HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (4) Citing Articles via Google Scholar Google Scholar Articles by Williams, M. R. Articles by Camargo, P. B. Search for Related Content PubMed PubMed Citation Articles by Williams, M. R. Articles by Camargo, P. B. Agricola Articles by Williams, M. R. Articles by Camargo, P. B. Related Collections Surface Water Quality Air Pollution Water Pollution

TECHNICAL REPORT
Surface Water Quality

Precipitation and River Water Chemistry of the Piracicaba River Basin, Southeast Brazil

^a The Ecosystems Center, Marine Biological Lab., Woods Hole, MA 02543
^b Univ. of São Paulo, Centro de Energia Nuclear na Agricultura, Ave. Centenário 303, CEP 13416-000, Piracicaba, SP, Brazil

Corresponding author (mikewilliams{at}mbl.edu)

Received for publication January 24, 2000.

	ABSTRACT

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION REFERENCES

 
Annual precipitation and river water volumes and chemistry were measured from 1995 to 1998 in a mesoscale agricultural area of southeast Brazil. Precipitation was mildly acidic and solute concentrations were higher in the west than in the east of the basin. Combustion products from biomass burning, automobile exhaust, and industry typically accumulate in the atmosphere from March until October and are responsible for seasonal differences observed in precipitation chemistry. In river waters, the volume-weighted mean (VWM) concentrations of major solutes at 10 sites across the basin were generally lower at upriver than at downriver sampling sites for most solutes. Mass balances for major solutes indicate that, as a regional entity, the Piracicaba River basin was a net sink of H⁺, PO^3-₄, and NH⁺₄, and a net source of other solutes. The main stem of the Piracicaba River had a general increase in solute concentrations from upriver to downriver sampling sites. In contrast, NO^-₃ and NH⁺₄ concentrations increased in the mid-reach sampling sites and decreased due to immobilization or utilization in the mid-reach reservoir, and there was denitrification immediately downriver of this reservoir. Compared with tributaries of the Chesapeake Bay estuary, the Piracicaba River is affected more by point-source inputs of raw sewage and industrial wastes than nonpoint agricultural runoff high in N and P. Inputs of N and C are responsible for a degradation of water quality at downriver sampling sites of the Piracicaba River drainage, and water quality could be considerably improved by augmenting sewage treatment.

Abbreviations: ANC, acid neutralizing capacity • BOD, biological oxygen demand • CSS, coarse suspended solids • DIC, dissolved inorganic carbon • DOC, dissolved organic carbon • FSS, fine suspended solids • SE, standard error • VWM, volume-weighted mean

	INTRODUCTION

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION REFERENCES

 
MOST studies of water chemistry in South America have been done in relatively large, undeveloped systems such as the Amazon (Stallard and Edmond, 1981, 1983, 1987; Richey et al., 1988; Edmond et al., 1995), Orinoco (Hamilton and Lewis, 1990), Paraná (Depetris and Kempe, 1993), and Paraguay Rivers (Hamilton et al., 1995). Little is known about the dynamics of solutes in rivers of mesoscale, developed areas, such as the Piracicaba River basin (12400 km²), where there is a combination of natural and managed ecosystems punctuated by urban and industrialized centers (Krusche et al., 1997; Martinelli et al., 1999). Until recently, reliable hydrochemical data were unavailable for large river basins in southern Brazil. Moreover, other than several recent studies from the Amazon Basin (Lesack and Melack, 1991; Williams et al., 1997a; Filoso et al., 1999), chemical data are scarce on wet deposition that include a large portion of annual events, all major solutes, and daily or event-based samples from areas of South America.

In contrast, a great deal of hydrochemical information exists from mesoscale basins in North American and European temperate zones. For example, there is a large body of literature pertaining to the effects of development and agriculture on solute inputs to the Chesapeake Bay, USA (Correll et al., 1992; Boynton et al., 1995; Jordan et al., 1997a,b; Fisher et al., 1998), and on a global basis (Peierls et al., 1991).

The flux of major solutes from developed basins can result in economic loss and a degradation of water quality. Algal blooms, reduced water transparency, a loss of submerged aquatic vegetation (SAV), and a general degradation of water quality are associated with excessive fertilizer applications and agricultural activities in other studies of developed basins in the temperate zone (Staver and Brinsfield, 1990, 1996; Staver et al., 1996; Jordan et al., 1997a; Owens et al., 1998). In humid climates of the temperate zone, greater amounts of nutrients and major solutes are moved via subsurface flow than in surface runoff (Correll et al., 1992; Jordan et al., 1997a). However, this may not be true in the humid tropics where precipitation and soil depths are generally much greater, and the types and management of agroecosystems are fundamentally different. For the purpose of water resources management, it is valuable to make comparisons among subtropical and temperate-zone studies to elucidate possible differences in the factors that regulate the chemical composition of precipitation and river water in South America, as well as the applicability of the findings of other studies to the Piracicaba River basin.

Our main objective in this study was to determine the factors that regulate solute concentrations in precipitation and the river water of tributaries in the Piracicaba River drainage. We used mass balances to evaluate the relative influence of different land use (i.e., agriculture [sugarcane (Saccharum spp.) and citrus] and pasture) and point-source inputs on solute export. We compared upriver (less polluted) with downriver (more polluted) sites to evaluate anthropogenic inputs and possible processes responsible for changes in solute fluxes, and we compared the major factors regulating river water chemistry of the tributaries of the Piracicaba River basin with those of temperate watersheds.

	STUDY AREA

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION REFERENCES

 
The Piracicaba River basin is located in the state of São Paulo in southeast Brazil (Fig. 1) and has three main watersheds: the Jaguari (3400 km²), the Atibaia (3000 km²), and the Piracicaba (6000 km²). The main stem of the Piracicaba River flows from east to west for approximately 250 km before discharging into the upper Tietê River (a tributary of the Paraná River). Soils of the Piracicaba River basin are primarily Oxisols and Ultisols (Table 1), and the topography declines from an elevation of 2000 m in the eastern basin to 460 m in the western basin. About 80 to 95% of the total area of each subbasin (1–10, Fig. 1) was in a form other than natural forest in 1993; areas devoted to silviculture, sugarcane, and pasture are prominent in almost all subbasins (Table 1). The Piracicaba River basin had about 2.8 million inhabitants in 1993 (226 inhabitants km^-2), 92% in urban centers and 8% in rural areas (São Paulo, 1994). The central and western subbasins have higher population densities responsible for raw sewage and industrial waste inputs to the main stem of the Piracicaba River (Table 1).

View larger version (24K): [in this window] [in a new window]   Fig. 1. Map of the Piracicaba River basin indicating the locations of the stations used for the collection of precipitation and river water samples. Precipitation collection sites are (A) Bragança, (B) Campinas, and (C) Piracicaba. River sampling sites are Bairro da Ponte (1), Morungaba (2), Desembargador Furtado (3), Fazenda da Barra (4), Paulínia (5), Usina Ester (6), Carioba (7), Copersucar (8), Recreio (9), and Artemis (10). Subbasin boundaries are marked by dotted lines.

	METHODS

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION REFERENCES

Precipitation was collected using Aerochem wet/dry collectors (Aerochem Metrics, Bushnell, FL), and all the precipitation that occurred over a 24-h period was pooled to represent a daily total. Plastic buckets used to collect precipitation were rinsed daily with deionized water. River water samples were collected from just below the surface in the middle of the river channel using a Niskin bottle. Chemical data of river water prior to June 1995 were collected by Companhia Estadual de Tratamento de Esgoto e Saneamento Básico (CETESB) and analyzed using the protocols outlined by the American Public Health Association (1976). Precipitation and river samples collected after June 1995 were analyzed at Centro de Energia Nuclear na Agricultura (CENA).

Temperature, electrical conductance, and pH (Orion [Beverly, MA] 250A) were measured in the field. All conductance and pH measurements were standardized to 25°C. Aliquots of water were filtered immediately after collection through Millipore (Bedford, MA) HA filters (nominal pore size 0.4 µm) for analysis of major ions and precombusted glass-fiber filters for dissolved organic and inorganic carbon analyses (DOC and DIC, respectively). Samples were stored in high-density polyethylene bottles that had been acid-washed and rinsed with deionized water until the rinse water had conductivity <1 µS cm^-1. Acid neutralizing capacity (ANC) was measured by titration (Gran, 1950, 1952). Nitrate, SO^2-₄, Cl^-, and organic acids were determined using either a Shimadzu (Kyoto, Japan) LC-10AD or Dionex (Sunnyvale, CA) ion chromatograph. Base cations and NH⁺₄ were determined by atomic emission spectroscopy using samples preserved with 50 µM of HgCl₂ for river water samples and by ion chromatography (Dionex 2010i) for precipitation samples. Concentrations of DOC and DIC were determined by gas chromatography, and those of SiO₂ by a silico-molybdate technique (Strickland and Parsons, 1972). Coarse (60 µm) and fine (0.1 and <60 µm) suspended solids (CSS and FSS, respectively) were collected with a depth-integrating sampler and determined gravimetrically.

Data Analysis
Volume-weighted mean (VWM) concentrations were calculated for solutes in precipitation and river water. The equation for calculating VWM concentrations for n measured precipitation events can be represented as:

[1]

where Ci = the observed concentration of combined events for a daily sample i, Vi = the size of combined events i (mm), and the denominator is the V. The same equation is used for calculating VWM concentrations in river water, where Ci = the observed concentration of instantaneous river flow i, Vi = discharge volume (liters) for the sampling interval with sample data as the midpoint of the period i, and the denominator is the total discharge volume for period i.

Annual fluxes of precipitation (inputs) were calculated as:

[2]

where Q_p is the precipitation volume (mean of the of the 1995 and 1996 calendar years) for each of 10 subbasins, and C_p is the VWM solute concentration for each of three sampling sites. Equation [2] was also used to calculate river fluxes (outputs), where Q_r is river discharge (mean of the of the 1995 and 1996 calendar years) for each subbasin, and C_r is the VWM solute concentration measured at the sampling site representing a particular subbasin. Precipitation depth for the collectors in each subbasin was averaged. Due to limited spatial sampling, chemical data from Bragança (Precipitation Collection Site A) were used to calculate wet deposition for Subbasins 1 and 2, the mean of the VWM solute concentrations for Bragança and Campinas (Precipitation Collection Site B) was used to calculate wet deposition for Subbasins 3 through 7, and that of the three precipitation collection sites was used for Subbasins 8 through 10.

Mass balances for the entire Piracicaba River basin and its subbasins were calculated using inputs from precipitation and outputs from river fluxes as:

[3]

where precipitation chemistry from the three sampling sites was used for the different subbasins as described above. The quantity of precipitation for each subbasin was combined with precipitation chemistry data to calculate inputs. River chemistry and runoff for each river sampling site were used to calculate outputs. Solute inputs and outputs from each river sampling site are representative of the watershed area above the sampling site. For example, precipitation inputs for the subbasin area upriver of Sampling Site 7 are the product of the mean precipitation chemistry for Sites A and B (May 1997 to May 1998), and the mean volume that occurred using all the precipitation measurements available for the subbasin (annual average for 1995 and 1996). River inputs for the subbasin area upriver of Sampling Site 7 were calculated as the product of the river chemistry and runoff volume for Site 7 (VWM solute concentrations and annual mean runoff volume for June 1995 to December 1997).

	RESULTS

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION REFERENCES

 
Precipitation
Precipitation is seasonal with the largest deposition generally occurring from the months of October to March (Fig. 2). A total of 56 to 83 daily measurements of precipitation >0.9 mm occurred at the three collection sites in the Piracicaba basin during the period of observation (May 1997 to May 1998) yielding a mean total of 1226 mm, or 191 mm less than the long-term average for the entire Piracicaba River basin (i.e., Site 10, Table 2). Volume-weighted mean (VWM) concentrations were calculated for all solutes (Table 3). Mean pH was about 4.5, and ranged from 3.5 to 6.1 at the three sampling sites. Hydrogen ion typically had the highest concentration of any ion and represented about 54% of the cations. Hydrogen ion was followed, in descending order, by NH⁺₄, Ca²⁺, K⁺, Na⁺, and Mg²⁺. Sulfate had the highest concentration of any anion and accounted for about 40% of the anion total, followed by NO^-₃, Cl^-, PO^3-₄, and ANC. Regression analyses of solute concentrations against the quantity of daily precipitation indicate that dilution occurred with increasing precipitation for most solutes, as shown in the comparison of ionic sum against daily precipitation ( major ions) for the three sampling sites (Fig. 3).

View larger version (33K): [in this window] [in a new window]   Fig. 2. Monthly precipitation and runoff for Site 10 of the Piracicaba River basin. Long-term records indicate that precipitation and runoff are generally highest from October to March (wet season).

View larger version (17K): [in this window] [in a new window]   Fig. 3. The relationship of storm size to the sum of major solutes (ionic sum) in the precipitation of the three sample collection sites (A, B, and C). Site A, located in a more rural setting than the other sampling sites, has the strongest relationship and commonly lower ionic sum values for equivalent storm sizes at the other sampling sites.

Hydrology
Water yield coefficients (the ratio of runoff to precipitation) from 1936 through 1996 were 0.29, 0.33, and 0.34 for the Atibaia, Jaguari, and Piracicaba watersheds, respectively (Table 2). Water yield coefficients (referred to as runoff coefficients) were lower in 1995 and 1996 for all three basins (about 0.26), in part due to the removal of an average of 33 m³ s^-1 from headwater tributaries for the city of São Paulo over the period of study (Moraes et al., 1998). The runoff coefficient for the Piracicaba Basin was 0.39 and included drinking water withdrawals. The annual average discharges of the Atibaia, Jaguari, and Piracicaba Rivers at Sites 5, 6, and 10 from 1943 to 1996 were 36, 53 (from 1947), and 140 m³ s^-1, respectively (Table 2). The high and low water periods of river stage are generally from October to March and April to September, respectively (Fig. 2).

River Chemistry
Mean pH was about 7.0, and ranged from 6.2 to 8.1 among the sampling sites. Almost all of the carbonate alkalinity is present as HCO^-₃ (referred to herein as ANC) at the pH range of the tributaries in the Piracicaba River basin (Stumm and Morgan, 1996). The ANC had the highest concentration of any ion, and on average represented about 57% of the anion total. The ANC concentrations ranged from 17 to 79% of the anion total, decreasing from upriver (1–4 and 9) to downriver (5–8 and 10) sites. Chloride and SO^2-₄ were typically higher in concentration downriver compared with upriver sites; both were followed, in descending order, by NO^-₃ and PO^3-₄. Calcium and Na⁺ were highest in concentration among the cations; both were higher in concentration downriver compared with upriver sites. The VWM concentrations of Ca²⁺ and Na⁺ were followed, in descending order, by Mg²⁺, K⁺, and NH⁺₄ (Table 4). The mean cation deficit was 5% of the total, indicating that there was a positive bias toward excess anions. We attribute part of this unusual excess to our calculation of ANC; substituting DIC concentrations for ANC improved our cation deficit (2%).

View this table: [in this window] [in a new window]   Table 4. Volume-weighted mean concentrations of river water samples collected during the study period. All concentrations are expressed in µeq L-1, except for silicate, dissolved organic carbon (DOC), and dissolved inorganic carbon (DIC) (µM); coarse suspended solids (CSS) and fine suspended solids (FSS) are in mg L-1. The ions required to balance the anions with the cations are designated in the Deficit category.

View larger version (25K): [in this window] [in a new window]   Fig. 4. Relationships of river discharge to nitrate concentrations at the 10 river sampling sites of the Piracicaba River basin. Discharge–concentration relationships are generally inverse and logarithmic at upriver locations, and become less discernible at downriver locations. Numbers in parentheses are sampling site locations indicated in Fig. 1.

View larger version (61K): [in this window] [in a new window]   Fig. 5. a. Comparison of inputs (precipitation), outputs (riverine), and the resulting mass balance (net export) of selected major solutes. Values are annual means in Mg yr-1 for the period of study. Numbers are ranked from upriver (1–4 and 9) to downriver (5–8 and 10) sampling sites to show possible trends in river chemistry. b. Comparison of inputs (precipitation), outputs (riverine), and the resulting mass balance (net export) of selected major solutes. Values are annual means in Mg yr-1. Numbers are ranked from upriver (1–4 and 9) to downriver (5–8 and 10) sampling sites.

View larger version (66K): [in this window] [in a new window]   Fig. 6. a. Residual export of solutes obtained by subtracting background concentrations from an upland catchment stream (Fazenda Bonfim). These values were divided by a silicate modification factor (described in text), which eliminates the effects of dilution, increasing basin area, and runoff on solute export. Values are in Mg yr-1. b. Residual export of solutes obtained by subtracting background concentrations from an upland catchment stream. These values were divided by a silicate modification factor and are in units of Mg yr-1.

We assume that the mass balances for the Piracicaba River basin are representative of the mean of the 1995 and 1996 calendar years. However, because our precipitation data overlap with our discharge data by only 7 months, there was a certain amount of error associated with our mass balances that would not have occurred had we used data of precipitation and runoff for the same time periods. We analyzed a 17-yr database of runoff volume and chemistry (CETESB) from some of the same locations in the Piracicaba River basin used in our study. The 17-yr trends show some significant variability in the VWM concentrations observed at each site, especially with regard to increases in NO^-₃. However, the maximum variability using the SE of NO^-₃ at the sampling sites was 8% of the VWM. An error of 10% for our annual estimates of discharge are reasonable for rivers with well-known rating curves such as those in this study (Winter, 1981; Linsley et al., 1982).

Precipitation chemistry may have resulted in more error compared with runoff because we do not have estimates of annual variability. However, based on other studies of precipitation chemistry, it would be unusual to expect an interannual variability of VWM concentrations >10% (Williams et al., 1997a; Filoso et al., 1999), barring a significant change in natural or anthropogenic inputs (Likens and Bormann, 1995). Even extreme wet or dry years may not affect VWM concentrations. Williams and Melack (1997a) in Sequoia National Park, California observed an interannual variability (SE) of NO^-₃ that was 7% of the VWM. This 10-yr record included a combination of a 6-yr drought period and two extremely wet years. Since NO^-₃ concentrations in precipitation are generally affected by anthropogenic factors more than concentrations of other major solutes (Likens and Bormann, 1995), we used a surrogate for temporal changes in atmospheric chemistry by calculating the variability of NO^-₃ concentrations for our precipitation data; this error (SE) was 9% of the VWM. Given that we calculated the annual precipitation volume for the Piracicaba Basin with a grid of up to 31 collection sites, the error associated with our annual estimates of precipitation volume are probably <5% (Winter, 1981; Linsley et al., 1982). As a surrogate for volume errors, we used the spatial variability (SE) calculated from the collection sties 1995 to 1996 (Table 2), which was about 2% of the annual precipitation volume.

Our estimates of the individual errors in the net solute flux from the Piracicaba River basin at Site 10 were propagated to give an estimate of overall error. Propagation of the component errors to obtain an approximate estimate of the overall error associated with the total flux of solutes was performed as in Williams and Melack (1997b), and is briefly outlined here. The reduced term of the propagation equation (Reckhow and Chapra, 1979; Taylor, 1982) is:

[4]

where V is the volume of water, C is the concentration of water, and S is the standard error with the subscripts of volume (v), concentration (c), and total flux (T). To calculate this error, we used the surrogate errors associated with our measurements of annual precipitation and runoff volumes and chemistry mentioned above. Since variances are additive, propagating the uncertainties for the fluxes of major solutes from precipitation and runoff using Eq. [2] gives a maximum error of 24% in our estimates of net solute flux.

Correlation Analysis
A Spearman correlation analysis of SiO₂ concentrations with those of other major solutes was done to determine the relative influence of the ground water contribution to river water versus that of overland flow or point-source inputs at different sites of the Piracicaba River and its tributaries. Assuming that the ground water contribution to river water is more affected by equilibrium reactions due to mineral weathering in soils, we would expect to see a higher SiO₂ signal from the ground water seepage component in river water than from overland flow or point-source inputs. Hence, higher positive correlations with SiO₂ suggest that there is a stronger ground water component. Sodium, Ca²⁺, Mg²⁺, and NO^-₃ had higher positive correlations with SiO₂ concentrations than the other solutes, and the relationships were significant (p < 0.05; Fig. 7). The correlations of these solutes with SiO₂ concentrations decreased with distance downstream. These were positively correlated with the percent areas of silviculture and natural forest for the different subbasins (larger source of ground water to rivers than overland flow), and inversely correlated with those of sugarcane plus citrus (larger overland flow component than forested areas) and biological oxygen demand (BOD) (large point source) (Table 1).

View larger version (19K): [in this window] [in a new window]   Fig. 7. Spearman correlations of major solutes with SiO2 concentrations for six sampling stations along the main stem of the Piracicaba River. Numbers are ranked from upriver to downriver sampling sites to show possible trends in river chemistry. The designation of "Cat" refers to the mean correlation value for Na+, Ca2+, and Mg2+, whereas "An" refers to the mean correlation value for Cl- and SO2-4. SS is suspended sediments. Numbers 1, 3, and 5 are upstream sites, and 7, 8, and 10 are downstream sites.

	DISCUSSION

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION REFERENCES

 
Precipitation Chemistry and Solute Origins
Particulates, gases, and aerosols in the atmosphere are commonly regarded as determinants of wet deposition chemistry. Anthropogenic emissions can be oxidized to produce C, S, and N compounds in the atmosphere (Cogbill and Likens, 1974) that are balanced stoichiometrically by a net production of H⁺. Gaseous sulfuric and nitric acids are partially neutralized by gaseous NH₃ to form NH₄NO₃ and (NH₄)₂SO₄, both common components of rainfall influenced by anthropogenic sources. Nitric and sulfuric acids are regarded as the sole contributors of NO^-₃ and SO^2-₄ in precipitation (Chan et al., 1987) and are assumed to be the major sources of acidity in precipitation influenced by combusted hydrocarbons (Likens et al., 1979). In our study, there was a positive relationship (r² = 0.70) between the sum of NO^-₃ and SO^2-₄ against H⁺ concentrations at Site B only (p < 0.05), which suggests that considerable H⁺ is contributed by the dilute mineral acids of nitrogen and sulfur compounds. The correlation value improved slightly in the wet season (r² = 0.74). The residual H⁺ concentrations contributed to moderate anion deficits (Table 3), but the potential H⁺ concentration contributed by strong acid anions was not balanced by that of formic and acetic acids, suggesting that there are other important sources of free acidity.

Dissolved organic carbon usually arises from organic aerosol emissions of vegetation (Isidorov et al., 1985; Greaves et al., 1987), and the oxidation of biogenic compounds contributes to the presence of formic, acetic, and pyruvic acids (Keene and Galloway, 1986). These organic acids exist predominantly as gaseous aerosols (Andreae et al., 1988b). The equivalent concentration of carbon from measured formic and acetic acids (arithmetic mean of 1.1 µM, Table 3) was about 1% of the DOC concentration measured in the precipitation of the Piracicaba River basin. Urban areas have point sources of hydrocarbons and other organic industrial pollutants (Kawamura and Kaplan, 1986). Our study indicates that only a small amount of DOC in precipitation was from formic and acetic acids, and that DOC was derived mostly from other sources of organic carbon.

The dry season in the Piracicaba River basin (April through September) is generally associated with the intense burning of sugarcane biomass (Vallis and Keating, 1997, p. 105–113). This burning results in an accentuated buildup of combusted materials in the atmosphere, similar to that caused by biomass-burning in the Amazon basin (Crutzen et al., 1985; Andreae et al., 1988a; Harriss et al., 1988), and is probably responsible for the higher concentrations of solutes measured in the dry season than in the wet season (Table 3). Moreover, the Piracicaba River basin is one of the most populated regions in the state of São Paulo, and particulate concentrations in the atmosphere are commonly greater than those measured in the city of São Paulo (population 13 x 10⁶ in 1991) in the winter months (P. Artaxo, Univ. São Paulo, personal communication, 1999). Particulates are probably responsible for large solute inputs from dry deposition.

Sea salt appears to be an important source of some solutes in precipitation. Assuming that all of the Na⁺ in precipitation was derived from sea salt, and using the molar ratios of Na⁺:K⁺:Ca²⁺:Mg²⁺:Cl^-:SO^2-₄ = 1 : 0.022 : 0.044 : 0.227 : 0.167 : 0.121 (Riley and Chester, 1971; Williams et al., 1997a), we calculated that sea salt accounted for an average of 100, 2, 6, 77, 56, and 1% of the solutes above, respectively. This suggests that the majority of SO^2-₄, K⁺, and Ca²⁺, almost half of the Cl^-, and about 20% of the Mg²⁺ in precipitation are anthropogenic or terrestrial in origin. For instance, there are point sources of HCl inputs to the atmosphere from pulp mill industries in the basin, some of which are located close to Piracicaba (Precipitation Site C) and may be a source of some Cl^- in the precipitation of our study due to scavenging of gaseous HCl (Keene et al., 1986).

Despite the indication that a large portion of some solutes had anthropogenic or terrestrial origins, our data were collected from metropolitan centers and probably overestimated the influence of these factors in the precipitation of more rural areas of the Piracicaba River basin. However, any overestimation appears to be slight since the influence of cyclic salts at the three precipitation sites was similar, and Site A is located in a much smaller urban setting than Sites B and C. This suggests that the factors affecting the regional air mass that regulate precipitation chemistry are similar in different parts of the basin.

Prominent Factors Affecting River Chemistry
Mass balances for each sampling site were calculated to estimate the influence of anthropogenic factors on the aquatic system, and how these rivers process inputs of major solutes. Chemistry of the Piracicaba River and its tributaries may be influenced by a combination of factors, such as agricultural practices (fertilizer applications and seasonal burning of sugarcane), citrus plantations (pesticide applications), pastures (runoff), as well as anthropogenic sources of pollutants in precipitation and runoff (raw sewage and industrial wastes). Point-source pollutants in the Piracicaba River have been shown to decrease O₂ concentrations, and increase respiration rates and CO₂ concentrations in areas bordering heavily developed areas (Ballester et al., 1999). Chronic anaerobic conditions in rivers may result in strong reducing environments that are conducive to sulfate reduction, denitrification, and ammonification processes (Wetzel, 1975).

The net export of major solutes was common and increased with distance downriver due to higher discharge and concentrations of major solutes downriver than those upriver (Fig. 5a,b). Exceptions were observed with NO^-₃, PO^3-₄, and NH⁺₄. There was net retention of NO^-₃ in the headwater basins, but net export occurred downriver. In contrast, the net retention of PO^3-₄ at downriver sites was due to higher PO^3-₄ concentrations in the precipitation of the western Piracicaba River basin (Site C) compared with those of the eastern part of the basin. Ammonium was net-retentive at all sites, suggesting that there was terrestrial immobilization and/or riverine utilization of NH⁺₄. Concentrations of DOC in precipitation were about 30% of those in the rivers, suggesting that there were large inputs of DOC to river water from other sources.

The gross export of most solutes and suspended sediments increased from Sites 1 to 5, decreased at Site 7 relative to Site 5, and increased from Sites 7 to 10 (Fig. 6a,b). The more conservative solutes, such as Ca²⁺ and Na⁺, clearly show this trend, and suggest that inputs from a combination of sources increased the VWM concentrations and export of solutes from upriver to downriver sites. For example, Ca²⁺ concentrations along the main stem from Site 3 to Site 10 had much higher variance than that of SiO₂ and generally increased from upriver to downriver sampling sites (mean Ca²⁺ concentration = 302.4 µeq L^-1; SE = 30.9). The exception was Site 7 (downriver of the Americana Reservoir), where we observed some solute retention.

Ammonium and NO^-₃ concentrations along the main stem had similar trends and variability to those of Ca²⁺. Ammonium had its largest input at Site 5, and subsequently had a large decrease in concentration that cannot be attributed to the differences in concentrations of river water after the mixing of Piracicaba and Jaguari Rivers downstream (Site 7). Because there was a decrease in the concentrations of solutes from Site 5 to Site 7 (Fig. 6a,b), and the Americana Reservoir is just downriver of Site 5 (Fig. 1), we determined if the Americana Reservoir was functioning as a solute sink by conducting an analysis of river mixing. We were able to predict the expected solute concentrations that would occur from Sites 5 to 7 along the main stem of the Piracicaba River due to inputs from the Jaguari River (Table 4) assuming that SiO₂ acts conservatively and there were equal volumes of water at Sites 5 and 6 that mixed to form the water composition at Site 7. The mixture gave a 96% recovery (i.e., the SiO₂ concentration of the mixture was 96% of the measured VWM concentration at Site 7). Most other solutes had slight gains and losses from the SiO₂ reference mixture, although SO^2-₄ had moderate removal (81% recovery), whereas PO^3-₄, NH⁺₄, and suspended sediment fractions had relatively large losses from Sites 5 to 7 (1 to 38% recovery), presumably due to biological uptake (for nutrients) and siltation (for sediments). The utilization of N in the Atibainha Reservoir would also account for the low inorganic N concentrations at Site 1 compared with the other sites.

A decrease in the concentrations of most solutes occurred after the Americana Reservoir, yet there was an additional decrease in NO^-₃ concentrations from Sites 7 to 8 that was not observed with the other solutes. This decrease suggests that either the inputs of NO^-₃ along this river reach were less than other downriver sites, or there was some denitrification. The concomitant increase in the relative export of NH⁺₄ from Sites 7 to 8 suggests that denitrification was coupled with NH⁺₄ inputs from point sources. Anoxic conditions observed at these sampling sites suggest also that the decomposition of organic-rich sewage supported some ammonification (Martinelli et al., 1999).

Sewage treatment in this area of Brazil is primary (i.e., remove particulate organic matter only), and only 4% of raw sewage from the city of Campinas is treated (population 1.5 million; IBGE, 1996). Of the other larger metropolitan areas responsible for raw sewage inputs to the main stem, only the city of Piracicaba treats a large percentage of its wastes (about 60%), albeit treatment is also primary (E. Fischer, Companhia Estadual de Tratamento de Esgoto e Saneamento Básico, personal communication, 1999). We calculated the N inputs that would be expected in the river water between Sites 7 to 8 to infer the relative influence of land use versus raw sewage and industrial wastes. Assuming annual inputs of 3.3 kg of total nitrogen (TN) per person from raw sewage (Meybeck, 1982) and a population of 632000 in the watershed area representing Site 8 (Table 1), inputs to the river from sewage were about 2100 Mg TN per year in 1993. Subtracting the fluvial outputs for Site 8 from Site 7 gives 400 and 1600 Mg of N–NO^-₃ and N–NH⁺₄, respectively. Since 92% of the residences in the Piracicaba River basin are sewered, the amount of inorganic N in river water from raw sewage inputs alone accounts for about 87% of the estimated TN export. Because the BOD per year of industrial wastes was about 20% that of raw sewage at Site 8, industrial wastes probably account for less inorganic N than in raw sewage. Nevertheless, including inorganic N from industrial wastes with the value calculated above for raw sewage, the amount of inorganic N attributable to nonpoint sources at Site 8 was negligible. Since inorganic N is about 50% of TN at Site 8, this indicates that there were high losses of inorganic N via denitrification or utilization between Sites 7 and 8.

Unlike diffuse-source N concentrations, which are inversely related to discharge volume in the upper tributaries of the Piracicaba River (Fig. 4), FSS concentrations are positively related to discharge. Although PO^3-₄ export was small (Fig. 5a), most of total P export was probably in the particulate fraction, and the large export of fine (FSS) and coarse (CSS) suspended solids observed in the Piracicaba River basin suggests that P fluxes associated with particle transport were high also. Since particle transport is generally associated with erosion (Staver et al., 1996), the high export of suspended sediments (and associated P) suggests that there was a large overland flow component to runoff in some watersheds (Jordan et al., 1997a, b).

Discharge–concentration relationships were analyzed to help elucidate the factors regulating solute concentrations during the rising and falling limbs (trajectories) of the discharge hydrographs at different sampling sites. Trajectories for Sites 1, 2, 3, and 8 were used to evaluate the prominent factors upriver and downriver, respectively. At upriver sites, NO^-₃, Na⁺, and Ca²⁺ had similar trajectories over the rising and falling limbs of the discharge hydrograph. Most trajectories at upstream sites were inverse and logarithmic with no obvious hysteresis, indicating that the concentrations of these solutes were primarily regulated by discharge and that there was a relatively constant source of these solutes to river water. In contrast, the trajectories of most solutes at Site 8 had clockwise hysteresis (Fig. 8), indicating that the sources of these solutes were different than those at upriver locations. Gradual changes in inorganic nitrogen concentrations with discharge have been attributed to nonpoint sources in other studies (Chesterikoff et al., 1992; Lewis et al., 1999), and since overland flow and point source inputs create more variability and hysteresis in trajectories, respectively, the primary factor regulating these solute concentrations at upstream sites was probably the dilution of ground water seepage to the river, whereas point sources and urban runoff were probably more influential downriver. The trajectories of other solute concentrations at upriver sites were less definitive, tending toward dilution with increasing discharge. Exceptions were DOC, H⁺, K⁺, and FSS, which had some positive relationships with increasing discharge, especially for FSS, suggesting that overland flow and point sources were more influential than inputs of ground water seepage for these solutes.

View larger version (32K): [in this window] [in a new window]   Fig. 8. Trajectories of the rising and falling limbs of the discharge hydrograph in 1995–1996 for selected solutes at Site 8. Rising limbs of the discharge hydrograph are designated by lines, whereas the falling limbs are designated by dashed lines with open circles. Most trajectories at Site 8 are inverse with clockwise hysteresis.

At Site 8, the discharge–concentration relationships were much different than at upstream sites. In 1995–1996, most solutes had inverse logarithmic relationships with clockwise hysteresis, indicating that there was a stronger source of solutes to the river during rising than falling water. In an undisturbed system, clockwise hysteresis can be caused by exhaustion of a particular solute (Williams and Melack, 1997a). In the heavily disturbed Piracicaba River basin, clockwise hysteresis may instead be caused by large inputs from point sources or overland flow (i.e., urban runoff) during the rising limb of the discharge hydrograph. In contrast, in the 1996–1997 water year, trajectories had counterclockwise hysteresis and most trajectory limbs were not logarithmic. Peak discharge was about a factor of three times less in 1996–1997 than in 1995–1996, and was probably partly responsible for less discernible discharge–concentration relationships. However, the interannual differences also suggest that the timing and magnitude of point source and overland flow inputs to the downriver sites of the Piracicaba River confound discharge–concentration relationships, as we observed with those of NO^-₃ (Fig. 4).

A correlation analysis supported the findings of our trajectory analysis. Spearman correlations of SiO₂ concentrations with those of Na⁺, Ca²⁺, Mg²⁺, and NO^-₃ were positive (Fig. 7) and decreased from upriver sites (1, 3, and 5) to downriver sites (7, 8, and 10), suggesting that these solutes had a proportionally larger ground water component at upriver than at downriver sites. These trends were positively correlated with the percent areas of silviculture and forest for the different subbasins, and inversely correlated with that of sugarcane and BOD (Table 1), which agrees with other studies indicating that a large portion of these solutes is potentially derived from raw sewage or industrial wastes (Bluth and Kump, 1994; Cameron, 1996). The relationship of Na⁺ to Cl^- was significantly positive (p < 0.01), suggesting a common source, and Cl^- and SO^2-₄ had higher positive correlations with SiO₂ concentrations downriver than upriver. Hence, there may be a stronger ground water source of Cl^- and SO^2-₄ associated with the larger percentage of developed areas downriver. Potassium had low correlation coefficients at all locations, suggesting that it is derived from a mixture of sources and that the correlations may be confounded by biological processes, such as assimilation. The correlations of DOC to SiO₂ concentrations were generally weak, suggesting that there was a small ground water source of DOC and that point-source inputs were more influential. This observation is supported by data indicating that raw sewage and industrial waste inputs are substantial in this river system (Table 1). Suspended sediments generally had significant negative relationships with SiO₂ concentrations because these are generally derived from erosion due to overland flow (Staver et al., 1996). Correlations of DOC and suspended sediments with SiO₂ were highly negative and the relationships significant (p < 0.01) at Site 5. For SS, this is probably due to siltation at the upper reaches of the Americana Reservoir that influences this sampling site (Table 4).

The results of our correlation analysis indicate that the relative influence of solute inputs from nonpoint and point sources changed from upriver to downriver. Sources of NO^-₃, Na⁺, Ca²⁺, and Mg²⁺ were more influenced by ground water upriver, whereas concentrations of these solutes were more influenced by point sources downriver. This agrees with our finding that only about 13% of the inorganic N inputs between Sites 7 and 8 were derived from nonpoint sources. In contrast, ground water seepage appears to make a proportionally larger contribution to Cl^- and SO^2-₄ concentrations at downriver than at upriver sites. Dissolved organic carbon and K⁺ concentrations had poor correlations with SiO₂, suggesting that point sources of these solutes and/or the interplay of biological processes are generally more influential than nonpoint sources both upriver and downriver.

Comparison of the Piracicaba River with a Temperate System
There are fundamental differences in hydrology between subtropical and temperate systems. For instance, the tributaries of the Chesapeake Bay receive large inputs of solutes during the spring due to a combination of snowmelt and rainfall, although rainfall is evenly distributed throughout the year and has less of an effect than snowmelt. The flushing of ground water in the spring can accentuate the leaching of overwintering products, such as nutrients and major solutes, and agrochemicals to receiving waters (Staver and Brinsfield, 1990, 1996). Nutrient inputs to most of the tributaries of the Chesapeake Bay are currently more a function of nonpoint runoff from agriculture and animal husbandry (Boynton et al., 1995; Jordan et al., 1997a,b; Fisher et al., 1998). For example, the Chesapeake Bay watershed contributes approximately two-thirds of the N and one-quarter of the P to Chesapeake Bay (Correll, 1987), and agriculture is the main source of N discharge from the watershed (Jaworski et al., 1992). In contrast, the Piracicaba River basin has discrete wet and dry seasons (Fig. 2), and there is extensive burning of sugarcane during the dry season.

Depending on the depth of the water table, intact riparian buffer systems reduce nutrient and solute fluxes in some tributaries of the Chesapeake Bay (Jordan et al., 1997a,b; Lowrance et al., 1997). Although there is presently no documentation for the Piracicaba River basin, we speculate that riparian corridors function similarly by reducing terrestrial nutrient fluxes to receiving waters. However, soils of the Piracicaba River basin are generally deeper and have larger hydraulic conductivities than those of the Chesapeake Bay watershed. Hence, there may be larger leaching losses of NO^-₃ and other major solutes in the Piracicaba River basin than in the Chesapeake Bay watershed because ground water may circumvent hyporheic areas that are more effective at nutrient uptake than deeper areas of riparian zones (Jordan et al., 1997a,b).

In the Chesapeake region, corn (Zea mays L.) is usually fertilized at 150 to 250 kg N ha^-1 yr^-1 and 20 to 50 kg P ha^-1 yr^-1 (T. Fisher, Univ. Maryland, personal communication, 1999). Recommended N inputs from fertilizer for sugarcane, citrus, and silviculture for the state of São Paulo are about 80, 120, and 6 kg ha^-1 yr^-1, respectively, albeit application appears to be sporadic because of the high cost of fertilizers. Nitrogen inputs to pasture were assumed to be negligible because fertilization of pasture is generally not done in Brazil. Hence, the bulk application rates of fertilizers in agricultural areas of the Chesapeake Bay are much higher than those of the Piracicaba River basin. Lower rates of fertilizer application in the Piracicaba River basin than in temperate zone agricultural watersheds suggest that leaching losses to ground water are much smaller in the Piracicaba River basin. Although larger leaching losses of solutes may occur in some areas of the Piracicaba River basin than in the Chesapeake Bay watershed, nutrient leaching to ground water from sugarcane monocultures is small (Vallis and Keating, 1997, p. 105–113). The leaching of N from fertilizers in sugarcane and citrus plantations have been estimated to be about 5% of the N applied, and that from silviculture areas is negligible (Lima et al., 1999). Small leaching losses may be attributed to the greater immobilization efficiency of inorganic N and K⁺ by sugarcane compared with crops such as corn, and smaller applications of fertilizer to agroecosystems of the Piracicaba River basin than those used in the Chesapeake Bay watershed. The volatilization and denitrification of N in fertilizers can also reduce leaching losses to ground water (Vallis and Keating, 1997, p. 105–113).

Small leaching losses of solutes to receiving waters from agricultural areas in the Piracicaba River basin have management implications. In the Chesapeake Bay, anthropogenic increases in watershed inputs of both N and P have lead to excessive plankton production (Boynton et al., 1982; Malone et al., 1988; Jordan et al., 1991), which has contributed to the demise of submerged aquatic vegetation (SAV) and an increase in the extent of hypoxic waters (Officer et al., 1984). Unlike the Chesapeake Bay watershed, agriculture and pasture are apparently not responsible for much of the degradation of water quality observed in the Piracicaba River during the last several decades. Instead, the factors responsible for the high inorganic N and DOC export downriver, and most of the other solutes measured in this study, are the point-source inputs of raw sewage and industrial wastes dumped into the Piracicaba River each year. Hence, improving the water quality of the Piracicaba River basin may not require a reduction of nonpoint nutrient fluxes to receiving waters, as has been the case in the Chesapeake Bay watershed (Staver et al., 1996). Rather, reducing inputs of industrial effluents and raw sewage by augmenting waste treatment will be the most effective short-term approach to improve the water quality of receiving waters.

Nevertheless, quantifying the relative inputs of various solute (viz. N and P) delivery pathways on riverine water quality will be critical for developing effective strategies for reducing anthropogenic solute inputs to the Piracicaba River basin and its tributaries. In addition to continued monitoring of precipitation and river chemistry, this will require (i) reliable measurements of flow and chemical data from all point sources, (ii) data of leaching losses from the major land use types in the basin, (iii) research on the implementation and effectiveness of erosion control practices, and (iv) studies to determine the effectiveness of riparian corridors in mitigating diffuse solute inputs to receiving waters.

	ACKNOWLEDGMENTS

 
We thank J. Moraes for assistance with obtaining hydrological data, A. Krusche for coordinating sampling and lab analyses of river chemistry, and T. Fisher, J. Melack, and one anonymous reviewer for helpful comments. This research was partially funded by the following grants: 94.0529-9 and 98.10269-5 (FAPESP), 62.0363/92.4 (CNPq/PADCT), and INT-9901158, DEB-9726862, and OCE-9726921 (NSF).

	REFERENCES

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION REFERENCES

 

• American Public Health Assocation. 1976. Standard methods for the examination of water and wastewater. 14th ed. APHA, Washington, DC.
• Andreae, M.O., E.V. Browell, M. Garstang, G.L. Gregory, R.C. Harriss, G.F. Hill, D.J. Jacob, M.C. Pereira, G.W. Sachse, A.W. Setzer, P.L. Silva Dias, R.W. Talbot, A.L. Torres, and S.C. Wofsy. 1988a. Biomass-burning emissions and associated haze layers over Amazonia. J. Geophys. Res. 93:1509–1527.
• Andreae, M.O., R.W. Talbot, T.W. Andreae, and R.C. Harriss. 1988b. Formic and acetic acid over the central Amazon region, Brazil. 1. Dry Season. J. Geophys. Res. 93:1616–1624.
• Andreae, M.O., R.W. Talbot, and S. Li. 1987. Atmospheric measurements of pyruvic and formic acids. J. Geophys. Res. 92:6635–6641.
• Ballester, M.V., L.A. Martinelli, A.V. Krusche, R.L. Victoria, M. Bernardes, and P.B. Carmago. 1999. Effects of increasing organic matter loading on the dissolved O₂, free dissolved CO₂ and respiration rates in the Piracicaba River basin, southeast Brazil. Water Res. 33:2119–2129.
• Bluth, G.J.S., and L.R. Kump. 1994. Lithologic and climatologic controls of river chemistry. Geochim. Cosmochim. Acta 58:2341–2359.
• Boynton, W.R., J.H. Garber, R. Summers, and W.M. Kemp. 1995. Inputs, transformations, and transport of nitrogen and phosphorus in Chesapeake Bay and selected tributaries. Estuaries 18:285–314.
• Boynton, W.R., W.M. Kemp, and C.W. Keefe. 1982. A comparative analysis of nutrients and other factors influencing estuarine phytoplankton production. p. 69–90. In V. Kennedy (ed.) Estuarine comparisons. Academic Press, San Diego, CA.
• Cameron, E.M. 1996. Hydrogeochemistry of the Fraser River, British Columbia: Seasonal variation in major and minor components. J. Hydrol. 182:209–225.
• Chan, W.H., A.J.S. Tang, D.W.S. Chung, and N.W. Reid. 1987. An analysis of precipitation chemistry measurements in Ontario. Environ. Sci. Technol. 21:1219–1224.
• Chesterikoff, A., B. Barba, G. Billen, and M. Poulin. 1992. Inorganic nitrogen dynamics in the River Siene downstream from Paris (France). Biogeochemistry 17:147–164.
• Cogbill, C.V., and G.E. Likens. 1974. Acid precipitation in the northeastern United States. Water Resour. Res. 10:1133–1137.
• Correll, D.L. 1987. Nutrients in Chesapeake Bay. p. 298–319. In S.K. Majumdar et al. (ed.) Contaminant problems and management of living Chesapeake Bay resources. Pennsylvania Acad. of Sci., Philadelphia.
• Correll, D.L., T.E. Jordan, and D.E. Weller. 1992. Nutrient flux in a landscape—Effects of coastal land-use and terrestrial community mosaic on nutrient transport to coastal waters. Estuaries 15: 431–442.
• Crutzen, P.J., A.C. Delany, J. Greenberg, P. Haagenson, L. Heidt, R. Lueb, W. Pollock, W. Seiler, A. Wartburg, and P. Zimmerman. 1985. Tropospheric chemical composition measurements in Brazil during the dry season. J. Atmos. Chem. 2:233–256.
• Depetris, P.J., and S. Kempe. 1993. The impact of the El Niño 1982 event on the Paraná River, its discharge and carbon transport. Palaeogeogr. Palaeoclimato. Palaeoecol. 89:239–244.
• Edmond, J.M., M.R. Palmer, C.I. Measures, B. Grant, and R.F. Stallard. 1995. The fluvial geochemistry and denudation rate of the Guayana Shield in Venezuela, Colombia, and Brazil. Geochim. Cosmochim. Acta 59:3301–3325.[ISI]
• Filoso, S., M.R. Williams, and J.M. Melack. 1999. Composition and deposition of throughfall in a flooded forest archipelago (Negro River, Brazil). Biogeochemistry 45:169–195.
• Fisher, T.R., K.Y. Lee, H. Berndt, J.A. Benitez, and M.M. Norton. 1998. Hydrology and chemistry of the Choptank River basin. Water Air Soil Pollut. 105:387–397.
• Gran, G. 1950. Determination of the equivalent point in potentiometric titrations. Acta Chem. Scan. 4:559–577.
• Gran, G. 1952. Determination of the equivalent point in potentiometric titrations: Part 2. Analyst 77:661–671.
• Greaves, R.C., R.M. Barkley, R.E. Sievers, and R.R. Meglen. 1987. Covariations in the concentrations of organic compounds associated with springtime atmospheric aerosols. Atmos. Environ. 21: 2549–2561.
• Hamilton, S.K., and W.M. Lewis, Jr. 1990. Basin morphology in relation to chemical and ecological characteristics of lakes on the Orinoco River floodplain, Venezuela. Arch. Hydrobiol. 119:393–425.
• Hamilton, S.K., S.J. Sippel, and J.M. Melack. 1995. Oxygen depletion and carbon dioxide and methane production in waters of the Pantanal wetland of Brazil. Biogeochemistry 30:115–141.
• Harriss, R.C., S.C. Wofsy, M. Garstang, E.V. Browell, L.C.B. Mollon, R.J. McNeal, J.M. Hoell, R.J. Bendura, S.M. Beck, R.L. Navarro, J.T. Riley, and R.L. Snell. 1988. The Amazon boundary layer experiment (ABLE-2A): Dry season 1985. J. Geophys. Res. 93:1351–1360.
• IBGE–Fundação Instituto Brasileiro de Geografia e Estatística. 1996. Atlas Nacional do Brasil. Directoria de Geociências. 2nd ed. Rio de Janeiro.
• Isidorov, V.A., I.G. Zenkevich, and B.V. Ioffe. 1985. Volatile organic compounds in the atmosphere of forests. Atmos. Environ. 19:1–8.
• Jaworski, N.A., P.M. Groffman, A.A. Keller, and J.C. Prager. 1992. A watershed nitrogen and phosphorus balance: The upper Potomac River basin. Estuaries 15:83–95.
• Jordan, T.E, D.L. Correll, J. Miklas, and D.E. Weller. 1991. Nutrients and chlorophyll at the interface of a watershed and an estuary. Limnol. Oceanogr. 36:251–267.
• Jordan, T.E., D.L. Correll, and D.E. Weller. 1997a. Effects of agriculture on discharges of nutrients from coastal plain watersheds of Chesapeake Bay. J. Environ. Qual. 26:836–848.[Abstract/Free Full Text]
• Jordan, T.E., D.L. Correll, and D.E. Weller. 1997b. Relating nutrient discharges from watersheds to land use and streamflow variability. Water Resour. Res. 33:2579–2590.
• Kawamura, K., and I.R. Kaplan. 1986. Biogenic and anthropogenic organic compounds in rain and snow samples collected in Southern California. Atmos. Environ. 20:115–124.
• Keene, W.C., and J.N. Galloway. 1986. Considerations regarding sources for formic and acetic acids in the troposphere. J. Geophys. Res. 91:14466–14474.
• Keene, W.C., A.A.P. Pszenny, J.N. Galloway, and M.E. Hawley. 1986. Sea-salt corrections and interpretation of constituent ratios in marine precipitation. J. Geophys. Res. 91:6647–6658.
• Krusche, A.V., F.P. Carvalho, J.M. Moraes, P.B. Camargo, M.R.V. Ballester, S. Hornink, L.A. Martinelli, and R.L. Victoria. 1997. Spatial and temporal water quality variability in the Piracicaba River basin, Brazil. J. Am. Water Resour. Assoc. 33:1117–1123.
• Lesack, L.F.W., and J.M. Melack. 1991. The deposition, composition, and potential sources of major ionic solutes in rain of the central Amazon basin. Water Resour. Res. 27:2953–2977.
• Lewis, W.M., Jr., J.M. Melack, W.H. McDowell, M. McClain, and J. Richey. 1999. Nitrogen yields from undisturbed watersheds in the Americas. Biogeochemistry 46:149–162.
• Likens, G.E., and F.H. Bormann. 1995. Biogeochemistry of a forested ecosystem. 2nd ed. Springer–Verlag, Berlin.
• Likens, G.E., R.F. Wright, J.N. Galloway, and T.J. Butler. 1979. Acid rain. Sci. Am. 241:43–51.
• Lima, W.P, M.J. Zakia, C.D. Camara, F.F.B. Ferraz, and C.H.G. Cezare. 1999. Indicators of soils and water conservation for sustainable forest management in the neotropics. In Conf. on Sust. For. Manage. in the Neotropics, Turrialba, Costa Rica. 9–12 Nov. 1999.
• Linsley, R.K., M.A. Kohler, and J.L.H. Paulhus. 1982. Hydrology for engineers. McGraw–Hill, New York.
• Lowrance, R., L.S. Altier, J.D. Newbold, R.R Schnabel, P.M. Groffman, J.M. Denver, D.L. Correll, J.W. Gilliam, J.L. Robinson, R.B. Brinsfield, K.W. Staver, W.L. Lucas, and A.H. Todd. 1997. Water quality functions of riparian forest buffers in Chesapeake Bay watersheds. Environ. Manage. 21:687–712.