Litter chemistry explains contrasting feeding preferences of bacteria, fungi, and higher plants

Litter decomposition provides a continuous flow of organic carbon and nutrients that affects plant development and the structure of decomposer communities. Aim of this study was to distinguish the feeding preferences of microbes and plants in relation to litter chemistry. We characterized 36 litter types by 13C-CPMAS NMR spectroscopy and tested these materials on 6 bacteria, 6 fungi, and 14 target plants. Undecomposed litter acted as a carbon source for most of the saprophytic microbes, although with a large variability across litter types, severely inhibiting root growth. An opposite response was found for aged litter that largely inhibited microbial growth, but had neutral or stimulatory effects on root proliferation. 13C-CPMAS NMR revealed that restricted resonance intervals within the alkyl C, methoxyl C, O-alkyl C and di-O-alkyl C spectral regions are crucial for understanding litter effects. Root growth, in contrast to microbes, was negatively affected by labile C sources but positively associated with signals related to plant tissue lignification. Our study showed that plant litter has specific and contrasting effects on bacteria, fungi and higher plants, highlighting that, in order to understand the effects of plant detritus on ecosystem structure and functionality, different microbial food web components should be simultaneously investigated.

phytotoxic compounds dynamically change during litter decay as a result of chemical-physical transformations and biological processes driven by microorganisms 24,28 . While a rapid degradation of most allelochemicals into non-toxic molecules has often been reported 19,29,30 , the persistence of extracellular fragmented plant DNA in decomposed litter can produce species-specific effects at medium-long term, ranging from inhibition of conspecifics to stimulation of heterospecifics 31 .
In addition to complex effects on plant development, litter dynamics critically affect the structure and composition of decomposer communities. A large number of studies which describe the successional dynamics of bacterial 10 and especially fungal communities [32][33][34] that occur during litter decomposition, report several compositional shifts and a general increase of microbial diversity at decreasing litter quality. Several mechanisms have been proposed to explain the observed patterns, including limitation of spore dispersal, inter-specific competition 11 , time constraints for fruiting body development 35 , grazing pressure by arthropods 36 , and shifts of substrate suitability owing to litter biochemical dynamics 37,38 . In this context, it is relevant that heterotrophic saprotrophic microbes, by using plant litter as a source of energy and nutrients, progressively modify the biochemical quality of organic substrates 39,40 which, in turn, affects microbial turnover, community composition and diversity. For example, the rapid disappearance of weak parasites and "sugar fungi" replaced by cellulolytic and ligninolytic fungi, (a well recognized successional pattern 11,32 ), has been associated with the rapid depletion of labile C compounds and the concomitant relative increase of cellulose and lignin content in litter 41 . In addition, according to Prescott 42 , stable organic matter would be produced by microbial and biochemical transformation of litter materials into novel recalcitrant compounds, rather than by selective preservation of the pre-existing resistant fraction. Several studies based on throughput analytical techniques such as pyrolysis-gas chromatography/mass spectrometry 43 , near infrared reflectance spectroscopy 44 , and 13 C-CPMAS NMR spectroscopy 2 have clarified major molecular traits 45 and changes 46,47 corresponding to litter ageing. These include a rapid reduction of aminoacids, polypeptides, O-alkyl and di-O-alkyl fractions corresponding to carbohydrates, and a relative increase of lignin content and the associated methoxyl and N-alkyl compounds. Correspondingly, litter suitability for the growth of opportunistic saprophytic fungi progressively decreases with litter age 40 .
Taken together, such evidence indicates litter molecular composition is a main driver of bacteria, fungi and higher plant development on litter materials. Previous studies have separately tested litter effects on higher plants [18][19][20]25 , bacteria 10 , or fungi 40 , while tests simultaneously targeting representatives of all these microbial food web components over the same materials have not yet been reported. In addition, most previous evidence relies on over-simplified systems including a single or at most few sensitive target species of little ecological relevance. Such study conditions clearly do not match those in nature, where plant litter simultaneously affects different, highly diverse trophic levels. The strength of the conclusions that can be drawn at this stage has thus been limited. Moreover, the use of single-species bioassay prevents the assessment of species-specific responses possibly related to differences in sensitivity and in functional traits, even within the same trophic level. Many field and greenhouse experiments limited to plants, however, report that a thick litter layer mostly hampers the establishment of short-lived, small seeded species 48,49 , owing to the scarce penetration capability of small seedlings or to the requirements of light for germination of small seeds 50 . Surprisingly, in this context, the hypothesis of a higher sensitivity to litter inhibitory effects in small-seeded seedling compared to large-seeded one has not yet been tested.
In this study, we combined litter characterization by 13 C-cross-polarization magic angle spinning nuclear magnetic resonance ( 13 C-CPMAS NMR) spectroscopy 2 , proximate analysis 51 and C and N content determination, with a multi-species bioassay. In particular, the effect of 36 litter types spanning a wide range of biochemical quality was assessed on 14 target plants, 6 fungi and 6 bacteria species. We expect a progressive decrease of bacterial and fungal growth with litter age. Differently, for plants, we expect a shift of litter effects on root growth during litter decay, from inhibitory to enhancing, in relation to changes in litter chemistry. We also explicitly hypothesize a negative association between seed size and sensitivity to litter inhibitory effects, expecting that seedlings of small-seeded plants are more prone to litter inhibitory effects compared to seedlings of large-seeded plants. Finally, based on the finer molecular resolution of 13 C-CPMAS NMR spectroscopy compared to traditional indices based on C-to-N, we expect that specific spectral data are associated to litter functional quality in relation to bioassay results.

Results
Effects of litter materials. The effect of litter on plants, bacteria and fungi was largely determined by litter type and age (Fig. 1). Bacteria response over undecomposed litter showed on average an inhibitory effect compared to the control (Fig. 1). However, the effect largely varied according to litter type, being positive for Populus, Arbutus, Fraxinus, and Hedera litters, markedly inhibitory for Quercus, Festuca, Cupressus, Picea and Fagus materials, and showing an intermediate pattern for the remaining species (Fig. 1). Fungi, on average, showed higher growth on undecomposed litter compared to the control (+18%). Consistent with bacteria response, also for fungi the effect of undecomposed material was largely affected by litter type. A positive response was found for 9 litter types (e.g. Coronilla, Medicago, Salix) and a moderate inhibition over the remaining 9 materials (Fig. 1). Interestingly, both bacteria and fungi showed a response pattern to litter age opposite to that of higher plants (Figs 1 and 2). In particular, bacteria and fungi showed a significantly higher growth on undecomposed compared with 180-days old litter (Fig. 2a), corresponding to a remarkable growth inhibition (>50%) over all aged litter types (Fig. 1).
Over undecomposed litter, plants showed on average a root inhibition of 48% compared with control. While all undecomposed materials showed a significant inhibitory effect on target plant species (Fig. 1), the response varied according to litter type, being very high for Medicago, Picea, and Arbutus material, relatively low for Populus, Fagus and Cupressus litter and intermediate for the remaining plant residues (Fig. 1). Decomposition for 180 days greatly reduced the inhibitory effect, with 9 litter types that stimulated root growth of the target species compared to untreated control (Fig. 1). The only exceptions were Acanthus mollis and Medicago sativa materials  Table 1). Among bacteria, Rhizobium was the most inhibited over aged litter while Bacillus showed the highest growth over undecomposed substrates (Fig. S1a). Concerning fungi, all species thrived over fresh litter, with very high growth for Mucor and Trichoderma (Fig. S1b). Differently, over aged substrates, only Trichoderma showed a relatively positive performance, while the remaining five species were severely inhibited, as compared to the control over PDA as well as the performance over undecomposed litter (Fig. S1b).
Concerning higher plants, species-specific responses to litter treatments were observed (Fig. S1c). Over undecomposed litter, the five annual species were the most inhibited together with the deciduous tree Populus (Fig. S1c), with Arabidopsis, Lycopersicon, Trifolium and Lepidium almost completely inhibited over fresh, undecomposed litter. In contrast, some perennial and woody species were slightly affected compared to the untreated control (Fig. S1c), also showing, in the cases of Hedera, Pinus and Quercus, a not significantly different response to fresh and aged litter. Concerning root growth over aged litter, Arabidopsis were the most inhibited species (Fig. S1c) but the difference among target plants was largely reduced, with most species showing minor variations, of both signs, compared to the untreated control (Fig. S1c). Interestingly, seedling root growth over undecomposed litter was positively associated with seed weight, while such relationships did not hold over aged litter (Fig. 3).
Litter chemistry and target species sensitivity. Litter proximate chemical parameters showed a contrasting pattern of correlations with the response of microbes and plants ( Table 2). In particular, labile C was positively associated with fungal growth, and negatively with root growth of annual, perennial, and woody plant species (Table 2). Cellulose and N content did not show significant relationships with the response of target organisms (Table 2). Finally, C/N ratio was negatively and positively correlated with growth of annual plant and fungi, respectively ( Table 2). Growth is expressed as percent difference compared to the controls (Nutrient Broth, PDA, and water for bacteria, fungi, and plants, respectively). For each bar, data refer to mean ± standard deviation of different target species (N in brackets) over18 litter types. Asterisks indicate within-group statistically significant differences (***p < 0.001; **p < 0.01; *p < 0.05), according to Tuckey's HSD post-hoc tests for the interactive effects of litter age and target organisms (a) and litter age and plant functional type (b) from corresponding GLMMs models in Table 2.
Scientific RePoRTS | 7: 9208 | DOI:10.1038/s41598-017-09145-w Considering 13 C-CPMAS NMR reference regions and growth data pooled for either bacteria or fungal species, no significant correlations were found between litter molecular composition and microbial growth ( Table 2). The only two exceptions were the O-alkyl C and methoxyl C litter fractions, positively and negatively related to fungal growth, respectively (Table 2). Considering higher plant responses, we observed a general pattern of non-significant relationships between seedling root growth and C-types corresponding to spectral regions, with the exceptions of annual and woody species, positively related to methoxyl C and O-substituted aromatic C, respectively ( Table 2). The widely used alkyl C/O-alkyl C and CC/MC ratios were not significantly related to bacteria, fungal and plant growth ( Table 2).
An extensive correlation analysis of all signals from the 13 C-CPMAS NMR litter spectra with the response of the 26 target species on the same materials showed an interesting pattern of association between the growth of target organisms, and some restricted spectral regions (Fig. 4). In the case of bacteria, the relationships between litter chemistry and species response along the spectrum, considering data averaged over the six target species, did not show a significant correlation (Fig. 4a). However, such a profile showed high inter-specific variation (Fig. 4a), with restricted spectral regions significantly related to the growth of different species (Supplementary  Table S1). Among these, we observed negative correlations with spectral signals resonating in the regions 4-14 ppm (for Bacilus and Escherichia), 53-60 ppm (Bacillus, Escherichia, Lysobacter and Pseudomonas), 123-126 ppm (Escherichia, Erwinia and Pseudomonas), 132-138 ppm (i.e. Bacillus, Escherichia, Erwinia, and Pseudomonas), 151-153 ppm (Bacillus, Escherichia, Erwinia, Pseudomonas), and positive association with signals in the 68-78 ppm region (i.e. Bacillus, Escherichia, Erwinia and Pseudomonas) (Supplementary Table S1). Further, single species were positively associated with specific signals, as in the cases of Bacillus and Lysobacter with the signals at 97-101 ppm and 144-146 ppm, respectively. Finally, Rhizobium did not show significant correlations along the whole spectrum (Supplementary Table S1).
The six target fungi showed an exceptionally similar pattern of correlation with litter chemistry defined by 13  In the case of higher plant growth, a different pattern of association with litter chemistry was observed for annual, perennial and woody species (Fig. 4c-e). Annual plants showed a very similar response among the five species (Supplementary Table S3) with an outstanding opposite behaviour compared to fungi (Fig. 4b,c). In particular, all annual plants were positively associated with the signals resonating within the 3-24 ppm, 45-60 ppm, 122-126 ppm, 133-138 ppm and 151-153 ppm, and negatively with those at 67-80 ppm and 96-106 ppm regions (Fig. 4c, Supplementary Table S3). The only exceptions were Lepidium and Trifolium, showing marginally significant correlations with the signals at 133-138 and 151-153 ppm (Supplementary Table S3). Root growth of perennial species was positively correlated with the signals resonating within the 61-65 ppm, 110-120 ppm and 145-155 ppm regions, and negatively with those within the 35-42 ppm region (Fig. 4d). However, response differences among perennial species were recorded, as in the case of Acanthus growth, not related with the 145-155 ppm region, and Ampelodesmos and Hedera, only marginally associated with the 61-65 ppm and 35-42 ppm regions (Supplementary Table S4). Finally, woody plant species showed a general trend of non-significant correlation between root growth and litter quality defined by 13 C NMR (Fig. 4e), with high inter-specific differences (Supplementary Table S5). The only spectral region associated with root growth included the signals resonating at 151-155 ppm, positively correlated to root growth for Quercus and Robinia, and marginally related for Populus, but unrelated to Alnus and Pinus responses. Robinia growth was also positively associated with two single signals, resonating at 114 pm and 134 ppm, respectively (Supplementary Table S5).
Principal component analysis (PCA) provided satisfactory ordination of the 13 C-CPMAS NMR spectral signals across the 36 litter types, with the first four eigenvalues accounting for 93.2% (52.1%, 22.3%, 13.2%, and 5.6%) of the total variance, respectively. The bi-dimensional space defined by the first two components provided a synthetic picture of the litter chemistry-dependent effects of treatments on the target species responses (Fig. 5). It confirmed the above-described relationships between the relative abundance of different organic C types in the litters and the growth response of bacteria, fungi and plant functional types over such materials, based on the loading vectors of 13 C-NMR signals measured in the litter materials (i.e. relative abundance of each signal and how they relate to the PC axes) and those of supplementary variables, including target species response and litter proximate parameters (Fig. 5a). In the same bi-dimensional space, the trajectories of the litter materials based on their factorial scores (Fig. 5b) reflected the chemical changes occurring during the decomposition process as defined by spectral data.

Discussion
In this work, we show that plant litter has an age-dependent, contrasting effect on the growth of bacteria, fungi and higher plants. In particular, undecomposed litter has a major inhibitory effect on plants, but promotes bacterial and especially fungal growth, although the effect on microbial response largely varied according to litter type. On the contrary, as decomposition proceeds, litter becomes progressively more suitable for plant growth, while bacterial and fungal responses are largely inhibited. This is clearly related to biochemical quality shifts in decomposing materials, with a progressive depletion of easily degradable C sources and enrichment of recalcitrant compounds. Our results demonstrate that defining litter chemistry by 13 C-CPMAS NMR provides useful insight for characterizing the substrate preference of bacteria, fungi and higher plants.
Litter effects on higher plants. We found that undecomposed litter caused a severe inhibition on root growth of most target plants. The prevalence of inhibitory effects by undecomposed litter recorded in this survey is consistent with previous studies carried out in different ecosystems 18,20,25,52,53 . Thus, root inhibition by undecomposed litter seems to be a rather general phenomenon not restricted to few "allelopathic" plants. However, such inhibitory effect largely varied among litter types, and almost disappeared after 180 days of decomposition for all the tested litters, with the exception of two species (i.e. Acanthus and Medicago). Previous studies aimed at assessing litter allelopathic effects were criticized for using one or only a few sensitive target species of little ecological relevance. Here, by using 14 different bioassay plants belonging to different functional groups, we were able to generalize these findings and to highlight species-specific response differences, with annual plants far more sensitive than perennial and woody species. In addition, the magnitude of seedling root growth inhibition was inversely proportional to seed size, with small-seeded plants being more sensitive to litter chemical effects.
Our findings raise two main questions: why is undecomposed litter broadly inhibitory to root growth and why does this harmful effect lessen during decomposition? About the first issue, we have to consider that the initial phase of decomposition basically consists of plant tissue breakdown and subsequent release of cell contents. At this stage, the inhibitory effect showed a large variability among litter types, being high for materials from nitrogen fixing species such as C. emerus and M. sativa, but also for P. abies, A. unedo, C. sativa, and A.mollis, and low for other species (e.g. C. sempervirens, P. nigra and F. sylvatica). Root growth inhibition by undecomposed plant litter has been attributed to litter N immobilization 23 Table 1. Summary of the Generalized Linear Mixed Models (GLMM) testing for main and 2 nd order interactive effects of litter species (random effect) and age (fixed effect, either 0 or 180 days of decomposition) on bioassay results (i.e. growth of plants, fungi and bacteria, expressed as percentage of the untreated control). GLMMs were tested for all data pooled, including a fixed effect of the group of organisms (3 levels), and separately for plants, fungi and bacteria, including a random effect of target species. In the case of plants, a further GLMM was tested, including the effect of target functional type (3 levels: annual, perennial and woody).
compounds 17 . The nutrient immobilization hypothesis sustains that litter with high C/N ratio, by reducing nitrogen relative availability, would limit root system development. However, such hypothesis is not supported by the present work, because litter N content and C/N ratio in our bioassays was unrelated to root proliferation for all target plant species. This indirectly suggests that other causal factors should be taken into account to explain the  general inhibitory effects of undecomposed plant litter, although N limitation may still hold under specific ecological conditions, as in the case of litter with high C/N ratio in N poor soils 22,54,55 .
The alternative hypothesis states that plant litter can release, either directly or mediated by microbial activity, phytotoxic compounds such as phenols, aromatic rings, and tannins with harmful effects on root growth 17 . The use of 13 C-CPMAS NMR allowed the depiction of biochemical differences among litter materials, thereby providing useful insights to explaining the variable effect on root growth. Two positive correlations were found between root growth of annuals and woody plants with the methoxyl and aromatic regions, respectively. Both regions, being based on lignin-related NMR signals 2, 56 , indicate that highly lignified plant materials have limited inhibitory effects on root proliferation. Moreover, our extensive correlative analysis applied along the whole 13 C NMR spectrum provided a considerably improved definition of litter biochemical quality, compared with reference regions from literature 2, 57 , in relation to ecological functions. Specifically, we identified two restricted spectral intervals within the alkyl C, and the methoxyl C regions (3-24 ppm and 50-60 ppm, respectively), positively associated with root growth, and two other intervals within the O-alkyl C and di-O-alkyl C regions (67-80 ppm and 96-106 ppm, respectively) negatively correlated with root growth. These spectral signals correspond to specific molecular components of plant litter: the aliphatic chain in lignin and the acetate of some hemicelluloses, the methoxyl C of lignin, the C2, C3, and C5 of labile carbohydrates, and the C1 of cellulose 2, 41 .
Our correlative analysis also helps to explain the specific sensitivity to undecomposed litter of different plant functional groups (annuals≫ perennials ≥ woody). The five annual species showed a remarkably similar response to litter biochemical quality, with root growth being positively and negatively associated to methoxyl C and O-alkyl C, respectively. Proximate analysis confirmed evidence from NMR data, with root growth of annual plants, but not that of perennial and woody species, negatively associated to labile C. These results indicate that short-lived plants are far more inhibited, compared to long-lived species, by simple carbohydrates and other compounds that rapidly disappear during decomposition, such as condensed tannins 45,58 . Root growth of short-lived plants was also positively related to specific spectral signals corresponding to different aromatic C types (122-126 ppm, 133-138 ppm and 151-153 ppm). The signals resonating at 151-153 ppm were previously attributed to N-substituted unsaturated carbons from heterocyclic aromatic amines, diagnostic of the N bases of nucleic acids 59 . These were previously associated with a species-specific effect on plant root growth, with inhibition and stimulation on conspecific and heterospecific species, respectively 31 .
The absence of sensitivity to different litter fractions in perennial and woody plants is yet to be clarified. In this regard, it is worth noting that some species (e.g. Pinus) showed a limited sensitivity to plant litter effects, as well as the absence of significant relationships between root growth and litter quality as defined by 13 C NMR.
The higher sensitivity of short-lived and small-seeded plants to the chemical component of undecomposed litter is consistent with several field studies reporting the rapid disappearance of short-lived species in old fields, as a result of litter accumulation 12,13,48 . Considering the molecular and physiological basis of plant sensitivity to undecomposed litter, in relation to seed size and lifespan, the reserve effect hypothesis 60 may provide a partial explanation. According to such hypothesis, small seeds contain too little reserve tissue to face environmental stresses such as litter phytotoxic effect. Previous studies related this negative effect of the litter layer to the light requirement and sensitivity to mechanical impediment of small-seeded plants 50 . In this context, we add a further possible mechanistic hypothesis, based on a direct negative effect of litter chemicals on seedling root growth of species with small seeds. At the field scale, most experiments deal with seedling emergence or establishment and not with the underlying processes 12,13 . Therefore, future studies are required to reveal the relative contributions of mechanical impediment, light requirement and chemical interference by litter on the inhibition of small seeded, short-lived plant species.
Litter preference by bacteria and fungi. In our experiments, about half of the tested fungi and bacteria thrived when exposed to the majority of undecomposed litter materials, with growth remarkably higher, not different, or slightly lower to that recorded on the controls over rich, standard microbiological substrates. However, a large response variability was found depending on litter type, with higher growth over materials from herbaceous, nitrogen-fixing (i.e. Coronilla and Medicago) and fast decomposing species (e.g. Hedera, Salix, and Fraxinus). A significant inhibition over litter from coniferous (i.e. Cupressus, Picea, and Pinus), slow decomposing deciduous trees (e.g. Quercus, Fagus), and perennial grasses (i.e. Ampelodesmos and Festuca) was recorded, but also in these cases all the bacteria and fungi were able to use these substrates for growth ( Supplementary Fig. S1). In this regard, it is interesting to note the higher saprophytic capability of the selected fungi compared to bacteria. Opposite to higher plant response, a very steep decline of microbial growth was recorded over all litter decomposed for 180 days, compared to fresh materials, consistent for all bacterial and fungal species. These results confirm previous finding 40,61 , reporting a high growth rate of 18 fungal species over several undecomposed leaf litter, followed by a rapid reduction of fungal development as litter decomposition proceeded. At first glance, these results could appear surprising, because several undecomposed litter used in this study are known to be rich in antifungal and antibacterial compounds. For instance, antimicrobial activity of pure molecules and/or leaf extracts obtained with organic solvents (e.g. butanol, ethyl acetate, methanol, etc.) has been reported for A. mauritanicus 62 , A. unedo 63 , H. helix 64 , P. nigra 65 , R. pseudoacacia 66 and Q. ilex 67 . However, the use of organic solvents, compared to the water extracts used in our bioassays, allows a more complete extraction of organic compound from the starting leaf material, which is useful for phytochemical and toxicology assessments, but it also provides a less reliable reproduction of ecological processes in field conditions.
The dramatic decline of bacterial and fungal growth observed on aged litter, compared to undecomposed materials, can be explained by a progressive decrease in litter biochemical quality, with the substrate becoming less suitable to microbial exploitation. Such changes could be related to a decrease of easily degradable C sources, and to the accumulation of toxic and/or recalcitrant organic compounds. Results of proximate chemical analyses and 13 C-CPMAS NMR highlight that both processes occur in ageing litter, with a sharp decrease of the labile C fraction and a relative increase of aromatic compounds 19,40 . Surprisingly, proximate cellulose was unrelated to microbial growth, despite targeted fungi, and especially Trichoderma harzianum, are able to use it as a C source. Differently, 13 C NMR data showed a positive association between signals within the di-O alkyl C region, diagnostic of the C1 of cellulose 41 , and fungal growth, although with species-specific differences. These contradictory results could be partially explained considering that a large cellulose proportion is entrapped with lignin, so that decomposition cannot proceed independent of lignin degradation 68 .
Microbial growth was also independent of litter N content. Previous studies reported a dual, contrasting effect of N on both litter decay 69,70 and fungal growth 40 , with high N content in litter promoting decomposition at the early stage (up to 30-40% of mass loss) and slowing it thereafter 41 . We suggest that high N content initially sustains a rapid microbial growth, while at later decay stages high N concentration could favor the formation of recalcitrant compounds with potential inhibitory effect 41 . However, the underlying microbiological and biochemical processes have not yet been clarified. In general terms, we found a pattern of association, between microbial growth and restricted 13 C-CPMAS NMR spectral regions in litter, generally consistent among species, likely emerging from the balance between labile C sources availability and the occurrence of recalcitrant and/or toxic compounds. However, inter-specific response variability was higher for bacteria, suggesting that our approach could be more appropriate to address species level questions, and lower among fungi with the six tested species showing an exceptionally similar pattern. Such clear-cut behavior showed the same magnitude but an opposite direction to what was observed for higher plants, and for annual species in particular. Although our study design was not specifically aimed to compare microbes and annual plants coexisting in nature, our results suggest a complementary effect of litter on these two groups of organisms, whose occurrence in field conditions should be assessed. Along this line, saprotrophs would exploit fresh leaf materials being limited by aged and lignified litter, while plants would be inhibited by phytotoxic fresh litter with seed size-and lifespan-dependent magnitude, then thriving over aged substrates.

Conclusions
Our multi-species bioassay approach revealed that plant litter has a species-specific effect on bacteria, fungi and higher plants. It is noteworthy that we found undecomposed plant litter mainly acts as a C source for saprophytic microbes, although microbial responses largely varied across litter types. In contrast, fresh litter hampers plant root growth, with severe inhibition observed for most plant species. An opposite response was found in bioassays with aged litter that largely inhibits microbial growth, but has neutral or stimulatory effects on root proliferation. The use of 13 C-CPMAS NMR provides an improved definition of litter molecular quality, helping to explain the variable effects of plant litter on different microbial food web levels. 13 C-CPMAS NMR revealed that restricted resonance intervals within the alkyl C, methoxyl C, O-alkyl C and di-O-alkyl C spectral regions are crucial for understanding litter effects, with plant root growth negatively affected by labile C, but positively associated with signals related to plant tissue lignifications. The opposite response was observed for bacteria and especially fungi to the corresponding 13 C-CPMAS NMR signals. Despite limitations of species-specific bioassay methodology, we demonstrate that litter molecular quality can promote or inhibit functional species, and potentially control successional dynamics through time. These results emphasize the importance of simultaneously studying different microbial food web components in order to fully understand the effects of plant detritus on ecosystem structure and functionality.

Materials and Methods
Target bacteria, fungi and plants. Twenty-six target species were selected to assess substrate preference in bioassays (Supplementary Table S6; all species were obtained from collections at the Department of Agriculture, University of Naples Federico II). The species pool included six bacteria, six fungi with different functional traits and fourteen plants including five annuals, four perennial forbs and grasses and five evergreen or deciduous trees. Besides differing by life span and growth form, the selected plant species also showed a wide range of seed size (Supplementary Table S6). Plant litter types. Plant litter derives from a previous decomposition experiment focused on mass loss dynamics 47 , based on the litterbag method 41 . The litterbag experiment, based on eighteen plant species selected to represent a wide range of leaf litter quality, was carried out in controlled optimal conditions of temperature and water availability (further details in Supplementary Methods S1). The experiment produced 36 litter materials, either undecomposed (thereafter indicated as 0 days) or decomposed for 180 days. Biochemical quality of the 36 materials was previously reported 47 and used as a reference dataset for plant residue biochemical quality in relation to bioassay results. Data include total C and N contents, labile C, proximate cellulose 51 and spectral data from 13 C-CPMAS NMR in solid state (for details see Supplementary Methods S1 and Bonanomi et al. 19 ). Statistical analyses were done according to Supplementary Methods S5.
Plant and microbe bioassays. For higher plants, "root proliferation" bioassays were conducted in root observation chambers 19,71 . The aim of the bioassay was to assess the capability of seedling roots of the 14 target species to colonize the 36 different litter types, independent of the germination process. The experimental set up is fully reported in Supplementary Fig. S2 and in Supplementary Methods S2. Treatments, each with 10 replicates, included application of dry, powdered litter (separately for each of the 18 plant litter species at the 2 decomposition stages, for a total of 36 treatments) at a rate within the range observed in natural ecosystems considering the amount of litterfall and standing litter 72 , or distilled water as the control (Supplementary Methods S2). At the end of the experiment seedling root length was measured by digital photography and the total root length was measured by image analyser software (LUCIA; Laboratory Imaging Ltd. version 4.51).
Bioassays with bacteria and fungi aimed at assessing the effects of plant litter on target species growth in absence of inter-specific competition, thus allowing the evaluation of the relationships between litter biochemistry and microbial saprophytic growth. Treatments for bacteria were litter water extracts, obtained and applied as specified in Supplementary Methods S3, while Nutrient Broth, a common microbiological substrate, was used as control. Bacterial growth was spectrophotometrically measured at 6, 12, 24 and 48 hours of incubation after the inoculum (Supplementary Methods S3) using a Thermomax microtitre plate reader (Molecular Devices, Wokingham, UK). For data analysis, we used growth recorded after 48 hours and species response was expressed as percentage of the untreated control.
Treatment solutions for fungi were prepared as described in Supplementary Methods S4, in order to test the capability of fungi to use plant litter as the only source of nutrient, with Potato dextrose agar (PDA, Fluka), a common mycological substrate, used as the control. Seven days after the inoculum (Supplementary Methods S4), hyphal density and radial growth of each colony were assessed as described in Supplementary Methods S4. A fungal growth index was calculated as the product of the area of fungal colony, calculated from the measured colony radius, and hyphal density, following Tuitert et al. 73 .
Growth data from all bioassays were expressed as either percent of the untreated controls (i.e. control value = 100) or percent difference compared to the controls (i.e. control value = 0) and statistically analysed as described in Supplementary Methods S5