Nine NPOAs were selected because represent materials with a wide range of organic carbon chemistry and total nitrogen (N) content: F.O.R.S.U., is the organic fraction of municipal solid waste (Naples municipality, Southern Italy), wood chips from sawmill, Medicago sativa L. hay, Zea mays L. stalks, a compost from cattle manure, and four freshly fallen leaf litter types (Quercus ilex L., Fagus sylvatica L., Robinia pseudoacacia L., and Populus alba L.). In the case of leaf litter, for each species freshly abscised leaves were collected from randomly selected plants (N > 10), air dried for 25 days until reaching constant weight, and then was stored at room temperature.

Ten biochar types were used in this study: two were purchased and eight produced in our laboratory. The eight biochar types were made from four feedstock (F.O.R.S.U., Medicago hay, Zea stalks, and wood chips) that were subject to pyrolysis at two temperatures (300 and 550°C) for 5 h in a muffle furnace. Activated carbon (AC) and a biochar from Fagus wood were purchased from Sigma–Aldrich Co., and a local market, respectively. All NPOAs and biochars were powdered in a blender to obtain <2 mm particles and then stored in air-tight containers.

Electrical conductivity (EC) and pH of the NPOAs and biochars were measured by using a pH-meter (Basic 20 CRISON) and conductivity meter (CRISON) in suspensions of 1:2.5 and 1:5 organic material:water, respectively. NPOAs and biochars were characterized for total H, C, and N content by flash combustion of micro samples (5 mg) using an Elemental Analyser NA 1500 (Fison 1108 Elemental Analyzer, Thermo Fisher Scientific).

All materials were characterized by ¹³C-CPMAS NMR obtained in the solid state under the same conditions to allow a quantitative comparison of spectra. A Bruker AV-300 spectrometer equipped with a 4 mm wide-bore MAS probe was used. NMR spectra were obtained with MAS of 13000 Hz of rotor spin, 1 s of recycle time, 1 ms of contact time, 20 ms of acquisition time, 2000 scans. Samples were packed in 4 mm zirconium rotors with Kel-F caps. The pulse sequence was applied with a 1H ramp to account for non-homogeneity of the Hartmann-Hahn condition at high spin rotor rates. The ¹³C-CPMAS NMR spectrum was automatically integrated to calculate the area of the peaks which appeared in the chosen region. Seven spectral regions were selected as identified by previous studies (Kögel-Knabner, 2002; Bonanomi et al., 2016a): 0–45 ppm = alkyl C; 46–60 ppm = methoxyl and N-alkyl C; 61–90 ppm = O-alkyl C; 91–110 ppm = di-O-alkyl C; 111–140 ppm = H- and C- substituted aromatic C; 141–160 ppm O-substituted aromatic C (phenolic and O-aryl C); 161–190 ppm carboxyl C.

A plant bioassay was carried-out to determine the effect of NPOAs and biochars on the growth of lettuce (Lactuca sativa L.), a vegetable cultivated worldwide. Seeds of lettuce were purchased at a local market.

The bioassay was carried-out with 9 NPOAs and 10 biochars used pure plus the untreated control. In addition, all possible NPOA-biochar mixtures were made, resulting in 90 organic matter combinations, replicated 10 times with a total of 1,100 experimental units. Each organic amendment was applied in a pot alone at 2% (10 g^-1 per pot of dry matter), and carefully mixed with the soil; the unamended soil was also mixed in the same manner. The pots of the two-organic matter mixtures (90 combinations) were amended with 2% of dry organic amendment (10 g^-1 per pot) having a loading ratio of 50:50. The unamended soil was set as the control.

For this bioassay, about 800 kg of a previously characterized fertile soil were collected from the top 20 cm layer in an agricultural field. Soil was collected in a very productive area of about 5,000 ha cultivated under greenhouses located in the Salerno area (Southern Italy). The study site is characterized by a Mediterranean climate with a mean annual temperature of 15.9°C and monthly temperatures ranging from 23.6°C in August to 9.0°C in January. Mean annual rainfall does not exceed 988 mm with a relatively dry summer (84 mm). Low-technology, unheated polyethylene-covered greenhouses (height ∼4 m) are the main crop protection structures used in this area (Bonanomi et al., 2016b). According to Soil Survey Staff classification (Soil Survey Staff, 1998), the soil was Lithic Haplustolls (Supplemenatry Table S1). The soil was sieved in the laboratory (mesh size < 2 mm). Pots (14 cm diameter and 16 cm height) were filled with 500 g of air-dried soil and planted with 1 pre-germinated 7-day-old seedlings. Then, pots were placed in a greenhouse (20 ± 5°C during the day and 14 ± 5°C at night) following a completely random design with regular rotation every 10 days. Pots were wetted with distilled water every 3 days until water holding capacity was reached. After 90 growing days, above-ground plant part were harvested, dried (80°C in a ventilated oven until constant weight was reached), and dry weighted.

Initial nutrient concentrations and organic C chemical fractions significantly varied between biochars and NPOAs (Table 1). The highest N concentration was found in leaf litter from N fixer species (i.e., Medicago and Robinia), in compost, F.O.R.S.U. and biochars derived from N rich feedstocks (Table 1). C/N ratio showed large variations, ranging from the 639 value of wood biochar made at 300°C to the 6.41 value of Medicago hay (Table 1). EC was high for compost, F.O.R.S.U., Medicago hay and biochars derived from these feedstocks (Table 1). For NPOAs the pH showed small variations, ranging from 5.35 to 7.10 values of Fagus and Medicago litter, respectively. Biochar, instead, showed a much larger variation of pH, ranging from 5.67 of AC to 11.60 of Medicago biochar made at 550°C (Table 1). The H/C ratio of NPOAs was in all case >1.71 with the only exception of compost (1.32). In contrast, pyrolyzation drastically reduced the H/C ratio giving values ranging from 0.21 of AC to 1.11 of Medicago biochar made at 550°C (Table 1).

The ¹³C-CPMAS NMR data of NPOAs and biochars, with regard to C molecular types corresponding to reference spectral regions, revealed consistent changes of the material organic chemistry (Table 1). Among NPOAs, ¹³C NMR spectra showed consistent differences of C molecular types, with major chemical changes observed for the alkyl-C fraction (0–45 ppm), significantly lower for Zea and wood, and the methoxyl and N-alkyl C fraction (46–60 ppm) that was higher in F.O.R.S.U., compost, Medicago, and Robinia litter (Table 1). The relative abundance of the O-alkyl-C fraction (61–90 ppm) was higher for wood and Zea, and quite low for compost (Table 1). The carboxylic C fraction (161–190 ppm) ranged from 11.24 to 2.63% of Medicago hay and Zea residues, respectively. The H-C-substituted aromatic C region (111–140 ppm) was very high for lignin rich plant litter (i.e., Fagus, Populus, and Quercus) (Table 1). Interestingly, the chemistry of biochars was dramatically different from that of NPOAs. In particular, di-O-alkyl-C, methoxyl and N-alkyl C, and carboxylic C fractions were very low compared to those of NPOAs (Table 1). The O-alkyl-C region was abundant only in biochars made at 300°C, but disappeared in all biochars made at higher temperatures (Table 1). In contrast with the other spectral regions, the aromatic C fractions (141–160 ppm, but especially that at 111–140 ppm) showed a sharp increase in biochars compared to NPOAs, with the highest values recorded for biochar made at temperatures >300°C (Table 1).

Cluster analysis of NPOA and biochar spectra showed a remarkable difference between the two groups (Figure 1). NPOAs were mostly grouped together at low distance values, with only wood and Zea that had major distances, indicating some species-specific chemical differences (Figure 1). The dendrogram of biochars based on ¹³C NMR spectra showed two main clusters segregating at high distances and including materials made at either low (300°C) and high-intermediate (550°C, AC and commercial biochar) temperatures (Figure 1). Moreover, within the low-temperature cluster, most samples belonging to the same feedstock were consistently aggregated at the lowest distance levels, while within the high-temperature cluster sample aggregation was independent of the feedstock of origin (Figure 1).

The effect of NPOAs and biochars used alone on plant growth was largely affected by the type material (Figures 2A,B). With NPOAs lettuce showed various responses, being inhibited by Fagus, Quercus, and Zea, not affected by wood powder and promoted by the remaining materials, especially Medicago hay and F.O.R.S.U. (Figure 2A). Among biochars, five materials stimulated lettuce growth compared to untreated control, three had no significant effects and only two showed a moderate inhibition (i.e., F.O.R.S.U. and Zea biochar made at 300°C) (Figure 2B).

The PCA results provided a synthetic picture of the chemistry-dependent effects on the target plant responses caused by NPOAs and biochars (Figures 2C,D). PCA provided a satisfactory ordination of the NPOA and biochar chemical parameters in relation to the sensitivity of the tested plant, with the eigenvalues of the first two components accounting for 76.53% (53.13 and 23.40%) and 62.47% (35.82 and 26.65%) of the total variance for NPOAs and biochars, respectively. Figures 2C,D reports the loading vectors that represent the organic matter chemistry parameters plotted in the bi-dimensional space. For NPOAs, the first and second PCA components highlight the importance of pH, N content, EC, C/N ratio, and several NMR regions to affecting plant growth. Lettuce growth was largely explained by the negative correlations with C/N ratio as well as O-alkyl C and di-O-alkyl C regions, and positive correlations with EC, N content, and the carboxylic C, methoxyl C, and alkyl C NMR regions (Figure 2C). For biochars, instead, lettuce growth was explained to its negative correlations with EC and H/C ratio and positive correlations with O-alkyl C, di-O-alkyl, and carboxylic C NMR regions (Figure 2D).

Non-additive and additive effects on plant growth were found in 87.78 and 12.22% of the tested mixtures, respectively (Figure 3). Antagonistic responses were more common than synergistic interactions, with 59 and 20 cases, respectively. The magnitude of synergistic and antagonistic interactions ranged from -87.99% to +156.51% of the expected plant growth (Figure 3).

Biochar and NPOA types in the mixture affected the outcome of non-additive interactions. For instance, when Medicago hay and F.O.R.S.U. were used as NPOA, synergistic interactions with biochars were recorded in 7 and 6 out of 10 cases, respectively (Figure 3). The deviation between observed and expected growth among all biochar types was +36.74%, +18.72%, and +15.74%, for Medicago, F.O.R.S.U. and compost, respectively (Figure 4). Mixtures with Populus and Robinia litter showed synergistic, non-additive, and antagonistic interactions, with the latter being more common than the other two (Figure 3). In contrast, the mixtures containing Quercus litter as NPOA gave an antagonistic interaction in 90% of the cases, with no synergistic responses (Figure 3). Similar results were found for Fagus and Zea litter as well as with wood which showed 80% of antagonistic interactions and 20% of non-additive responses (Figure 3). The deviation between observed and expected growth with all biochar types was negative for Robinia and wood powder (Figure 4). For Quercus, Fagus, and Zea litter, the deviation between observed and expected growth was -64.05%, -63.15%, and -58.42%, respectively (Figure 4).

With biochars, the interactions were additive, i.e., the deviation from expected values was not significant, for AC, Zea, and wood biochars made at 300 and 550°C and for Medicago char made at 550°C. For the remaining biochars (commercial, Medicago 300°C, and F.O.R.S.U. 300 and 550°C), the deviation from expected values was negative with all NPOA types, meaning that antagonist interactions were more important (Figure 4).

Deviation from expected values are adequately described by the linear model [F_(9,80) = 8.92, P < 0.001, Figure 5]; half of the total variance in observed-expected ratio of plant growth is explained by the chemistry of the mixtures. In particular, the significance of individual coefficients indicated a positive influence of carboxylic C (161–190 ppm), C/N ratio, and N content (for P < 0.001, P < 0.01, and P < 0.05, respectively) and a negative effect of H/C ratio (for P < 0.01) on the observed-expected ratio of growth. On the other hand, EC and the other C-types corresponding to NMR spectral regions did not reveal statistical significance. Moreover, the relative explained variance vary drastically among chemical parameters, i.e., between 3.5 and 20.3% (Figure 5). In this regard, it is noteworthy that the significant variables that contributed the most (20.3% of the relative explained variance) were the carboxylic C (161–190 ppm) followed by the H/C ratio (15.0%), N content (14.8%), and C/N ratio (7.2%). The residual portion of the relative variance is attributed to the non-significant variables.

Biochars and NPOAs used pure have contrasting effects on the growth of lettuce, showing clear relationships with their chemical characteristics. Effects of NPOAs on lettuce growth range from intense inhibition (-67% by Zea residues) to strong stimulation (+226% by F.O.R.S.U.). The major inhibitory effect was found with N poor materials characterized by high C/N and H/C ratios. Our finding of a severe plant root inhibition by undecomposed litter of Fagus, Quercus, and Zea is consistent with previous studies carried-out in natural ecosystems (Dorrepaal et al., 2007; Lopez-Iglesias et al., 2014; Meiners, 2014). Lettuce growth inhibition by undecomposed leaf litter can partially be attributed to microbial N immobilization (Hodge et al., 2000) and/or to the phytotoxic activity of putative allelopathic compounds (Bonanomi et al., 2011). The N immobilization hypothesis sustains that when organic matter with high C/N ratio is incorporated in soil reduce N availability, thus impairing root system development. The main mechanism is related to the behavior of saprophytic microorganisms in soil. In fact, to efficiently exploit soil organic matter microbes require both organic C and N in a relatively constant stoichiometric ratio. Organic C or N limit microbial growth, and when C/N ratio of organic matter is above the threshold value of ∼30–35 mineral N is taken from the soil, incorporated in microbial biomass and temporarily immobilized (Hodge et al., 2000). Such a hypothesis is indirectly supported by the present work, because plant growth is inversely related to substrate C/N ratio that, in turn, is directly correlated to N immobilization (Palm et al., 2001). However, C/N ratio can only partially explain the observed inhibitory effect because the wood, characterized by the highest C/N ratio, had a neutral effect on plant growth. This discrepancy could be explained by the quality of organic C of the NPOA. We found, in fact, that plant growth was inversely related to the relative abundance of the O-alkyl and di-O-alkyl NMR regions that are mainly associated with sugar and polysaccharides (Kögel-Knabner, 2002). This result suggests that organic amendments rich in labile C, e.g., sugars and relatively simply polysaccharides, may induce a faster N immobilization with a more negative impact on crop growth than materials rich in lignin (e.g., powdered wood). An alternative hypothesis is that plant litter can release, either directly or mediated by microbial activity, allelopathic compounds with harmful effects on root growth (Bonanomi et al., 2011). The identification of the ultimate causes of litter inhibitory is out of the aim of this work, but our approach by using ¹³C NMR indicates that lettuce growth is negatively affected by O-alkyl and di-O-alkyl C, but positively associated with signals related to plant tissue lignification. On the other hand, we found that lettuce growth was promoted by NPOAs with a relatively high content of methoxyl C and carboxyl C NMR regions, as well as a high N content. These results confirm the well-known positive effect of mineral and organic N sources on plant development (Marschner, 2012).

Concerning biochars, the effect on lettuce growth was less variable than that observed with NPOAs. Lettuce response ranged from moderate inhibition with F.O.R.S.U. biochar made at 300°C, to a highly significant stimulation observed with AC and the three biochars derived from wood. ¹³C NMR spectra showed that during pyrolysis all materials are subjected to a decrease of labile C molecular compounds, especially at high temperatures, and a relative increase of aromatic fractions. Noteworthy, all biochars pyrolyzed at 300°C showed a peak in the alkyl fraction, mainly related to aliphatic compounds, including the so-called bio-oils (Amonette and Joseph, 2009). However, the alkyl C region was absent in biochar made at 550°C or higher temperatures (e.g., AC and commercial wood biochar). Biochars made at high temperature or derived from wood sources showed a quite low H/C atomic ratio. This index, which has been proposed to define biochar quality in the context of bioremediation (Xiao et al., 2016), seems to be the best parameter to identify the biochar with the ability to promote plant growth. Finally, it is notable that total N content of biochars (ranging from 0.13 to 6.26% of wood 300°C and F.O.R.S.U. 550°C, respectively) had a limited power to explain its effect on lettuce growth. Probably, during pyrolysis most of the organic N was gradually transformed into pyridine-like compounds, with a consequent drastic reduction of its bio-availability for plants (Bagreev et al., 2001; Chan and Xu, 2009).

In our experiments non-additive interactions, either synergistic or antagonistic, were prevalent. However, the majority of these interactions showed antagonistic effects, i.e., plant growth was lower than that expected by using each organic amendment alone. When biochars were mixed with N rich materials (i.e., Medicago hay, compost or F.O.R.S.U.), synergistic interactions were prevalent.

According to the recent review of Kammann et al. (2016), co-composted biochars have a remarkable plant growth-promoting effect compared to biochars used pure, but no-systematic studies were done to understand the interactive effects of biochars with NPOAs. In our experiment, additive effects were rare considering the expected vs. observed pooled data (11 cases out of 90 combinations). Contrarily to the general trends reported in the literature (Kammann et al., 2016), antagonistic interactions were more common than synergistic ones. However, some important methodological differences between the present and previous studies should be considered. In this study, the organic substrates mixed with biochars span an ample range of N content and C chemistry, while previous studies used only compost (Joseph et al., 2013; Schulz et al., 2013), or selected nutrient-rich materials such as urine (Schmidt et al., 2015).

Our results, being based on a wide range of NPOA types, provided a complete picture to understand the complex relationships between biochar and organic amendments. First, our data highlight that NPOA identity is more important than biochar types as a determinant of plant growth in the mixture. In fact, the deviation from expected values ranged from -29 to -5% for biochar, while for NPOAs the deviation spans from -64 to +36%. This means that most biochars appear suitable as a component of the mixture, with the exception of biochar made from F.O.R.S.U. and Medicago at 300°C. The same is not true for NPOAs, because only some materials are able to synergize when are used in combination with biochars. Synergistic interactions were observed with N rich, lignin poor substrates (i.e., compost, F.O.R.S.U., and Medicago hay). This result is in agreement with previous findings concerning biochar co-composting, where the best synergism was achieved blending C rich biochars with N rich organic materials (Kammann et al., 2016). The mechanisms proposed to explain the synergism between biochars and compost is based on the improvement of soil structure and water availability (Novak et al., 2012), the promotion of beneficial microbes (Warnock et al., 2007), and the enhanced plant-available nutrients (Liu et al., 2012). A recent study (Kammann et al., 2015) showed that co-composted biochars became highly enriched with nutrients (i.e., nitrate, ammonium, and phosphate) and dissolved organic carbon that is slowly released in soil. In addition to nutrients, low- and high-molecular weight organic C compounds can accumulate on biochar surfaces and in pores, having the potential to affect plant growth. In fact, many low-molecular weight organic C compounds are known to be phytotoxic at high concentrations, but have hormonal effects at low concentrations (Joseph et al., 2013). In this regards, Medicago sativa hay and F.O.R.S.U. could be phytotoxic if applied at high rate (Chefetz et al., 1996; Chung et al., 2000), but in presence of biochars such a negative effect could be reversed. Our study, however, was not designed to identify the mechanisms behind antagonistic and synergistic interactions between NPOAs and biochars. In the soil-biochar-NPOA mixtures, all the aforementioned cited mechanisms could be operative, but further studies are needed to clarify their relative contribution.

This study is the first reporting antagonist interactions between NPOAs and biochars. The occurrences of antagonistic interactions are well predictable based on the chemistry of the original components. Antagonisms were observed with wood powder and all leaf litter types, materials which are characterized by low N content combined with a high lignin content and high H/C ratio. Noteworthy, the combination among biochars with Robinia leaf litter showed an intermediate behavior, with several additive interactions and some moderate antagonistic interactions. This leaf litter is the only material characterized by a high N concentration and a relatively high lignin content. These results clearly indicate that the combination of biochars with low quality litter types (i.e., high lignin, high H/C ratio, and low N content) would enhance the probability of antagonistic interactions on plant growth. The mechanisms underlying such antagonist interactions are unknown, although a short-term, intense N immobilization could be hypothesized.