Long-term collar deployment leads to bias in soil respiration measurements

Accurate measurements of soil respiration (R-s) are critical for understanding how soil carbon will respond to environmental changes. However, a commonly used method for R-s measurements, the collar deployment method, may introduce artefacts that cause bias in R-s measurements. Our objective was to quantify the effect of long-term collar deployment on R-s and to unravel potential causes due to changes in the soil environment. A field experiment (2017-2019) including short-term (2-3 days before the measurement) and long-term collar deployment (lasting three consecutive growing seasons) was conducted to assess the methodological effect on R-s in an alpine grassland of the northeastern Tibetan Plateau. Soil incubation was used to further explore the mechanisms underlying the effects of collar deployment. The effect of long-term collar deployment on R-s varied over time. In the first one and a half growing seasons, no significant difference in R-s was noted under short- and long-term collar deployment. This may be attributed to the negative effects of lower root biomass inside long-term collars and the positive effects of higher temperature and pulse input of dead roots following collar deployment. Under the long-term collar, R-s decreased rapidly in the middle of the second growing season and remained low until the end of the experiment, resulting in an 18.2% decrease relative to short-term collar deployment in the third growing season. Higher soil bulk density and lower root and microbial biomass inside long-term collars may explain the decrease in R-s and temperature sensitivity (Q(10)). Soil incubation experiments revealed that the soil organic carbon (SOC) decomposition rate and Q(10) were significantly reduced after long-term collar deployment. Long-term collars led to substantial underestimates of R-s after more than 2 years. Our findings suggest that such potential artefacts should be considered when interpreting R-s data based on long-term collar deployment. Long-term collars should be relocated every 1-2 years to avoid artefacts if feasible. Alternatively, periodic measurements using short-term collars are recommended to quantify the magnitude of collar artefacts.