Soil fertility, usually defined as the ability of a soil to promote plant growth and yield by integrating different soil functions1including nutrient availability, microbial activity and physical properties is fundamental for determining the productivity of all farming systems. Consequently, the knowledge of chemical, physical, and biological properties of a given soil is fundamental to reach a high standard production. Chemical parameters of the multifaceted soil fertility can be predicted approximately through an NIRS examination of samples in laboratory2,3,4or directly in a field in a precision agriculture framework5. However, information on chemical fertility is useless if it is not combined with the knowledge of the microbial fertility of the cultivated soil. Increasing interest in microorganisms, such as endophytes, symbionts, pathogens and plant growth promoting rhizobacteria, was observed in the literature, while less attention was paid to the larger community of soil microorganisms, or soil microbiome, which may have more far-reaching effects. Each organism in the community of soil microorganisms acts in coordination with the overall soil microbiome to influence plant health and crop productivity6.

The use of litterbags is a technique that has long been adopted in soil studies on the evolution of microfauna in the bulk soil7. However, there is still a lack of rapid measurement techniques that can assess the microbial status of cultivated soils. The integrated use of NIRS and litterbag techniques could be a functional and rapid solution, as demonstrated by the fact that a change caused by a biofertilizer is reflected in the biochemical functioning mechanisms, and that such a change can be easily testified8. The coupled use of these two techniques (intended as a quality evolution of the litterbags swamps and not as mass decay), can be modeled as a valid fingerprinting of the studied field conditions, a process that results in data validation and predictive models (such as the random forest model algorithm). Furthermore, this combination technique could be used as a rapid and cost-efficient method, especially when compared with more complex methods, such as molecular metabarcoding, which is time consuming and expensive, as well as requiring a great deal of knowledge for the data analysis9.

Rapid predictors of yield are necessary tools to advance biofertilizer tuning and management. Classic microscopic techniques are usually necessary for arbuscular mycorrhizal (AM) studies in this sector10 and were recently improved through molecular techniques thanks to semi-automated digital image analyses, that are used to measure root length and fungal hyphae of dense mycelia11. However, fungal-plant phenotypes are only weakly related or even not related to yield. For example, the correlation between AM and yield in fifteen species grown in pots, after elimination of leeks as an outlier, shifted from +0.60 (P ≤ 0.018) to -0.16 (P ≤ 0.58)12. In a study on potato, treated twice with four commercial AM (Rizophagusirregularis) inoculants13, root length colonization was measured 120 days after planting, and a correlation between root length and yieldof +0.31 (P ≤ 0.18) was observed.

In a two-year project based on maize field trials14 it was concluded that just a few rapid NIRS analyses of litterbags and leaves, together with foliar pH measurements, are able to explain over 87% of the variation in yield from biofertilized or non-biofertilized fields. Validation of the spectroscopic and foliar pH model was carried out on a different biofertilized cultivar, which provided five AM sources. The yield results expressed negative symbiosis effects, and a tendential negative response to bio-fertilizers was correctly predicted in 84% of the cases in the negative quadrant (P ≤ 0.0012). The results of the rapid analyses were confirmed to be correlated to maize quality15 and to potato yield16.

The aims of the present experiment are to confirm the Litterbag-NIRS and the Foliar-NIRS methods in a different species under soil microbial enrichment, and to search spectral correlation with biochemical parameters of the soil and with the phenotype of the litterbags in order to extend the "model learning" of the statistical analyses applied to soil microbial fertility.

A field trial, aimed to evaluating the response of tomato plants to a regeneration of the mineral soil (C), as a result of using an organic amendment (O), and by adding a biofertilizer complex with a mycorrhizal source (OM), was conducted in an over utilized soil. The Litterbag-NIRS method was used, together with chemical analyses, to testify, by means of fingerprinting, the microbiological activities of the treatments in the electromagnetic spectra.

From the present experiment we were able to observe an increase of tomato yield by 18% thanks to the use of organics amendments able to partially restore soil sickness due to several successions of tomato cultivation. Moreover, the activity of the organic amendment was little (+5%) enhanced by the supply of the biofertilizer complex. It should be also noted that the control yield (60.5 t ha^-1)was very low for the area that is usually characterized by production ranging around 65 to 95 t ha^-1.

As summarized in the model reported in Table 7, a greater availability of both N species in the first period and of NO₃^--N in the second period was the lever that led to a higher yield; it should be noticed that these two parameters had a high CV of around 130%; microbial C was the least variable parameter (CV= 48%), but it did not show covariation with the treatments, even though it registered the highest value in the first month. Microbes stimulated by organic amendments are only aerobic, thus an increase in microbial biomass could be expected. However, the response could be the result of a complex polydromic function, resulting from the multitude of autochthonous microorganisms and as a response to the glucose added to the soil in the test. Soil microorganisms have a great potential to adaptively vary their growth traits, according to the energy source and its availability20. Although, according Fontaine et al.21, a wide range of microbial types are present in soil, only a few of them can adapt to the dominant soil organic resource, while the others remain dormant i.e. in a viable but non-culturable state. After fresh organic matter (FOM) is supplied to soils, or vital newborn roots start to develop into the soil, many dormant microorganisms are triggered into activity, and this leads to dramatic changes in the structure of the microbial community. It is important to recall that at first time FOM specialized microorganisms, which are commonly classified as r-strategists, adapt to rapid growth intervals, become dominant. After substrate exhaustion, r-strategists die or become dormant because they are unable to use soil organic matter (SOM). At that point other microorganisms labelled as k-strategists occurred to attack SOM. They grow slowly and only dominate in the last stages of FOM decomposition. In the present experiment we have observed a general reduction of the microbe r-strategist from point zero (SIR 362) to point A-56 d (181; -50% from 0) and to point B-92 d (142; -21% from B), this mainly because these microbes easily attack the glucose added to soil microbial biomass determination. However, a polydromic function must describe level and evolution of SOM and microbes at different times. What did the O and OM treatments do that was different from the Control? The C and OM treatments lost more rapidly the r-strategist population vs. O, which only collapsed in the final period. The SIR technique is not enough to distinguish between the C and OM treatment; consequently, it is necessary to take into consideration the evolution of the litterbag composition which more closely reflects the population of the slow k-strategists that were able to attack the cellular walls of the hay in litterbags. Hence, it is possible to observe, in Figure 4, that the crop maturity index of the degrading hay increased to a great extent as the digestible fraction of the NDF reduced. As far as the hypothesis of a polydromic function is concerned, the organic amendment delayed the functionality of the r-strategists, but at the same time activated the k-strategists to feed the walls. Something similar occurred for the biofertilized OM treatment which, however, antagonized the rapid microbes at the control (C) level early on, as clearly shown by the presence of the saddle in Figure 2.

A new finding of hay litterbags has emerged concerned their intrinsic basal respiration. In fact, the values were nearly tenfold higher than the SIR of the soil at period zero. Sapronov and Kuzyakov22, in a laboratory, separated the soil CO₂ flux into the root respiration of maize plants and the respiration of the rhizosphere and nonrhizosphere microorganisms, and concluded that the contribution of the roots alone was 8–19%, while that of the non-rhizosphere microorganisms was 51–82%.

The results of the present experiment agree with previous results obtained from Litterbags-NIRS and from Foliar-NIRS in biofertilizer experiments. The anomaly of the pH - which increased by 3% and 14% - instead of reducing by about 1-3%, as usually observed in other experiments involving AM23,24,25 - may be due to an excessive delay between the sampling and pH measurement, but this cannot have affected the general structure of the leaves in the optical path traversed by the NIR beam. Notice that the reflectance of the leaves measured by the researchers’ version of the SCiO device (linked to a repository of collections and spectra providing a downloading management) is more useful than a Leaf Chlorophyll Meter, which has a similar price, when the influence of the factors exceeded from the chlorophyll - protein domain, for example, when an experiment is aimed at modifying ontogenic traits, in particular the development of the cell walls and to vary the Crop Maturity Index or similar complex traits18,26. The relationship with the Sentinel-2 b6 band, which on average explained 39% of the yield variations, is interesting. Therefore, this tool could be proposed for remote agriculture purposes.

The most meaningful result of the current experiment was that obtained from proximal spectroscopy, which established significant and useful correlations of several components of the litterbags with the short region of the NIR rays. The development of spectroscopy led to concrete interest in portable NIR devices27,28, but especially in SCiO systems, for the prediction of the horticultural products quality29, to maize single kernel quality33 and the application of NIR tomoscopy to animals30and the application of NIR tomoscopy to animals30. This Foliar-NIRS method therefore remains an underestimated tool for the rapid phenotyping of plants, especially when the relationships between plants and AM are studied, contextually with a lowering of the foliar pH14,19,24,25.

According to van Veen et al31, the greatest problem of inoculating soil for beneficial purposes is the general obstinacy of the soil ecosystem, which normally acts as a buffer against any incoming microorganisms. From the perspective of the attackers, provided by strong ecological selectivity, this recalcitrance of the system can be overcome, and the selected organism can become established and active. The greatest chance for the success of microbial introductions for unselected or poorly selected inoculants might be when the normal homeostasis of the system is (temporarily) disturbed, as this can result in an alleviation of anti-invader pressure. In our opinion, when considering the minimal dose\effect results that have sometimes been obtained in real fields, that is, far from artificial (and often aseptic) sterile laboratory conditions, and with very high (non-economic) doses, we can hypothesize a collateral multi-purposes player that restores some strategic functionality that is blocked in the soil. We must draw inspiration from the Van Nood et al.32 fecal transplant model first to solve the deadly recurrent Clostridium difficile diarrhea in humans. In the current experiment, the complex nature of the crude inoculum inherent to the commercial microbial consortium Micosat F may have catalyzed a symbiotic profile, as testified by the yield results and by the phenotype effects on the tomato plants, and foreseen for the selective degradation of the litterbags. The debate on multitude (quantity) vs. handful (quality and variability) remains open. But the right answers to a good question are in the normal soils. As far as rapid methods are concerned, there is only one answer: when the aim of an experiment is to obtain knowledge on the effect of an organic amendment or a biofertilizer complex, considering a multitude of unknown soil microorganisms, it is easier and quicker to scan litterbags and leaves than to perform laborious analyses of the roots, soil and leaves. The close correlation between the short NIR SCiO spectra of the cell wall compounds and of other components of the litterbags is further confirmation of the results of the first paper of the litterbag series8. However, this paper can open new scenarios pertaining to the planning of biofertilizer \ bioinoculants experiments and the interpretation of the results, whether in bottom-up or top-down mode, as Litterbag-NIRS models will grow from experiments and symbiotic field-testing operations. In short, SIR alone cannot fully explain the soil fertility orchestrated by a complex and very assorted Biofertilizer: traces of the microbial work and footsteps can be found - almost archaeologically- in the NIR spectra of the litterbags and from the inherent substances, in primis in the fibers declared to be digestible, such as in rumen, as well as in the free sugar variations of degrading hay. Rumen obviously can host only anaerobic microbes, but the soil has equivalent aerobic priming and factors for SOM attack, where AM are the corner stones for long-life microbial fertility. Rapid spectroscopic analyses, and Litterbag-NIRS in particular, are therefore recommended for a rational assessment of microbial soil fertility before and during the use of biofertilizers or bioinoculants.

Thanks to Roberto Dalpozzo (Demetra Italia S.r.l., Castel Guelfo, Italy), Fabio Pelliconi (Consorzio Agrario di Ravenna, Cotignola, Italy) and Giusto Giovannetti (CCS-Aosta S.r.l., Quart, Italy) for the valuable support to ideation and execution of the experiment.