Elliott Kellner

A comparison of the spatial distribution of vadose zone water in forested and agricultural floodplains a century after harvest

To improve quantitative understanding of the long-term impact of historic forest removal on floodplain vadose zone water regime, a study was implemented in fall 2010, in the Hinkson Creek Watershed, Missouri, USA. Automated, continuously logging capacitance-frequency probes were installed in a grid-like formation (n=6) and at depths of 15, 30, 50, 75, and 100 cm within a historic agricultural field (Ag) and a remnant bottomland hardwood forest (BHF). Data were logged at thirty minute intervals for the duration of the 2011, 2012, and 2013 hydrologic years. Results showed volumetric water content (VWC) to be significantly different between sites (p<0.01) during the study, with site averages of 33.1 and 32.8% at the Ag and BHF sites, respectively. Semi-variogram analyses indicate the presence of strong (<25%) horizontal and vertical spatial correlation of VWC at the Ag site, and a relatively short-range (25 cm) vertical spatial correlation at the BHF, but only indicate horizontal VWC spatial correlation in the top 30 cm of the BHF profile. Likely mechanisms contributing to patterns of observed differences are contrasting rates and depths of plant water use, and the presence of preferential flow paths in the below ground BHF. Results suggest historic forest removal and cultivation of the Ag site lead to an effective homogenization of the upper soil profile, and facilitated the development of strong VWC spatial dependency. Conversely, higher hydraulic conductivity of the more heterogeneous BHF subsurface likely results in a wetting of the deeper profile (75 cm) during climatically wet periods, and thus a more effective processing of hydrologic inputs. Collective results highlight the greater extent and degree to which forest vegetation impacts subsurface hydrology, relative to grassland/agricultural systems, and point to the value of reestablishing floodplain forests for fresh water routing, water quality, and flood mitigation in mixed-land-use watersheds.

A comparison of forest and agricultural shallow groundwater chemical status a century after land use change.

Considering the increasing pace of global land use change and the importance of groundwater quality to humans and aquatic ecosystems, studies are needed that relate land use types to patterns of groundwater chemical composition. Piezometer grids were installed in a remnant bottomland hardwood forest (BHF) and a historic agricultural field (Ag) to compare groundwater chemical composition between sites with contrasting land use histories. Groundwater was sampled monthly from June 2011 to June 2013, and analyzed for 50 physiochemical metrics. Statistical tests indicated significant differences (p<0.05) between the study sites for 32 out of 50 parameters. Compared to the Ag site, BHF groundwater was characterized by significantly (p<0.05) lower pH, higher electrical conductivity, and higher concentrations of total dissolved solids and inorganic carbon. BHF groundwater contained significantly (p<0.05) higher concentrations of all nitrogen species except nitrate, which was higher in Ag groundwater. BHF groundwater contained significantly (p<0.05) higher concentrations of nutrients such as sulfur, potassium, magnesium, calcium, and sodium, relative to the Ag site. Ag groundwater was characterized by significantly (p<0.05) higher concentrations of trace elements such as arsenic, cadmium, cobalt, copper, molybdenum, nickel, and titanium. Comparison of shallow groundwater chemical composition with that of nearby receiving water suggests that subsurface concentration patterns are the result of contrasting site hydrology and vegetation. Results detail impacts of surface vegetation alteration on subsurface chemistry and groundwater quality, thereby illustrating land use impacts on the lithosphere and hydrosphere. This study is among the first to comprehensively characterize and compare shallow groundwater chemical composition at sites with contrasting land use histories.

Continuous and event-based time series analysis of observed floodplain groundwater flow under contrasting land-use types

There is an ongoing need to improve quantitative understanding of land-use impacts on floodplain groundwater flow regimes. A study was implemented in Hinkson Creek Watershed, Missouri, USA, including equidistant grids of nine piezometers, equipped with pressure transducers, which were installed at two floodplain study sites: a remnant bottomland hardwood forest (BHF) and a historical agricultural field (Ag). Data were logged at thirty minute intervals for the duration of the 2011, 2012, 2013, and 2014 water years (October 1, 2010-September 30, 2014). Results show significant (p<0.001) differences between Darcy-estimated groundwater flow at the two study sites. Although median flow values at the two sites were similar (0.009 and 0.010mday(-1) for the Ag and BHF sites, respectively), the BHF displayed a more dynamic flow regime compared to the Ag site. Maximum flow values were 0.020 and 0.049mday(-1) for the Ag and BHF sites, respectively. Minimum flow values were -0.018 and -0.029mday(-1) for the Ag and BHF sites, respectively. The BHF showed greater magnitude, longer duration, and more frequent negative flows, relative to the Ag site. Event-based analyses indicated a more seasonally responsive flow regime at the BHF, with higher flows than the Ag site during the wet season and lower flows than the Ag site during the dry season. Notably, the seasonal pattern of relative site flow differences was consistent across a wide range of precipitation event magnitudes (i.e. 8-45mm). Results are by majority attributable to greater rates of plant water use by woody vegetation and preferential subsurface flow at the BHF site. Collectively, results suggest greater flood attenuation capacity and streamwater buffering potential by the BHF floodplain, relative to the Ag, and highlight the value of floodplain forests as a land and water resource management tool.

Quantifying loading, toxic concentrations, and systemic persistence of chloride in a contemporary mixed-land-use watershed using an experimental watershed approach

A nested-scale experimental watershed study was implemented to quantify loading and persistence of chloride in an urbanizing, mixed-land-use watershed. A Midwest USA (Missouri) watershed was partitioned into five sub-basins with contrasting dominant land use. Streamwater was tested for chloride concentration four days per week from October 2009 through May 2014 at each site. Monitoring sites included co-located gauging and climate stations recording variables at 30-minute intervals. Results indicate significant (p<0.01) differences in chloride concentrations and loading between sites. Loading consistently increased from the forested headwaters (average=507kgday-1) to primarily urban watershed terminus (average=7501kgday-1). Chloride concentrations were highest (average=83.9mgL-1) with the greatest frequency of acutely toxic conditions (i.e. 860mgL-1) mid-watershed. This finding is in-part attributable to the ratio of chloride application to streamflow volume (i.e. increasing flow volume with stream distance resulted in chloride dilution, offsetting increased percent urban land use with stream distance). Results highlight the important, yet often confounding, interactions between pollutant loading and flow dynamics. Chloride peaks occurred during late winter/early spring melting periods, implicating road salt application as the primary contributor to the chloride regime. Floodplain groundwater analysis indicated seasonal sink/source relationships between the stream and floodplain, which could contribute to chronic toxicity and persistent low Cl- concentrations in streamwater year-round. Results hold important implications for resource managers wishing to mitigate water quality and aquatic habitat degradation, and suggest important water quality limitations to stream restoration success in complex urban aquatic ecosystems.

Improving understanding of mixed-land-use watershed suspended sediment regimes: Mechanistic progress through high-frequency sampling

Given the importance of suspended sediment to biogeochemical functioning of aquatic ecosystems, and the increasing concern of mixed-land-use effects on pollutant loading, there is an urgent need for research that quantitatively characterizes spatiotemporal variation of suspended sediment dynamics in contemporary watersheds. A study was conducted in a representative watershed of the central United States utilizing a nested-scale experimental watershed design, including five gauging sites (n=5) partitioning the catchment into five sub-watersheds. Hydroclimate stations at gauging sites were used to monitor air temperature, precipitation, and stream stage at 30-min intervals during the study (Oct. 2009-Feb. 2014). Streamwater grab samples were collected four times per week, at each site, for the duration of the study (Oct. 2009-Feb. 2014). Water samples were analyzed for suspended sediment using laser particle diffraction. Results showed significant differences (p<0.05) between monitoring sites for total suspended sediment concentration, mean particle size, and silt volume. Total concentration and silt volume showed a decreasing trend from the primarily agricultural upper watershed to the urban mid-watershed, and a subsequent increasing trend to the more suburban lower watershed. Conversely, mean particle size showed an opposite spatial trend. Results are explained by a combination of land use (e.g. urban stormwater dilution) and surficial geology (e.g. supply-controlled spatial variation of particle size). Correlation analyses indicated weak relationships with both hydroclimate and land use, indicating non-linear sediment dynamics. Suspended sediment parameters displayed consistent seasonality during the study, with total concentration decreasing through the growing season and mean particle size inversely tracking air temperature. Likely explanations include vegetation influences and climate-driven weathering cycles. Results reflect unique observations of spatiotemporal variation of suspended sediment particle size class. Such information is crucial for land and water resource managers working to mitigate aquatic ecosystem degradation and improve water resource sustainability in mixed-land-use watersheds globally.

Confounded by forgotten legacies: Effectively managing watersheds in the contemporary age of unknown unknowns

School of Natural Resources, University of Missouri, 203 ABNR Building, Columbia, MO 65211, USA 2 Institute of Water Security and Science, Davis College, Schools of Agriculture and Food, and Natural Resources, West Virginia University, 3107 Agricultural Sciences Building, Morgantown, WV 26506, USA Correspondence Elliott Kellner, School of Natural Resources, University of Missouri, 203 ABNR Building, Columbia, MO 65211, USA. Email: elliott.kellner@mail.wvu.edu

Spatiotemporal variability of suspended sediment particle size in a mixed-land-use watershed

Given existing knowledge gaps, there is a need for research that quantitatively characterizes spatiotemporal variation of suspended sediment particle size distribution (PSD) in contemporary watersheds. A five-year study was conducted in a representative watershed of the central United States utilizing a nested-scale experimental watershed study design, comprising five gauging sites partitioning the catchment into five sub-watersheds. Streamwater grab samples were collected four times per week, at each gauging site, for the duration of the study period (Oct. 2009-Feb. 2014). Samples were analyzed using laser particle diffraction. Significantly different (p<0.05) suspended sediment PSDs were observed at monitoring sites throughout the course of the study. For example, results indicated greater proportions of silt at site #5 (65%), relative to other sites (41, 32, 29, and 43%, for sites #1-#4, respectively). Likewise, results showed greater proportions of sand at sites #2 and #3 (66 and 68%, respectively), relative to other sites (57, 55, and 34%, for sites #1, #4, and #5, respectively). PSD spatial variability was not fully explained by hydroclimate or sub-watershed land use/land cover characteristics. Rather, results were strengthened by consideration of surficial geology (e.g. supply-controlled spatial variation of particle size). PSD displayed consistent seasonality during the study, characterized by peaks in the proportion of sand (and aggregates) during the winter (i.e. 70-90%), and minimums during the summer (i.e. 12-38%); and peaks in the proportion of silt particles in the summer (i.e. 61-88%) and minimums in the winter (i.e. 10-23%). Likely explanations of results include seasonal streamflow differences. Results comprise distinct observations of spatiotemporal variation of PSD, thereby improving understanding of lotic suspended sediment regimes and advancing future management practices in mixed-land-use watersheds.

Land use impacts on floodplain water table response to precipitation events

Two floodplain sites were instrumented, a bottomland hardwood forest and a historic agricultural field. Continuous, automated, in situ shallow groundwater monitoring was paired with a gridded sampling design to facilitate detailed spatiotemporal analysis of water table response to precipitation events. Data were collected at 30 minute intervals for the duration of the 2011, 2012, 2013, and 2014 water years. A total of 403 precipitation events were observed, with median depth of 3 mm. Results show significantly (p < 0.001) greater study site median water table response at the agricultural site (0.02 m), relative to the forest (0.01 m). The ratio of water table response magnitude to precipitation depth was 8.3 and 3.3 at the agricultural and forest sites, respectively, a difference of more than 150%. Median shallow groundwater level lag time was shorter with respect to both precipitation and streamflow at the agricultural site (10 and 2 hrs. for precipitation and flow, respectively), relative to the forest (11.5 and 4.5 hrs. for precipitation and flow, respectively). Longer lag times indicated slower water table response and potentially increased flood attenuation capacity by the forest site. Results show the agricultural site water table responded to a greater number of precipitation events (including smaller events), reflecting potentially reduced relative flood attenuation capacity. Observed differences highlight the various impacts of forest vegetation on water table dynamics and emphasize the benefits of floodplain forests as a water resource management tool, including flood mitigation, in mixed-land-use watersheds.