whose presence had been verified in the highly complex crude inoculum of the Micosat® containing 350 species of fungi and 1500 of other species 32.