TL;DR
Winter respiration (RECO) is a crucial but understudied component of the global carbon cycle. This paper uses eddy covariance data from 57 northern hemisphere sites to reveal that while temperature and litter availability are key drivers of spatial RECO variation, temporal and spatial temperature sensitivities differ significantly. This suggests that models projecting future warming impacts on winter CO2 fluxes might overestimate effects if they rely solely on spatial relationships.
What Problem Does This Paper Address?
The carbon cycle of northern ecosystems is heavily influenced by winter CO2 fluxes, yet these processes are poorly understood compared to growing season activity. Accurately quantifying winter respiration (RECO) is vital for assessing the overall terrestrial carbon budget and predicting climate feedbacks. Previous studies have often focused on specific sites, making it difficult to generalize findings or understand how factors like temperature and vegetation productivity influence RECO across diverse northern ecosystems. This paper aims to fill this knowledge gap by analyzing a large dataset to identify the key drivers of spatial and temporal RECO patterns and to understand how temperature sensitivity varies between spatial and temporal scales.
Methods
The study utilized eddy covariance flux data from the La Thuile FLUXNET synthesis database, encompassing 57 sites across the northern hemisphere. Two definitions of "winter" were employed: D1, based on a period of sustained sub-zero air temperatures, and D2, a fixed 90-day period. Key methods included:
- Flux Partitioning: Net ecosystem exchange (NEE) was partitioned into gross primary productivity (GPP) and ecosystem respiration (RECO) using a flux-partitioning algorithm that leverages short-term temperature sensitivity.
- Arrhenius Equation: The temperature dependency of daily winter RECO was analyzed using the Arrhenius equation, allowing for the estimation of site-specific parameters like reference respiration rate ($R_{ref}$) and activation energy ($E_0$).
- Spatial and Temporal Analysis: The study compared the relationships between RECO and environmental variables (temperature, leaf area index) across sites (spatial) versus over time at individual sites (temporal/inter-annual).
- Cross-validation: Regression models were validated using a cross-validation approach to assess their predictive power.
Datasets
The primary dataset used was the La Thuile FLUXNET synthesis database, which provided eddy covariance flux data (NEE, GPP, RECO), air temperature, upper soil temperature, and precipitation from 965 site-years.
Additionally, MODIS-Aqua Leaf Area Index (LAI) data (MYD15A2) were used to derive the seasonal amplitude of LAI ($\Delta$LAI), serving as a proxy for substrate availability.
Key Insights and Ideas
- Ecosystem Type Matters: Deciduous broadleaf forests showed the highest winter RECO rates, while wetlands had the lowest. This highlights the importance of ecosystem structure in winter carbon cycling.
- Temperature and Litter are Key Spatial Drivers: Both winter air/soil temperature and the seasonal amplitude of leaf area index ($\Delta$LAI) were identified as the primary factors explaining spatial variation in winter RECO. Together, they explained a significant portion (69-71%) of this variation.
- Spatial vs. Temporal Temperature Sensitivity: A critical finding is that the temperature sensitivity of winter RECO observed across different sites (spatial) is significantly higher than that observed over time at individual sites (temporal/inter-annual). This suggests that extrapolating spatial relationships to predict future warming impacts might overestimate CO2 losses.
- Soil Temperature Dominates Inter-annual Variability: Soil temperature, rather than air temperature, better explained inter-annual variations in winter RECO, likely due to the insulating effect of snow cover.
- Potential Warming Dampening Effect: The study found that the activation energy ($E_0$) for winter respiration decreases with increasing soil temperature. This implies that as the climate warms, the temperature sensitivity of winter respiration might decrease, potentially dampening the overall increase in CO2 release.
Key Results
- Winter RECO Rates: Deciduous broadleaf forests exhibited the highest winter RECO rates (0.97–1.05 g C m⁻² d⁻¹), while wetlands had the lowest (0.21–0.23 g C m⁻² d⁻¹).
- Spatial Variance Explained: Temperature (air or soil) and $\Delta$LAI together explained 71% (Tair, $\Delta$LAI) or 69% (Tsoil, $\Delta$LAI) of the spatial variance in daily mean winter RECO.
- Temperature Sensitivity Difference: A 1°C warming was estimated to increase winter CO2 loss by 0.12–0.14 g C m⁻² d⁻¹ based on spatial relationships, compared to only 0.07–0.09 g C m⁻² d⁻¹ based on temporal (inter-annual) relationships.
- Model Fit: Regression models incorporating temperature and $\Delta$LAI explained 67% and 62% of winter RECO variance, respectively, in cross-validation.
Limitations and Future Work
- Winter Definition: The study acknowledges that the choice of winter definition (D1 vs. D2) can influence results, particularly for high-latitude sites with shorter winter durations.
- Snow Cover Complexity: While snow cover's insulating effect is highlighted, the study notes that detailed snow data (accumulation, density, structure) are needed for a more comprehensive understanding of soil thermal regimes.
- Model Overestimation: The significant difference between spatial and temporal temperature sensitivities suggests that models relying solely on spatial relationships may overestimate the impact of future warming on winter RECO.
- Future Research: The authors emphasize the need for more in-situ snow data and further exploration of interactions between soil carbon, soil temperature, moisture, and snow cover effects. They also suggest that future research should consider the distinct temporal and spatial temperature sensitivities when developing models for climate change projections. The inclusion of more data from cold regions, particularly permafrost areas, is also recommended for robust analysis.
Summary updated