Using temporal patterns in vapor pressure deficit to explain spatial autocorrelation dynamics in tree transpiration

We measured 120 trees with constant-heat sap flux sensors in a subalpine forest in southern Wyoming, USA to quantify the relationship between temporal and spatial variation in tree transpiration. The forest stand was located along a soil moisture gradient from a stream side to near the top of a ridge. The entire stand was dominated by lodgepole pine ( Pinus contorta Dougl.) with Engelmann spruce (Picea engelmannii Parry) and subalpine fir (Abies lasiocarpa (Hook.) Nut.) present near the stream and scattered individuals of trembling aspen (Populus tremuloides Michx.) throughout the stand. We utilized a cyclic sampling design in space that maximized the number of point pairs at each spatial lag for semivariogram analyses. All four tree species exhibited previously established responses to environmental variables in which the dominant driver was a saturating response to vapor pressure deficit (D). This response to D is predictable from tree hydraulic theory in which stomatal conductance declines as D increases to prevent excessive cavitation. The degree to which stomatal conductance declines with D is dependent on both species and individual tree physiology and increases the variability in transpiration as D increases. We quantified this variability spatially by calculating the spatial autocorrelation within 0.2 kPa D bins. Across 11 bins of D, spatial autocorrelation in individual tree transpiration was inversely correlated to D and dropped from 45 to 20 m. Spatial autocorrelation was much less in transpiration per unit leaf area and not significant in transpiration per unit sapwood area suggesting that spatial autocorrelation within a particular D bin could be explained by tree size. Future research should focus on the mechanisms behind tree size spatial variability, and the potentially broad applicability of the inverse relationship between D and spatial autocorrelation in tree transpiration.