How climate warming interacts with atmospheric nitrogen (N) deposition to affect carbon (C) release from soils remains largely elusive, posing a major challenge in projecting climate change‒terrestrial C feedback. As part of a five-year (2006–2010) field manipulative experiment, this study was designed to examine the effects of 24-hour continuous warming and N addition on soil respiration and explore the underlying mechanisms in a semi-arid grassland on the Mongolian Plateau, China. Across the five years and all plots, soil respiration was not changed under the continuous warming, but was decreased by 3.7% under the N addition. The suppression of soil respiration by N addition in the third year and later could be mainly due to the reductions in the forb-to-grass biomass ratios. Moreover, there were interactive effects between continuous warming and N addition on soil respiration. Continuous warming increased soil respiration by 5.8% in the ambient N plots, but reduced it by 6.3% in the enriched N plots. Soil respiration was unaffected by N addition in the ambient temperature plots yet decreased by 9.4% in the elevated temperature plots. Changes of soil moisture and the proportion of legume biomass in the community might be primarily responsible for the non-additive effects of continuous warming and N addition on soil respiration. This study provides empirical evidence for the positive climate warming‒soil C feedback in the ambient N condition. However, N deposition reverses the positive warming‒soil C feedback into a negative feedback, leading to decreased C loss from soils under a warming climate. Incorporating our findings into C-cycling models could reduce the uncertainties of model projections for land C sink and global C cycling under multifactorial global change scenarios.

2.1. Study site

The study site is located in a semi-arid grassland in Duolun County (42°02ʹ N, 116°17ʹ E, 1324 m a.s.l.), Inner Mongolia, China. Long-term (1953–2019) mean annual temperature and precipitation are 2.4°C and 382 mm, respectively (China Meteorological Data Sharing Service System). The sandy soil is classified as chestnut soils (Chinese classification) and Xeric Haplocalcids according to the US soil taxonomy classification system (Soil Survey Staff, 1999). Plant community at the study site is dominated by perennial species such as Agropyron cristatum, Artemisia frigida, Artemisia pubescens, Lespedeza davurica, and Stipa krylovii (Zheng et al., 2021).

2.2. Experimental design and manipulations

A randomized complete block design was used in this experiment with four treatments including a control (C; no warming and without N addition), 24-hour continuous warming (W), N addition (N), and continuous warming plus N addition (WN). There were six replicates (block) for each treatment (see also Xia et al., 2009). Twenty-four 3-m × 4-m plots were arranged in 4 × 6 matrix, with a 3-m buffer zone between any two adjacent plots.

The continuous warmed plots were heated continuously using MSR–2420 infrared radiators (Kalglo Electronics Inc., Bethlehem, PA, USA) suspended 2.25 m above the ground from 23 April to 15 November in 2006 and from 15 March to 15 November during 2007–2010. In the unwarmed plots, one ‘dummy’ heater with the same shape and size as infrared radiators was suspended 2.25 m high to mimic the shading effects of infrared radiators. The level of N addition was set at 10 g N m^-2 year^-1 as ecosystem responses to N enrichment in this area could reach the saturation point at a rate of 10.5 g N application m^-2 year^-1 (Bai et al., 2010). NH₄NO₃ was applied in the enriched N plots once on 19 July in each of the five experimental years from 2006 to 2010.

2.3. Soil microclimate and plant biomass measurements

In each growing season from May to October, soil temperature (°C) at 10-cm depth was measured three times per month using a portable temperature probe (LI-8100-201) attached to a LI-8100 Soil CO₂ Flux System (Li-Cor Inc., Lincoln, Nebraska, USA). Soil volumetric water content (% V/V) at 10-cm depth was measured weekly with a Diviner-2000 Portable Soil Moisture Probe (Sentek Pty Ltd, Balmain, Australia). Both the measurements of soil temperature and moisture were conducted adjacent to collars used for soil respiration measurements on clear and sunny days between 09:00 and 12:00 a.m. (local time).

2.4. Soil respiration measurements

Two 5-cm tall polyvinyl chloride collars (10 cm in diameter) were permanently inserted 3 cm into the soil at two opposite corners of each plot. A CO₂ flux chamber attached to the LI-8100 Soil CO₂ Flux System was put on each collar for 90 secs to measure soil respiration, and then moved to the next collar. To eliminate influences of plant respiration, the aboveground parts of living plants inside collars were cut at least one day prior to the measurement. The aboveground cut materials were left inside the collars. Soil respiration was measured three times per month on clear and sunny days between 09:00 and 12:00 a.m. (local time).