Site description: The experimental site (35º35'N, 109º?'38E) was located in the centre of Shaanxi Province. The climate of the area is semi-arid temperate continental. The mean annual temperature and precipitation of the study site were 13.28C and 550 mm respectively. Nearly 60% of the annual rainfall is distributed from June to September, coinciding with the hottest months of the year. Soil in the study area was uniform in texture down the profile (16% clay, 50% silt and 34% sand) and classified as Calciustepts based on the USDA Soil Taxonomy (Soil Survey Staff 1999). Wheat (Triticum aestivum), maize (Zea mays) and sweet potato (Ipomoea batatas) were the major crops grown at the study site. Soil was ploughed twice a year after harvest and before seeding. Animal (oxen and donkeys) and manual tillage are the prevalent tillage systems in this area. Among the traditional manual tilling methods, three types of hand tools are the most popular: hoeing for removing weeds over the growing season, mattocking for preparing the seedbed for soybean or autumn crops and spading for preparing the seedbed for winter wheat (Zhu 1989).
Experimental design: The experiment was set up on three short slopes, typical of the cultivated landscape on the Loess Plateau. The length (~7 m), gradient (10–15%), aspect and soil texture of the three slopes that served as replicates were similar. For each slope, four experimental plots (2 m 3 m) were allocated at both upslope (erosion (E) sites) and downslope (deposition (D) sites). Treatments assigned to the experimental plots included control (undisturbed; no soil removal or deposition) and three types of manual tillage practices (hand-hoeing, mattocking and spading) that removed upslope soil at different depths (2, 6 and 10 cm for hand-hoeing, mattocking and spading respectively). Each tool is traditionally designed to remove soil at a certain depth to meet a special purpose. Soil was subsequently deposited downslope. In all, 24 plots were established on the three slopes and there were three replicates for each treatment. The experiment was conducted from April to October 2013. The tillage operations were all performed by experienced local farmers. On the steep hillslopes of the Loess Plateau, local farmers rarely use organic fertilisers. Both shoots and roots of crops were removed from these experimental sites. Moreover, during the experiment, any new vegetation that appeared was also removed to prevent interference with the heterotrophic respiration in soil CO2 emission measurements.
Soil sampling and analysis: Core samples (100 cm3) were collected from each plot before plot establishment to determine soil properties, including soil moisture, bulk density and SOC content. During the experiment, soils were sampled once a month. At each plot, soil samples were collected at nine depths (0–5, 5–10, 10–15, 15–20, 20–25, 25–30, 30–35, 35–40 and 40–45 cm) using a metal 100-cm3 cylinder, followed by volumetric sampling at equal depths for determination of soil bulk density and gravimetric water content. Soil samples were air dried and ground to pass through a 0.15-mm sieve, which provides a more uniform particle size and reduces the variability in C values estimated using a high temperature combustion method (Cihacek and Jacobson 2007). Soil organic matter (SOM) was measured by dry combustion (Nelson and Sommers 1996) and SOM was converted to SOC by multiplying by a factor of 0.58 for topsoil (0–20 cm) and a factor of 0.40 for subsoil (20–45 cm; Mann 1986; Tabatabai 1996). Soil bulk density samples were obtained using the core method (Blake and Hartge 1986) with a metal cylinder (5 cm diameter, and 5 cm high).
Measurements of soil CO2 emission, temperature and moisture: During the experimental period, soil CO2 emission, soil temperature and soil moisture were monitored three times per month (20 observations in total), at both E and D sites. Before the measurement, a polyvinyl chloride (PVC) collar was installed 2–3 cm into the soil at each plot. Soil CO2 emission was measured using a Li-8100 infrared gas analyser equipped with an 8100-103 survey chamber (Li-Cor, Lincoln, NE, USA). The volume of the chamber was 4843 cm3. Details regarding the design and operation of the equipment are available elsewhere (https://www.Li-Cor.com/env/pdf/soil_flux/8100A_brochure.pdf, accessed 15 March 2017). Typically, soil CO2 emission rate was recorded three times at 0800, 1200 and 1700 hours (Beijing Standard Time). At each plot, individual measurements were conducted for approximately 3 min. Mean values for this period were given by the Li-Cor software. The daily mean soil CO2 emission rate for each plot was calculated by averaging the three measurements (H. Q. Yu, Y. Li, Y. Geng, S. Saggar, unpubl. data). Cumulative CO2 emission was calculated by interpolating the measurements from each day and then integrating. Soil temperature at 5 cm depth was monitored simultaneously with soil CO2 emission measurement by inserting three mercury thermometers into the soil close to the collars. In addition, soil volumetric water content was measured close to the collars for the 0–10 cm layer using a portable TDR probe (Campbell Scientific, Logan, UT, USA)