Three categories were considered: i) Arbuscular Mycorrhizae (AM), with thirteen theses of crude inocula belonging to the Glomus genre from the USA (AM01-AM13); ii) Microbial Biota of the soil (MB), with two commercial approved bio-fertilizer consortia (MB1-MB2); iii) a non-inoculated Control thesis (C) (Table 1).

Sorghumsudanensis seeds from cv Puma (Padana Sementi Elette S.r.l. Tombolo, Italy) were inoculated at seeding in 10 L pots (three seeds per pot), using 12 g pot^-1 of crude inoculum. The growth substrate was volcanic lapillus rock (5 L) and peat (5 L). The substrate was kept in the greenhouse and drip irrigation was adopted.

The leaves were harvested after 66 d and measured to establish the raw pH of the leaves by means of direct contact of the split central vein with a combined plastic-glass Double Pore Knick® electrode (Hamilton, Reno, USA) using an XSpH 70 pH meter (Giorgio Bormac S.r.l., Carpi, Italy).

The whole leaves from each of the 16 theses were chopped and dried at 60 °C to a constant weight, air-equilibrated, ground in a Cyclotec mill and stored for later analysis by means of NIRS.

The ground leaves were repeatedly examined (three scans) with a Spectrum IdentiCheck^TM FT-NIR/MIR system (Perkin-Elmer, Beaconsfield, Buck, England) and the chemical composition was predicted via NIRS, using equations that were established on twelve different forage crops, as reported in Tassone et al.20.

The raw pH of the leaves appeared as a very narrow-distributed variable, which was characterized by a variation coefficient of 3.8% (Table 2).

The normal mean value of 5.18±0.03 in the leaves from the C non-inoculated plants was significantly decreased to 5.10±0.02 in the MB category (-1.5%, P=0.0336), and decreased to a value of 5.00±0.01 in the AM category (-3.5%, P<0.001). Moreover, the AM vs. MB difference was highly significant, with the MBs being less acidifying. When the individual theses (Table 3a and Table 3b; Figure 1) were considered, the differences appeared more scaled.

As far as the biofertilizers are concerned, MB1 (-1.1%) was more acidifying than MB2 (-2.0%), and the AM were scaled according to three gradients of acidity: the most acidic strains, that is, below -4%, were the AM01, AM13, AM05, AM02 and AM08 strains.

The weight of the leaves harvested from the plants was much more reduced in the non-inoculated Control than in all other theses (Table 3a). Moreover, the foliar composition was affected by the category of the inoculant, as calculated from Table 3a and Table 3b. The MB category scored 7 significant increments and 3 decrements vs. the C, while the AM category scored 5 and 1 variations, respectively.

As can be observed in Figure 2, the constituents augmented in the leaves of the inoculated Sorghumsudanensis were: crude protein, ash, digestible NDF, hemicellulose and the ether extract; the constituents reduced in contents and properties, namely dry matter, crude fiber, cellulose and crop maturity index, were increased less.

The PLS standard model (Figure 3), with R²=0.81, showed that a low pH steadily favoured the dry mass weight and, to a lesser extent, the hemicellulose, digestible NDF and ether extract contents; on the other hand, a high pH tendentially increased the dry matter percentage and the cellulose content of the leaves.

In general, the discriminations of the three categories when the SCIO instrument was used were highly significant (Table 4). The leaves from plants inoculated with AM were individuated more correctly, in the scans on the fresh leaves (96%) as well as in the scans on the dry leaves (94%); the performances pertaining to the MB leaves were significant, albeit not so high (59 and 51%, respectively), while the Control leaves were individuated more clearly (74 and 65%, respectively).

A luxuriating effect has been displayed in foliar development. A meta-analysis of the field studies on the responses of wheat to AM has highlighted that field AM inoculation can be proposed as an effective agronomic practice for wheat production7, with aboveground biomass increases of around 20%, as already assessed in maize under high11,15222324or low25Italian input conditions and under Indian conditions21

As expected, a more juvenile ontogenic status emerged for the leaves of inoculated plants but graded among the inocula types.

Claps et al.26, in field conditions, after the autumn sowing of barley (Hordeumvulgare) and clover (Trifoliumalexandrinum) tanned at 1 kg ha^-1 with the MB1 of the present work, observed a slight increase in yield (not significant), but also significant variations in composition (Figure 4), albeit only in the barley grass. As in the present work, the advantage was centred on crude protein (+18%), but in that case a strong reduction in fibrousness was found, which enhanced the digestibility coefficients of the NDF, crude protein and organic matter. On the other hand, a reduction in ash (-10%) disagreed with the increase observed in the present work.

Other AA have conducted experiments on AM at more mature stages, but hay and silages have rarely been studied. Uzun22 observed that no reductions in forage yield occurred in Sorghumsudanensis or maize, when the diminished supply of fertilizers was balanced with the action of Glomus spp. and the forage produced using an AM treatment showed better chemical characteristics when harvested at an appropriate phenological stage.

As far as maize forage is concerned, Sibi et al.27 observed that an AM fungus was significant, at a level of 5%, on water soluble carbohydrates, which increased by 14%. The Authors hypothesised that inoculation with an AM fungus increased the cytokinins, and, moving nutrients from other parts to the leaves tended to delay their senescence.

In our short modification experiment, low pH steadily favoured the dry mass weight and, to a lesser extent, the hemicellulose, digestible NDF and the ether extract contents of the leaves. If the raw pH is a real key-variable, the modifier direction in Sorghumsudanensis appears quite attractive for ruminants. Phyto-evolution studies17,28 of a subarctic flora suggest that tissue pH itself is controlled to a great extent for a given species, because of its direct or indirect functions in the plant; in fact, a low pH corresponds to poor digestibility and may therefore act as an anti-herbivore defence in the same subarctic flora.

The field surveys of AM communities over a wide range of soil pH suggest that it is also the major driving force in structuring these communities29.

It was found, in the microcosm experiments6, that N₂O emissions were cumulatively 42.4% higher without AM than with AM microcosms. However, it should be recalled that the Authors, in their tomato experiment, observed an increase in soil pH after an AM inoculum of 7.75 vs. 7.62 (+1.6%; P=0.004). Unfortunately, the raw pH was not measured in that experiment. In light of our present and previous results15,16, it is possible to hypothesize that the transfer of [H⁺] from the soil to the AM plants could have effectively reduced the circulating acidity of the soil. Several results support this hypothesis. Bago et al.30 demonstrated, in a sterile culture with Glomus intraradices, that extra-radicalmycelium utilised NO₃-N and increased pH of the medium by up to one unit. Li and Christie31 showed that soil solution Zn concentrations were lower and pH values were higher in mycorrhizal treatments than non-mycorrhizal controls, and the AM effects were more pronounced at higher Zn application rates.

The direct NIRS discrimination of bio-fertilized crops is rare. A first work on Ocimumbasilicumcv. Genovese 32 assessed that three AMs were modifiers of growth, and distributors of glandular trichomes and of the essential oil composition, with a homologous relative pattern that was appreciated by NIRS and Electronic Nose. A smart NIRS technique that has recently been applied to hay-litter-bags19has produced promising results in discriminating the fingerprint effects of bio-fertilizers. The results of direct NIR discrimination have instead been obtained for several cases. Szuvandzsiev et al.33 suggested that the effect of very different irrigation regimes on tomato leaves is detectable by means of NIRS, irrespective of the cultivar and phenological stage; around 960 nm emerged a good correlation between different irrigation regimes and the reflectance of the raw spectra of tomato leaves taken from the 900-1000 nm representative spectral range. Foliar UV-Vis-NIR spectra (325-1075 nm) was used to discriminate high oil and high protein from a normal maize type on five sampling dates over two years34; the results showed that the qualitative models constituted a robust classification method; however, the best method on average attained only 58.3±5.5% of validation, a superior value to the hazard threshold of 33%, but of no use for traceability purposes. The somewhat optimistic values reported in the present work, obtained using the smart NIR-SCIO (Tab. 4), need to be confirmed over some years and under different conditions.