We continuously measured the transpiration of different tree differentiation classes and ecosystem evapotranspiration using the sap flow method and eddy covariance technique in a natural Larix gmelinii-dominated boreal forest located in the Northern Great Hinggan Mountains.The sapwood area of stem was used for scale expansion from single stand to forest.We examined variations in total transpiration of the forest (T_tot) and its three components,namely,the transpiration of dominant (T_d),intermediate (T_i),and suppressed (T_s) trees,and measured ecosystem evapotranspiration (ET).We also analyzed the regression correlations between T_tot or ET and vapor pressure deficit (VPD) and net radiation (R_n).The results showed that daily variation in sap flow velocity in different tree differentiation classes exhibited a unimodel pattern on both rainy and non-rainy days.T_tot on rainy days (24.7 mm) was lower than that on non-rainy days (50.6 mm),and T_d was higher than T_i and T_s on both rainy and non-rainy days.ET on rainy days (24.7 mm) was lower than that on non-rainy days (50.6 mm);however,the ratio of latent heat to R_n on rainy days (31%) was higher than that on non-rainy days (25.1%),indicating that environmental conditions on non-rainy days were more conducive to water vapor exchange over the interface between the atmosphere and plant stoma.The values of T_tot/ET,T_d/ET,T_i/ET,and T_s/ET on rainy days (38.1%,27.2%,8.5%,and 2.4%,respectively) were lower than those in non-rainy days (65.0%,45.5%,15.3%,and 4.2%,respectively).This finding demonstrates that evaporation represented the principal water loss in ET on rainy days,whereas on non-rainy days,ET was dominated by transpiration.This,however,would overestimate the forests'transpiration capacity when scaled up considering only dominant tree transpiration consumption.In general,it was found that the correlation between T_tot and environmental variables (i.e.,VPD and R_n) is better than that between ET and these variables,which indicates that T_tot is more sensitive to environmental variations.Furthermore,the correlation between R_n and water vapor fluxes (i.e.,T_tot and ET) is better than that between VPD and water vapor fluxes,indicating that R_n is the primary driver of an ecosystem's water vapor fluxes.