Xylem sap flow and environmental variables were measured on seven consecutive midsummer days in a 130-year-old Larix gmelinii (Rupr.) Rupr. forest located 160 km south of Yakutsk in eastern Siberia, Russia (61 degrees N, 128 degrees E, 300 m asl). The site received 20 mm of rainfall during the 4 days before measurements, and soil samples indicated that the trees were well watered. The tree canopy was sparse with a one-sided leaf area index of 1.5 and a tree density of 1760 ha(-1). On a clear day when air temperature ranged from 9 to 29 degrees C, and maximum air saturation deficit was 3.4 kPa, daily xylem sap flux (F) among 13 trees varied by an order of magnitude from 7 l day(-1) for subcanopy trees (representing 55% of trees in the forest) to 67 l day(-1) for emergent trees (representing 18% of trees in the forest). However, when based on xylem sap flux density (F'), calculated by dividing F by projected tree crown area (a surrogate for the occupied ground area), there was only a fourfold range in variability among the 13 trees, from 1.0 to 4.4 mm day(-1). The calculation of F' also eliminated systematic and large differences in F among emergent, canopy and subcanopy trees. Stand-level F', estimated by combining half-hourly linear relationships between F and stem cross-sectional area with tree size distribution data, ranged between 1.8 +/- 0.4 (standard deviation) and 2.3 +/- 0.6 mm day(-1). These stand-level F' values are about 0.6-0.7 mm day(-1) (30%) larger than daily tree canopy transpiration rates calculated from forest energy balance and understory evaporation measurements. Maximum total tree conductance for water vapor transfer (G(tmax), including canopy and aerodynamic conductances), calculated from the ratio of F' and the above-canopy air saturation deficit (D) for the eight trees with continuous data sets, was 9.9 +/- 2.8 mm s(-1). This is equivalent to a leaf-scale maximum stomatal conductance (g(smax)) of 6.1 mm s(-1), when expressed on a one-sided leaf area basis, which is comparable to the published porometer data for Larix. Diurnal variation in total tree conductance (G(t)) was related to changes in the above-canopy visible irradiance (Q) and D. A saturating upper-boundary function for the relationship between G(t) and Q was defined as G(t) = G(tmax)(Q/[Q + Q(50)]), where Q(50) = 164 +/- 85 micro mol m(-2) s(-1) when G(t) = G(tmax)/2. Accounting for Q by excluding data for Q < Q(85) when G(t) was at least 85% of G(tmax), the upper limit for the relationship between G(t) and D was determined based on the function G(t) = (a + blnD)(2), where a and b are regression coefficients. The relationship between G(t) and D was curvilinear, indicating that there was a proportional decrease in G(t) with increasing D such that F was relatively constant throughout much of the day, even when D ranged between about 2 and 4 kPa, which may be interpreted as an adaption of the species to its continental climate. However, at given values of Q and D, G(t) was generally higher in the morning than in the afternoon. The additional environmental constraints on G(t) imposed by leaf nitrogen nutrition and afternoon water stress are discussed.