Trees allocate carbon belowground to fuel the functioning of roots and mycorrhizal fungi, which affect litter decomposition, but the direction and magnitude of this effect are variable. While tree mycorrhizal type is often suggested to mediate this aboveground-belowground linkage, previous studies yield mixed results. In this study we investigated how absorptive root traits, soil conditions and litter type influence the response of litter decomposition to altered belowground carbon allocation, both within and across mycorrhizal types. We girdled transport roots of seven subtropical tree species to eliminate carbon allocation to distal absorptive roots. We monitored leaf and root litter decomposition surrounding girdled and un-girdled root branches. We found that girdling generally slowed leaf litter decomposition. However, the effect of girdling on root litter decomposition depended on soil moisture, stimulating decomposition in dry soil but suppressing it in moist soil. Absorptive root traits did not influence the girdling effect on either leaf or root litter decomposition. These findings suggest that disturbance in carbon allocation can impact litter decomposition, with the outcome largely contingent on litter type and soil moisture.

Synthesis. Our findings highlight the importance of accounting for local soil variability in understanding the relationship between aboveground and belowground carbon dynamics. This study underscores a critical need for comprehensive assessment of belowground ecosystem responses to aboveground disturbances, as it is essential for accurately predicting future forest carbon and nutrient cycles.

We conducted our study in a subtropical forest at Tianmu Mountain (30°18′ N, 119°26′ E). We selected 5 AM tree species (Aphananthe aspera, Phoebe chekiangensis, Cunninghamia lanceolata, Cryptomeria japonica var. sinensis Miquel, and Torreya grandis) and 2 EM tree species (Cyclobalanopsis gilva, Pseudolarix amabilis) . Three individuals of each species were randomly selected, with about 100m elevational distance between two adjacent replicates. We used leaf and root litter homogenized from a variety of local tree species, including but not limited to the selected ones. Leaf litter was collected from litter traps (1 m × 1 m) near the selected trees from November 2020 to January 2021. Root material was excavated from soil monoliths (1 m × 1 m, 10 cm depth) near the litter traps. After field sampling, leaf and root litter were prepared in the laboratory. The twigs and seeds were excluded from the collected leaf litter. Leaf litter was air-dried at room temperature. The water content of the air-dried leaf litter was determined by oven-drying a subsample at 50°C for 48 hours. Roots were gently washed. Absorptive roots (1^st-3^rd order) were dissected and oven-dried at 50°C for 48 hours. Dried leaf and root litter were carefully homogenized. Leaf litter (5 g oven-dry weight equivalent) and root litter (1 g oven-dry weight) were placed in 10 × 10 cm litter bags. Litter bags were made by heat-sealing two pieces of nylon mesh (10 × 10 cm). The leaf litter bags had a nylon mesh aperture of 2 mm on both sides, while the root litter bags had a mesh aperture of 2 mm on top and 0.15 mm on the bottom to reduce matter from escaping the bags.

In late May 2021, we carefully excavated two root branches (including the 1st to 6th order roots) in the 0-10 cm soil near each target tree, one to be girdled and the other to serve as the control. We traced the roots to the base of the stem to confirm the tree. The paired root branches were of comparable size and were directly connected to a 7th-order root. We inserted each root branch into a nylon mesh root pocket (20 × 35 cm, 0.15 mm aperture) through a zipper edge. We then filled all root pockets with nearby sieved soil (2-mm mesh) to a thickness of 3 cm, and placed three pairs of root litter and leaf litter bags within each root pocket. Leaf litter bags were closed to the upper side of the root pocket, while root litter bags were in the bottom. We modified the insertion position and angle of the 6th order root to maximize the spatial occupancy of the distal absorptive roots within the root pocket. The root pockets were then zipped up, reburied at ca. 10 cm depth and covered with the original duff layer. We waited two months for the roots to get over initial disturbance before the girdling treatment. We conducted a separate, accompanying study to investigate the response of microbial decomposers to girdling. For this study, we installed two additional root pockets with each of the trees used in the decomposition study.

Two months after litter-bag incubation and prior to girdling, the first set of leaf and root litter bags were collected from both the decomposition and microbial studies. We opened the zipper of the root pocket and used long forceps to carefully remove the litter bags from the root pocket without disturbing the remaining litter bags. We placed all collected litter bags on ice in a cooler and transported them to the laboratory within a few hours for further processing. We then applied root girdling immediately after the litter bag harvest. We girdled the transport root (6th order) by removing a 1-cm ring of phloem close to the root pockets. Varying depth of root tissue was removed, depending on the phloem thickness of the target species. (Fig. S2). Four months after installation (i.e. two months after girdling), we harvested the second set of litter bags from the litter decomposition study only. Six months after installation (i.e. four months after girdling), we harvested all root pockets and litter bags.

For litter bags collected from the decomposition study, we carefully brushed and wiped off as much adhering soil as possible. Infrequent encroaching roots were removed. We then oven-dried the litter at 50°C for 48 hours and weighed it. We also quantified the nitrogen (N) concentration of the initial and harvested litter samples to investigate the litter N dynamics in response to girdling. The N concentration were determined with an elemental analyzer (FLASH 2000 CHNS/O Elemental Analyzer, Thermo Scientific, USA). The litter samples collected for microbial study were stored at -80°C for later processing.

At the time of root pocket installation, as well as after root pocket harvest, we collected soil samples in each root pocket to determine the soil properties. The soils were sieved (2-mm mesh) to remove coarse roots and large detritus. Soil pH was measured using a pH meter (Metter-FE28, Mettler Toledo, Greifensee, Switzerland) after suspending the soil in water (soil: water ratio was 1:2.5) and shaking for 2 minutes. Soil moisture was quantified by oven-drying the soil at 105 °C for 24 hours. The dried soil was ground and analyzed for total C and nitrogen (N) concentration with the FLASH 2000 elemental analyzer.

At the time of root pocket installation, we collected living absorptive roots from each target tree near the root pocket. We carefully washed the roots in water and scanned them on a desktop scanner at 300 DPI (Epson 12000XL, Epson America, Inc., California, USA). Scanned images were processed with WinRHIZO (Regent Instruments, Inc., Quebec, Canada) to determine the root diameter and total root length. The scanned roots were oven-dried (50 °C for 48 h) and weighed. Specific root length, the ratio between total root length and root dry weight, was calculated. Root tissue density was calculated as the root mass divided by turgid tissue volume determined by WinRHIZO. Dried root samples were ground and root tissue nitrogen concentration (N) was determined using the FLASH 2000 elemental analyzer. For the recovered root pockets, the 1^st-3^rd order roots were dissected from the inserted root branch and distinguished into living and dead roots, according to their color, morphological characteristics, tissue density and strength. All living and dead roots were washed, scanned and weighed. The morphological traits were determined only on living roots.

Using the litter bags collected 4 months after the girdling treatment, we compared the microbial biomass of saprotrophic fungi and bacteria with and without girdling. First, we collected approximately 0.5 g of fresh litter (well-mixed, with adherent soil) from each litter bag for DNA extraction. Each subsample was placed in a 15 ml centrifuge tube with 10 ml of nuclease-free water. We vortexed each subsample at top speed for 10 seconds and centrifuged it at 4000x g for 2 minutes to separate the litter from the adherent soil. The clean litter in the supernatant was collected. We extracted DNA from litter using the DNeasy PowerSoil Pro Kit (QIAGEN, Valencia, CA, USA) following the manufacturer's instructions. Quantitative polymerase chain reaction (qPCR) was used to quantify the abundance of fungi and bacteria (i.e., DNA copy numbers) in litter. Mass-specific abundance of bacteria and saprotrophic fungi in litter was calculated as the corresponding qPCR copy numbers divided by the litter dry weight. To calculate the abundance of saprotrophic fungi, we sequenced the fungal community and estimated the saprotrophic proportion. We used primer sets ITS1F-ITS2R to amplify the ITS region of the fungal communities. Successful PCR products were sequenced on the Illumina Novaseq platform (2 x 250bp). Detailed protocols for the PCR are provided in supplementary methods. The fungal amplicon sequences were analyzed using the EasyAmplicon pipeline . The pipeline first merged and filtered pair-end reads using the VSEARCH algorithm . Next, the amplicons were denoised into amplicon sequence variants (ASVs) in de novo mode. The UNITE database (Abarenkov et al., 2010) were used as reference databases for fungal taxonomy assignment, with a 0.5 confidence threshold. Finally, the FungalTraits database was used to determine the primary lifestyle of the fungal ASVs. The abundance of saprotrophic fungi was calculated as the total fungal abundance determined by qPCR multiplied by the saprotrophic proportion of the fungal community based on the FungalTraits database. 

We calculated the simple exponential decay rate constants for both leaf and root litter:

ln (M_t/M₀) = -kt 

where k is the decomposition constant (yr^-1), and M₀ is the initial dry weight of the sample. We then calculated ∆k of leaf and root litter as the difference between k in the girdled (k_girdled) and un-girdled (k_un-girdled) root pockets of the same tree:

∆k = k_girdled - k_un-girdled