Michael R. Williams

Solute export from forested and partially deforested catchments in the central Amazon

The hydrochemical responses to slash-and-burnagriculture in a small rainforest catchment of thecentral Amazon were investigated for one year. Disturbances in the partially deforested catchmentbegan in 1987, and during the study a 2-ha plot was cut(July 1989) and burned (October 1989) in preparationfor the cultivation of manioc; the partially deforestedcatchment was approximately 80% deforested at the timeof this study. Solute fluxes exported by base flowwere estimated from solute concentrations of stream watermeasured at least once per week. Solute fluxesfor storm flow were estimated by measuring streamwaterconcentrations during two storms. Baseflow runoffrepresented about 94% of the water outflow from thestudy basin and was the dominant pathway of soluteexport. Total rainfall during the study period was2754 mm of which 2080 mm was exported from thepartially deforested catchment as stream runoff. Theratio of surface runoff to annual rainfall for asimilar study conducted in the same catchment whilecompletely forested in 1984 was lower than after thecatchment was 80% deforested in 1990 (0.57 versus0.76), while evapotranspiration (ET) was lower by about afactor of two in 1990 compared to 1984. Particulateremoval from the partially deforested catchment was 151kg ha−1 yr−1. Nutrient losses from thepartially deforested catchment were higher than thosemeasured when the catchment was undisturbed in 1984 byfactors of 1.4, 1.8, and 2.1 for total inorganicnitrogen (TIN), total dissolved nitrogen (TDN), and totalnitrogen (TN); and by factors of 4.0, 6.6, and 7.9 for solublereactive phosphate (PO3−4), total dissolvedphosphorus (TDP), and total phosphorus (TP),respectively. These data show that deforestation andcolonization in upland catchments of the central Amazonalter the hydrochemical balance of streams bydecreasing ET, thereby increasing discharge and soluteexport.

Chemical composition and deposition of rain in the central Amazon, Brazil

Abstract Major solute concentrations in wet deposition were measured for 115 individual events from October 1988 to June 1990 at Lake Calado, Amazonas, Brazil. A continuous record from July 1989 through June 1990 included 210 events that ranged in size from 0.2 to 85 mm (annual total, 2754 mm). The 95 events chemically analyzed during this period (45% of the total) were evenly distributed over all storm sizes. The volume-weighted mean (VWM) of the measured ionic sum was low (36 μeq l −1), and H+ was the most abundant ion (VWM pH = 4.8). Organic acids made substantial contributions to both measured dissolved organic carbon and H+. Small interannual differences are apparent in a comparison with rain data collected from the same study site in 1984. The composition of rain appeared to be derived from a constant source of oceanic and rain-forest aerosols, and was not significantly influenced by local or regional biomass burning.

Solute dynamics in soil water and groundwater in a central Amazon catchment undergoing deforestation

Hydrochemical changes caused by slash-and-burnagricultural practices in a small upland catchment inthe central Amazon were measured. Soluteconcentrations were analyzed in wet deposition,overland flow, shallow throughflow, groundwater andbank seepage in a forested plot (about 5 ha) and anadjacent plot (about 2 ha) which had been deforestedin July 1989 and planted to manioc, and in streamwater in partially deforested and forested catchments. Measurements were made from November 1988 to June1990. The effects of slash-and-burn agriculturalpractices observed in the experimental plot includedincreased overland flow, erosion, and large losses ofsolutes from the rooted zone. Concentrations ofNO3-, Na+, K+, SO42-,Cl- and Mn in throughflow of the experimentalplot were higher than those of the control plot bymore than a factor of 10. Extensive leaching occurredafter cutting and burning, but solute transfers werediminished along pathway stages of throughflow togroundwater, and particularly within the riparian zoneof the catchment. High concentrations of N and P inoverland flow indicate the importance of usingforested riparian buffers to mitigate solute inputs toreceiving waters in tropical catchments.

Development and evaluation of a spatially-explicit index of Chesapeake Bay health.

In an effort to better portray changing health conditions in Chesapeake Bay and support restoration efforts, a Bay Health Index (BHI) was developed to assess the ecological effects of nutrient and sediment loading on 15 regions of the estuary. Three water quality and three biological measures were combined to formulate the BHI. Water quality measures of chlorophyll-a, dissolved oxygen, and Secchi depth were averaged to create the Water Quality Index (WQI), and biological measures of the phytoplankton and benthic indices of biotic integrity (P-IBI and B-IBI, respectively) and the area of submerged aquatic vegetation (SAV) were averaged to create the Biotic Index (BI). The WQI and BI were subsequently averaged to give a BHI value representing ecological conditions over the growing season (i.e., March-October). Lower chlorophyll-a concentrations, higher dissolved oxygen concentrations, deeper Secchi depths, higher phytoplankton and benthic indices relative to ecological health-based thresholds, and more extensive SAV area relative to restoration goal areas, characterized the least-impaired regions. The WQI, P-IBI and BHI were significantly correlated with (1) regional river flow (r=-0.64, -0.57 and -0.49, respectively; p<0.01), (2) nitrogen (N), phosphorus (P) and sediment loads (all positively correlated with flow), and (3) the sum of developed and agricultural land use (highest annual r(2)=0.86, 0.71 and 0.68, respectively) in most reporting regions, indicating that the BHI is strongly regulated by nutrient and sediment loads from these land uses. The BHI uses ecological health-based thresholds that give an accurate representation of the health conditions in Chesapeake Bay and was the basis for an annual, publicly released environmental report card that debuted in 2007.

Processes regulating the solute concentrations of snowmelt runoff in two subalpine catchments of the Sierra Nevada, California

Geochemical processes regulating solute concentrations in snowmelt runoff were investigated for the snowmelt periods of 1992 and 1993 in two subalpine catchments in Sequoia National Park, California. The catchments, 0.5 and 0.2 ha, have 10 and 25% soil coverage, respectively; the remaining area is exposed granodiorite bedrock. Mineral weathering was the major source of solutes in runoff. Calcium export in excess of stoichiometric plagioclase weathering was attributed to dry deposition and the weathering of mafic minerals and disseminated calcite. Cation exchange was important in the regulation of K and increased base cation export in response to rain-on-snow events and LiBr tracer. Sulfate fluxes were attributed to snowpack elution and SO4 desorption after the beginning of melt. The selective retention of Li during tracer experiments suggests that acid-neutralizing capacity is linked to soil stocks along convoluted flow paths and is not necessarily compromised in areas of sparse soil cover.

The Efficacy of Constructed Stream-Wetland Complexes at Reducing the Flux of Suspended Solids to Chesapeake Bay.

Studies documenting the capacity of restored streams to reduce pollutant loads indicate that they are relatively ineffective when principal watershed stressors remain intact. Novel restorations are being designed to increase the hydraulic connectivity between stream channels and floodplains to enhance pollutant removal, and their popularity has increased the need for measurements of potential load reductions. Herein we summarize input-output budgets of total suspended solids (TSS) in two Coastal Plain lowland valleys modified to create stream-wetland complexes located above the head-of-tide on the western shore of Chesapeake Bay. Loads entering (input) and exiting (output) the reconfigured valleys over three years were 103 ± 26 and 85 ± 21 tons, respectively, and 41 ± 10 and 46 ± 9 tons, respectively. In both cases, changes in loads within the reconfigured valleys were insignificant relative to cumulative errors. High variability of TSS retention among stormflow events suggests that the capacity of these systems to trap and retain solids and their sustainability depend on the magnitude of TSS loads originating upstream, design characteristics, and the frequency and magnitude of large storms. Constructed stream-wetland complexes receiving relatively high TSS loads may experience progressive physical and chemical changes that limit their sustainability.

An integrated modelling system for management of the Patuxent River estuary and basin, Maryland, USA

The Patuxent River watershed is a heavily impacted basin (2290 km2) and estuarine tributary (120 km2) of the Chesapeake Bay, USA. To assist management of the basin, we are testing a coupled modelling system composed of a watershed model (HSPF), an estuarine circulation model (CH3D), and an estuarine water‐quality model (CE‐QUAL‐ICM). The modelling system is being tested to guide the development of Total Maximum Daily Loads (TMDLs), and therefore errors in the models must be carefully evaluated. A comparison of daily total nitrogen (TN) concentrations simulated in HSPF with observations indicated that there was no significant bias, with an rms error of 37%. In contrast, modelled total phosphorus (TP) and total suspended solids (TSS) had significant bias with larger rms errors (65% and 259%, respectively). In the estuary, CH3D accurately simulated tides, temperature, and salinity. CE‐QUAL‐ICM overestimated nitrogen (N) and phosphorus (P) in the upper estuary and underestimated in the lower estuary, primarily because intertidal marshes are not currently a model component. Model errors declined from short (⩽1 day) to long (multi‐year) timescales as under‐ and overestimations cumulatively cancelled. Watershed model errors propagate into the estuarine models, interacting with each subsequent models errors, which limits the effectiveness of this TMDL management tool at short timescales.