2.5 Sap flux estimates
One to five of each overstory tree species (loblolly pine, oak spp., tulip poplar, sweetgum, Virginia pine, and red maple) were instrumented with a heat dissipation sensor to measure sap flux in the riparian buffer, mid-hillslope, or upland-hillslope zones. In total, there were 68 monitored trees across the watershed. Each 20 mm sap flux sensor was installed 1.4 meters above the ground surface (i.e., at dbh) on the north face to avoid direct sun influence. Sensor signals were converted from temperature difference to tree sap flux density (J s, g cm-2day-1, g sap cm-2 sapwood area day-1) for each overstory species according to the computation method outlined in Granier (1987).
American beech, hickory spp, black cherry (Prunus serotina ), scarlet oak (Quercus coccinea ), blackgum (Nyssa sylvatica ), black oak (Quercus velutina ), sourwood (Oxydendrum arboretum ), and shortleaf pine (Pinus echinata ) accounted for only 8% of co-dominant canopy species and were not monitored at any of the three zones for transpiration due to limited field resources and long distance between trees. Red maple and tulip poplar were also not monitored at the upland-hillslope for the same reasons. Mean species-specific transpiration (Ts , liters day-1) for these non-monitored trees was quantified based on a linear model developed from diameter at breast height (dbh, cm) and Ts from monitored trees:
For 2015, Ts = 3.1*dbh – 34, r2 = 0.76, p < 0.001 (Eq. 1)
For 2016, Ts = 3.8*dbh – 48, r2 = 0.70, p < 0.001 (Eq. 2)
Ts was then converted from liters day-1 to mm day-1 (equivalent to liters m-2 day-1) by dividing the total sap flow volume of the survey plot trees by plot area (78.5 m2).
Zone watershed-level transpiration (T z, mm day-1) was estimated from tree-scaledJ s for all monitored trees, estimates ofT s from the non-monitored trees (equation 1), stand density (tree counts), and species composition and sapwood area. The same species sapwood area and tree counts were used for each zone to scale J s to T z to avoid over- or under-emphasizing watershed sapwood area for each zone. Weighted area percentages for each zone was estimated based on landscape units that included percent slope, soil type, and unsaturated zone depth (Dreps et al., 2014) and field observations of species across the watershed (Boggs et al., 2015). The weighted area average (i.e., 10% buffer, 45% mid-hillslope, and 45% upland-hillslope) ofT z from each zone was then used to compute the average riparian buffer, mid-hillslope, and upland-hillslopeT w.
Measurements of tree J s in the buffer have been continuous since 2010 as part of other studies (Boggs et al., 2015; 2016). However, monitoring in mid-hillslope and upland-hillslope did not begin until May 2015. Despite these two zones having only two months (November – December 2015) of non-growing seasonJ s data in 2015, average daily non-growing seasonT w based on J s from this limited dataset is representative of the longer-term non-growing record for Piedmont hardwoods (~ 0.25 mm day-1, Oishi et al., 2010). Therefore, we estimated the other four months (January – April 2015) of missing daily non-growing season T w and filled other data gaps based on averaged daily T w that was generated during the monitoring periods.
To correct for changes in sap flux density throughout the sapwood area, (as these changes can influence the relative comparisons of species sap flux density), we used the predictions of the radial profile of sap flux density with depth for each wood type from the gamma-type model in Berdanier et al., 2016. Based to their model, we developed correction factors for each species and tree anatomy/type. After the correction factor was applied to sap flux density, annual stand transpiration was computed. The trees in our study were smaller than the ones in the Berdanier et al., 2016 which means that the sap flux density corrections were likely not going to be large, and would become even smaller when corrected for the area represented by the outer 20 mm probes in relation to the other sections of the tree. The corrected transpiration values (sap flux density*sapwood area*correction factor) are reported in this manuscript.