The Santa Lucia River basin provides drinking water to 60% of Uruguay population. The input of excessive amounts of nitrogen and phosphorus both from point and nonpoint sources could impair surface water quality and prevent its use for human consumption. The objective of this work was to evaluate surface water quality in small catchments under agricultural use located within this river basin. To this end, 5 streams and 4 small polders were surveyed between September 2008 and December 2009. The median concentrations of total phosphorus and total nitrogen in the streams across all sites and periods were 770 and 1659 µg L−1, respectively, exceeding in the case of total phosphorus national and international thresholds. Furthermore, soluble phosphorous, the most readily available form to algae, represented 88.3% of total phosphorus. Concentrations of phosphorous and nitrogen in polders were also high. The eutrophication process of this water bodies could be linked to the intensive agricultural land use in the area because the nutrient input from cities and industries was not relevant in these catchments.
Introduction
The Santa Lucia River basin, located in the south of Uruguay, provides drinking water to 60% of the country’s population.1 In the northern part of this watershed (zone A), about 50% of the area is occupied by grazing-based dairy farms, whereas in the southern area (zone B), vegetable farms associated with poultry and swine production predominate.2 These farms use large amounts of organic and chemical fertilizers to increase production. However, the perception that an increase in nitrogen (N) is needed to increase agronomic productivity has resulted in an excessive input of phosphorus (P) to the system. This is due to the low N:P ratio of the organic fertilizer used (poultry manure), resulting in a positive balance of P and in a soil P enrichment.3 Nutrient enrichments from point or nonpoint sources may lead to eutrophication, degrading the aesthetic and recreational value of surface water.4
International nutrient thresholds can be used as guidelines for water quality assessments. According to Daniel et al,5 to control eutrophication, total phosphorus (TP), and soluble phosphorus (SP) in streams should not exceed 20 and 10 µg L−1, respectively. Dodds et al6 compiled a database with TP and total nitrogen (TN) log-mean concentrations in temperate streams and set the boundary between mesotrophic and eutrophic conditions in the lowest third of the cumulative frequency plots (70 and 1500 µg P L−1, respectively). In Uruguay, Decree 253/79 established a TP critical level of 25 µg P L−1 for surface waters,7 but recently, the National Water Authority has been considering raising this limit to 70 µg P L−1.8
The results of several surveys conducted in this basin have shown that the water quality in the area has been deteriorating, affecting the possibility of using this water resource as a source for drinking water. One of these surveys reported TP values that ranged between 80 and 735 µg P L−1, whereas those of N in the forms of nitrate and ammonium fluctuated from 0.014 to 0.25 and 0.01 to 0.25 µg N L−1, respectively.9 In another report of 15 sampling points, the TP and TN annual means in 9 of these points were above the eutrophic limit (TP: 170-440 µg P L−1 and TN: 4200-10 500 µg N L−1), whereas 4 of the remaining 5 were classified as supereutrophic, and 1 as hypereutrophic.10 In 2013, a harmful algal bloom in this area caused changes in the odour and taste of the water supplied by the water authority and generated concern and alarm among the public.11 Because of this event, an ‘Action Plan’ based on several best management practices (BMPs) was implemented by the relevant ministry.2
According to the latest report, this situation has not changed substantially; the TP and TN values at a measurement station located at the confluence of the Santa Lucia and San Jose rivers were 310 and 1280 µg P L−1, respectively.8 There are conflicting opinions about the main sources of this pollution; although some stakeholders primarily hold the point sources of cities and industries responsible for this situation, others blame the agriculture-related sources of contamination.
Due to the steady and worrying decline in water quality of some watersheds in the country and the scarce availability of information in horticultural areas, the aim of this work is to evaluate the impact of agricultural land use on surface water quality in horticultural catchments located within or near zone B of the Santa Lucia River basin, where there were no point sources of pollution (cities and industries). In addition, with this information it is possible to identify the effect on the water quality of this productive system and to implement a national control plan and an evaluation of the effectiveness of the control mechanisms and control systems based on the results of official water controls.
Materials and Methods
A total of 9 ‘pilot’ farms located within zone B of the Santa Lucia River basin2 were selected for surface water quality monitoring (Figure 1 and Table 1), including 5 natural streams and 4 artificial polders (Table 2). Sampling was conducted from September 2008 to December 2009 with a variable frequency.
Table 1.
Location and production system at each site.
Table 2.
Number of samples collected per site, type of water, and year.
Soils in this area are mostly fertile-clay soils with a relatively low textural differentiation (Pachic or Vertic Argiudolls and Typic Hapluderts) developed above 2 Pleistocene formations (Libertad and Raigón). These are mixed with less-fertile and more differentiated loamy-clay soils (Abruptic Argiudolls) formed on top of a late Oligocene formation (Fray Bentos).12 Further details of the soil are given in Table 3.
Table 3.
Information about the predominant geological formations and soil types at each site.
The main land use in this area is intensive vegetable production, mixed with poultry and swine farms. The climate is classified as temperate, with a mean annual temperature and rain of 17°C and 1200 mm, respectively.14
Grab samples were collected from mid-stream into polyethylene bottles (1 L). To prevent biological changes to the water composition, all bottles were previously spiked with 1 mL of H2SO4 5.5 M. All samples were analysed for TN, nitrate (N-NO3−), TP, soluble reactive phosphorus (SRP), and turbidity (Tb). Total nitrogen was analysed using H2SO4 digestion, according to the method of Bremmer and Mulvaney15 and colorimetry determination according to the method of Rhine et al.16 Before digestion, a 100-mL water subsample was acidified with 2 mL of H2SO4 5.5 M and concentrated to 10 mL by evaporation at 90°C. Analysis of N-NO3− was conducted according to the method of Keeney and Nelson.17 Total phosphorus and SRP were analysed using the method of Pote and Daniel.18 All nutrient colorimetric determinations were made with a Unicam 5675 Spectrometer. Turbidity was determined using an Oakton T-100 portable turbidity meter.
Concentrations of TP and TN measured in water samples were compared with thresholds used in different countries (including Uruguay) to assess the freshwater eutrophication risk (Table 4).
Table 4.
International reference values for total nitrogen (TN), total phosphorus (TP), and turbidity (Tb) in streams.
Nitrate concentration levels were compared with the Uruguayan drinking water threshold7 enforced by Decree 253/79. This parameter, however, only allows the identification of non-drinkable water – it cannot be used to evaluate potability.
Regression procedures (Microsoft Excel, 1997) were used to evaluate the relationships between the water quality variables. Nutrient concentrations were not normally distributed according to the Shapiro-Wilks test, so differences between sites were evaluated by the Kruskal-Wallis nonparametric median test using the statistical package InfoStat version 2011.22 Differences were considered statistically significant when P ⩽ .05.
Results and Discussion
Streams
The median concentration across all sites and periods (MCAS) of TP in streams was 770 µg P L−1; a value higher than all national and international thresholds presented in Table 1. The median concentrations for the study sites (MCSS) varied between 27 and 3080 µg P L−1 (Figure 2). In most cases, these MCSS values were above the previously mentioned thresholds but comparable with TP concentrations reported from international studies performed under similar land use conditions. For example, in commercial citrus groves in the United States, Yu et al23 reported that mean TP concentrations in surface water ranged from 510 to 2640 µg P L−1, whereas He et al24 found that this range was between 1510 and 3920 µg P L−1 for vegetable farms. In turn, in our study, the MCAS of SRP was 554 µg P L−1, whereas MCSS ranged between 10 and 3078 µg P L−1, within the ranges of 200 to 2130 and 470 to 1980 µg P L−1 reported by Yu et al23 and He et al,24 respectively. In the 2 studies mentioned above, SRP, which is the most bioavailable form of phosphorus, represented between 74% and 45% of TP, respectively. In our study, this range was between 40% and 99%.
Figure 2.
Box plots of total phosphorus, soluble reactive phosphorus, total nitrogen, and nitrate in streams (S) and polders (P) from all sites and years. Centre lines show the medians, box limits indicate the 25th and 75th percentiles, and whiskers indicate minimum and maximum values, and outliers are represented by dots.
In the case of TN, MCAS was 1659 µg N L−1: higher than the limits established for Europe but lower than those established for New Zealand and the United States (Table 4). The range of MCSS values varied from 87 to 8724 µg N L−1, exceeding the limits established for the US Environmental Protection Agency (USEPA) agricultural ecoregion at only 2 sites (#3 and #4) (Table 4). These values were lower than the median TN concentration obtained by Bartley et al25 for Australian basins used for horticulture (32 000 µg N L−1), but the MCSS range of TN found here was wider than that cited by Smith et al26 for US dairy basins (690-3270 µg N L−1). The average N: P relation for streams (across all sites and periods) was 16:1, equal to the Redfield ratio.27 This implies that neither N nor P was limiting primary productivity, or in another words, that TN and TP were in equilibrium with the phytoplankton demand for both nutrients. This equilibrium, however, was more apparent than real – in four of these sites (#1, #3, #4, and #6), this ratio was lower than 16:1 (1.2:1, 2.8:1, 4.2:1, and 4.9:1, respectively), indicating that P was excessive, whereas in the other site (#9), this relation was higher than 16:1 (20:1), indicating that P was limiting.
In addition to the TN and TP values presented above, the MCAS of Tb (6.19NTU) was lower than most of the international reference values; however, it was just above the threshold for the most pristine of USEPA’s ecoregions (Table 4). This indicates that the visual aspect of ‘clear water’ does not necessarily indicate the absence of nutrient pollution problems. In the case of N-NO3−, MCAS was 340 µg N L−1, and this value fell within the expected range, given that in the absence of artificial drainage, this anion would preferentially move to groundwater.28
Among the sites examined here, there were significant statistical differences in MCSS values in the cases of both TP and SRP. The highest concentrations of both P forms were found at sites #4 and #6, and the lowest at sites #1 and #9, respectively. In the cases of TN, N-NO3−, and Tb, these differences were not statistically significant, although the probability value (P > .074) for Tb was close to that of the significance limit.
Polders
The MCAS of TP was 350 µg P L−1 (Figure 2) – higher than the threshold suggested by Dodds et al6 to separate oligotrophic from mesotrophic waters in lakes and reservoirs (lentic systems). This value also exceeded the maximum allowable concentration in the United States and New Zealand for different ecoregions.19,20 As in surface waters, the MCAS for SRP was high (230 µg P L−1), and the greater part of TP was in soluble form. Therefore, as in streams, the polders were enriched with P, although in a lesser extent. This is expected because there is more sediment deposition in dams (lentic systems) than in streams (lotic systems). This is why lotic systems have higher thresholds, which are translated into less demanding values than those established for lentic systems.19 It may be noted that international thresholds established for lentic systems19 are the same as the Uruguayan national water quality standards for freshwater (25 µg P L−1), which suggests that this threshold is too demanding and more applicable to lakes and reservoirs.
In the case of TN, observed MCAS (1567 µg N L−1) was higher than the less demanding reference value (1270 µg N L−1) for ecoregion #XIII in the United States.19 In Uruguay, there is no reference for this variable. These waters were more enriched with P than N (N:P ratio < 9:1) but to a lesser extent than surface waters. The value of MCAS for Tb (20.6 NTU) was higher than the reference value for the US agricultural ecoregion but lower than other international reference values (Table 4). However, this variable is not included in the standards for lentic systems.19 The MCAS of NO3 (14.5 µg N L−1) was lower than the international and local critical level concentrations for drinking water (10 000 µg N L−1). Moreover, all samples were below this threshold.
No significant relationships between the analysed water quality parameters were observed, but when the MCAS values from streams and polders were grouped together, there was a statistically significant positive relationship between TN and Tb (r = 0.85, P < .0076). Vidon et al29 showed similar results gathered from surface water monitoring sites located both upstream and downstream from a cattle grazing area. They reported that when cattle had direct access to the stream channel (during summer-fall), Tb values tended to be higher downstream than those observed upstream at a 13:1 ratio, but NO3 concentrations were not significantly affected. According to these authors, the reason for these high Tb values could be related to the deposition of cattle manure in surface water sources. Part of the organic matter from the manure remains as a suspension in the water.
Relationships between TP and SRP concentrations in surface waters
The concentration ranges of TP and SRP observed in the 2 types of surface water (streams and polders) evaluated in this study were relatively similar. Therefore, these 2 water sources were grouped together, and the relationship between TP and SRP in this new group was analysed, resulting in a linear and statistically significant relationship between them (Figure 3). The linear equation did not include an intercept term because it was not significantly different from 0. Based on this equation, and on the legal Uruguayan threshold (25 µg P L−1) established in Decree 253/79, an SRP threshold value of 22 µg P L−1 was estimated, which would be applicable to both streams and polders. This value was higher, and less strict, than the corresponding SRP threshold of 9 µg P L−1 established for Australia30 but lower than the MCAS values observed in streams and polders (340 and 230 µg P L−1, respectively), confirming the environmental deterioration of these surface waters.
Figure 3.
Relationship between median concentration of total phosphorus and soluble reactive phosphorus in streams and polders by site. The filled boxes represent streams and the empty boxes represent the polders. The grey cross represents a stream value that was not included in the regression.
It should be noted that the existence of high SRP concentrations in streams seems to contradict the conclusions of Correll27 that SRP always tends to be a small fraction of TP because of the strong adsorption of orthophosphate onto stream sediments, which causes the TP:SRP balance to be strongly shifted towards the TP side. In the intensive farming conditions of this study, however, this balance may have been modified. This change could have been caused, in part, by a larger input of soluble P to these surface water bodies. This P is likely to originate from single-point sources, such as untreated dairy-farm effluent, but also from more diffuse inputs, like P carried by water runoff.
The diffuse P input tends to increase when there is a boost in soil labile P,31 which is the P fraction most related to P uptake by crops. At these study sites, soil labile P, as measured by the Bray-P1 method at a soil depth of 0 to 20 cm, has increased over many years from a basal value32 of 4 mg P kg−1 to an average of 69 mg P kg−1, with a range of 22 to 113 mg P kg−1 (Figure 4). This increase was more than was needed to maximize crop productivity. For example, the optimum Bray-P1 value required for lucerne (Medicago sativa), the relevant species with the higher P requirement is 25 mg P kg−1for this soil depth.33
Figure 4.
Soil test values for phosphorus (Bray I) at each site. The dotted line represents the critical value required to maximize lucerne (Medicago sativa) yield.
Since 2000, these farms have shifted from conventional to no-till farming, and nowadays, almost all farms have made this transition.34 Under no-till planting, phosphate fertilizer is usually applied to the soil surface without being mixed in, which leads to a high P stratification at the soil surface, further increasing the concentration of soluble P in runoff.35 Perdomo et al36 have also observed strong labile P stratification in soils under intensive management in other areas of the Santa Lucia River basin.
A lower rate of soil erosion under no-till farming also reduces the fraction of particulate P carried away by runoff, leading to a decrease in sediment transport; thus, the specific soil surface capable of P adsorption also diminishes, leading to an increase in SRP.37 According to James and Barko,38 soluble forms of P comprised between 48% and 57% of the flow-weighted concentration of TP between May and September 2002, at Upper Eau Galle River (Wisconsin, USA), whereas all bioavailable P (including SP plus labile particulate phosphorus) represented 79% of TP. Goyenola et al39 also reported a high proportion of soluble P (86.4%) in streams located in areas undergoing intensive farming in another zone of the Santa Lucia River basin. In our case, this median-estimated value was 88%.
It should be noted that the contribution of point sources of water pollution from cities and industries located within the Santa Lucia River basin do not affect the water quality of the catchments under study because there is no direct hydrologic connection between these 2 zones. Thus, the most plausible explanation of the observed concentration rise of P, and other nutrients in these water bodies, is that it is mainly related to the intensification of agriculture in the area. The implications of this conspicuous case of surface water contamination are obvious because these catchments are in the area that provides potable water to more than half of the Uruguayan population. Although Uruguay has already implemented farm policies to protect water quality in this area, they are primarily based on preventing soil erosion and restricting P fertilization to fields that are below agronomic soil test phosphorus (STP) thresholds. However, as Baker et al40 has noted, one of the BMPs that show larger potential to reduce dissolved reactive phosphorus in runoff in similar situations is the reduction of STP stratification by a one-time soil inversion, as well as the application of P fertilizer below the soil surface instead of on top.
Conclusions
The MCAS of TP (770 and 350 µg P L−1) and TN (1659 and 1567 µg N L−1) in courses and polders, respectively, were higher than the most permissive thresholds established by USEPA for both water types, even though these water bodies mainly receive discharge from agricultural sources and are not influenced by cities or industries.
Soil labile P (Bray 1) at the surveyed sites ranged from 22 to 113 mg P kg−1 (0-20 cm), well above the original value of 4 mg P kg−1. This suggests that part of the increase in P concentration in water could be related to P carried away through runoff. Another source of pollution still present at many sites comes from the direct discharge into waterways of dairy and swine effluents. Consequently, BMPs should be enforced to minimize the impact of these 2 contamination sources.
Even with the application of these BMPs, water quality improvement would be slow because current labile P levels in both soil and stream sediments are high, and their decrease would be sluggish because P equilibrium is strongly displaced towards the solid phase.
Acknowledgements
The authors are grateful to Victoria Mancassola, Sebastian Peluffo, and José Pedro Dieste for their help on collecting water samples from the different sites.
REFERENCES
Notes
[1] Five peer reviewers contributed to the peer review report. Reviewers’ reports totalled 545 words, excluding any confidential comments to the academic editor.
[2] Financial disclosure The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was partially funded by the European Union, Project INCOCT-2006-032387, and by INIA-Uruguay, Project FPTA 209.
[3] Conflicts of interest The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
[4] PB, CP, and SD conceived and designed the experiments; contributed to the writing of the manuscript; agree with manuscript results and conclusions; jointly developed the structure and arguments for the paper; and reviewed and approved of the final manuscript. PB analysed the data and wrote the first draft of the manuscript. CP and SD made critical revisions and approved final version.
