1 INTRODUCTION
About 50-75% of annual precipitation that falls in southern forests in the U.S. returns the atmosphere through the process of evapotranspiration (ET), the sum of tree transpiration, canopy interception, and soil evaporation (Sun et al., 2001; Sun et al., 2016). Natural (e.g. climate change and variability, drought) and anthropogenic stressors (e.g. land use change, urbanization), and forest management (e.g., thinning, prescribed burning) affect water quantity and water quality, and ecosystem productivity (Sun et al., 2011) by directly altering forest transpiration process, a key component of ET in forests (Domec et al., 2012). Water use by trees or stand is naturally variable due to the large differences in tree species, age, stand structure (leaf area index), and climate (Sun et al., 2011). Given the issues with landcover change, urbanization, and an increase in extreme events from climate change facing the Piedmont region in the U.S. (Wear and Greis, 2013), an improved understanding of tree transpiration at the species level within a forested watershed is needed. Such information is useful for developing foundational steps toward improving watershed-level estimates of transpiration in mixed forests and reliable ecohydrological model to predict the effects of environmental change on water and carbon resources (Liu et al., 2020; Li et al., 2020a, 2020b).
Land use change and fire regimes have produced highly variable species composition in the U.S. southeastern Piedmont. The region supports about 12.5 million hectares of forest land or 62% of the total land area, with 52% of forest lands covered by upland hardwoods and 34% by pines (Rummer and Hafer, 2014). The upland hardwood forest types consist of a mixture of white oak (Quercus alba ), red oak (Quercus rubra ), hickories (Carya spp. ), sweetgum (Liquidambar styraciflua ), and tulip poplar (Liriodendron tulipifera ). The oak-pine forest type is dominated by loblolly (Pinus taeda ), Virginia (Pinus virginiana ) and shortleaf (Pinus Echinata ) pines, and southern red oak (Quercus falcata ). Although many state and federal regulatory agencies require forests in the source water supply zone of a watershed be protected to slow runoff and maintain minimal discharge levels, 17% of total forest area could be lost by 2060 due to high urbanization and low timber prices with most of the losses occurring in the upland hardwood (Rummer and Hafer, 2014). Given the variability in species transpiration, the type of trees removed from the landscape will likely impact soil moisture, streamflow in headwater catchments, and downstream water supply (Moore et al., 2004; Swank and Vose, 1994).
Quantifying spatial variations in species-specific sap flux density (i.e., g H2O cm-2day-1) and daily tree-level transpiration (i.e., kg day-1 or liters day-1) is essential to improve stand-scaled transpiration estimates and total ET at the watershed level. Sap flux density and transpiration rates can vary widely by species (Boggs et al., 2015; Yi et al., 2017). For example, tree transpiration in a temperate pine-hardwood riparian buffer forest can range from 2 to 142 liters day-1 and can increase nonlinearly with increasing tree diameter (Bosch et al., 2014). Transpiration can vary across plant species due to tree xylem structure (Ford et al., 2011), responses to vapor pressure deficit (VPD) (Moore et al., 2017), and age (Brantley et al., 2019). Tulip poplar, a species with diffuse porous xylem and large amount of sapwood, can use up to three-fold more water than oaks (Quercus spp. ), a species with a narrow sapwood characterized by a ring porous structure (Ford et al., 2011).
Species level transpiration can also differ across topography or zones due to landscape level variations in microclimate and soil moisture (Emanuel et al., 2010; 2011). Hawthorne and Miniat (2016) found that transpiration per unit leaf area in hickory species was sensitive to the topographic position during a wet year, producing 56% less water at a cove site when compared to an upland site. However, chestnut oak (Quercus prinus ) decreased 41% from a wet year to a dry year but was not influenced by topographic position. Bosch et al. (2014) also found that species topographical position does not consistently influence tree transpiration rates across species.
At the stand level, linking species transpiration and soil moisture helps further refine watershed-level estimates of transpiration (Oishi et al., 2010) and improve our understanding of how forests respond to water stresses, including drought (Vose et al., 2016, Yi et al., 2017). Sap flux density and tree-level transpiration measurements have advanced dramatically during the last decade (Poyatos et al., 2020). However, questions are still being raised about how to improve watershed-level transpiration, given the variations in species composition, topography, and soil moisture. In addition, variability in tree transpiration across heterogeneous watersheds is typically not considered (or well captured) by models and may be one contributing factor to over or under predictions of water loss from the forest at local and regional scales. Therefore, the objectives of this study are to 1) quantify tree sap flux density and transpiration across three zones (riparian buffer, mid-hillslope, and upland-hillslope) in a wet and dry year, 2) analyze the relationships between sap flux density and vapor pressure deficit, and 3) compare watershed-level transpiration derived from three zones.