Introduction

Forests store vast reserves of soil carbon in the form of soil organic matter (OM) but the long-term fate of these carbon stores is uncertain (Trumbore and Czimczik 2008; Schmidt et al. 2011; Lehmann and Kleber 2015). Globally, net soil carbon losses have been reported for forests (Bond-Lamberty et al. 2018). In addition, changes in climate and atmospheric carbon dioxide levels will not only alter primary productivity of forests, but alter the quality and quantity of litter that enters soil (Prescott 2010; Bradford et al. 2016; Kohl et al. 2018). The quantity and quality of litter are also tied to the subsequent microbial decomposition of these inputs (Billings and Ziegler 2008; Schimel and Schaeffer 2012). Interestingly, litter addition has been observed to enhance soil priming and respiration, resulting in no net gain in soil carbon or even a decrease in soil carbon (Crow et al. 2009a; Sayer et al. 2011). The stimulated loss of carbon via priming may negate the increased carbon inputs through litter addition because this litter is rapidly degraded by soil microbiota (Scholes et al. 1997; Fang et al. 2015; Miao et al. 2019). Furthermore, several investigations reported that the addition of above-ground litter such as leaves or needles as well as woody debris in forests did not result in net carbon gains (Lajtha et al. 2014; Bowden et al. 2014; vandenEnden et al. 2018). As such, there is a need to further investigate how changes in litter quantity and quality alter soil carbon in forests, especially since there is growing evidence that additional litter inputs do not equate to increased soil carbon storage.

It is also widely established that litter quality and quantity can alter soil OM composition and stability, especially in forests where a major component of this OM is plant-derived (Quideau et al. 2001b, a; Prescott 2010). The Detrital Inputs and Removals Treatment (DIRT) experimental network aims to understand the role of plant detritus inputs and litter (above- versus below-ground) quality on the accumulation and dynamics of soil OM in temperate forests (Lajtha et al. 2018). The DIRT experimental design, where litter has been chronically doubled or excluded, has facilitated several novel insights into changes in soil OM biogeochemistry at the molecular-level. For example, 20 years of doubling litter at the Harvard Forest, Massachusetts resulted in accelerated decomposition of soil OM as measured using several molecular proxies (Pisani et al. 2016). Interestingly, treatments that excluded either or both above- and below-ground litter, also accelerated soil OM decomposition likely due to microbial adaptation to changes in available substrates (Pisani et al. 2016). This study emphasized that even when faced with substrate limitations, soil microbes were still able to maintain their metabolic processes and respire carbon (Pisani et al. 2016). Similar findings were also observed after 20 years of experiment at the Bousson Experimental Forest, Pennsylvania further suggesting that double litter inputs resulted in fueled microbial processing of plant-derived OM and these extra inputs were not stabilized in soil (Wang et al. 2017). Both studies also reported higher concentrations of microbial biomass indicating that these processes and resulting changes in soil OM chemistry were driven via microbial processing of carbon across addition and exclusion treatments (Pisani et al. 2016; Wang et al. 2017).

The H. J. Andrews Experimental Forest in Oregon is an old-growth conifer forest which is also part of the DIRT network. This forest differs from the Harvard Forest and Bousson Experimental Forest because there are two addition treatments using contrasting litter qualities, as determined by the carbon/nitrogen, and include double litter which is predominantly high quality conifer needles (lower carbon/nitrogen) and double wood, comprised of Douglas-fir wood (higher carbon/nitrogen). Pierson et al. (2021a) examined the proportion of carbon in both particulate and mineral associated fractions after 20 years of DIRT treatments at the H.J. Andrews Experimental Forest and reported that above-ground litter was more influential to stabilized carbon in the mineral-associated fraction. This study also reported that low quality wood additions increased the particulate OM fraction only (Pierson et al. 2021a), suggesting that the higher quality double conifer needles were degraded more rapidly and contributed to the stabilized soil carbon pool more than extra wood. As such, it is important to assess how different quality litter additions (conifer needles versus wood) may alter soil OM molecular composition and how these results compare to the results from the Harvard Forest and Bousson Experimental Forest after 20 years of litter manipulation. These results are especially critical as it is proposed that higher quality litter is decomposed faster by soil microbiota preferentially over lower quality litter (Hicks Pries et al. 2017) but there is limited information about how these contrasting litters alter soil OM composition at the molecular-level, especially in the long-term.

In this study, a suite of molecular-level approaches was used to better understand soil OM dynamics after 20 years of litter additions and exclusions at the H. J. Andrews Experimental Forest. Based on previous DIRT studies conducted after 20 years of experiment (Pisani et al. 2016; Wang et al. 2017), we hypothesize that increased above-ground additions accelerated soil OM decomposition at the H. J. Andrews Experimental Forest. Furthermore, Crow et al. (2009b) reported that conifer-derived lipids were preferentially preserved in soil after 5 years of double litter addition, which was consistent with specific plant-derived OM compounds being more long-lived in soil (Schmidt et al. 2011). As such, we also hypothesize that above-ground additions will alter the selective preservation of specific soil OM compounds after 20 years of litter addition. We further hypothesize that shifts in the microbial biomass and substrate utilization will be significant after 20 years of litter manipulation and that differences in litter quality will alter the production of microbial-derived compounds. To test these hypotheses, we employed targeted and non-targeted molecular methods to assess the composition of specific soil OM components and the overall chemistry, respectively. Overall, the objectives of this study are to better understand the fundamental biogeochemical processes that govern soil OM molecular-level composition with 20 years of litter manipulation and to determine how litter quality and quantity impacts may control these processes.

Materials and methods

Experimental site description and sample collection

The litter addition and exclusion experiment started in 1997 at the H. J. Andrews Experimental Forest in the Cascade Mountains of west-central Oregon, USA (44°15ʹ N, 122°10ʹ W). The H. J. Andrews Experimental Forest is a natural old-growth (> 500 years) temperate coniferous rainforest dominated by Douglas-fir (Pseudotsuga menziesii) and western hemlock (Tsuga heterophylla) with other deciduous tree species including western red cedar (Thuja plicata) and vine maple (Acer circinatum). The trees are all distributed in a relatively even manner throughout a uniform forested area and the litter is predominantly comprised of Douglas-fir needles. The plots were established on a level and stone-free ground extensively covered with coarse woody debris and moss on the forest floor. The soil type is coarse loamy mixed mesic Typic Hapludands (Lajtha et al. 2005) and is classified as an Aluandic Andosol according to the World Reference Base (WRB) system (FAO 2014). The mean annual temperature is 9.4 °C and annual precipitation is 2080 mm (average from year 1999–2014; Pierson et al. 2021b).

The DIRT experimental plots include: control (ambient inputs) and six different addition or exclusion treatments which are depicted in Fig. S1. The exclusion treatments include: No Litter, No Roots and No Inputs where above-ground leaves/needles were excluded with a mesh screen, below-ground roots were excluded with a plastic barrier, and both above- and below-ground detritus were excluded from plots, respectively. To exclude above-ground litter, a 1 mm mesh screen was placed on the soil surface all year to collect and remove the above-ground detritus once per year in the late summer months while still allowing water and gas exchange. In the below-ground exclusion treatments, roots were excluded by installing vertical plastic, permeable barriers from surface soil to the top of C horizon (140 cm). It is important to note that the exclusion treatments (No Litter, No Roots and No Inputs) are not complete removals, but rather litter exclusion treatments. We further note that there may be some minor litter decomposition and solubilization of OM in the months before the litter is moved. The addition treatments include: Double Litter and Double Wood. For Double Litter plots, above-ground leaves/needles from the No Litter treatments were screened to remove large debris (i.e., branches and stems) and transferred to Double Litter plots once per year in the late summer months. For Double Wood, fresh and decayed shredded Douglas-fir wood pieces were added annually to the soil surface equivalent to the mass of litterfall. In the OA-less plots, the surface 0–30 cm (O and A horizons) soils were removed and then backfilled with B horizon material obtained from a hillslope immediately adjacent to the experimental area. The OA-less treatment provides insight into soil carbon genesis as the B horizon soil has mineral surfaces that are not yet saturated with soil carbon. This treatment therefore represents new soil carbon accumulation alongside treatments which have either litter additions or removals, but may already have reached carbon-mineral saturation. After the initial removal of the O and A horizons, the OA-less plots received ambient inputs continuously. Each treatment consists of three randomized field plots. The plot size is approximately 150 m2 (15 m × 10 m) for all the treatments except No Roots and No Inputs. The No Roots and No Inputs plots are approximately 75 m2, as plots of these two treatments are located adjacent to each other within a ~ 150 m2 root exclusion zone to provide a buffer region between the barriers and the experimental plots. All study plots are located in close proximity, spaced approximately 1–2 m apart across a vegetatively and topographically uniform area.

Soil samples were collected in July 2017 after 20 years of experiment. In each plot, soils were sampled using a 5.8 cm diameter Oakfield style soil core sampler. Six soil cores from random locations within each plot were collected and homogenized (Pierson et al. 2021a) for the O horizon layer (except for No Inputs and No Litter which do not have O horizons due to the experimental design) and mineral layer (0–10 cm). The soil carbon stocks, bulk density and pH for each treatment in this study were measured and reported in Pierson et al. (2021b), and were not significantly different between treatments and the control. Soil samples were frozen after sampling and transported to the laboratory. Samples were then freeze-dried, sieved with 2 mm mesh, and ground into a fine powder for analyses. Our study focused on how litter manipulation altered the soil OM composition relative to the control plots to evaluate the changes relative to ambient inputs after 20 years of experiment.

Soil carbon and nitrogen content measurement

The bulk carbon and nitrogen contents of the O horizon soil samples were determined with combustion by a Thermo Flash 2000 elemental analyzer (Thermo Scientific, Hudson, NH, USA). The bulk carbon and nitrogen contents of the mineral layer soil samples were measured using combustion on an Elementar Vario Macro Cube 219 (Elementar Analysensysteme GmbH, Langenselbold, Germany). Standard reference samples (> 90% accuracy) and sample replication (> 90% consistency) were used to confirm the accuracy of the elemental analyses (Pierson et al. 2021b).

Targeted soil organic matter (OM) extractions and analyses

Targeted OM compounds were extracted by solvent extraction, base hydrolysis and copper (II) oxide (CuO) oxidation to analyze total solvent extractable, cutin-, suberin-, microbial-derived, and lignin-derived compounds, respectively (Hedges and Mann 1979; Goñi and Hedges 1990; Otto and Simpson 2005, 2006a, b). Soil samples from each field replicate were extracted in duplicate (two analytical replicates). The detailed procedure was described in Supplementary Materials. Briefly, to isolate the solvent extractable molecules, soil samples were extracted with dichloromethane, dichloromethane:methanol (1:1, v:v) and methanol sequentially (Otto and Simpson 2005). The soil samples after solvent extraction were used for base hydrolysis to release the lipids in leaf/needle-derived cutin and root-derived suberin compounds. Approximately half of the soil mass after solvent extraction was hydrolyzed with methanolic potassium hydroxide at 100 °C for 3 h in Teflon-lined bombs (Goñi and Hedges 1990; Otto and Simpson 2006a). To release lignin-derived compounds, soil samples after base hydrolysis were oxidized with CuO, ammonium iron (II) sulfate hexahydrate [Fe(NH4)2 (SO4)2·6H2O] and sodium hydroxide in Teflon-lined bombs at 170 °C for 2.5 h (Hedges and Mann 1979; Otto and Simpson 2006b). Solid phase extraction was used to collect the released lignin-derived phenols (Pinto et al. 2010). Phospholipid fatty acids (PLFAs) were extracted using a modified Bligh–Dyer method (Bligh and Dyer 1959). Briefly, lipids were extracted from soil samples with methanol, chloroform, and sodium citrate buffer (acidified to pH 4). The collected extracts in the chloroform phase were fractionated into nonpolar lipids, glycolipids and polar lipids by elution with chloroform, acetone and methanol, respectively on a silicic acid chromatography column. The PLFAs in the methanol fraction were then converted into fatty acid methyl esters through a mild alkaline methanolysis.

Prior to gas chromatography-mass spectrometry (GC–MS) analysis, the extracts from targeted soil OM extractions were derivatized with N,O-bistrifluoroacetamide and pyridine except that base hydrolyzed extracts were first methylated using N,N-dimethylformamide dimethyl acetal. To quantify the targeted compounds, external standards were used: tetracosane, 1-docosanol, methyl tricosanoate and cholesterol for solvent extracts; methyl tricosanoate for base hydrolysed extracts; syringic acid and syringaldehyde for CuO oxidized extracts; and methyl oleate for PLFAs. The GC–MS analysis was conducted on an Agilent 7890B gas chromatograph attached to a 5977B mass spectrometer. The instrument was equipped with a HP-5MS fused silica capillary column (30 m × 0.25 mm i.d., 0.25 μm film thickness) with helium as the carrier gas at a flow rate of 1 mL min−1. The temperature gradient program in the gas chromatograph was set as follows: 65 °C for 2 min., elevated temperature to 300 °C at a rate of 6 °C min−1, held for 2 min. at 300 °C. The m/z ranged from 50 to 650 with 70 eV electron impact ionization. Agilent Mass Hunter GC–MS Acquisition (version B.07.03.2129) and the Agilent Enhanced ChemStation software (version E.02.02.1431) were used for data acquisition and processing, respectively. Mass spectra were compared with the National Institute of Standards and Technology (NIST) and Wiley 275 MS library data to identify the targeted compounds. To quantify different groups of compounds, peak areas of the compounds were compared with those of external standards (Otto and Simpson 2005, 2006a, b; Spielvogel et al. 2014; Angst et al. 2016). The organic compound concentrations normalized to soil mass (μg/g) were used for the discussion in this study (Tables S1-S2), which is consistent with other DIRT studies (Mayzelle et al. 2014; Pisani et al. 2016; Wang et al. 2017; Reynolds et al. 2018; vandenEnden et al. 2018; Pierson et al. 2021b, a). The organic compound concentrations normalized to soil carbon content (μg/g C) are also available in Tables S3-S4. An overview of comparison between concentrations normalized to soil mass (μg/g) and concentrations normalized to soil carbon content (μg/g C) is available in the Supplementary Materials.

Solvent extracted acyclic lipids mainly including n-alkanes, n-alkanoic acids and n-alkanols were categorized into short-chain (< 20) and plant-derived (long-chain (≥ 20)) lipids as these two types of compounds are considered to be mainly derived from microbial and plant origins, respectively (Lichtfouse et al. 1995). The plant-derived steroid degradation ratio, calculated as the ratio of altered steroids over their steroid precursors, was used to determine the degradation state of plant-derived steroids (Otto and Simpson 2005). Cutin-derived compounds (arise from plant leaf/needle waxes) were mainly comprised of short-chain, mid-chain hydroxy acids, C16 mono- and dihydroxy acids and dioic acids (Otto and Simpson 2006a). Suberin-derived compounds (derived from plant roots) primarily contained long-chain ω-hydroxy-alkanoic and dioic acids, and 9,10-epoxy-ω-hydroxy C18 acid (Otto and Simpson 2006a). As the ratio of C16 ω-hydroxy-alkanoic acids to all hydrolysable C16 fatty acids (ω-C16/∑C16) increased with enhanced cutin degradation (Goñi and Hedges 1990), this ratio was calculated to determine the degradation status of cutin. Microbial-derived hydrolysable lipids (microbial-derived lipids hereafter) primarily consisted of C14–C19 branched alkanoic acids and short-chain (C10–C18) β-hydroxy-alkanoic acids (Cai et al. 2017; Jia et al. 2019). Lignin-derived compounds included: vanillyl, syringyl, and cinnamyl phenols (Hedges and Mann 1979; Otto and Simpson 2006b; Kaiser and Benner 2012). To assess lignin oxidation (degradation), the acid to aldehyde ratios for lignin-derived vanillyl and syringyl compounds ((Ad/Al)v and (Ad/Al)s, respectively) were calculated, as these ratios increased with progressive lignin degradation (Hedges and Mann 1979). Microbial PLFAs were categorized as Actinomycetes (10Me16:0, 10Me17:0, 10Me18:0), Gram-positive bacteria (i14:0, a14:0, i15:0, a15:0, i16:0, i17:0, a17:0, Actinomycetes), Gram-negative bacteria (16:1ω7c, cy17:0, 18:1ω7c and cy19:0) and fungi (18:2ω6,9c; Harwood and Russell 1984). Because more cyclopropane PLFAs were produced when bacteria experience substrate limitations, the ratio of cyclopropane PLFA over its monoenoic precursor (cy17:0/16:1ω7c) was used as a microbial stress indicator. This ratio (cy17:0/16:1ω7c) has also been used to assess changes in microbial community composition and growth (Frostegård et al. 2011; Wixon and Balser 2013). Gram-negative/Gram-positive bacterial PLFAs and fungal/bacterial PLFAs were also determined to assess changes in microbial community structure with litter manipulation (Lyu et al. 2018; Chen et al. 2019).

Solid-state 13C nuclear magnetic resonance (NMR) spectroscopy

The overall OM composition of the soil samples was characterized using solid-state 13C nuclear magnetic resonance (NMR) spectroscopy. As soil sample masses were limited from long-term experiments and NMR instrument time was constrained, composite samples (one per each treatment for both O horizon and mineral layer soils) were prepared and used for NMR analysis. This approach represents average responses between different treatments. This method was applied at other DIRT sites by Pisani et al. (2016) and Wang et al. (2017), and was used here to maintain data consistency and facilitate the direct comparisons between sites. It is also important to note that NMR spectra were not replicated because each NMR spectrum represents an average of thousands of individual replicate scans. Moreover, solid-state 13C NMR is highly reproducible with measurement errors ranging between 1–5% (Dria et al. 2002; Sun et al. 2019; Usman and Simpson 2021). Composite soil samples from the mineral layer were first pretreated with hydrofluoric acid (HF, 10%) repeatedly with the same number of treatments and for the same period of time to concentrate soil OM and to remove paramagnetic minerals. The HF-treated samples were then rinsed with deionized water approximately 15 times to remove salts. The excess deionized water was removed by centrifugation and the samples were then frozen and freeze-dried. Around 200 mg of each sample was packed into a 4 mm zirconium rotor with a Kel-F cap and characterized by solid-state 13C cross polarization with magic angle spinning (CP-MAS) NMR spectroscopy. The spectra were acquired on a 500 MHz Bruker BioSpin Advance III spectrometer (Bruker BioSpin, Rheinstetten, Germany) with a 4 mm H–X MAS probe. The main operating and processing parameters were set as follows: 11 kHz spinning rate, 1 ms ramp-CP contact time, 1 s recycle delay and 50 Hz line broadening. The spectra were integrated into four chemical shift regions including alkyl carbon (0–50 ppm) that is mainly derived from cutin, suberin side chains, aliphatic side chains and lipids; O-alkyl carbon (50–110 ppm) that mainly arises from carbohydrates, peptides and methoxyl carbon in lignin; aromatic and phenolic carbon (110–165 ppm) that primarily comes from lignin, aromatic amino acids observed in peptides and black carbon; carboxyl and carbonyl carbon (165–215 ppm) that is mainly from fatty acids and amino acids found in peptides (Baldock et al. 1992; Simpson et al. 2008).

Statistical analyses

As each treatment contains 6 replicates: 3 field replicates and 2 analytical replicates, a mixed model (with treatments as the fixed factor and plots (field and analytical replicates) as the random factor, respectively) analysis of variance (ANOVA) with Tukey’s post hoc comparison was used to determine the differences in soil OM and microbial indices relative to the control as well as between treatments. Significant differences relative to the control and amongst treatments are noted using different letters in Tables S1-S4. Because our interpretation focused on differences relative to the ambient inputs (control), significant differences were only denoted between the manipulation treatments and the control in the figures. A t-test was used to determine the differences in soil carbon, nitrogen content and carbon/nitrogen between litter manipulation treatments and the control. Differences were considered significant at the level of P < 0.05. Statistical analyses were conducted using IMB SPSS Statistics version 22. The full dataset and statistical analysis results were available in Tables 1 and S1-S4. A summary of percentage changes in soil OM components from detrital manipulation treatments relative to the control in O horizon and mineral soils is listed in Table 2.

Table 1 Carbon, total nitrogen concentrations and solid-state 13C nuclear magnetic resonance (NMR) characteristics of O horizon and mineral layer soil samples from detrital manipulation treatments
Full size table
Table 2 Summary of concentration changes and degradation of organic matter compounds in O horizon and mineral soils from detrital manipulation treatments relative to the control
Full size table

Results

Soil carbon, nitrogen, organic matter (OM) compositional changes and degradation

O horizon

In the O horizon, soil carbon increased slightly with Double Wood and decreased with OA-less, although no significant change in soil carbon was found among treatments compared with the control (Table 1). Soil nitrogen contents did not significantly differ from the control across most litter manipulation treatments but decreased with OA-less (P < 0.05; Table 1). Carbon/nitrogen was only significantly higher for Double Wood and significantly lower for OA-less relative to the control (P < 0.05; Table 1). Plant-derived lipids (mainly from plant leaf waxes) were 59% lower under OA-less relative to the control (P < 0.05; Fig. 1a and Table 2). Short-chain (< 20) acyclic lipids and cyclic lipids were significantly lower with Double Litter, No Roots and OA-less compared to the control (P < 0.05; Table S1). Simple sugars were 55% lower under OA-less relative to the control (P < 0.05; Fig. 1b and Table 2). Plant-derived steroid degradation significantly declined with Double Wood compared to the control (P < 0.05; Fig. 3a and Table 2). Cutin-derived compounds (from the waxy coating of plant leaves/needles) were significantly higher with Double Litter and Double Wood but lower under OA-less and No Roots than the control (P < 0.05; Fig. 2a and Table 2). Suberin-derived compounds (root-derived compounds) were significantly higher with Double Wood but lower under OA-less and No Roots relative to the control (P < 0.05; Fig. 2b and Table 2). Cutin degradation (ω-C16/∑C16) significantly declined across treatments relative to the control (P < 0.05; Fig. 3b). Microbial-derived lipids, which stem from both cellular and extracellular compounds (Zelles 1999; Cai et al. 2017; Jia et al. 2019), were 94% higher under Double Wood and 69% lower with OA-less than the control (P < 0.05; Fig. 1c and Table 2).

Fig. 1
figure 1

Plant-derived lipid, simple sugar and microbial-derived lipid concentrations of the O horizon and the mineral layer (0–10 cm) soil samples from detrital manipulation treatments. Error bars represented the standard errors of both field (n = 3) and analytical replicates (n = 2). Significant differences (P < 0.05) between treatments and the control were denoted with “*”. No Litter and No Inputs treatment plots lacked O horizons as they excluded the above-ground litter. Dashed lines represented the ambient concentrations

Full size image
Fig. 2
figure 2

Cutin-, suberin- and lignin-derived compound concentrations of the O horizon and the mineral layer (0–10 cm) soil samples from detrital manipulation treatments. Error bars represented the standard errors of both field (n = 3) and analytical replicates (n = 2). Significant differences (P < 0.05) between treatments and the control were denoted with “*”. No Litter and No Inputs treatment plots lacked O horizons as they excluded the above-ground litter. Dashed lines represented the ambient concentrations

Full size image
Fig. 3
figure 3

Organic matter degradation indicators of the O horizon and the mineral layer (0–10 cm) soil samples from detrital manipulation treatments. Error bars (except alkyl/O-alkyl carbon ratio) represented the standard errors of both field (n = 3) and analytical replicates (n = 2). Significant differences (P < 0.05) between treatments and the control were denoted with “*”. Plant-derived steroid degradation ratio was the ratio of altered steroids over their steroid precursors. ω-C16/∑C16 ratios were the ratios of C16 ω-hydroxy-alkanoic acids over all hydrolysable C16 fatty acids. (Ad/Al)v and (Ad/Al)s ratios were the acid to aldehyde ratios for lignin-derived vanillyl and syringyl compounds, respectively. No Litter and No Inputs treatment plots lacked O horizons as they excluded the above-ground litter. Dashed lines represented the ambient degradation state. Results above and below the dash lines represented higher and lower soil organic matter (compound) degradation relative to the control, respectively

Full size image

Lignin-derived compounds increased by 24% under Double Wood in the O horizon relative to the control (P < 0.05; Fig. 2c and Table 2). Lignin-derived vanillyl degradation ((Ad/Al)v) was significantly lower under Double Wood and OA-less, higher with No Roots relative to the control (P < 0.05; Fig. 3c and Table 2). Lignin-derived syringyl oxidation ((Ad/Al)s) was significantly higher under No Roots relative to the control (P < 0.05; Fig. 3d and Table 2). Lignin-derived syringyls/vanillyls significantly decreased under Double Wood and increased with OA-less and No Roots relative to the control (P < 0.05; Table S1). Lignin-derived cinnamyls/vanillyls significantly declined under Double Wood and OA-less and increased under No Roots compared with the control (P < 0.05; Table S1). The plotted lignin-derived syringyls/vanillyls and cinnamyls/vanillyls demonstrated that lignin composition under Double Wood and No Roots were significantly altered from the control (P < 0.05; Fig. S2a).

Solid-state 13C NMR analysis with the O horizon soil samples (Fig. S3) revealed that alkyl carbon (mainly from lipids and aliphatic side chains found in cutin and suberin) was higher under Double Litter and OA-less and lower under Double Wood, but similar with No Roots relative to the control (Table 1). O-alkyl carbon (from cellulose, peptides and methoxy groups) was generally lower across treatments than the control (Table 1). Aromatic and phenolic carbon was similar across treatments relative to the control except under Double Wood where it was higher (Table 1). Carboxyl and carbonyl carbon was generally higher among the treatments than the control (Table 1). Alkyl/O-alkyl carbon was calculated to assess soil OM degradation as O-alkyl carbon forms are preferred substrates for microbes compared with alkyl carbon (Baldock et al. 1992). Soil OM degradation (alkyl/O-alkyl carbon) was higher under detritus manipulation treatments except under Double Wood where it was similar with the control (Table 1; Fig. 3e).

Mineral (0–10 cm) soil layer

Soil carbon concentration in the mineral layer (0–10 cm) was slightly higher under Double Wood, but this change was not statistically significant. No significant differences were observed in soil carbon concentrations across treatments compared with the control except under OA-less where soil carbon was lower (Table 1). Soil total nitrogen concentrations did not show significant differences among the treatments relative to the control (Table 1). Carbon/nitrogen was significantly lower under No Roots and OA-less compared with the control (Table 1). Plant-derived lipids were significantly higher under Double Litter and Double Wood (P < 0.05; Fig. 1d) and lower with No Roots and OA-less compared to the control (P < 0.05; Fig. 1d and Table 2). Short-chain (< 20) acyclic lipids were significantly lower under OA-less, No Roots, No Litter and No Inputs than the control (P < 0.05; Table S2). Simple sugars significantly increased under Double Litter and Double Wood compared with the control (P < 0.05; Fig. 1e and Table 2). Cyclic lipids were significantly lower with OA-less, No Roots, No Litter and No Inputs relative to the control (P < 0.05; Table S2). Plant-derived steroid degradation (ratio) was slightly higher under Double Litter and slightly lower with No Roots, No Litter and No Inputs than the control, although these changes were not significant (Fig. 3f).

In the mineral layer (0–10 cm) cutin-derived compounds increased by 21% with Double Litter (P < 0.05) and significantly decreased under Double Wood, OA-less, No Litter and No Inputs compared with the control (P < 0.05; Fig. 2d and Table 2). Suberin-derived compounds were higher under Double Litter (+ 25%) and Double Wood (+ 33%) and lower with OA-less (-36%) and No Litter (− 19%) relative to the control (P < 0.05; Fig. 2e and Table 2). The total concentration of cutin- and suberin-derived compounds significantly increased with Double Litter and Double Wood, and declined under OA-less, No Litter, No Roots and No Inputs (P < 0.05; Table S2). Cutin degradation (ω-C16/∑C16) was significantly higher with Double Wood and lower under No Roots, No Litter and No Inputs compared with the control (P < 0.05; Fig. 3g). Microbial-derived lipids were significantly higher with Double Litter and No Roots, and significantly lower under OA-less, No Litter and No Inputs relative to the control (P < 0.05; Fig. 1f and Table 2). Lignin-derived compounds were significantly lower under all the treatments than the control (P < 0.05; Fig. 2f and Table 2). Lignin-derived vanillyl oxidation ((Ad/Al)v) was lower with Double Wood, OA-less, No Roots and No Inputs than the control (P < 0.05; Fig. 3h and Table 2). Lignin-derived syringyl degradation ((Ad/Al)s) did not significantly differ across treatments compared with the control (Fig. 3i and Table 2). Lignin-derived syringyls/vanillyls significantly declined across treatments (P < 0.05) except under No Litter where it remained similar compared with the control (Table S2). Lignin-derived cinnamyls/vanillyls increased under Double Litter, decreased under Double Wood, OA-less and No Inputs compared with the control (P < 0.05; Table S2). A comparison of lignin-derived syringyls/vanillyls and cinnamyls/vanillyls showed that lignin composition under Double Wood, OA-less, No Roots and No Inputs was significantly distinct from the control (P < 0.05; Fig. S2b).

Solid-state 13C NMR (Fig. S4) showed that alkyl carbon in the mineral soil layer was similar under Double Wood, OA-less, No Roots and No Litter, higher with Double Litter and No Inputs than the control (Table 1). O-alkyl carbon content was generally lower among all the detrital manipulation treatments relative to the control (Table 1). Aromatic and phenolic carbon was similar under Double Litter, Double Wood, OA-less and No Litter, and increased under No Roots and No Inputs compared with the control (Table 1). Carboxyl and carbonyl carbon content was slightly higher across treatments compared with the control (Table 1). Soil OM degradation, as indicated by alkyl/O-alkyl carbon, was higher with all the detrital manipulation treatments compared with the control except under OA-less where it showed a lower ratio (Table 1; Fig. 3j).

Microbial biomass and community composition

O horizon

In the O horizon, fungal, bacterial and total microbial biomass were significantly higher under Double Wood and No Roots and lower with OA-less compared with the control (P < 0.05; Fig. 4a–c and Table 2). Microbial stress (cy17:0/16:1ω7c) significantly increased under Double Litter and Double Wood relative to the control (P < 0.05; Fig. 4d and Table 2). Gram-negative/Gram-positive bacterial PLFAs significantly declined under Double Wood and No Roots relative to the control (P < 0.05; Tables 2 and S1). Fungal/bacterial PLFAs was similar across treatments relative to the control in the O horizon (Tables 2 and S1).

Fig. 4
figure 4

Microbial phospholipid fatty acid (PLFA) indices for the O horizon and the mineral layer (0–10 cm) soil samples from detrital manipulation treatments. Error bars represented the standard errors of both field (n = 3) and analytical replicates (n = 2). Significant differences (P < 0.05) between treatments and the control were denoted with “*”. No Litter and No Inputs treatment plots lacked O horizons as they excluded the above-ground litter. Dashed lines represented the ambient PLFA concentrations and microbial stress

Full size image

Mineral (0–10 cm) soil layer

In the mineral soil layer, fungal biomass was significantly higher with Double Litter, Double Wood, No Roots and No Inputs compared with the control (P < 0.05; Fig. 4e and Table 2). Bacterial and total microbial PLFAs were significantly higher with Double Litter, Double Wood and No Inputs (P < 0.05) and significantly lower with OA-less compared with the control (P < 0.05; Fig. 4f, g and Table 2). The microbial stress (cy17:0/16:1ω7c) was significantly higher with all the treatments relative to the control (P < 0.05; Fig. 4h and Table 2). Gram-negative/Gram-positive bacterial PLFAs was not significantly different for most treatments but was lower with Double Wood and No Inputs (P < 0.05; Tables 2 and S2). Fungal/bacterial PLFAs was significantly higher under Double Litter, Double Wood and OA-less, and significantly lower with No Litter relative to the control (Tables 2 and S2).

Discussion

Added litter quality drives soil organic matter (OM) molecular biogeochemical pathways

After 20 years of Double Litter treatment, no net gain in surface soil carbon concentration (Table 1) nor in carbon stock (Pierson et al. 2021b) was observed relative to the control. The soil carbon results were consistent with other experimental sites that also reported no significant increase in soil carbon with the addition of litter inputs (Lajtha et al. 2014; Bowden et al. 2014; vandenEnden et al. 2018; Sayer et al. 2019, 2021), suggesting that increased litter quantity did not necessarily result in soil carbon accrual (Pierson et al. 2021a). The lack of net carbon gains with annual doubled litter (higher quantity and quality) was attributed to enhanced soil priming and enhanced carbon dioxide production (Crow et al. 2009a; Sayer et al. 2011; Bréchet et al. 2018). The increase in simple sugars with Double Litter was likely from decomposition products of biopolymeric cellulose, which are relatively fast-cycling OM components that may provide additional preferred substrates for microbes (Gunina et al. 2014; Gunina and Kuzyakov 2015) and stimulate microbial priming. Greater availability of simple sugars in Double Litter plots would also explain the higher microbial biomass in the mineral soil. The decreased relative abundance of cellulose (O-alkyl carbon; Table 1) indicated enhanced microbial decomposition of cellulosic components, which was consistent with the overall higher stage of soil OM decomposition (alkyl/O-alkyl carbon) from solid-state 13C NMR analysis. The lower cyclic lipids in the O horizon and slightly enhanced steroid degradation in the mineral layer under Double Litter (Fig. 3 and Table 2) also suggested higher microbial decomposition of specific forms of soil OM (Chen et al. 2014; Pisani et al. 2016; Lyu et al. 2018), and supported our hypothesis that increased above-ground additions accelerated soil OM decomposition after 20 years. These results are consistent with continued and sustained soil priming with Double Litter observed earlier at this site (Crow et al. 2009a).

Lower cutin degradation in the O horizon with increased quantity of above-ground litter (Double Litter) was consistent with extra inputs of conifer needles. The accumulation of plant cuticular wax inputs with added litter agreed with increased cutin-derived compounds in the O horizon and mineral soil layer (Fig. 2) and higher plant-derived lipids (long-chain acyclic compounds) in the mineral layer (Fig. 1 and Table 2) of the Double Litter plots. Higher concentrations of root-derived (suberin) compounds with Double Litter reflected a potential shift in microbial substrate use because of added preferred substrates from litter. These results agreed with a change in microbial community composition as indicated by an increase in fungi relative to bacteria (higher fungal/bacterial PLFAs; Table S2) in the mineral soil and an increase in microbial stress (cy17:0/16:1ω7c; Fig. 4), which may relate to microbial growth and community change. There was no evidence for lignin degradation with Double Litter in both the O horizon and mineral layers (insignificant differences in lignin oxidation ((Ad/Al)v and (Ad/Al)s) and no change in aromatic carbon content from NMR analysis).

Previous studies have indicated that increased microbial biomass coupled with increased degradation of specific OM components may enhance the contribution of microbial-derived compounds to soil OM (Miltner et al. 2012; Wieder et al. 2014; Kallenbach et al. 2016). Microbial-derived lipids were significantly higher with Double Litter in the mineral soils (Fig. 1 and Table 2), suggesting the higher production of microbial-derived OM with needle addition. As above-ground litter (mainly conifer needles) is relatively nitrogen-rich, our results suggested that the addition of high quality litter enhanced the production of microbial-derived OM, likely in relation to altered microbial physiological status with high quality substrates (Córdova et al. 2018; Ni et al. 2020). Pierson et al. (2021a) also reported increased mineral-associated carbon with Double Litter which also suggested the stabilization of soil carbon with needle addition. Interestingly, we did not observe any increase in soil carbon with needle addition. However, these results demonstrated that biogeochemical processes, as well as the production of microbial-derived compounds were shifted.

The potential for soil carbon accumulation under Double Wood was likely due to slower decomposition and related to the lower quality of woody debris (Melillo et al. 1982; Bradford et al. 2016; Pierson et al. 2021b). However, the addition of cellulose-rich woody debris (increased quantity) may provide preferred substrates for microbes and increase microbial processing selectively. The higher microbial biomass (PLFAs) and decrease of cellulose (O-alkyl carbon) in both O horizon and mineral layers suggested enhanced microbial decomposition of high quality substrates. The higher decomposition coincided with the overall enhanced soil OM decomposition (alkyl/O-alkyl carbon) in the mineral soil from the NMR analysis and agreed with our hypothesis that above-ground additions enhanced soil OM decomposition. The higher lignin-derived compound and aromatic carbon (mostly from lignin) content in the O horizon and similar aromatic carbon content in the mineral layer under Double Wood suggested that lignin concentrations did not change. This result is likely due to higher lignin inputs from wood, which was consistent with the compositional changes in lignin-derived compounds (syringyls/vanillyls and cinnamyls/vanillyls; Fig. S2) that pointed towards increased gymnosperm wood inputs compared with the control (Hedges and Mann 1979). Moreover, in tandem with the decreased or similar lignin degradation ((Ad/Al)v and (Ad/Al)s) under Double Wood relative to the control, these data collectively suggested that lignin in the Double Wood treatment was not significantly oxidized. This result is consistent with a previous study at the same site that reported similar lignin-derived compound concentrations relative to the control in the mineral soil layer after 5 years of wood addition (Crow et al. 2009b) and suggested that lignin accumulation and stability were unaltered after 20 years of experiment.

Interestingly, Double Wood enhanced cutin degradation in the mineral soil layer but not in the O horizon. This observation is likely because the cellulose-rich wood inputs in the O horizon provided sufficient energy sources for microbial growth (higher microbial PLFAs) and reduced the need to degrade cutin. The decomposition products of cellulose (such as simple sugars; Fig. 1 and Table 2) may translocate into the mineral layer (Lajtha et al. 2005) and stimulate soil priming (Sayer et al. 2011), which then resulted in the enhanced decomposition of cutin-derived compounds which are typically long-lived and slow-cycling (Riederer et al. 1993; Lorenz et al. 2007). The increased root-derived compounds under Double Wood were likely associated with microbes shifting to use added cellulose from wood over other available soil OM components. A shift in microbial community structure, relatively higher Gram-positive over Gram-negative bacteria in the O horizon and the mineral layer and higher proportion of fungi over bacteria in the mineral layer were also observed (Tables S1-S2), further suggesting changes in substrate use. This shift also agreed with an increase in microbial stress (cy17:0/16:1ω7c), which may be associated with alterations in microbial community and growth as well as changes in available substrates (Frostegård et al. 2011; Wixon and Balser 2013). Moreover, another study at the same site reported that the mass of fine roots increased under Double Wood relative to the control (Pierson et al. 2021a), which may also account for the higher root-derived compounds observed in this study. Despite higher microbial biomass in the mineral soil under Double Wood, this treatment did not result in higher microbial-derived lipids. These results suggested that wood debris of low quality (higher carbon/nitrogen) may not increase the microbial carbon use efficiency and the production of microbial-derived OM (Lekkerkerk et al. 1990; Manzoni et al. 2012). Comparisons between the microbial-derived lipids under Double Litter and Double Wood corroborated litter quality as a key controlling factor for microbial-derived OM production. These results confirmed our hypothesis that higher litter quality resulted in a production of higher microbial-derived lipids in addition to higher microbial biomass.

Detrital exclusion alters soil organic matter (OM) molecular biogeochemical trajectories

Litter exclusion (No Roots, No Litter, No Inputs) did not significantly change surface soil carbon content and concurred with observations made after 5 years of detritus exclusion (Crow et al. 2009b). Interestingly, the soil OM composition was sensitive to various forms of detrital exclusion (No Roots, No Litter, and No Inputs), which all exhibited an overall increase in soil OM decomposition (alkyl/O-alkyl carbon) relative to the control. Microbial biomass (total PLFA concentrations) also increased with litter exclusion but the increase was only statistically significant with No Inputs, suggesting that microbial biomass adapted to substrate limitations to sustain their metabolic activity (Brant et al. 2006; Yarwood et al. 2013). The increase in microbial biomass (total PLFA concentrations) likely enhanced the decomposition of soil OM with litter exclusion, even when faced with changes in inputs and substrates (Yarwood et al. 2013; Pisani et al. 2016). These results coincided with an increase in microbial stress (cy17:0/16:1ω7c PLFAs; Fig. 4 and Table 2) with long-term changes in substrate quality (Bossio and Scow 1998; Zhao et al. 2017). For all litter exclusion treatments (No Roots, No Litter, and No Inputs), the decreased cellulose (O-alkyl carbon from solid-state 13C NMR) suggested that this substrate was critically important for microbial processing of soil OM with exclusion of above- and/or below-ground detritus (Štursová et al. 2012). Other long-term DIRT studies also reported similar patterns (Pisani et al. 2016; Wang et al. 2017). The decreased or similar degradation of plant-derived steroids, cutin-derived compounds, and lignin-derived compounds collectively indicated that cellulose decomposition was sustained with litter exclusion (Štursová et al. 2012; Kotroczó et al. 2020).

Twenty years after the removal of the O and A horizons (OA-less), the soil carbon concentration for both the surface layer (O horizon) and mineral soil (0–10 cm) remained lower than the control, which was consistent with observations that soil carbon accrual required several decades (Huang and Spohn 2015; Lajtha et al. 2018; Sayer et al. 2019). Moreover, less degraded cutin (ω-C16/∑C16) was observed in the O horizon of the OA-less treatment. Lignin degradation ((Ad/Al)v and (Ad/Al)s) was either lower or similar relative to the control in the O horizon and mineral soil layers with OA-less. These results suggested slower cycling of specific soil OM components (i.e., cutin and lignin; Pisani et al. 2016). The lower total microbial biomass (PLFAs) in the O horizon and the mineral soil demonstrates that 20 years was insufficient to restore the soil to the same levels as ambient conditions (control; Lajtha et al. 2014; Pisani et al. 2016). The decreased relative abundance of cellulose (O-alkyl carbon) under OA-less suggested that cellulose was the main driver for microbial processing in nutrient-limited soils (Štursová et al. 2012).

Several unique differences between the exclusion treatments (No Roots, No Litter, and No Inputs) were also observed. Above-ground litter exclusion (No Litter and No Inputs) decreased the concentration of cutin-derived compounds. Below-ground litter exclusion (No Roots and No Inputs) did not decrease the concentration of suberin-derived compounds, likely due to the microbial shift towards other substrates, such as cellulose (Štursová et al. 2012; Wang et al. 2017). Furthermore, these results suggested that in this forest, the exclusion of above-ground litter exerted a greater control on the molecular biogeochemistry of soil OM than below-ground detritus. With No Litter and No Inputs, a decline in microbial-derived lipids may be associated with the lack of high quality litter substrates for microbes and thus lower microbial carbon use efficiency (Manzoni et al. 2012). With No Roots, the increased microbial-derived lipids in the mineral layer also demonstrated the shift towards microbial use of higher quality above-ground litter as substrates. The presence of high quality litter may promote microbial carbon use efficiency, and stimulate the production of microbial-derived OM (Manzoni et al. 2012; Cotrufo et al. 2013). Particularly, with microbial shifts towards using above-ground litter, litter-derived (cutin) compound concentrations were similar under No Roots. These results suggested that litter-derived compounds did not decrease and were preserved, which is consistent with the preferential preservation of needle-derived rather than root-derived OM observed earlier at this site (Crow et al. 2009b).

Conclusions

Annual litter addition over a 20 year period, in the form of conifer needles (Double Litter) or wood (Double Wood) did not enhance soil carbon sequestration. Collectively, our results demonstrated that these litter addition treatments continued to enhance soil priming via continued and sustained microbial decomposition over time. The lack of net carbon gain due to soil priming was consistent with observations made at the Harvard Forest and Bousson Forest DIRT experiments after 20 years of litter additions. Interestingly, the contrasting litter quality between Double Litter (high quality) and Double Wood (low quality) also governed how soil microbes processed these additions as well as native soil OM composition. For the first time, we have established a direct molecular-level relationship between high litter quality (Double Litter) and the production of cellular and extracellular microbial-derived lipids, which subsequently contributed to soil OM. The exclusion of above- and/or below-ground litter (No Roots, No Litter, No Inputs) also increased the decomposition of soil OM, suggesting that microbes adapted to changes in substrate quantity and quality to maintain their metabolic activity over time. Our findings also indicated that above-ground litter-derived carbon was more dominant in altering soil OM biogeochemistry in comparison to below-ground litter inputs. Moreover, this study for the first time provided direct evidence regarding how litter quality regulated the contribution of microbial-derived OM to soil at the molecular-level. In addition, there was a clear link between microbial biomass adaptation to changes in litter deposition and soil OM chemistry over 20 years which further supported the importance of a more holistic approach to studying soil carbon biogeochemistry. Our results also revealed that above-ground litter quality and quantity governed soil OM biogeochemical cycling and dynamics more than below-ground litter in forests.