2.1 Study site description
This study was conducted at the Ailaoshan Station for Subtropical Forest Ecosystem Studies (24°32'N, 101°01'E; 2480 m above sea level), which is located in Jingdong County, Yunnan Province. It is an old-growth subtropical evergreen broad-leaved forest that has been deeply studied (Tang et al. 2007). It is characterized by a subtropical mountain climate with an annual average temperature of 11.3 oC and annual precipitation of 1840 mm (Wu et al., 2014). There is an obvious rainy season from May to October every year. The forest soil is rich in nitrogen and acid (pH < 5) (Chan et al. 2006). The main tree species of the forest are Machilus gamblei, Castanopsis wattii, and Hartia sinensis and the litterfall is 864 g m− 2 year− 1 (Wu et al. 2014).
2.2 Experimental design and setup
In this study, we selected 12 subplots (1×1 m) in a 30×40 m experimental plot. We conducted two factors with two levels of experiments: control (CK), warming (W), nitrogen addition (N), and warming together with nitrogen addition (WN). Each treatment had three repeats and contained treatment and recovery periods. Three lamps with a length of 90 cm were set up in the warming area for warming treatments. The lampshade was 100 cm and 1.2 m from the ground. The three lights form a triangle tilted and fixed at a 30o angle to the bottom to collect heat. We added NH4NO3 solution at 50 kg N ha− 1 yr− 1 for nitrogen treatments, divided into 12 months per year. Therefore, we weighed 1.19 g NH4NO3, dissolved it in 500 ml ultrapure water, and sprayed it evenly into the subplots once a month. We also spread the same volume of ultrapure water for the subplots with no nitrogen addition.
We collected fresh litter leaves from the subtropical forest during the dry season from October to December 2017, air-dried them, and dried them in an oven at 60 oC for 72 hours to constant weight. We prepared 12 litter bags (2 mm mesh, size 10 × 15 cm) for each subplot with an initial mass (Mi) of approximately 4 g. After removing the visible litter, we placed the litter bags on the soil surface 10 cm apart on December 31, 2017. Here we intend to carry out the decomposition experiment for four years. Therefore, the decomposed litter samples were collected three times per year. Litter decomposition during the treatment period was terminated for many reasons, and a recovery experiment was conducted instead. We prepared 6 litter bags (2 mm mesh, size 15 × 30 cm) for each subplot with approximately 14 g initial mass for the recovery period (fresh litter leaves were collected from early January to mid-March 2021). The recovery experiment started on March 31, 2021, and lasted for two years.
2.3 Data collection and analysis
As mentioned above, we collected litter bags every four months for two years for a total of 6 times in the treatment and recovery period. After each collection of litter bags, we cleaned, washed, and dried them at 60 oC for 72 hours to constant weight to obtain the litter dry weight (Mf).
We calculated the decomposition coefficient (k value) during the study period using the following decay model (Olson, 1963).
$$RR=\frac{{M}_{f}}{{M}_{i}} \left(1\right)$$
$$RR={e}^{-kt} \left(2\right)$$
where RR is the residual rate, and t (yr) is the corresponding decomposition time.
We divided litter decomposition into six stages and calculated the stage decomposition rate (SDR, %/month).
$${SDR}_{s}=\frac{{RR}_{s-1}-{RR}_{s}}{{RR}_{s-1}} ÷{T}_{stage}÷12\times 100\text{\%} \left(3\right)$$
where is the stage number from 1 to 6, and Tstage is each stage time (yr).
Every month, we measured the soil temperature (oC) and soil water content at a 5-cm depth using a digital thermometer (6310; Spectrum, Illinois, USA) and time domain reflectometry (MP-KIT; Beijing Channel, Beijing, China) three times from January 2017 to March 2023. We divided them into three periods: before treatment (2017.01-2017.12, background of each subplot), treatment period (2018.01-2019.12, treatment effect), and recovery period (2021.04-2023.03). The averages of each period were used to test the treatment effect, taking the before-treatment period as the background value and calculating the ratio value of the treatment and recovery period. We also divided the soil temperature and water content into six stages (SST and SSW) corresponding to the SDR in the treatment and recovery periods.
One-way ANOVA and Duncan’s test were used to test the difference in each period’s soil temperature and water content, ratio values, and k values among treatments. Two-way ANCOVA was used to test each period’s k values and SDR difference. Warming and nitrogen addition treatments were the fixed factors, and soil water content was the covariate for k values. SST, SSW, and initial decomposed degree (IDD) were the covariates for SDR. All the data passed the normality and homoscedasticity test. The effects of warming, nitrogen addition, IDD, SST, and SSW on the SDR were analyzed by variance partitioning analysis (VPA). The variance inflation factor (VIF) test was conducted, and all VIF values were less than 4. All statistical charts were generated using R software (version 4.3.0) and R Studio with the readxl, ggplot2, lubricate, ggpubr, car, gtsummary, agricolae, and ggbeeswarm packages.