COS 193-3 - Snowpack-mediated temperature responses of winter ecosystem respiration are linked to differences in plant functional type across snow-covered forests.

Background/Question/Methods

Winter ecosystem respiration (R_eco) in seasonally snow-covered forests constitutes a significant proportion of the carbon (C) that is fixed annually. For these ecosystems, snowpack dynamics (i.e., snowpack depth and timing of snow arrival/melt) play a critical role in the C cycle by regulating belowground temperature as well as supplying water for plant and microbial demand. As warming trends continue, the associated declines in snowfall amount and snowpack depth could have enormous consequences for respiratory C losses via climate-C feedbacks. In this study, we analyzed the temperature response (Q₁₀) of winter ecosystem respiration using flux tower data across 49 sites across the FLUXNET2015 and AmeriFlux networks. Here we hypothesized that R_eco (which includes above- and belowground sources) would exhibit differential responses to either air or soil temperature, owing to direct snowpack effects that disrupt the exchange of energy between the soil and sky above. We also hypothesized that evergreen needleleaf forests (ENF) would exhibit higher sensitivity to air temperature fluctuation than deciduous broadleaf forests (DBF) due to maintenance cost of evergreen leaves that are exposed to the air.

Results/Conclusions

Here, we found that the location of temperature measurements as a model predictor (i.e., air vs. soil) can lead to dramatically different interpretations of winter temperature sensitivity, which lends support to our first hypothesis related to energy decoupling by the snowpack. When using air temperature to predict nighttime NEE (equivalent to R_eco), the derived Q₁₀ values were surprisingly conserved between different forest types (air T models: Q_10,ENF = 1.8 ± 0.1, Q_10,DBF = 1.7 ± 0.1). In contrast, the soil temperature models produced Q₁₀values that were 2 – 3.7x higher on average, with evergreen forests exhibiting the highest overall temperature sensitivity (soil T models: Q_10,ENF = 6.5 ± 1.1, Q_10,DBF = 3.6 ± 0.3) (p = 0.056 for forest ´ T-source interaction;p < 0.001 for T-source main effect). These results suggest that while air temperature can be a reliable predictor of winter C fluxes across different forest types, the independent effects of soil and air temperature on whole ecosystem C balance must also be regarded carefully, especially in the context of a declining snowpack and future shifts in plant community composition.