Partitioning of ecosystem respiration in winter wheat and silage maize—modeling seasonal temperature effects

• Varying temperature sensitivity improved models of ecosystem respiration components.
• Aboveground autotrophic respiration was the main component of ecosystem respiration.
• Labile soil carbon accounted for half of heterotrophic respiration in one season.
• Temperature sensitivity of different components has to be considered in modeling.

The response of agroecosystem carbon (C) respiration fluxes to environmental changes needs to be better understood as respiration subcomponents may respond differently to management and seasonal weather dynamics, which is important for soil organic matter (SOM) modeling. Respiration measurements at two different spatial and temporal scales (eddy covariance (EC) and soil chambers) were used to ascertain the relationship between temperature and CO2 flux of different ecosystem respiration components (ecosystem (Reco), soil and root combined, and soil). Further, different model approaches (static versus dynamic reference CO2 rate (rb) and activation energy type parameter (E0) with an Arrhenius-like function) in order to partition Reco into above- and belowground autotrophic (RA_above, RA_below) and heterotrophic respiration (RH_SOM) were tested. Canopy level CO2 fluxes in winter wheat and silage maize were measured by EC stations and soil surface CO2 flux by a handheld chamber analyzer in arable fields in Southwest Germany over a period of three growing seasons (2009, 2010, and 2012). Additionally, successive bare fallow plots were installed at the beginning of each growing season to partition soil respiration between autotrophic and heterotrophic sources (including “labile” soil C (newest bare fallow) as the difference to the oldest bare fallow). Stepwise model building was tested with keeping rb and E0 constant (static method) and then by varying rb and E0 each individually or together by time period (dynamic method) over the whole growing season (15, 10 or 7 days for Reco, measurement periods for soil chamber measurements). The dynamic models were superior as measured by Aaike Information Criteria (AIC) and coefficient of determination (average R2, 0.15 for the static model and 0.50 for the dynamic model). In the best fitting model for each crop-year (lowest AIC), rb was successfully estimated in each time period (relative standard error <50%), while seasonally variable E0 estimates were found in half of the crop years. Estimated Q10 values were between 1 to 6.01 between different components and seasons. Estimated Reco components during 2012, autotrophic above ground respiration accounted for the largest component during the intense measurement periods under both winter wheat (50%) and maize (60%), with root respiration accounting for 19% and 21%, respectively. Additionally under winter wheat 31% of Reco was estimated as heterotrophic respiration, with 15% from labile soil C. The results highlight the need to apply individual temperature response functions when using temperature as a driving force for ecosystem respiration components (autotrophic and soil heterotrophic).