Soil Moisture Shapes How Temperature Drives Forest Soil Carbon Loss

In a temperate deciduous broadleaved forest in Gongju, South Korea, two years of continuous measurements show that the way soil respiration responds to rising temperature depends strongly on how wet the soil is. When soil moisture was relatively low, the expected increase in soil carbon dioxide efflux with warming was constrained, and a model based on temperature alone systematically overestimated the measured flux. The findings, published in Ecological Processes, suggest that soil respiration models built on temperature alone may misrepresent forest carbon losses under changing precipitation regimes.

Soil respiration, the combined release of carbon dioxide from root and microbial activity, is one of the largest carbon fluxes in forest ecosystems. It generally rises exponentially with soil temperature, and temperature-only exponential models have long served as a standard approach across diverse forests. Yet temperature-only models ignore moisture conditions and can overestimate or underestimate respiration under excessively dry or wet soils. The research team, led by Dongmin Seo and colleagues, set out to determine whether the temperature response of soil respiration becomes more dependent on soil moisture as soils dry within the observed range.

The study site is a preserved natural forest dominated by Quercus serrata, located at about 218 meters elevation and shaped by the Asian monsoon climate. Mean annual precipitation there is 1256.6 millimeters, more than 90 percent of it falling between April and October. This seasonality produces hot, humid summers and cold, dry winters, creating the kind of climatic variability the researchers needed to examine how moisture modifies the temperature response of respiration across a wide range of field conditions.

From January 2022 through December 2023, the team monitored soil respiration continuously with an automatic open and closed chamber system using five automated chambers distributed across the site. Each chamber ran a 30-minute cycle, closing for five minutes while air circulated through an infrared gas analyzer. Soil temperature was measured at 5 centimeters depth beside each chamber, and soil moisture content at 15 centimeters. Rainfall, photosynthetic photon flux density, and air temperature were recorded simultaneously, with quality-control criteria used to exclude periods affected by rainfall artifacts, condensation, sensor contamination, or instrument inspection.

Over the study period, soil temperature ranged from minus 2.3 to 27.5 degrees Celsius, air temperature from minus 18.2 to 32.1 degrees Celsius, and soil moisture from roughly 6.1 to 32 percent, rising sharply after rainfall events and gradually declining between them. Mean soil respiration across the entire period was 515.6 plus or minus 87.6 milligrams of carbon dioxide per square meter per hour, following a clear seasonal pattern that tracked soil and air temperature, increasing after February, peaking in summer, and declining afterward.

Structural equation modeling identified soil temperature and soil moisture as the only environmental variables directly associated with soil respiration in both years. The standardized path coefficient for soil temperature, at 0.78 to 0.82, exceeded that for soil moisture, at 0.32 to 0.35, confirming temperature as the stronger overall control. Rainfall and light were linked to respiration only indirectly, through their effects on temperature and moisture. Variance inflation factors of 1.037 in 2022 and 1.001 in 2023 indicated negligible multicollinearity between the two soil variables, and the correlations held when data were aggregated from hourly to daily means.

The clearest evidence for moisture dependence came from the residuals of the temperature-only model. These residuals were significantly related to soil moisture at the hourly scale, with a coefficient of determination of 0.34, strengthening to 0.60 at the daily scale. Under relatively low soil moisture, residuals tended to be more negative, meaning the temperature-only model overestimated observed respiration when soils were relatively dry. The temperature-only model achieved a slope of 0.87 and a coefficient of determination of 0.89 in 2022, but a slope of 0.76 and a coefficient of determination of 0.83 in 2023, with prediction error rising sharply in the second year, from a root mean square error of 137.71 to 303.92 milligrams of carbon dioxide per square meter per hour.

Models that included soil moisture performed better. The additive temperature-plus-moisture model reached a slope of 0.94 and a coefficient of determination of 0.92 in 2022 and 0.86 and 0.90 in 2023, while the interaction model, in which the temperature response itself varies with moisture, achieved a slope of 1.01 and a coefficient of determination of 0.92 in both years. The interaction model had the highest coefficient of determination and the lowest root mean square error, mean absolute error, Akaike information criterion, and Bayesian information criterion among the three models, and it produced fewer biologically implausible negative respiration predictions under cold, dry conditions than the additive model.

The apparent temperature sensitivity, expressed as Q10, the proportional increase in respiration for a 10-degree Celsius warming, varied nonlinearly across five soil moisture classes. Mean Q10 rose from 3.67 in the 5 to 10 percent moisture class to 3.93 at 10 to 15 percent, 4.50 at 15 to 20 percent, and 4.94 at 20 to 25 percent, then fell sharply to 2.32 at 25 to 30 percent. In 2022, Q10 peaked at 5.26 in the 20 to 25 percent class and dropped to 2.66 in the wettest class; in 2023 the corresponding values were 4.62 and 1.97. The authors caution that the wettest class contained few observations and a narrow temperature range, so its estimate should be interpreted carefully.

The researchers emphasize that the study was conducted at a single site, so the relationships may differ under other soil types, forest structures, and climates, and that the empirical models did not directly represent substrate availability, root biomass, microbial activity, or oxygen diffusion. The moisture classes also arose from naturally occurring field conditions rather than balanced sampling. Even so, the pattern was consistent across both years: increasing soil temperature did not translate into a uniform increase in soil carbon dioxide efflux across moisture conditions. The authors conclude that respiration models should explicitly account for moisture-dependent variation in the temperature response to better predict forest soil carbon efflux under future changes in temperature and precipitation.

Subject of Research: Moisture dependence of the temperature response of soil respiration in a temperate deciduous broadleaved forest

Article Title: Greater moisture dependence of the soil respiration response to temperature under relatively lower soil moisture conditions in a temperate deciduous broadleaved forest

Article References: Seo, D., Lee, M., Lee, J., & Lee, J. (2026). Greater moisture dependence of the soil respiration response to temperature under relatively lower soil moisture conditions in a temperate deciduous broadleaved forest. Ecological Processes, 15(1), Article 94. https://doi.org/10.1186/s13717-026-00752-y

Image Credits: AI Generated

DOI: 10.1186/s13717-026-00752-y

Keywords: soil respiration, soil moisture, soil temperature, Q10 temperature sensitivity, temperate deciduous forest, carbon cycling, structural equation modeling, forest carbon flux, monsoon climate, climate warming, soil CO2 efflux, South Korea