Partitioning evapotranspiration in C3-C4 mixed tallgrass prairies: An inter-comparison of five methods and evaluation of seasonal parameterization

Accurate estimates of transpiration (T) remain particularly challenging in mixed C 3 -C 4 ecosystems, where the relative dominance of species and their physiological characteristics shifts dynamically throughout the growing season. Some methods for evapotranspiration (ET) partitioning, such as Flux Variance Similarity (FVS), typically require an explicit specification of either C 3 or C 4 photosynthetic pathways to parameterize intercellular carbon dioxide concentrations (c i ). Other methods, such as Modified Relaxed Eddy Accumulation (MREA), Conditional Eddy Covariance (CEC), Conditional Eddy Accumulation (CEA), and the FVS optimum approach, do not require prior c i approximations or designation of C 3 and C 4 vegetation. This study evaluated and inter-compared five ET partitioning methods: FVS, MREA, CEA, CEC, and a water-use efficiency-integrated modification of CEC (CECw) within two differently managed (grazing and hay harvest) C 3 -C 4 mixed tallgrass prairies in Central Oklahoma, USA. Specifically, we investigated how partitioning outputs varied when using a static, year-round C 4 parameterization compared to a seasonally dynamic C 3 -C 4 framework. Results indicated that T:ET ratios from FVS methods using constant c i values or c i /c a ratios were similar across seasonal and annual scales, regardless of whether a year-round C 4 or temporally dynamic C 3 -C 4 parameterization was applied. During peak growth, MREA and CEC produced the highest T:ET ratios (0.93–0.95), whereas FVS and CEA yielded more moderate values (<0.75), and CECw provided the lowest estimates (0.56–0.62). Consequently, MREA and CEC estimated T 10–20% higher than the biophysically reasonable estimates from FVS and CEA. Methodological discrepancies were most pronounced during periods when either stomatal or non-stomatal fluxes dominated, with closer agreement occurring when these exchanges were nearly equivalent. Validation against lysimeter measurements showed that FVS and CECw optimum approaches performed best (r = 0.84–0.85; RMSE = 0.23–0.25 mm d −1 ), while other methods deviated substantially. As the first intercomparison of five partitioning methods in mixed C 3 -C 4 tallgrass prairies, this study demonstrates that the choice of method significantly influences the resulting ecosystem water budget.