A meta-analysis reveals increases in soil organic carbon following the restoration and recovery of croplands in Southwest China
Data files
Dec 04, 2023 version files 21.04 KB
-
README.md
7.41 KB
-
SOCc_data.csv
6.16 KB
-
SOCs_data.csv
7.47 KB
Abstract
In China, the Grain for Green Program (GGP) is an ambitious project to convert croplands into natural vegetation, but exactly how changes in vegetation translate into changes in soil organic carbon remains less clear. Here we conducted a meta-analysis using 734 observations to explore the effects of land recovery on the soil organic carbon and nutrients in 4 provinces in Southwest China. Following GGP, the soil organic carbon content (SOCc) and soil organic carbon storage (SOCs) increased by 33.73% and 22.39%, respectively. Likewise, soil nitrogen increased, while phosphorus decreased. Outcomes were heterogeneous, however, depending on variation in soil and environmental characteristics. Both the regional land use and cover change indicated by landscape type transfer matrix and net primary production from 2000 to 2020 further confirmed that GGP promoted the forest area (2.95%) and regional mean net primary production (52.94%). Our findings suggest that GGP could enhance soil and vegetation carbon sequestration in Southwest China and help to develop carbon neutral strategy.
README
Title of Dataset:
A meta-analysis reveals increases in soil organic carbon following the restoration and recovery of croplands in Southwest China
https://doi.org/10.5061/dryad.51c59zwfs
Here, we collected data by searching published articles in the Web of Science from https://www.webofscience.com/, and supplemented this with a more specific search on Chinese dissertations and other Chinese-language studies in the China National Knowledge Infrastructure Database from https://www.cnki.net/. Studies published up to 23 June 2022 that contained the following keywords phrases (themes), were kept: ‘Southwest China’, ‘landscape type transfer matrix’, ‘vegetation net primary production’, ‘Grain for Green Program’, ‘soil organic carbon’, ‘afforestation’, ‘plantation on cropland’, ‘total nitrogen’, ‘total phosphorus’, ‘available nitrogen’ and ‘available phosphorus’. In order to ensure the integrity and precision of the investigation, we also searched the references of the papers to obtain more comprehensive relevant data.
The study area located in 4 provinces in SouthwestChina and a total of 57 original research articles meeting the inclusion criteria were identified in this paper. Research data such as soil physicochemical properties were directly collected from the text and tables in the published studies and were extracted from the literature charts using GetData (V.2.25). In total, we extracted 734 observations from the studies to calculate the effect size of the GGP on SOCc, SOCs and other properties (please see Appendix S1, SOCc_data.csv and SOCs_data.csv).
Description of the data and file structure
Articles data selected for this meta-analysis had to meet the following criteria: (a) the main requirements for studies were that they must include field experiments in China, and the experiments had to include at least three replicates. When the same experiment was reported in multiple publications (e.g., a scientific article and dissertation from the same author), only one dataset was retained. Units of measurement were adjusted to be the same across experiments, only univariate analysis data were selected; (b) the study indicated the soil nutrient changes in ecosystem types, soil depth, experimental duration, soil texture, and slope direction; (c) studies up to 10 years in duration were selected; (d) SOCc or SOCs were reported since both are important indicators, and the changes in both SOCc and SOCs were from control (current adjacent croplands) and treatment and (e) standard deviation (SD), publication biases and sample size were reported or could be calculated from the data presented in the publications.
We used the Landscape Type transfer Matrix for Southwest China from 2000-2020 (unit: km2) to describe the transform of landscape type in our paper (please see Appendix S2 Table S1) and raster numbers to represent changes in NPP in 2000 and 2020 (please see Appendix S2 Table S2). We also listed Effects of Grain for Green Program on pH (Appendix S2 Figure S1a), total nitrogen (TN, Appendix S2 Figure S1b), available nitrogen (AN, Appendix S2 Figure S1c), total phosphorus (TP, Appendix S2 Figure S1d), available phosphorus (AP, Appendix S2 Figure S1e), available potassium (AK, Appendix S2 Figure S1f), total potassium (TK, Appendix S2 Figure S1g) in Southwest China. Error bars are the 95% confidence intervals (CIs) of the weighted response ratios. The dashed line indicates points where the weighted response ratio equals 0. If any 95% CIs did not overlap with zero, the effect was considered significant. The numbers in parentheses represent the sample sizes of each observation. , , and ** indicate P < 0.05, < 0.01, and < 0.001, respectively. Regression models of the response ratio of soil total nitrogen content (TN) and soil total nitrogen stock (TNs) to latitude (a, d), elevation (b), mean annual precipitation (MAP, c), and longitude (e) were listed in Appendix S2 Figure S2. The relative importance of each variable in the responses of total nitrogen (TN, a), total phosphorus (TP, b), total potassium (TK, c), available nitrogen (AN, d), available phosphorus (AP, e) and available potassium (AK, f) experimental duration, soil depth, ecosystem types, soil texture and slope direction of transform caused by Grain for Green Program (GGP) were listed in (Appendix S2 Figure S3), the importance values were calculated using a model selected with corrected Akaike’s information criteria, the cutoff was set to 0.8 to differentiate between important and non-significant variables. In Appendix S2 Figure S4, we described the net primary productivity (NPP) distribution of Yunnan, Guizhou, Sichuan, and Chongqing in 2000 (Figure S4a) and 2020 (Figure S4b), different composition and flow direction represent different NPP value percentage from each city. The Technical flowchart of China Land Cover Data (CLCD) was also listed (please see Appendix S2 Figure S5), we used 30m annual land cover and its dynamics in China from 1990 to 2019, (Earth System Science Data, 2021) and the data set has the land use classification with 30m × 30m and has been updated to 1985 - 2020. Furthermore, to consider the publication biases, we used funnel plots of published offsets in Southwest China, showing the effect of main elements before and after GGP, including SOC content (SOCc, a), SOC stock (SOCs, b), total nitrogen (TN, c), total phosphorus (TP, d), total potassium (TK, e), available nitrogen (AN, f), available phosphorus (AP, g) and available potassium (AK, h). The Egger method was used to perform sensitivity analysis and test for publication bias on the funnel plot, and the stability of the results was evaluated. In order to reduce the existence of publication bias, the estimated value of combined effect is obtained by using the shearing method and compared with the original effect size (please see Appendix S3 Figure S1).
Code/Software
Research data such as soil physicochemical properties were directly collected from the text and tables in the published studies and were extracted from the literature charts using GetData (V.2.25).
All the statistical analyses were performed with R from https://r-project.org/. The ‘Metafor’ package was used to conduct mixed-effects meta-analysis. The ‘glmulti’ package was used to calculate the relative importance of all predictors after GGP effects and a cutoff of 0.8 was set to differentiate between important and nonsignificant predictors. The ‘ggplot’ package was used to draw figures. The ‘stats’ package was used for linear regression analysis and set model significance at P < 0.05.
We obtained the NPP data from the annual China Land Cover Data set (CLCD) from Landsat. Land use dataset and classification method from 2000 to 2020 were obtained from Resources and Environment Sciences and Data Center, Chinese Academy of Sciences from http://www.resdc.cn/, with the resolution was 30m × 30m. The annual land cover data set of China from Landsat was produced on the Google Earth Engine (GEE) platform in this paper.
We used the ArcGIS 10.8 software Geostatistical analyst method, input the corresponding longitude and latitude value, SOCc and SOCs values of the research sites and then performed Kriging interpolation to predict the overall content of SOCc and SOCs in the Southwest region in 2000 and 2020.