Land use and topography as predictors of nutrient levels in a tropical catchment

Limnologica 40 (2010) 322–329

Contents lists available at ScienceDirect

Limnologica journal homepage: www.elsevier.de/limno

Land use and topography as predictors of nutrient levels in a tropical catchment Marı´a M. Castillo  ´n Bolı´var, Apartado 89000 Baruta, Caracas 1080, Venezuela Departamento de Estudios Ambientales, Universidad Simo

a r t i c l e in f o

a b s t r a c t

Article history: Received 19 November 2008 Received in revised form 11 August 2009 Accepted 30 September 2009

The influence of landscape on nutrient concentration and yield was analyzed in a tropical catchment, the Guare River in northern Venezuela. Spatial and temporal variation in nitrate, SRP and total P were determined in 15 sites located along the river mainstem and tributaries. Higher nitrate concentrations and yields were reported from upper sites and both decreased in the downstream direction along the river mainstem. These trends appear to be related to more pronounced slopes and larger proportions of agricultural and forest lands in subcatchments located in the upper part of the basin, and dense algal mats in the lower reaches. Nitrate values were higher during periods of high discharge, suggesting that nitrate is primarily transported by shallow subsurface flow. SRP represented between 60 and 80% of total P. Phosphorus concentrations were homogeneous along the river mainstem and showed little seasonal variation, while yields were higher in the upper basin. Multiple regression identified slope, runoff and agriculture as primary predictors of nitrate and phosphorus across the watershed, which suggests that both natural and anthropogenic landscape characteristics have a strong influence on nutrient levels in the Guare catchment. & 2009 Elsevier GmbH. All rights reserved.

Keywords: Tropical stream Nitrate Phosphorus Land use Nutrient Topography Runoff

Introduction Nutrients transported by rivers have a profound effect on water quality, habitat condition and trophic state of aquatic ecosystems. In streams, increased nutrient inputs can enhance algal growth which can lead to hypoxia, alter the structure of food webs and affect environmental services such as water supply and recreation (Dodds 2006; Mallin et al. 2006). In temperate streams, changes in land use and nutrient loading are strongly correlated (Jordan et al. 1997; Boyer et al. 2002; Filoso et al. 2003; Brodie and Mitchell 2005). Agriculture, atmospheric deposition and point source inputs from human population in the catchment have been identified as strong predictors for N exports (Howarth et al. 1996; Jones et al. 2001; Kemp and Dodds 2001), while discharges from urban centers and wastewater treatment plants and runoff from agricultural fields, can represent important sources of P (David and Gentry 2000, Little et al. 2003; Hart et al. 2004). Other variables such as riparian cover, catchment geology and topography can also influence nutrient loading (Castillo et al. 2000; Ekholm et al. 2000; Jones et al. 2001). Several studies have investigated nutrient transport in undisturbed tropical streams and rivers (Lewis 1986; McDowell and Asbury 1994; Castillo 2000; Saunders et al. 2006), but information on impacted tropical

 Tel.: + 58 212 9063070.

E-mail address: [email protected] 0075-9511/$ - see front matter & 2009 Elsevier GmbH. All rights reserved. doi:10.1016/j.limno.2009.09.003

systems is limited. Streams located in the humid tropics can show a wide range of physical and biological features (Boulton et al. 2008); however, they are mostly located in developing countries, where socioeconomic factors and conservation and management practices can result in different impacts of land use changes on streams compared to those observed in developed nations (Ramı´rez et al. 2008). In developed countries, most efforts have focused on controlling the input from point sources through wastewater treatment implementation and phosphate detergent bans (Carpenter et al. 1998; Litke 1999), and as a result diffuse sources remain as the main source of pollution (Hart et al. 2004; Royer et al. 2004). To reduce the export of nutrients from diffuse sources, particularly from cropland, measures such as conservation tillage, fertilizer management and conservation of highly erodible land have been proposed (Baker and Richards 2002; Moog and Whiting 2002). These practices have contributed to a reduction in N and P loading in some catchments of North America and Europe (Baker and Richards 2002; Kronvang et al. 2005). In many developing countries, where wastewater management has not been implemented, both diffuse and point source pollution constitute major nutrient sources. In Latin America, about 73% of the population lives in cities and less than 14% of the municipal wastewater receives treatment (Howarth and Ramakrishna 2005; CEPAL 2006). In addition, deforestation is increasing in many developing regions, with land being converted primarily to agricultural use (F.A.O. 2007). Although N fertilizer use in Latin

M.M. Castillo / Limnologica 40 (2010) 322–329

America represents only 6.3% of the world consumption compared to 13.5% of North America, it is expected to increase at an annual rate of 2.4% in the next four years (F.A.O. 2008). In tropical regions, such a change in land use can have a higher impact than in less humid areas because the magnitude and duration of precipitation can increase nutrient losses from the land as well as the vulnerability of soils to erosion (Pringle et al. 2000). In addition, implementation of regulations and conservation practices are usually poor, increasing the impacts of human activities (Ramı´rez et al. 2008). Since changes in land use in these regions probably will result in greater impacts on nutrient loading to streams, it is important to understand how land use at the catchment level influences nutrient export and concentration in streams. In this study, the relationship between landscape variables and spatial patterns in nitrate and phosphorus were examined in the Guare catchment in Venezuela. Although a large proportion of the study catchment is covered by forests, human activities have disturbed much of the upper basin, influencing nutrient patterns along the river mainstem and some tributaries. By using multiple linear regression, percent agriculture in the catchment, runoff and slope were identified as predictors of nitrogen and phosphorus levels, suggesting that land use and topography can influence nutrient export in the Guare catchment.

Study area The Guare River is a tributary of the Tuy River, which drains into the Caribbean Sea in northern Venezuela. The Guare basin has a humid tropical climate with a dry season from October to March and a rainy season from April to September. The Guare catchment covers an area of 182 km2 and it is located between the interior and the coastal branches of the Coast Mountain Range, exhibiting a hilly topography. The study watershed is heterogeneous in terms

323

of land use, showing different types of vegetation (forest and shrubland), crops (horticulture and pasture) and livestock (cattle and poultry farms) (Carnemolla et al. 1990). Although the watershed is mostly forested (40%), agriculture and pasture represent 10% and 12% of the total area, respectively. Agriculture is primarily concentrated in the upper watershed, representing up to 45% of the area of some headwater catchments. Nickel mining is conducted in the catchment of the Mesia River, one of the main tributaries of the Guare. In addition, the small urban centers of ˜ a, Las Dolores and San Daniel Ta´cata, Altagracia de la Montan account for a total population of 6900 (I.N.E. 2001).

Methods Fifteen sites were selected to cover spatial variation in land use in the study watershed. The sites were visited between September 2001 and November 2002 to follow seasonal variation in discharge. Sampling was conducted in September and November 2001 and January, March, May, June, July, September and November 2002. All sites were sampled at the same date. Eight sites were located on the mainstem of the Guare River and seven sites on the tributaries Las Dolores, Agua Frı´a, Quebrada Seca, Emilia and Mesia (Fig. 1). Mainstem sites Guare 1 and 2 represented the conditions prevailing in the upper part of the basin, where agriculture is mostly concentrated, particularly at Las Dolores catchment (Table 1). In contrast, Emilia is less cultivated but a large proportion (74%) of the Guare population is located in this catchment. In Emilia, three sites were sampled so as to detect the influence of urban centers such as Bagre and Altagracia de la ˜ a, and the inputs of tributaries draining the west side of Montan the basin, like Quebrada Seca. The Mesia catchment is primarily covered by scrubland and secondary vegetation. Agriculture and pastureland are observed along the Guare mainstem.

Fig. 1. The catchment of the Guare River, Venezuela. Sampling sites at tributaries and mainstem of the river are shown.

324

M.M. Castillo / Limnologica 40 (2010) 322–329

Table 1 Landscape characteristics for the catchment above each sampling site. Site

Catchment area (km2)

Population (people km

2

)

Buildings (no. km

2

)

Forest (%)

Cropland (%)

Average slope (%)

Runoff (mm d

Tributary sites Las Dolores Agua Frı´a Quebrada Seca Emilia 1 Emilia 2 Emilia 3 Mesia

10.4 7.5 7.9 18.2 44.5 59.6 55.0

51.1 1.3 0.9 6.3 11.0 56.1 0.2

12.2 0.8 0.8 1.3 3.6 18.3 0.3

45.5 64.8 24.2 67.8 45.8 44.8 29.2

44.5 22.2 13.8 11.2 8.8 10.8 0.5

23.7 27.5 20.2 21.5 20.1 22.0 20.5

1.33 1.57 0.12 0.86 0.40 0.51 0.31

Mainstem sites Guare 1 Guare 2 Guare 3 Guare 4 Guare 5 Guare 6 Guare 7 Guare 8

30.8 35.1 94.9 105.2 115.7 116.3 120.6 175.6

28.2 24.7 44.4 42.0 38.5 38.8 37.5 25.8

7.2 6.3 13.8 13.3 12.3 12.2 11.8 8.2

53.2 51.0 47.0 44.2 44.2 44.0 44.8 39.8

29.9 26.8 16.8 15.3 15.0 15.0 14.6 10.2

22.9 19.2 20.2 19.9 20.5 20.5 25.4 22.1

1.05 0.33 0.50 0.47 0.44 0.44 0.42 0.38

Stream networks in the Guare River area were digitized from 1:25,000 topographic maps (Direccio´n de Cartografı´a Nacional, Repu´blica de Venezuela), which were used to outline the catchment area corresponding to each sampling site. Land use and cover were classified from digitized orthophotographs at 1:25,000 based on aerial photographs from 1994 (Servicio Auto´nomo de Geografı´a y Cartografia Nacional, Repu´blica de Venezuela). The following land use categories were employed: forest, scrubland, grassland, agroforestry land (temporarily abandoned land covered by secondary vegetation), cropland and urban land. Slope was derived from a digital elevation model. Five slope categories were used: 0–10, 11–20, 21–35, 36–50 and greater than 50%. Population and number of buildings were obtained from the 2000 national census (I.N.E. 2001). By using a GIS, land use, slope and population data were determined for the catchment area upstream from each sampling site. At each site, discharge was estimated from measurements of depth and current velocity (Swoffer 2100 flowmeter) at eight to ten points at each site cross section. Current velocity was determined at 40% of depth measured from the streambed. Temperature and specific conductance were measured at each site (Orion 122 conductivity meter). Grab water samples were collected at each site and transported on ice to the laboratory. Nitrate (plus nitrite) concentrations were analyzed by the cadmium reduction method (APHA 1995). As preliminary data indicated that nitrite concentrations were low ( o2 mg L 1), hereafter nitrate plus nitrite is referred to as nitrate. Soluble reactive phosphorus (SRP) was determined in 0.45 mm membranefiltered water samples by the ascorbic acid method (APHA 1995). Total phosphorus was determined in unfiltered water samples by digestion with persulphate to SRP (APHA 1995). For each sampling site and date, instantaneous loads (mg s 1) were calculated as the product of concentration and discharge. Daily yields (g ha 1 d 1) were calculated from instantaneous loads (mg s 1) converted to daily loads (g d 1), which were divided by catchment area (ha). Annual yields and loads could not be estimated because discharge records were not available for the study area. Stepwise multiple linear regressions with forward selection of variables were used to analyze the relationship of land use with nutrient concentrations and export. The dependent variables were concentration average and daily yield of nitrate, SRP and total P at each sampling station. Averages were calculated for the September 2001–November 2002 sampling period. The independent variables were land use, slope, runoff, population and building density in the watershed upstream from each sampling site. The

1

)

percent of the catchment under the land use categories and slope intervals described above were considered. In addition, the average slope was also calculated for each subcatchment. Specific runoff (mm d 1) was calculated from discharge estimates based on data collected in this study and catchment area, since discharge records were not available for the Guare River or its tributaries. Correlations and bivariate plots between spatial and nutrient variables were used to select the independent variables to be included in the stepwise multiple regression analyses. Variables that showed a significant correlation coefficient (Pearson’s) and a linear trend in the bivariate plots were included in the regression analysis. The regression analyses were conducted with two data sets: (1) all sampling sites included to examine the relationships between land use and nutrients across the watershed and (2) only the nine sampling sites located on the mainstem of the river. Logarithmic transformations were conducted to meet the normality assumption in some cases. The significance level used was 0.05. Plots of the residual against predicted values were used to test for homogeneity while normality was tested by cumulative probability plots and Kolmogorov–Smirnov test of the residuals. Durbin–Watson test was used to test for autocorrelation of the residuals.

Results Nitrate concentrations varied from 8 to 880 mg L 1. Higher values were reported in the upper basin tributary sites Quebrada Seca, Agua Frı´a and Las Dolores, where cropland represents a large proportion of the catchment (14–45%) and poultry farms are present (Table 1). The lowest concentrations were observed in the Mesia River, where the proportion of cropland is low (0.5%). Values decreased in the downstream direction, so that the concentration at mainstem site Guare 8 was, on average, onehalf of that at site 1. Differences in concentration between the upper and lower parts of the basin sites were more evident during the dry than during the wet season (Fig. 2A). Concentrations showed a marked seasonal variation, increasing in the rainy period and decreasing during the dry season at all sampling sites except Quebrada Seca, which most likely received sewage inputs from poultry farms (Fig. 3). At a mainstem site (Guare 8), significant and strong correlations were observed between discharge and nitrate concentration (r = 0.92 p o0.001), and between discharge and yield (r =0.97 p o0.001). Nitrate concentrations showed a higher seasonal variation at sites

M.M. Castillo / Limnologica 40 (2010) 322–329

325

Nitrate-N (µg L-1)

1000 dry season wet season

800 600 400 200

La

s

D ol o Ag res ua Fr ia G ua re Em 1 Q ue ili br ad a 1 a Se c Em a ilia 2 G ua re Em 2 ilia 3 G ua re 3 G ua re 4 G ua re 5 G ua re 6 G ua re 7 M es ia G ua re 8

0

Total P (µg L-1)

300 dry season wet season

200 100

Em 1 il ad ia 1 a Se ca Em ilia 2 G ua re Em 2 ilia 3 G ua re 3 G ua re 4 G ua re 5 G ua re 6 G ua re 7 M es ia G ua re 8

ia

Q

ue

br

re

Fr

ua G

ua Ag

La

s

D

ol

or

es

0

Nitrate-N (µg L-1)

Fig. 2. Nitrate (A) and Total P (B) average concentrations for the dry (January–May 2002) and wet (September–December 2001 and June–November 2002) seasons at each sampling site. Error bars represent one standard deviation.

Guare 1 Guare 8 Agua Fria Emilia 1 Quebrada Seca Las Dolores Mesia

1000 900 800 700 600 500 400 300 200 100 0 S O N D 2001

J

F M A M J 2002

J

A S O N

Fig. 3. Seasonal variation in nitrate concentration at some mainstem and tributary sites. Rı´o Mesia could not be sampled during September 2002 due to flood conditions.

located in the lower reaches as compared with the upper basin tributaries (Fig. 2A). Average nitrate load decreased in the downstream direction during the dry season while it showed only a 30% increase between mainstem sites Guare 1 and 8 during the wet season, despite the twofold increase in discharge (Fig. 4A). Load estimates indicated that the upper basin tributaries Agua Fria and Las Dolores provided 68% of the nitrate transported by the river at site Guare 8, indicating the importance of that portion of the basin as a source of nitrate. The sharp decrease in load between sites 1 and 2 probably was related to water withdrawal for irrigation. Nitrate carried by Rı´o Emilia (30% of the load transported at site 8) explained the load increase between mainstem sites 2 and 3 (Fig. 4A). During the dry season, nitrate load at main stem site 3 (18.9 mg s 1) was 5 times greater than nitrogen load at site 8

(3.6 mg s 1). Increase in nitrate load at site 6 during the wet season was probably related to the inputs of Quebrada Palma, an intermittent tributary of the Guare. Input of Rı´o Mesia during the wet season was proportional to increase in load between site 7 and 8; however, loads decrease between these sites during the dry season despite the nitrate contribution (0.45 mg s 1) of the Mesia. Nitrate yields ranged 0.12–39 g ha 1 d 1 among sampling sites and reached their highest values during the rainy season, particularly in September 2002. Highest average yields were registered at the catchments of Las Dolores (7.39 g ha 1 d 1) and Agua Frı´a (9.31 g ha 1 d 1). Lowest values occurred at Rı´o Mesia (0.12 g ha 1 d 1). The Guare at site 8 showed an average yield of 1.22 g ha 1 d 1. Soluble reactive phosphorus (SRP) concentrations ranged from 10 to 315 mg L 1 while total P concentrations varied from 14 to 386 mg L 1. SRP and total P values showed similar temporal and spatial patterns. P concentrations were homogeneous along sites located on the mainstem (Fig. 2B). The tributaries Agua Frı´a and Las Dolores showed the highest values while the average concentration at Emilia 1 was lower than in the rest of the sites. However, concentrations at this tributary doubled between sites 1 and 3. SRP and total P concentrations increased during the dry season and at the beginning of the rainy season (May–June) (Fig. 5). During a September 2002 storm, P concentrations, particularly SRP, decreased markedly. SRP represented between 64 and 80% of total P and this proportion was relatively constant among sampling sites and dates. However, this proportion decreased below 30% during the September 2002 storm. In contrast to nitrate, both SRP and total P loads showed a gradual increase along the mainstem sites during the wet season, which was consistent with the twofold increase in discharge and the low variation in P concentrations along the mainstem (Fig. 4B). As for nitrate, decrease in P loads during the dry season at site Guare 2 was probably related to water withdrawal for irrigation. Total P load increase between mainstem sites 2 and 3

500

30

400

25 20

Wet season Dry season

300

15

200

10

100

5

0

Nitrate Load Dry season (mg s-1)

M.M. Castillo / Limnologica 40 (2010) 322–329

Nitrate load wet season (mg s-1)

326

0 1

3

2

4

5

6

7

8

Mainstem site 15

80 10

60 40

5

Wet season Dry season

20 0

TP load dry season (mg s-1)

TP load wet season (mg s-1)

100

0 1

3

2

5 4 Mainstem site

7

6

8

Fig. 4. Nitrate (A) and total phosphorus (B) average loads for the dry and wet seasons at the mainstem sites.

3.0 SRP Total P Q

140 120

2.5 2.0

100 80

1.5

60

1.0

Q (m3s-1)

P concentrations (µgL-1)

160

40 0.5

20 0

0.0 S O 2001

N

D

J

F M 2002

A

M

J

J

A

S

O

N

Fig. 5. SRP and total P concentrations at main stem site Guare 8. The dotted line represents the discharge (Q) at the same site.

was proportional to inputs from Rı´o Emilia, which contributed on average with 35.8 and 7.0 mg P s 1 during the wet and dry season, respectively. Input from the Mesia was also evident between sites Guare 7 and Guare 8, and the increase in load was proportional to the tributary input. Similar amounts of phosphorus were found to be transported at the Guare mainstem site 2 and Rio Emilia site 3, and the sum of their inputs at mainstem site 3 represented 76% of the P transported by the river at site 8, indicating that most of the P transported by the Guare was supplied by the upper basin. Average SRP daily yields ranged from 0.07 to 1.12 g ha 1 d 1 and total P yields were between 0.11 and 1.68 g ha 1 d 1. Higher values were observed during the rainy season and at Agua Frı´a (1.68 g ha 1 d 1) and Guare 1 (0.72 g ha 1 d 1). Lower average yields were registered at Mesia (0.11 g ha 1 d 1) and Guare 8 (0.30 g ha 1 d 1). Slope, runoff, cropland and forest cover explained a large proportion of the variation in N and P, as indicated by the multiple

regression analyses (Table 2). The models for nitrate concentration suggested that catchments with slopes below 10% showed lower concentrations, probably related to the occurrence of higher nitrate levels in the upper part of the basin, where slopes are more pronounced ( 450%). Runoff explained a large proportion (87%) of the variation in nitrate yields indicating that catchments with higher runoff have higher nitrate yields. If Quebrada Seca was removed from the analysis, a high correlation was also obtained between nitrate concentration and runoff (r =0.840, po0.0001). Although the percent of forest explained much of the variation (91%) in nitrate concentrations among mainstem sites and was the variable included in the regression model, a significant correlation (Pearson, r =0.89, po0.0001) was also obtained between the proportion of the catchment covered by cultivated land and nitrate concentration at mainstem sites. In addition, the percent of forest and agricultural land at mainstem sites also showed a high correlation (Pearson, r= 0.977, p o0.0001), so it is not clear if the

M.M. Castillo / Limnologica 40 (2010) 322–329

Table 2 Multiple linear regression models between landscape variables and mean nitrate and phosphorus concentrations (C) and daily yields (Y). Dependent variables NITRATE All sampling sites Concentration Yield

Main stem sites Concentration Yield PHOSPHORUS All sampling sites Total P yield

SRP yield

Model

R2 change

C= 781.880–14.080*S10 R2 = 0.518 LnY= 0.896 Ln runoff + 0.646 LnAg R2 = 0.981

S10: 0.518 Ln runoff: 0.866 Ln Ag: 0.115

C= –405.500+ 15.041* FOR R2 = 0.909 LnY= 0.471+ 0.072*Ag R2 = 0.989

FOR: 0.909 Ag: 0.989

Y= 1.520+ 0.085*S_aver+0.078*S50 R2 = 0.872

S_aver: 0.708

Y= 1.110+ 0.058*S_aver+0.054*S50 R2 = 0.860

S50: 0.164 S_aver: 0.787 S50: 0.073

Mainstem sites Total P yield SRP yield

Y= 0.089+ 0.022*Ag R2 =0.983 Y= 0.079 + 0.0093*Ag R2 = 0.931

Ag: 0.983 Ag: 0.931

Analyses were conducted for two data sets: all sampling sites and mainstem sites. No significant correlations between landscape variables and phosphorus concentrations were obtained, so regression analysis was not performed for these variables. Variables included in the models are the percent of land in the catchment with slopes between 0 and 10% (S10), the average catchment slope (S_aver), the percent of land in the catchment with slopes greater than 50% (S50), the percent of cropland (Ag) and the percent of forest (FOR). R2 changes represent the amount of variation explained by each variable included in the model.

role of the forest was causal or if it was related instead to its position in the upper part of the catchment where agriculture also covers a large proportion of the landscape. Results indicated that agricultural lands have a strong influence on nitrate yields among mainstem sites. No significant correlations were obtained between P concentration and landscape variables, probably because both SRP and total P concentrations were relatively homogeneous across the catchment, compared to nitrate. The variables average slope and percent of land in the catchment with slopes higher than 50% were included in the regression models for SRP yields and total P at both tributary and mainstem sites (Table 2). As was seen for nitrate, the percent of cultivated land in the catchment was an important predictor for P yields, explaining a large proportion of the variation in SRP (93%) and total P (98%) yields at the mainstem sites.

Discussion Spatial and seasonal patterns in the amounts of nutrients in the Guare River are likely due to a combination of landscape characteristics and in-stream processes. The results of regression analyses indicate that slope, specific runoff, forest and cropland can influence spatial patterns of nitrate and phosphorus. A negative correlation between slope and nitrate concentration, and a positive influence of specific runoff on nitrate yield, suggest that areas with high slopes and specific runoff exhibit the highest nitrate levels. It is known that areas of high slope can have shallow soils with low water storage capacity and show a rapid hydrologic response to precipitation (Sidle et al. 2000). This can enhance the subsurface flow, which can move through the organic layers of the soil and transport nitrate flushed from these layers into streams (Welsch et al. 2001; Goller et al. 2005). In the Guare, a positive and significant correlation between average slope and

327

specific runoff was obtained (r =0.82 p o0.0001). For example, Agua Frı´a had one of the higher nitrate concentrations and yields despite not showing the largest proportion of cropland or highest population density. However, this site had the highest specific runoff (Table 1) and the largest proportion of catchment area (9%) with slopes greater than 36%. In studies of pristine tropical and North American watersheds, specific runoff also has been shown to be a strong predictor of nitrogen yields (Lewis et al. 1999; Lewis 2002). In these studies, variation in runoff among catchments was related to differences in vegetation, elevation and precipitation, with higher runoff observed in forest-dominated catchments. In the Guare, steep headwaters are covered primarily by forest (correlation between average slope and forest: r= 0.728 p o0.003) and this relationship is consistent with the high correlation observed between forest and nitrate concentration among mainstem sites (Table 2). In addition, occasional nitrate concentration measurements in headwater forested catchments of the Guare showed values of 600 mg L 1 (M. Castillo, unpublished data), which are similar to those found in Agua Frı´a and Las Dolores. This observation suggests that steep forested areas could represent an important N source, although further research is needed to confirm this result and identify the processes linked to N release, which are probably related to N dynamics (fixation and nitrification) and flow condition in forested slopes (McDowell and Asbury 1994; Castro et al. 2007; Dittman et al. 2007). Cropland can also contribute to the higher concentration of nutrients in the upper tributaries. The catchment of mainstem site 1 comprises 52% of the total cultivated surface and provided 79% of the nitrate transported by the Guare. In Las Dolores and Agua Frı´a, where nitrate concentrations and yields were highest, cropland represents 45 and 22% of the catchment land use, respectively. Cropland can be the main source of N in some catchments due to fertilizer application, N fixation by crops and imports of animal food (Howarth et al. 1996; Boyer et al. 2002; Borbor-Cordova et al. 2006; Lefebvre et al. 2007). In addition, vegetation removal, soil disturbance and decomposition of crop residues can also enhance N export (David et al. 1997; Mathers et al. 2007). Nitrate concentrations and yields in Agua Frı´a and Las Dolores were lower than those reported in some agricultural catchments in North America (David and Gentry 2000; Kemp and Dodds 2001; Royer et al. 2004); however, values were higher than the average yields reported by Lewis et al. (1999) for undisturbed tropical watersheds. In the Guare, the percent of cropland in the catchment explains a large part of the variation in nitrate yield along mainstem sites. This relationship has been observed in many studies conducted primarily in the temperate zone (Jordan et al. 1997; Castillo et al. 2000; Ekholm et al. 2000; Jones et al. 2001; Boyer et al. 2002; Filoso et al. 2003; Strayer et al. 2003; Brodie and Mitchell 2005); results for the Guare confirm the importance of croplands for nitrate export in tropical catchments. Longitudinal trends in N concentrations and loads can be explained by the location of N sources and in-stream removal processes. The high proportion of forested slopes and croplands in the upper basin can enhance N export, resulting in high concentrations of N in the upper sites. The longitudinal decrease in average concentrations is best explained by N removal by instream processes, such as benthic algal N uptake and denitrification, than by a dilution effect, because the N load decreased fivefold between mainstem sites 3 and 8 in the dry season (Fig. 4A). Dense algal mats developed during the dry season in the lower reaches of the Guare River, where the riparian canopy was more open and low flow conditions during this period favor algal growth. During the wet season, nitrate load did not increase proportionally to discharge along the mainstem sites, suggesting that nitrate retention can also be occurring during this period.

328

M.M. Castillo / Limnologica 40 (2010) 322–329

This study suggests that algae can reduce N concentrations, as have been reported elsewhere (Sabater et al. 2000; McKnight et al. 2004; Roberts and Mulholland 2007). Although denitrification rates for the Guare are unknown, dry season conditions may favor this process; low flows can enhance the interaction between sediments and water (Mulholland et al. 2004) and the growth of algal mats can provide anaerobic zones for denitrifying bacteria (Kemp and Dodds 2002a, b). Although in-stream processes probably influenced the seasonal variation of nitrate concentration and load, further research is necessary to assess the potential role of algae and denitrification. The marked temporal pattern of nitrate in the Guare following the changes in discharge suggests that watershed soils, rather than point source discharges, were the main source of N in the catchment. Similar nitrate concentrations and discharge patterns have been observed in montane forests in Ecuador and Peru, where catchment soils were the main sources of nitrate (Ramos et al. 2003; Goller et al. 2006). In agricultural catchments where groundwater represents a significant source of nitrate, higher concentrations are observed in streams at baseflow when groundwater contribution to the discharge is higher (Kemp and Dodds 2001). In the Guare, lower concentrations of nitrate were observed during the dry season, probably explained by lower nitrate inputs due to reduced rainfall and runoff. Under dry conditions, nitrate can accumulate in soils due to the mineralization of organic N, lower demand by the vegetation, and reduced percolation (Cambardella et al. 1999; Vink et al. 2007). In addition, in-stream processes such as decomposing leaf litter and increased algal growth can also contribute to decrease N concentrations during the dry season. In undisturbed streams in Puerto Rico, lower N concentrations were observed during periods of peak litter fall, suggesting that decomposing leaf litter can contribute to N immobilization in the forest floor and in the stream channel (McDowell and Asbury 1994). During the wet season, interception of shallow subsurface flow with soil organic layers can enhance the flushing of nitrate accumulated during the dry season or produced by enhanced mineralization due to more humid conditions during the rainy period (Creed et al. 1996; Welsch et al. 2001; Goller et al. 2006). The upper basin also represents the main source of P in the Guare, influencing the patterns of P concentration and yield. The catchment above mainstem site 3 provided on average 78% of the P transported by the Guare River at site 8, probably due to the inputs from agricultural activities and urban discharges combined with the high slopes of the upper basin. The influence of P inputs from the population was evident between Emilia 1 and 3 sites, where concentration increased twofold and load four times, probably due to wastewater discharges from Altagracia de ˜ a and other villages (Figs. 2B and 4B). Despite the la Montan influence of urban discharges on P concentrations at Rı´o Emilia, the regression models for SRP and total P yield included cropland and slope rather than population density. Slope and cropland have been correlated to P concentration in other studies (Ekholm et al. 2000; Tong and Chen 2002; Little et al. 2003). Steep areas can exhibit higher surface runoff, which can enhance the transport of particulate and soluble P from diffuse sources (Ekholm et al. 2000). Therefore, the location of croplands in steep areas of the Guare catchment likely enhances P export from the watershed soils. In the Guare, inputs from point and non-point sources probably change their relative importance throughout the year. An increase in P concentration, particularly SRP, during the dry season and the beginning of the wet season suggests that point source discharges could be relevant during this period, when inputs from diffuse sources are lower due to decreased runoff. Point source inputs, which are generally dominated by soluble

forms (Mainstone and Parr 2002), were primarily represented in the Guare by untreated wastewater discharges from houses and poultry farms along the river and its tributaries. During the periods of higher discharge, inputs from diffuse sources probably increased as a result of higher runoff. Inputs during this period were significant because P load increased several fold despite the small variation in concentrations. The change in the relative importance in P sources throughout the year probably explain the dominance of soluble forms during periods of lower discharge and the greater difference between total P and SRP concentrations as discharge increases (Fig. 5). In the Guare River, both natural and anthropogenic features of the upper catchment had a strong influence on nitrate and phosphorus temporal and spatial patterns. This influence is likely the result of the interaction of topography with land use and cover. Although cropland was identified as an important predictor of nutrient export, areas with high slopes and runoff influenced nutrient levels and it is not clear at this point how land use affects this influence. Seasonal patterns in concentrations suggest that the catchment soils are the main source of nutrients rather than point sources, although wastewater discharges from the population are likely to contribute to the P transported by the river. Besides the influence of landscape features, the longitudinal trend in nitrate suggests that this element is being removed in the lower reaches, so probably in-stream processes are also influencing nitrate patterns in the Guare River. Therefore, the role of both export mechanisms from the upper part of the basin and removal processes in the lower reaches need further research to understand the dynamics of nutrients in the Guare River.

Acknowledgments I am grateful to Meimalı´n Moreno, Armando Castro and Arrigo Lucchetti for their assistance in the field and in the laboratory. Mirady Sebastiani, Rodrigo Lazo and Gianni Papadakis provided assistance with land use analysis and GIS. David Allan provided helpful comments on the manuscript. This study was supported by the Decanato de Investigacio´n y Desarrollo (Universidad Simo´n Bolı´var) grant S1-CB-10-PN. References APHA, 1995. Standard Methods for the Examination of Water and Wastewater, twenty-first ed. APHA, Washington, D.C., USA. Baker, D.B., Richards, P.R., 2002. Phosphorus budgets and riverine phosphorus export in Northwestern Ohio watersheds. J. Environ. Qual. 31, 96–108. Boulton, A., Boyero, L., Covich, A., Dobson, M., Lake, P., Pearson, R., 2008. Are tropical streams ecologically different from temperate streams?. In: Dudgeon, D. (Ed.), Tropical Stream Ecology. Academic Press, Amsterdam, pp. 257–284. Boyer, E.W., Goodale, C.L., Jaworski, N.A., Howarth, R.W., 2002. Anthropogenic nitrogen sources and relationships to riverine nitrogen export in the northeastern USA. Biogeochemistry 57, 137–169. Borbor-Cordova, M., Boyer, E.W., McDowell, W.H., Hall, W.H., 2006. Nitrogen and phosphorus budgets for a tropical watershed impacted by agricultural land use: Guayas, Ecuador. Biogeochemistry 79, 135–161. Brodie, J.E., Mitchell, A.W., 2005. Nutrients in Australian tropical rivers: changes with agricultural development and implications for receiving environments. Mar. Freshwater Res. 56, 279–302. Cambardella, C.A., Moorman, T.B., Jaynes, D.B., Hatfield, J.L., Parkin, T.B., Simpkins, W.W., Karlen, D.L., 1999. Water quality in walnut creek watershed: nitratenitrogen in soils, subsurface drainage water, and shallow groundwater. J. Environ. Qual. 28, 25–34. Carnemolla, S., Cusimano, G., Di Natale, R., Giunta, G., 1990. La Cuenca hidrogra´fica del Rı´o Tuy (Norte de Venezuela): Introduccio´n a la geologı´a del sistema ˜ oso del Caribe y ana´lisis del riesgo ambiental en la subcuenca del Rı´o montan Guare. Quaderni IILA, Serie Scienza 2. Instituto Italo-Latino Americano, Rome. Carpenter, S.R., Caraco, N.F., Correll, D.L., Howarth, R.W., Sharpley, A.N., Smith, .H., 1998. Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecol. Appl. 8, 559–568. Castillo, M.M., 2000. The dynamics of bacterial production and abundance in tropical lowland rivers of the Orinoco basin. Ph.D. thesis, University of Michigan, Ann Arbor.

M.M. Castillo / Limnologica 40 (2010) 322–329

Castillo, M.M., Allan, J.D., Brunzell, S., 2000. Nutrient concentrations and discharges in a Midwestern agricultural catchment. J. Environ. Qual. 29, 1142–1151. Castro, M.S., Eshleman, K.N., Pitelka, L.F., Frech, G., Ramsey, M., Currie, W.S., Kuers, K., Simmons, J.A., Pohlad, B.R., Thomas, C.L., Johnson, D.M., 2007. Symptoms of nitrogen saturation in an aggrading forested watershed in western Maryland. Biogeochemistry 84, 333–348. CEPAL, 2006. Anuario estadı´stico de Ame´rica Latina y el Caribe, 2006. Publicacio´n de las Naciones Unidas, Santiago de Chile. Creed, I.F., Band, L.E., Foster, N.W., Morrison, I.K., Nicolson, J.A., Semkin, R.S., Jeffries, D.S., 1996. Regulation of nitrate-N release from temperate forests: a test of the N flushing hypothesis. Water Resour. Res. 32, 3337–3354. David, M.B., Gentry, L.E., 2000. Anthropogenic inputs of nitrogen and phosphorus and riverine export for Illinois, USA. J. Environ. Qual. 29, 494–508. David, M.B., Gentry, L.E., Kovacic, D.A., Smith, K.M., 1997. Nitrogen balance in and export from an agricultural watershed. J. Environ. Qual. 26, 1038–1048. Dittman, J.A., Driscoll, C.T., Groffman, P.M., Fahey, T.J., 2007. Dynamics of nitrogen and dissolved organic carbon at the Hubbard Brook Experimental Forest. Ecology 88, 1153–1166. Dodds, W.K., 2006. Eutrophication and trophic state in rivers and streams. Limnol. Oceanogr. 51, 671–680. Ekholm, P., Kallio, K., Salo, S., Pietilainen, O.P., Rekolainen, S., Laine, Y., Joukola, M., 2000. Relationship between catchment characteristics and nutrient concentrations in an agricultural river system. Water Res. 34, 3709–3716. F.A.O., 2007. State of the World’s Forests, 2007. Electronic Publishing Policy and Support Branch Communication Division. Rome, Italy. F.A.O., 2008. Current world fertilizer trends and outlook to 2011/12. Electronic Publishing Policy and Support Branch Communication Division. Rome, Italy. Filoso, S., Martinelli, L.A., Williams, M.R., Lara, L.B., Krusche, A., Victoria Ballester, M., Victoria, R., de Camargo, P.B., 2003. Land use and nitrogen export in the Piracicaba River basin, Southeast Brazil. Biogeochemistry 65, 275–294. Goller, R., Wilcke, W., Leng, M.J., Tobschall, H.J., Wagner, K., Valarezo, C., Zech, W., 2005. Tracing water paths through small catchments under tropical montane rain forest in south Ecuador by an oxygen isotope approach. J. Hydrol. 308, 67–80. Goller, R., Wilcke, W., Fleischbein, K., Valarezo, C., Zech, W., 2006. Dissolved nitrogen, phosphorus and sulfur forms in the ecosystem fluxes of a montane forest in Ecuador. Biogeochemistry 77, 57–89. Hart, M.R., Quin, B.F., Nguyen, M.L., 2004. Phosphorus runoff from agricultural land and direct fertilizer effects: a review. J. Environ. Qual. 33, 1954–1972. Howarth, R.W., Billen, G., Swaney, D., Townsend, A., Jaworski, N., Lajtha, K., Downing, J.D., Elmegren, R., Caraco, N., Jordan, T., Berendse, F., Freney, J., Kudeyarov, V., Murdoch, P., Zhao-Liang, Z., 1996. Regional nitrogen budgets and riverine N and P fluxes for the drainages to the North Atlantic Ocean: natural and human influences. Biogeochemistry 35, 75–139. Howarth, R., Ramakrishna, K., 2005. Nutrient Management. In: Chopra, K., Leemans, R., Kumar, P., Simons, H. (Eds.), Ecosystems and Human Wellbeing: Policy Responses. Volume 3 of the Millennium Ecosystem Assessment (MA). Island Press, Washington, D.C., pp. 295–2311. I.N.E., 2001. Nomenclador de Centros poblados y de Comunidades Indı´genas. Instituto Nacional de Estadı´stica. Caracas, Venezuela. Jones, K.B., Neale, A.C., Nash, M.S., Van Remortel, R.D., Wickham, J.D., 2001. Predicting nutrient and sediment loadings to streams from landscape metrics: a multiple watershed study from the United States Mid-Atlantic Region. Landsc. Ecol. 16, 301–312. Jordan, T.E., Correll, D.L., Weller, D.E., 1997. Nonpoint source discharges of nutrients from piedmont watersheds of Chesapeake Bay. J. Am. Water Resourc. Assoc. 33, 631–643. Kemp, M.J., Dodds, W.K., 2001. Spatial and temporal patterns of nitrogen concentrations in pristine and agriculturally-influenced prairie streams. Biogeochemistry 53, 125–141. Kemp, M.J., Dodds, W.K., 2002a. Comparisons of nitrification and denitrification in prairie and agriculturally influenced streams. Ecol. Appl. 12, 998–1009. Kemp, M.J., Dodds, W.K., 2002b. The influence of ammonium, nitrate, and dissolved oxygen concentrations on uptake, nitrification, and denitrification rates associated with prairie stream substrata. Limnol. Oceanogr. 47, 1380–1393. Kronvang, B., Jeppesen, E., Conley, D.J., Sondergaard, M., Larsen, S.E., Ovesen, N.B., Carstensen, J., 2005. Nutrient pressures and ecological responses to nutrient loading reductions in Danish streams, lakes and coastal waters. J. Hydrol. 304, 274–288.

329

Lefebvre, S., Clement, J.C., Pinay, G., Thenail, C., Durand, P., Marmonier, P., 2007. 15Nnitrate signature in low-order streams: effects of land cover and agricultural practices. Ecol. Appl. 17, 2333–2346. Lewis Jr., W.M., Melack, J.M., McDowell, W.H., McClain, M., Richey, J.E., 1999. Nitrogen yields from undisturbed watersheds in the Americas. Biogeochemistry 46, 149–162. Lewis Jr., W.M., 2002. Yield of nitrogen from minimally disturbed watersheds of the United States. Biogeochemistry 57/58, 375–385. Lewis Jr., W.M., 1986. Nitrogen and phosphorus runoff losses from a nutrient-poor tropical moist forest. Ecology 67, 1275–1282. Litke, D.W., 1999. Review of phosphorus control measures in the United States and their effects on water quality. US Geological Survey Water-Resources Investigations Report 99-4007. Little, J.L., Saffran, K.A., Fent, L., 2003. Land use and water quality relationships in the lower little Bow River Watershed, Alberta, Canada. Water Qual. Res. J. Canada 38, 563–584. Mainstone, C.P., Parr, W., 2002. Phosphorus in rivers-ecology and management. Sci. Total Environ. 282/283, 25–47. Mallin, M.A., Johnson, V.L., Ensign, S.H., Macpherson, T.A., 2006. Factors contributing to hypoxia in rivers, lakes, and streams. Limnol. Oceanogr. 51, 690–701. Mathers, N.J., Nash, D.M., Gangaiya, P., 2007. Nitrogen and phosphorus exports ¨ from high rainfall zone cropping in yustralia: issues and opportunities for research. J. Environ. Qual. 36, 1551–1562. McDowell, W.H., Asbury, C.E., 1994. Export of carbon, nitrogen, and major ions from three tropical montane watersheds. Limnol. Oceanogr. 39, 111–125. McKnight, D.M., Runkel, R.L., Tate, C.M., Duff, J.H., Moorhead, D.L., 2004. Inorganic N and P dynamics of Antarctic glacial meltwater streams as controlled by hyporheic exchange and benthic autotrophic communities. J. N. Am. Benthol. Soc. 23, 171–188. Moog, D.B., Whiting, P.J., 2002. Climatic and agricultural factors in nutrient exports from two watersheds in Ohio. J. Environ. Qual. 31, 72–83. Mulholland, P.J., Valett, H.M., Webster, J.R., Thomas, S.A., Cooper, L.W., Hamilton, S.K., Peterson, B.J., 2004. Stream denitrification and total nitrate uptake rates measured using a field N-15 tracer addition approach. Limnol. Oceanogr. 49, 809–820. ˜ ez-Ferrara, M., 2000. River Pringle, C.M., Scatena, F.N., Paaby-Hansen, P., Nu´n conservation in Latin America and the Caribbean. In: Boon, P.J., Davies, B.R., Petts, G.E. (Eds.), Global perspectives on River conservation: Science, Policy and Practice. Wiley & Sons, Chichester, pp. 41–77. Ramı´rez, A., Pringle, C., Wantzen, K.M., 2008. Tropical stream conservation. In: Dudgeon, D. (Ed.), Tropical Stream Ecology. Academic Press, Amsterdam, pp. 285–304. Ramos, O., Llerena, C., McClain, M., Quintanilla, J., 2003. Perdidas de nitro´geno del bosque de neblina en la sub-cuenca de San Alberto, Oxapampa, Peru´. Rev. Boliv. Quı´m. 20, 86–90. Roberts, B.J., Mulholland, P.J., 2007. In-stream biotic control on nutrient biogeochemistry in a forested stream, West Fork of Walker Branch. J. Geophys. Res. 112, G04002. Royer, T.V., Tank, J.L., David, M.B., 2004. Transport and fate of nitrate in headwater agricultural streams in Illinois. J. Environ. Qual. 33, 1296–1304. Sabater, F., Butturni, A., Marti, E., Munoz, I., Romani, A., Way, J., Sabater, S., 2000. Effects of riparian vegetation removal on nutrient retention in a Mediterranean stream. J. N. Am. Benthol. Soc. 19, 609–6290. Saunders, T.J., McClain, M.E., Llerena, C.A., 2006. The biogeochemistry of dissolved nitrogen, phosphorus, and organic carbon along terrestrial aquatic flowpaths of a montane headwater catchment in the Peruvian Amazon. Hydrol. Processes 20, 2549–2562. Sidle, R.C., Tsuboyama, Y., Noguchi, S., Hosoda, I., Fujieda, M., Shimizu, T., 2000. Stormflow generation in steep forested headwaters: a linked hydrogeomorphic paradigm. Hydrol. Processes 14, 369–385. Strayer, D.L., Beighley, R.E., Thompson, L.C., Brooks, S., Nilsson, C., Pinay, G., Naiman, R.J., 2003. Effects of land cover on stream ecosystems: roles of empirical models and scaling issues. Ecosystems 6, 407–423. Tong, S.T.Y., Chen, W.L., 2002. Modeling the relationship between land use and surface water quality. J. Environ. Manage. 66, 377–393. Vink, S., Ford, P.W., Bormans, M., Kelly, C., Turley, C., 2007. Contrasting nutrient exports from a forested and an agricultural catchment in south-eastern. Aust. Biogeochemistry 84, 247–264. Welsch, D.L., Kroll, C.N., McDonnell, J.J., Burns, D.A., 2001. Topographic controls on the chemistry of subsurface stormflow. Hydrol. Processes 15, 1925–1938.