Book Description

Agricultural land use impacts groundwater and surface water quality and many research programs have aimed to understand these impacts. For example, loading of agriculturally derived solutes such as nitrate can create hypoxic conditions in stream estuaries downstream of agricultural and human development. This problem exists in the Chesapeake Bay watershed, a 11,600 km2 watershed which encompasses six states: Delaware, Maryland, New York, Pennsylvania, Virginia, and West Virginia. The largest contributor of water and nutrients to the Chesapeake Bay is the Susquehanna River Basin. It is important to understand the transport dynamics within smaller tributary watersheds of the Susquehanna River, so that efforts can be made to mitigate the effects of farming on nutrient pollution in the Chesapeake Bay.The Susquehanna Shale Hills Critical Zone Observatory (SSHCZO) has been established to understand water quality in the context of natural geological background conditions within the critical zone. The SSHCZO is an observatory in the uplands of the Susquehanna River Basin. The SSHCZO spans the Shavers Creek watershed (120 km2), a HUC 10 watershed. This creek is a tributary of the Juniata River, which drains into the Susquehanna River. The SSHCZO contains three instrumented subcatchments: Shale Hills (0.08 km2), Garner Run (1.21 km2), and Cole Farm (0.34 km2). These three sites provide land use and lithology end members that serve as proxies for the rest of Shavers Creek watershed. In addition to using small catchment end-members to understand transport in Shavers Creek, three separate synoptic sampling campaigns were completed at high spatial resolution throughout the watershed. These three campaigns show how changes in hydrologic connectivity, land use, and lithology affect water quality in the watershed. Fall and winter synoptic campaigns highlight discrepancies in water and solute influxes between expected solute loads from tributary inputs to the mainstem and the measured solute loads at mainstem sites. These discrepancies suggest input from groundwater or runoff sources other than sampled tributaries. For example, both nitrate and chloride are both input from sources other than tributaries. Principal component analyses and a mixing analysis performed on the fall synoptic data (during the dry season) are best explained as documenting groundwater inputs to the mainstem of Shavers Creek as local interflow, i.e. shallow groundwater flow. Thus, tributaries and Shavers Creek become locally controlled during the dry period. Principal components for surface water chemistry in Shavers Creek vary in the main stem and in tributaries during this period. The components determined for the upland forested parts of Shavers Creek cluster more closely with the components which describe shallow interflow within the forested landscape as observed in the pristine forested subcatchments of Garner Run and Shale Hills while components determined for the lowland agricultural parts of Shavers Creek trend towards the components which describe shallow interflow in the agricultural catchment Cole Farm. Principal component analyses during the winter snow melt (a wet period) are consistent with a regional homogenization of water chemistry in Shavers Creek caused by the dominance of snow water inputs to interflow, i.e. nonlocal control. Specifially, principal components for surface water chemistry in Shavers Creek during snow melt are all clustered closely to components which reflect forested land use. This clustering highlights the homogenization of surface water chemistry from the wet winter synoptic.Two separate regression models for Shavers Creek, one using lithology and the other using land use as model inputs, show that each of these variables alone predict solute flux well in the fall during the local control exhibited in the dry period. However, the covariance of land use and lithology in the watershed that causes a confounding effect for the regression models, because land use varies with lithology. On the other hand, these same models, show that land use alone does not predict solute flux as well as lithology during the wet winter period of non-local control. Lithology is therefore the best predictor of solute flux spatially in Shavers Creek in both wet and dry periods. Land use is only as good of a predictor as lithology during periods of local control when water tables are low in the watershed. Variations in local versus nonlocal control on solute and water flux contributions to Shavers Creek is related to hydrologic connectivity. Local controls, defined as properties that control water transport at small spatial scales, are more significant during the dry season when water tables are low. In contrast to local controls such as soil properties, nonlocal controls dominate during wet periods in Shavers Creek. The nonlocal controls include larger scale characteristics such as watershed topography.In summary, both principal component analyses and regression modeling show distinct geochemical differences between the dry fall period and the wet winter period in Shavers Creek watershed. The dry fall period shows geochemical heterogeneity throughout the watershed, and solute flux can be predicted by lithology and land use. During this time, the watershed is hydrologically disconnected and surface water chemistry is controlled locally. In contrast, hydrologic connectivity in the watershed increases during the wet winter period and surface water chemistry is controlled largely by nonlocal properties. During this time, geochemical homogeneity is observed throughout the watershed and solute flux is better predicted by lithology than land use. Overall, lithology is the most consistent predictor of the spatial variation of solute flux in Shavers Creek and is therefore essential to understanding solute transport in the watershed. These findings provide a deeper insight into transport dynamics in Shavers Creek and could potentially inform an improved understanding of upland agricultural watersheds within the Susquehanna River Basin.