Data Package Metadata   View Summary

We collected surface mineral soils (0–10 cm, surface litter removed) from 10 North American forests spanning a broad range of edaphic characteristics (Appendix 1: Table S1). Samples were stored field-moist in polyethylene bags at 4 degree C for several weeks prior to the incubation. Standing senesced above-ground stem and leaf tissue (litter) from Andropogon gerardii Vitman, a C4 grass, was collected near Ames, IA, USA (42.04 degreeN, -93.60 degreeW), dried at 65 degreeC, and ground to pass a 250 microm mesh. The litter had 41.3% C and 1.1% N by mass (C:N ratio of 37.5) and a delta13C value of -13.1permille.

13C-labeled and unlabeled lignins

We prepared synthetic guaiacyl lignin following Kirk & Brunow (1988) which contained either a 13C label (99 atom%) or natural abundance C at the Cbeta position of each C9 substructure. Lignins were fractionated by gel permeation chromatography to obtain a high-molecular-weight fraction ( greater than 1000 Da). The average labeled phenylpropane subunit has a molecular mass of 197, so the 13C label represented 13/197 of total lignin mass. Lignins were precipitated on the A. gerardii litter in a 1:21 mass ratio by dissolution in a 1:4 solution of acetone and water and evaporation in a fume hood. The lignin-litter mixture had a C:N ratio of 38.3. Further preparation and characterization details are given in Hall et al. (2015). Soil incubation experiments

Each soil was gently homogenized (not sieved) after coarse roots and rocks were removed, and replicate subsamples (1 g dry mass equivalent) were used for incubations. Two treatments were imposed with three replicates for each soil: amendment with 100 mg A. gerardii litter + natural abundance lignin, or with 100 mg A. gerardii litter + 13Cbeta-labeled lignin. This litter:soil ratio is representative of moderate litter inputs to surface soil (1 kg m-2). Samples were incubated in 50 ml centrifuge tubes placed inside 946-ml glass jars. Deionized water was added as needed to achieve field moisture capacity, determined for each soil by saturating and draining a subsample through filter paper. Jars were sealed with Viton gaskets and stainless steel lids with butyl rubber septa for headspace sampling/flushing. Jars were vented to the atmosphere with a needle and purged with humidified CO2-free air. Accumulated CO2 was periodically sampled using a syringe with needle and stopcock and jars were again purged with CO2-free air. Sampling intervals increased from weekly to biweekly to monthly to bimonthly over the two-year incubation (Fig. 2) to ensure sufficient CO2 accumulation for precise analysis. Concentrations of CO2 were always less than 5000 ppm, indicating that oxygen limitation did not occur (assuming an approximately equimolar ratio of headspace oxygen consumed to CO2 produced). We thus measured the entire cumulative production of CO2 from each sample. Mole fractions and delta13C values of CO2 were analyzed with a tunable diode laser (TGA200A, Campbell Scientific, Logan UT, USA; Hall et al., 2017). The long-term mean precision of delta13C was 0.33permille (1 SD) and the relative SD of CO2 mole fractions was 1.1%. Water was added gravimetrically at 2–4-month intervals to replenish vapor lost during headspace flushing.

Data analysis

We calculated CO2 production from bulk litter, lignin Cbeta, and SOC using isotope mixing models (Appendix 1: Table S1). We assessed temporal trends in C losses using generalized additive mixed models (GAMMs) fit to each C source in each soil, including an autoregressive error term to account for temporal autocorrelation, using the mgcv package (Wood 2017) version 1.8.28 in R version 3.6.0. Differences in total decomposition among C pools within a given soil were tested with ANOVA and Tukey’s HSD. We used vector analysis of lignin:litter and lignin:SOC decomposition ratios (Moorhead et al. 2016; Appendix 1: Figure S2) to test the tripartite relationships of lignin, litter, and SOC decomposition with predictor variables (Appendix 1: Table S1).